system

Abstract

To provide a system.SOLUTION: Provided is a system comprising: means for collecting access logs on an advertiser site; means for storing the collected access logs in a database; means for pre-processing for cleaning and standardizing the stored access log data; means for building a generative AI model by making it learn a user's behavior patterns using the pre-processed data; means for generating a next action on the basis of an initial behavior of the user accessing the advertiser site using the generative AI model that has already learned; and means for dynamically changing the user's experience on the advertiser site on the basis of the generated next action.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background Art]

[0002] Patent Document 1 discloses a persona chatbot control method executed by at least one processor, the method comprising the steps of: receiving a user utterance; adding the user utterance to a prompt including an instruction associated with a description of a character of a chatbot; encoding the prompt; and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior Art Literature] [Patent Literature]

[0003] [Patent Document 1] Japanese Unexamined Patent Application Publication No. 2022-180282 [Summary of the Invention] [Problem to be Solved by the Invention]

[0004] Conventional advertiser sites can only provide a uniform user experience (UX), and have a problem in that it is difficult to provide an optimal UX according to the behavior pattern of each individual user. [Means for Solving the Problem]

[0005] The system of the present invention learns access logs of an advertiser site and constructs an AI (UX generation AI) that generates user actions leading to successful conversions. This UX generation AI generates and proposes a next action that leads to a successful conversion based on the initial behavior of a user who has visited the advertiser site. Then, the advertiser site dynamically changes the site UX based on the proposal. This makes it possible to provide an optimal UX for each user and improve the conversion rate. [Brief explanation of the drawing]

[0006] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11] This is a sequence diagram showing the processing flow of the data processing system in Embodiment 1 of Example 1. [Figure 12] This is a sequence diagram showing the processing flow of the data processing system in Application Example 1 of Form Example 1. [Figure 13] This is a sequence diagram showing the processing flow of the data processing system in Example 2 of Embodiment 2. [Figure 14] This is a sequence diagram showing the processing flow of the data processing system in Application Example 2 of Form Example 2. [Figure 15] This is a sequence diagram showing the processing flow of the data processing system in Embodiment 3 of Example 3. [Figure 16] It is a sequence diagram showing the processing flow of the data processing system in Application Example 3 of Embodiment 3. [Figure 17] It is a sequence diagram showing the processing flow of the data processing system in Example 1 of Embodiment 1 when an emotion engine is combined. [Figure 18] It is a sequence diagram showing the processing flow of the data processing system in Application Example 1 of Embodiment 1 when an emotion engine is combined. [Figure 19] It is a sequence diagram showing the processing flow of the data processing system in Example 2 of Embodiment 2 when an emotion engine is combined. [Figure 20] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 of Embodiment 2 when an emotion engine is combined. [Figure 21] It is a sequence diagram showing the processing flow of the data processing system in Example 3 of Embodiment 3 when an emotion engine is combined. [Figure 22] It is a sequence diagram showing the processing flow of the data processing system in Application Example 3 of Embodiment 3 when an emotion engine is combined. DETAILED DESCRIPTION OF EMBODIMENTS

[0007] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0008] First, terms used in the following description will be explained.

[0009] In the following embodiments, the labeled processor (hereinafter simply referred to as "processor") may be a single arithmetic device or a combination of a plurality of arithmetic devices. Further, the processor may be one type of arithmetic device or a combination of a plurality of types of arithmetic devices. Examples of arithmetic devices include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), APU (Accelerated Processing Unit), and TPU (TENSOR PROCESSING UNIT (registered trademark)).

[0010] In the following embodiments, the labeled RAM (Random Access Memory) is a memory that temporarily stores information, and is used as a work memory by the processor.

[0011] In the following embodiments, the labeled storage is one or more non-volatile storage devices that store various programs, various parameters, and the like. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), magnetic tapes, and the like.

[0012] In the following embodiments, the labeled communication I / F (Interface) is an interface including a communication processor, an antenna, and the like. The communication I / F manages communication between a plurality of computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), and the like.

[0013] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0014] [First Embodiment]

[0015] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0016] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0017] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0018] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0019] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0020] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0021] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0022] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0023] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0024] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0025] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0026] Next, the identification process performed by the identification processing unit 290 of the data processing device 12 will be described.

[0027] "Example of form 1"

[0028] The system of this invention constructs an AI (UX generation AI) that learns from the access logs of advertiser websites. Based on the initial actions of users who visit the advertiser website, this AI generates and proposes the next actions that will lead to a conversion. Specifically, based on the initial actions of a user when they visit the site (e.g., accessing a specific product page, searching for a specific keyword, etc.), it generates actions such as which page to guide the user to next and which products to recommend.

[0029] "Example of form 2"

[0030] The generated actions are reflected on the advertiser's website, dynamically changing the site's user experience (UX). Specifically, when a user visits the site, the displayed content changes based on the generated actions. For example, if an action recommending a specific product is generated, information about that product will be prioritized when a user visits the site. This makes it possible to provide an optimal UX for each user and improve conversion rates.

[0031] "Example of form 3"

[0032] The system of this invention learns from the access logs of advertiser websites to understand user behavior patterns and generates optimal actions based on them. Specifically, it learns user behavior patterns from past access logs and generates the next action based on those patterns. For example, if the system learns that a user who visits a particular product page is highly likely to purchase another specific product, it will generate the next action based on that pattern.

[0033] The following describes the processing flow for each example of the form.

[0034] "Example of form 1"

[0035] Step 1: Collect access logs from the advertiser's website. This includes initial user actions when they visit the site (e.g., accessing a specific product page, searching for specific keywords, etc.).

[0036] Step 2: The collected access logs are used to train the AI ​​(UX generation AI). The AI ​​learns user behavior patterns and generates the next action based on them.

[0037] Step 3: Reflect the generated actions on the advertiser's site. Specifically, when a user visits the site, the displayed content will change based on the generated actions.

[0038] "Example of form 2"

[0039] Step 1: Collect information on the initial actions of users when they visit the site.

[0040] Step 2: Based on the collected initial actions, the AI ​​(UX generation AI) generates the next action that will lead to a conversion.

[0041] Step 3: Reflect the generated actions on the advertiser's site and dynamically change the site's UX.

[0042] "Example of form 3"

[0043] Step 1: Collect access logs from the advertiser's website to understand user behavior patterns.

[0044] Step 2: Learn user behavior patterns from collected access logs and generate the next action based on those patterns.

[0045] Step 3: Reflect the generated actions on the advertiser's site and dynamically change the site's UX.

[0046] (Example 1)

[0047] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0048] Traditional advertiser websites lacked the means to effectively utilize user behavior data to improve conversion rates. In particular, it was difficult to appropriately suggest the next action based on the user's initial behavior and to dynamically change the user experience on the site. As a result, it was difficult to make suggestions that matched the user's interests and concerns, making it difficult to improve conversion rates.

[0049] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0050] In this invention, the server includes means for collecting access logs from advertiser sites, means for storing the collected access logs in a database, means for pre-processing the stored access logs, means for training a generation AI model using the pre-processed data, means for generating and proposing the next action based on the user's initial behavior, means for sending the generated next action to the user's terminal, and means for dynamically changing the user experience of the site based on the proposal. This makes it possible to effectively utilize user behavior data and improve the conversion rate.

[0051] An "advertiser site" is a website that displays advertisements and allows users to access information about products and services.

[0052] An "access log" is a record of user behavior when accessing a website, and includes information such as pages viewed, search keywords, and date and time of visit.

[0053] A "database" is a system for storing collected access logs, enabling efficient management and retrieval of data.

[0054] "Preprocessing" refers to processes such as cleaning and normalizing data to make it easier to analyze collected data.

[0055] A "generative AI model" is an artificial intelligence model that uses machine learning algorithms to learn user behavior patterns and generate the next action.

[0056] "Initial user behavior" refers to the first action a user takes when accessing a website, and includes actions such as viewing a specific product page or searching for specific keywords.

[0057] "Next action" refers to suggestions for the user's next course of action, generated based on the user's initial actions. This includes directing the user to a specific page or recommending products.

[0058] "User experience" refers to the overall experience a user has when using a website, and includes factors such as the ease of use of the site and the quality of the information provided.

[0059] "Dynamic modification" refers to changing the content and structure of a website in real time based on user behavior data.

[0060] Modes for carrying out the invention

[0061] System Overview

[0062] The system of this invention collects access logs from advertiser websites and builds a generative AI model that learns user behavior patterns based on these logs. This AI model generates and suggests the next action based on the user's initial actions. Specifically, it generates actions such as which page to guide the user to next and which products to recommend, based on the user's initial actions when visiting the site (e.g., accessing a specific product page, searching for a specific keyword, etc.).

[0063] Hardware and software to be used

[0064] The server collects, stores, and preprocesses access logs, trains the AI ​​model, and generates subsequent actions. Specifically, it uses the following hardware and software:

[0065] Hardware: High-performance server (CPU, GPU, memory, storage)

[0066] Software: Database management systems (e.g., MySQL®, PostgreSQL), machine learning libraries (e.g., TENSORFLOW®, PyTorch)

[0067] The device records user behavior data and sends it to the server. It also receives suggestions for the next action sent from the server and displays them to the user.

[0068] Hardware: User's device (e.g., smartphone, computer)

[0069] Software: Web browsers, mobile applications

[0070] Specific operation of the system

[0071] 1. Collection of access logs

[0072] The server collects real-time data on the behavior of users who visit advertiser websites. Specifically, it records information such as which pages users viewed and which keywords they searched for.

[0073] 2. Data Storage

[0074] The server stores the collected access logs in a database. The data stored includes user ID, visit date and time, viewed pages, search keywords, and more.

[0075] 3. Data preprocessing

[0076] The server preprocesses the stored data. Specifically, it performs data cleaning (imputing missing values ​​and removing outliers), data normalization (scaling), and other similar operations.

[0077] 4. Training the AI ​​model

[0078] The server uses pre-processed data to train a generative AI model. Machine learning libraries such as TensorFlow and PyTorch are used for training.

[0079] 5. Generate the next action

[0080] The server uses a pre-trained AI model to generate the next action based on the user's initial behavior. Specifically, it predicts which page the user should be directed to next and which products should be recommended.

[0081] 6. Submitting the proposal

[0082] The server then sends the generated next action to the user's device. This allows the user to be presented with appropriate pages and products.

[0083] Specific example

[0084] Suppose a user accesses an advertiser's website and views a page for a specific smartphone model. The device records this action and sends it to a server. The server stores this data in a database, preprocesses it, and then trains a generative AI model. The trained AI model determines that the user should next be directed to a page about smartphone accessories or related products and sends this suggestion to the device. The device displays the suggestion to the user, and the user views the suggested page.

[0085] Example of a prompt

[0086] "Please build an AI model that generates which page to direct users to next and which products to recommend, based on their initial user behavior data when they visit an advertiser's website and view a specific product page. Specifically, it should be a system that suggests the next action that will lead to a conversion."

[0087] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0088] Step 1:

[0089] Collection of access logs

[0090] The server collects user behavior data in real time from users who visit advertiser websites. Specifically, it records information such as which pages users viewed and which keywords they searched for. The input is user behavior data, and the output is the collected access logs.

[0091] Step 2:

[0092] Data storage

[0093] The server stores the collected access logs in a database. The data stored includes user ID, visit date and time, viewed pages, and search keywords. The input is the collected access logs, and the output is the access logs stored in the database.

[0094] Step 3:

[0095] Data preprocessing

[0096] The server preprocesses the stored data. Specifically, it performs data cleaning (imputing missing values ​​and removing outliers) and data normalization (scaling). The input is access logs stored in the database, and the output is the preprocessed data.

[0097] Step 4:

[0098] AI model training

[0099] The server uses pre-processed data to train a generative AI model. Machine learning libraries such as TensorFlow and PyTorch are used for training. The input is pre-processed data, and the output is a trained generative AI model.

[0100] Step 5:

[0101] Next Action Generation

[0102] The server uses a pre-trained AI model to generate the next action based on the user's initial behavior. Specifically, it predicts which page the user should be directed to next and which product should be recommended. The input is data on the user's initial behavior, and the output is the generated next action.

[0103] Step 6:

[0104] Submit a proposal

[0105] The server sends the generated next action to the user's device. This suggests appropriate pages and products to the user. The input is the generated next action, and the output is the suggestions sent to the user's device.

[0106] Step 7:

[0107] Receiving and displaying proposals

[0108] The terminal receives a suggestion for the next action sent from the server and displays it to the user. The input is the suggestion sent from the server, and the output is the suggestion displayed to the user.

[0109] Step 8:

[0110] User behavior

[0111] The user accepts the suggestions displayed on their device and takes the next action. For example, they might view the suggested product page or purchase the recommended product. The input is the suggestions displayed on the device, and the output is new user behavior data.

[0112] (Application Example 1)

[0113] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server," and the smart device 14 will be referred to as a "terminal."

[0114] Traditional advertiser websites struggled to appropriately suggest the next action that would lead to a conversion based on the user's initial behavior. As a result, they failed to increase user purchase intent and could not expect an improvement in conversion rates. Furthermore, they were unable to provide a site user experience optimized for each user, thus failing to improve user satisfaction.

[0115] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0116] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action that leads to a conversion based on the initial actions of a user who visits the advertiser site, means for dynamically changing the site user experience based on the suggestion, and means for analyzing the user's initial actions and suggesting in real time which product to view next and which page to proceed to. This makes it possible to appropriately suggest the next action that leads to a conversion based on the user's initial actions, and is expected to improve the conversion rate and user satisfaction.

[0117] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[0118] "Access logs" are data that records a user's behavior when they access a website, and include information such as page views and search keywords.

[0119] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to automatically perform specific tasks.

[0120] "Initial user behavior" refers to the first action a user takes when visiting a website, such as accessing a specific product page or searching for specific keywords.

[0121] "Contract completion" refers to a user purchasing goods or services on a website, signifying the completion of a transaction.

[0122] "Next action" refers to the next action a user should take on the website, suggesting actions that are highly likely to lead to a conversion.

[0123] "Website user experience" refers to the overall experience a user has when using a website, and includes factors such as ease of use and satisfaction.

[0124] "Dynamic modification" refers to changing the content and structure of a website in real time in response to user behavior and circumstances.

[0125] "Providing suggestions in real time" refers to instantly analyzing user behavior and immediately suggesting the next action based on the results.

[0126] The system for implementing this invention involves building artificial intelligence that learns from the access logs of advertiser websites and generates user actions that lead to conversions. A specific embodiment of this system is described below.

[0127] System Configuration

[0128] The system consists of the following main components:

[0129] 1. Server: Collects access logs from advertiser websites and stores them as training data.

[0130] 2. Artificial Intelligence (AI): Generates and suggests the next action based on the user's initial actions.

[0131] 3. User device: The device the user uses to access the advertiser's website (e.g., smartphone, tablet, PC).

[0132] 4. Database: Data storage for saving access logs and user behavior data.

[0133] Program processing

[0134] Data collection and learning

[0135] The server collects access logs of users who visit advertiser websites. These access logs include information such as which pages users viewed and which keywords they searched for. The collected data is stored in a database, which artificial intelligence uses as training data.

[0136] Analysis of user behavior

[0137] Artificial intelligence analyzes collected access logs and learns initial user behavior patterns. This generates a model that predicts what actions users should take to convert into sales.

[0138] Proposed next steps

[0139] When a user accesses an advertiser's website, artificial intelligence analyzes the user's initial behavior in real time and suggests which products to view next and which pages to proceed to. These suggestions are displayed on the user's device to guide their actions.

[0140] Hardware and software to be used

[0141] Hardware: Servers, user terminals (smartphones, tablets, PCs)

[0142] Software: Python, Pandas, Scikit-learn, database management systems (e.g., MySQL)

[0143] Specific example

[0144] For example, if a user views "Product A's page" on an advertiser's website, with 5 page views and the search keyword being "special offer," the artificial intelligence will suggest that the user then view "Product B's page." This suggestion increases the user's purchase intent and contributes to improving the conversion rate.

[0145] Example of a prompt

[0146] If a user views product A's page, has 5 page views, and searches for "special offer," predict which page they will view next.

[0147] In this way, it becomes possible to appropriately suggest the next action that will lead to a sale based on the user's initial actions, and an improvement in the conversion rate and user satisfaction can be expected.

[0148] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0149] Step 1:

[0150] The server collects access logs of users who visit advertiser websites. Specifically, it records information such as which pages users viewed and which keywords they searched for. The input is user behavior data, and the output is access log data.

[0151] Step 2:

[0152] The server stores the collected access log data in a database. The input is the access log data, and the output is the log data stored in the database. Specifically, the data is stored using a database management system (e.g., MySQL).

[0153] Step 3:

[0154] The server provides access log data stored in the database to the artificial intelligence, which uses it as training data. The input is access log data obtained from the database, and the output is a trained artificial intelligence model. Specifically, the data is preprocessed using the Python Pandas library, and the model is trained using the Scikit-learn library.

[0155] Step 4:

[0156] When a user accesses an advertiser's website, the device sends initial user behavior data to the server. The input is the user's initial behavior data, and the output is the data sent to the server. Specifically, the system captures the user's behavior in real time and sends it to the server.

[0157] Step 5:

[0158] The server inputs the received initial action data into artificial intelligence and predicts the next action. The input is the user's initial action data, and the output is a suggestion for the next action. Specifically, the data is input into a trained artificial intelligence model to obtain the prediction result.

[0159] Step 6:

[0160] The server sends the predicted next action to the user's terminal. The input is the prediction result, and the output is the suggestion displayed on the user's terminal. Specifically, the server formats the prediction result and sends it to the user's terminal.

[0161] Step 7:

[0162] The user's device displays the received suggestions to the user. The input is the suggestions sent from the server, and the output is the information displayed to the user. Specifically, the suggestion content is displayed in the user interface.

[0163] In this way, it becomes possible to appropriately suggest the next action that will lead to a sale based on the user's initial actions, and an improvement in the conversion rate and user satisfaction can be expected.

[0164] (Example 2)

[0165] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0166] Traditional advertiser websites have struggled to effectively utilize user behavior data to provide an optimized user experience (UX) for individual users. As a result, they have faced challenges in achieving sufficient improvements in conversion rates. In particular, they were unable to dynamically change the site's display content based on the user's initial behavior, making it difficult to address the diverse needs of each user.

[0167] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0168] This invention includes a server that includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, and means for generating and suggesting subsequent actions that lead to conversions based on the initial actions of users who visit advertiser sites. This enables a system that includes means for collecting user behavior data and storing the generated actions in a database, and means for acquiring the generated actions when a user revisits the site and dynamically changing the displayed content. This makes it possible to provide an optimized UX for each user and improve the conversion rate.

[0169] An "advertiser site" is a website that displays advertisements and allows users to access information about products and services.

[0170] "Access logs" are data that records a user's behavior when they visit a website, and include information such as page views, clicks, and time spent on the site.

[0171] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn and reason, and is particularly used to analyze user behavior data and generate optimal actions.

[0172] "User action" refers to the specific operations or actions that users perform on a website, including actions such as clicking on a product, adding it to a cart, or making a purchase.

[0173] "User experience (UX)" refers to the overall experience and satisfaction that users feel when using a website, and is influenced by factors such as the site's ease of use, design, and the quality of information provided.

[0174] "Dynamic modification" refers to changing the content and functionality of a website in real time, in order to provide the most relevant information based on the user's actions and circumstances.

[0175] "Behavioral data" refers to data about a series of actions and behaviors performed by a user on a website, including browsing history, click patterns, and purchase history.

[0176] A "database" is a system for efficiently storing, managing, and retrieving data, and is used to store generated action and user behavior data.

[0177] "Generated actions" refer to the optimal next actions or suggestions that artificial intelligence generates by analyzing the user's behavioral data.

[0178] "Displayed content" refers to the information and content displayed to users on a website, including text, images, videos, links, etc.

[0179] This invention is a system for dynamically optimizing the user experience (UX) of advertiser websites. Specific embodiments of this system are described below.

[0180] System Overview

[0181] This system consists of three main elements: a server, a terminal, and a user. The server collects user behavior data and generates optimal actions using a generative AI model. The terminal receives the actions generated from the server when the user visits the site and dynamically changes the displayed content. The user experiences an optimized UX by visiting the site.

[0182] Hardware and software to be used

[0183] Server: Use a high-performance server machine (e.g., an AWS® EC2 instance).

[0184] Database: Use a relational database such as MySQL or PostgreSQL.

[0185] Generative AI Model: We use advanced generative AI models such as OpenAI's GPT-4®.

[0186] Analytics tools: We use user behavior analysis tools such as Google Analytics and Mixpanel.

[0187] Data processing and data calculation

[0188] The server collects user behavior data (e.g., browsing history, click patterns, purchase history, etc.) in real time. The collected data is stored in a database. Next, the server inputs prompt messages into a generative AI model to generate the optimal action. The generated action is stored in the database, and when the user revisits the site, the device retrieves it and dynamically changes the displayed content.

[0189] Specific example

[0190] For example, suppose a user visits an online shopping site and frequently browses the sports equipment page. In this case, the server inputs the following prompt message into the AI ​​model:

[0191] Please recommend the best product for this user.

[0192] The generative AI model analyzes the input data and generates an action to "recommend sports equipment." The server saves this action in a database, and when the user revisits the site, the device retrieves this action. Based on the retrieved action, the device displays a banner for sports equipment on the homepage.

[0193] In this way, we can provide a user-optimized UX for each individual user and improve the conversion rate.

[0194] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0195] Step 1:

[0196] A user visits the site.

[0197] Input: The user opens a browser and enters the URL of the advertiser's website.

[0198] Output: A user accesses the site, and initial behavioral data is generated.

[0199] Specific action: The user accesses an online shopping site, and the homepage is displayed.

[0200] Step 2:

[0201] The server collects user behavior data.

[0202] Input: User behavioral data such as browsing history, click patterns, and purchase history.

[0203] Output: The collected behavioral data is stored in the database.

[0204] Specific operation: The server uses analytics tools such as Google Analytics and Mixpanel to collect user behavior data in real time and store it in a database.

[0205] Step 3:

[0206] The server inputs prompt messages into the generated AI model, which then generates the optimal action.

[0207] Input: Collected behavioral data and prompt text (e.g., "Recommend the best product for this user").

[0208] Output: The optimal action obtained from the generative AI model.

[0209] Specific operation: Based on the collected behavioral data, the server inputs a prompt message, "Recommend the best product for this user," into a generating AI model (e.g., GPT-4), and generates the optimal action.

[0210] Step 4:

[0211] The server saves the generated actions to the database.

[0212] Input: The optimal action obtained from the generative AI model.

[0213] Output: Actions saved in the database.

[0214] Specific operation: The server saves the generated actions to a database such as MySQL or PostgreSQL.

[0215] Step 5:

[0216] The device retrieves actions generated from the server when a user visits.

[0217] Input: A request made by a user when they visit the site again.

[0218] Output: Generated actions retrieved from the server.

[0219] Specific operation: When the user visits the site again, the device sends a request to the server and retrieves the generated action.

[0220] Step 6:

[0221] The displayed content is dynamically changed based on the actions taken by the device.

[0222] Input: Generated action retrieved from the server.

[0223] Output: Dynamically changed display content.

[0224] Specific operation: The device dynamically changes the displayed content based on the action it receives, such as displaying a banner for sports equipment on the top page.

[0225] Step 7:

[0226] Users experience an optimized UX.

[0227] Input: Dynamically changed display content.

[0228] Output: Improved user satisfaction and increased conversion rates.

[0229] Specific behavior: Users simply visit the site to experience a personalized UX, with information on sports equipment being displayed preferentially.

[0230] (Application Example 2)

[0231] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0232] Traditional advertiser websites struggled to provide an optimal user experience for each individual user, making it difficult to improve conversion rates. Furthermore, they lacked effective ways to utilize users' past purchase and browsing history, making it impossible to recommend appropriate products to users. This resulted in a failure to capture user interest, ultimately leading to lower conversion rates.

[0233] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0234] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action that leads to a conversion based on the initial actions of a user who visits the advertiser site, means for dynamically changing the user experience of the site based on the suggestion, and means for acquiring the user's past purchase and browsing history and dynamically changing the displayed content based on the generated actions. This makes it possible to provide an optimal user experience for each user and improve the conversion rate.

[0235] An "advertiser site" is a website used to display advertisements and is an online platform where users can visit to view and purchase products and services.

[0236] An "access log" is data that records a user's behavior when they visit a website, and includes information such as the date and time of the visit, the pages viewed, and the links clicked.

[0237] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn, reason, and make judgments, and in particular, it has the ability to analyze user behavior patterns and generate the next action.

[0238] "User experience" refers to the overall experience a user has when using a website or application, and includes elements such as ease of use, satisfaction, and convenience.

[0239] "Purchase history" refers to a record of products and services that a user has purchased in the past, including information such as the date and time of purchase, product name, and price.

[0240] "Browsing history" refers to a record of pages and products viewed by a user on a website, including information such as the date and time of viewing, pages viewed, and products viewed.

[0241] "Dynamically changing" refers to modifying the content and functionality of a website in real time in response to user behavior and circumstances, moving away from a fixed display to a more flexible one.

[0242] "Conversion rate" is an indicator that shows the percentage of users who actually purchased a product or service out of all users who visited a website, and it is an important indicator for measuring the effectiveness of marketing and sales.

[0243] A system for implementing this invention includes means for training an advertiser's website access logs, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action leading to a conversion based on the initial actions of a user visiting the advertiser's website, means for dynamically changing the user experience of the website based on the suggestion, and means for obtaining the user's past purchase and browsing history and dynamically changing the displayed content based on the generated actions.

[0244] System Configuration

[0245] The server will use the following hardware and software.

[0246] Hardware: Servers with high-performance processors, sufficient memory, and storage.

[0247] Software: Python, API request library (requests), Generative AI Model API

[0248] Data processing and data calculation

[0249] 1. Retrieving User Information: When a user visits an advertiser's site, the server retrieves the user's past purchase and browsing history. This is done using API requests.

[0250] 2. Sending prompts to the generating AI model: The server sends the acquired user information as prompts to the generating AI model. The generating AI model then generates the most suitable action for the user (such as recommended products).

[0251] 3. Dynamic changes to site content: The server dynamically changes the site's display content based on the generated actions. Specifically, it reflects recommended products in the site's content.

[0252] Specific example

[0253] For example, if a user has previously purchased "smartphone accessories" and recently viewed "smartwatches," the AI ​​model will generate an action recommending "smartwatch accessories."

[0254] Example of a prompt

[0255] User's past purchase history: Smartphone accessories, Browsing history: Smartwatches

[0256] By sending this prompt message to the AI ​​model, the system will recommend the most suitable products to the user. This allows for a personalized user experience and an improved conversion rate.

[0257] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0258] Step 1:

[0259] The server retrieves the user's past purchase and browsing history when the user visits the advertiser's website.

[0260] Input: User ID

[0261] Data processing: Use API requests to retrieve users' purchase and browsing history.

[0262] Output: User purchase history and browsing history data

[0263] Step 2:

[0264] The server sends the acquired user information as a prompt to the generated AI model.

[0265] Input: User's purchase history and browsing history data

[0266] Data processing: Convert user information into text-based prompt statements and send them to the API of the generated AI model.

[0267] Output: Recommended actions from the generated AI model (e.g., recommended products)

[0268] Step 3:

[0269] The server dynamically changes the site's display content based on the generated actions.

[0270] Input: Recommended actions from the generated AI model

[0271] Data processing: Retrieve the site's content data and update it to prioritize the display of recommended products.

[0272] Output: Updated site content data

[0273] Step 4:

[0274] When users visit a site, they view dynamically changed content.

[0275] Input: Updated site content data

[0276] Data processing: Display updated content in the user's browser.

[0277] Output: User-optimized display content

[0278] Step 5:

[0279] The server records the user's new behavior data in the access log.

[0280] Input: User behavior data (e.g., clicks, browsed pages)

[0281] Data processing: Add new behavior data to the access log.

[0282] Output: Updated access log

[0283] (Example 3)

[0284] Next, Example 3 of Embodiment 3 will be described. In the following description, the data processing device 12 is referred to as a "server", and the smart device 14 is referred to as a "terminal".

[0285] In conventional advertiser sites, it was difficult to sufficiently grasp user behavior patterns and propose optimal actions. As a result, there were problems that the user conversion rate was low and the advertising effect was not fully exhibited. Furthermore, it was not possible to provide optimized proposals for each user, and it was difficult to meet the needs of individual users

[0286] The specifying processing performed by the specifying processing unit 290 of the data processing device 12 in Example 3 is implemented by the following respective means.

[0287] In this invention, the server includes means for collecting access logs from advertiser sites, means for storing the collected access logs in a database, means for pre-processing the stored access logs, means for constructing a generative AI model that learns user behavior patterns using the pre-processed data, means for generating the next action based on the learned behavior patterns, means for transmitting the generated action to a terminal, means for the user to act based on the action displayed on the terminal, and means for recording that action again as an access log. This makes it possible to accurately grasp user behavior patterns and propose actions optimized for each individual user.

[0288] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[0289] "Access logs" are data that records a user's behavior when they visit a website, and include actions such as page views, clicks, and purchases.

[0290] A "database" is a system for storing collected access logs, and includes relational databases and NoSQL databases, among others.

[0291] "Preprocessing" refers to the process of preparing collected access logs into a format that is easy to analyze, and includes data cleaning and normalization.

[0292] A "generative AI model" is a model that uses machine learning algorithms to learn user behavior patterns and predict their next actions.

[0293] An "action" refers to the next step a user should take, such as purchasing a specific product or visiting a specific page.

[0294] A "device" refers to a device used by a user, and includes personal computers, smartphones, tablets, and other similar devices.

[0295] "Behavioral patterns" refer to the tendencies of a series of actions a user takes on a website, and are learned from past access logs.

[0296] A "suggestion" refers to the next action presented to the user by the generative AI model based on the behavioral patterns it has learned.

[0297] "Conversion rate" refers to the percentage of users who take the suggested action and achieve their objective, such as actually purchasing a product.

[0298] This invention is a system that collects access logs from advertiser websites, learns user behavior patterns, generates the optimal next action to take, and proposes it to the user. A specific embodiment of this system is described below.

[0299] Server Processing

[0300] The server collects access logs from advertiser websites. These access logs include information such as pages visited by users, links clicked, and products purchased. This data is collected using web server logs such as Apache® or Nginx.

[0301] Next, the server stores the collected access logs in a database. This database uses a relational database such as MySQL or PostgreSQL. The stored data includes user IDs, visited pages, action timestamps, and more.

[0302] The saved access logs undergo preprocessing, such as data cleaning (imputing missing values ​​and removing outliers) and normalization (scaling the data). The Python Pandas library is often used for this preprocessing.

[0303] Using the preprocessed data, the server builds a generative AI model. This generative AI model is constructed using machine learning frameworks such as TensorFlow and PyTorch. The model learns user behavior patterns from past access logs and predicts the next action to be taken by the user.

[0304] Based on the learned behavior patterns, the server generates the optimal next action that the user should take. For example, for a user who has visited a specific product page, the server proposes products that the user is highly likely to purchase next. This proposal is performed in real time using the generative AI model.

[0305] The generated action is transmitted to the terminal in JSON format via a REST API.

[0306] Processing at the terminal

[0307] The terminal receives the optimal action transmitted from the server and presents it to the user. Examples of such methods include displaying it as a pop-up on a web page, or sending a notification via email.

[0308] Processing by the user

[0309] The user checks the next action displayed on the terminal, and purchases the product if necessary. The user's reaction is also recorded as an access log again, and is used for the next training.

[0310] Specific example

[0311] For example, assume that a user visits the page of "Smartphone A" on an advertiser's site. The server has learned from past access logs and identified the pattern that users who have visited "Smartphone A" purchase "Smartphone Case B" with a high probability. In this case, the server proposes to the user that purchasing "Smartphone Case B" is the next action to take.

[0312] Example of a prompt

[0313] "When a user visits page A on their smartphone, suggest the next product they are most likely to purchase."

[0314] In this way, a system is realized in which the server, terminal, and user work together to generate and present optimal actions to the user based on the user's behavior patterns. The flow of specific processing in Example 3 will be explained with reference to Figure 15.

[0315] Step 1:

[0316] The server collects access logs from advertiser websites. Specifically, it records user actions such as page views, clicks, and purchases within the site. Input includes user behavior data, and output is this data stored in log files.

[0317] Step 2:

[0318] The server stores the collected access logs in a database. Specifically, it uses a relational database such as MySQL or PostgreSQL. The input includes the collected access logs, and the output is structured data stored in the database.

[0319] Step 3:

[0320] The server preprocesses the stored access logs. Specifically, it performs data cleaning (imputing missing values ​​and removing outliers) and normalization (scaling the data). This process uses the Python Pandas library. The input includes the raw data stored in the database, and the output is preprocessed, clean data.

[0321] Step 4:

[0322] The server builds a generative AI model that learns user behavior patterns using preprocessed data. Specifically, it trains a neural network using machine learning frameworks such as TensorFlow or PyTorch. The input includes preprocessed data, and the output is a trained generative AI model.

[0323] Step 5:

[0324] The server generates the next action based on learned behavioral patterns. Specifically, it uses a generative AI model to predict the next action the user should take. The input includes a trained generative AI model and real-time user behavior data, and the output is the optimal next action to take.

[0325] Step 6:

[0326] The server sends the generated action to the terminal. Specifically, it sends the data in JSON format via a REST API. The input includes the generated action, and the output is the action data sent to the terminal.

[0327] Step 7:

[0328] The device presents the user with the optimal action received from the server. Specifically, this can be done by displaying it as a pop-up on a web page or by sending a notification via email. The input includes action data sent from the server, and the output is the action presented to the user.

[0329] Step 8:

[0330] The user reviews the next action displayed on their device and purchases the product if necessary. Specifically, they purchase the product on the website according to the suggested action. The input includes the action displayed on the device, and the output is the user's purchase behavior.

[0331] Step 9:

[0332] The server records the user's response again as an access log. Specifically, it saves the actions taken by the user as a new access log. The input includes user behavior data, and the output is an updated access log.

[0333] (Application Example 3)

[0334] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0335] Traditional advertising delivery systems struggled to adequately understand user behavior patterns, making it difficult to display the right ads at the right time. As a result, they failed to capture user interest, and improvements in conversion rates were not expected. Furthermore, the lack of user-optimized ad display led to a decline in advertising effectiveness.

[0336] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means. In this invention, the server includes means for learning access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and proposing the next action that leads to a conversion based on the initial actions of a user who has visited the advertiser site, means for dynamically changing the site user experience based on the proposal, and means for learning the user's past website visit history and displaying advertisements that the user is likely to be interested in in real time. This makes it possible to grasp the user's behavior patterns in detail and display the appropriate advertisement at the optimal time.

[0337] An "advertiser site" is a website designed to display advertisements and is an online platform intended for user visits.

[0338] An "access log" is data that records a user's behavior when they visit a website, including information such as the pages visited and the time spent on each page.

[0339] Artificial intelligence is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to analyze user behavior patterns and generate the next action.

[0340] "User experience" refers to the experience and feelings a user has when using a website or application, and it affects the ease of use and satisfaction level of the site.

[0341] "Dynamic modification" means changing the content and layout of a website in real time according to user behavior and circumstances, providing an optimized display for each user.

[0342] "Website visit history" refers to a record of websites a user has visited in the past, including information such as pages visited, time spent on each site, and links clicked.

[0343] "Real-time display" means showing advertisements and content instantly based on the user's current actions and circumstances, providing information without delay.

[0344] The system for implementing this invention learns from the access logs of advertiser websites, analyzes user behavior patterns, and displays the most suitable advertisements in real time. A specific embodiment of this system is described below.

[0345] The server first collects access logs from advertiser websites and stores them in a database. These access logs include information such as the pages visited by users, the time spent on each page, and the links clicked. Next, the server uses these access logs to build an artificial intelligence model to learn user behavior patterns. This model is trained using libraries such as pandas or scikit-learn in Python.

[0346] When a user visits an advertiser's website, the server analyzes the user's initial behavior in real time and generates the optimal next action. This action is to display advertisements that the user is likely to be interested in. The server refers to the user's past website visit history and selects the most suitable advertisements based on information such as pages the user has visited in the past and the time spent on each page.

[0347] For example, if a user has previously visited a specific fashion brand's website and stayed on a particular product page for 120 seconds, the server will prioritize displaying new products and sales information from the same brand to that user. In this way, ads are optimized for each user, and an improvement in conversion rates can be expected.

[0348] As a concrete example, by inputting the following prompt into the AI ​​model, it is possible to generate advertisements based on user behavior patterns.

[0349] Example of a prompt:

[0350] Based on a user's past website visits, predict the next ads they should see. For example, if a user visited the website of "Fashion Brand A" and stayed on a specific product page for 120 seconds, then display new products and sales information from "Fashion Brand A" to that user.

[0351] This system allows for a detailed understanding of user behavior patterns and enables the display of appropriate advertisements at the optimal time. This is expected to improve advertising effectiveness and conversion rates.

[0352] The flow of the specific processing in Application Example 3 will be explained using Figure 16.

[0353] Step 1:

[0354] The server collects access logs from advertiser websites and stores them in a database. It receives information such as pages visited, time spent on the site, and links clicked as input, and stores this data in the database. Specifically, the web server collects access logs in real time and stores them in a database management system (e.g., MySQL or PostgreSQL).

[0355] Step 2:

[0356] The server uses collected access logs to build an artificial intelligence model that learns user behavior patterns. It uses access logs stored in a database as input and performs data processing such as feature extraction (e.g., page visit count, average time spent on page). The output is a trained artificial intelligence model. Specifically, it preprocesses the data using Python's pandas library and trains the model using tools like scikit-learn's RandomForestClassifier.

[0357] Step 3:

[0358] When a user visits an advertiser's website, the server analyzes the user's initial behavior in real time. It receives the user's current behavioral data (pages visited, time spent on each page, etc.) as input and feeds this into a previously built artificial intelligence model. The output generates the optimal next action to take (the advertisement to display). Specifically, it inputs real-time collected data into the model and obtains prediction results.

[0359] Step 4:

[0360] The server references the user's past website visit history and selects advertisements that are likely to interest the user. It uses the user's past visit history data as input and analyzes past behavioral patterns as data processing. The output is the selection of the most suitable advertisement. Specifically, it retrieves past visit history from a database and analyzes it using an artificial intelligence model.

[0361] Step 5:

[0362] The server displays selected advertisements to users in real time. It uses selected ad data as input and displays the advertisements on the user's screen as output. Specifically, it dynamically changes the content of the webpage and inserts the advertisements.

[0363] Step 6:

[0364] The server collects user responses after ad display as access logs and stores them in the database. It receives user response data to ads (clicks, time spent on site, etc.) as input and stores this data in the database. Specifically, the web server collects user responses in real time and stores them in the database management system.

[0365] This series of processes allows for a detailed understanding of user behavior patterns, enabling the display of appropriate advertisements at the optimal time. This is expected to improve advertising effectiveness and conversion rates.

[0366] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0367] "Example of form 1"

[0368] One embodiment of this system is a system that combines an emotion engine. This system combines an AI (UX generation AI) that learns from the access logs of advertiser sites and generates the next action that leads to a conversion based on the user's initial behavior, with an emotion engine that recognizes the user's emotions. Specifically, the emotion engine estimates the user's emotions from their actions and reactions when they visit the site, and the UX generation AI generates the next action based on those emotions. For example, if the emotion engine detects the user's excited state while they are browsing a product page, the UX generation AI takes that excited state into account and generates an action such as highlighting the "Buy Now" button. This provides a personalized UX that responds to the user's emotions and contributes to improving the conversion rate.

[0369] "Example of form 2"

[0370] Another embodiment of the present invention involves a system in which an emotion engine and a UX generation AI work together. In this system, the emotion engine recognizes the user's emotions in real time and feeds that information back to the UX generation AI. Based on this feedback, the UX generation AI generates the next action, dynamically changing the site's UX. For example, if the emotion engine detects the user's anxiety while the user is reading a product review, the UX generation AI generates an action to alleviate that anxiety, such as highlighting information about the "stress-free return guarantee." This provides a UX that responds to the user's emotions, thereby improving user satisfaction and conversion rates.

[0371] "Example of form 3"

[0372] Furthermore, another embodiment of the present invention is a system that combines an emotion engine and a UX generation AI. In this system, the emotion engine recognizes the user's emotions, and the UX generation AI generates the next action based on those emotions. Specifically, the emotion engine estimates the user's emotions from their actions and reactions when they visit the site, and the UX generation AI generates the next action based on those emotions. For example, if the emotion engine detects the user's excited state while they are browsing a product page, the UX generation AI takes that excited state into account and generates an action such as highlighting the "Buy Now" button. This provides a personalized UX that responds to the user's emotions and contributes to improving the conversion rate.

[0373] The following describes the processing flow for each example of the form.

[0374] "Example of form 1"

[0375] Step 1: The user visits the advertiser's website.

[0376] Step 2: The emotion engine estimates emotions from the user's behavior and reactions.

[0377] Step 3: The UX generation AI generates the next action based on the emotional information from the emotion engine.

[0378] Step 4: Dynamically change the UX based on the actions the site generates.

[0379] "Example of form 2"

[0380] Step 1: The user visits the advertiser's website.

[0381] Step 2: The emotion engine recognizes the user's emotions in real time.

[0382] Step 3: The emotion engine feeds back the recognized emotion information to the UX generation AI.

[0383] Step 4: The UX generation AI generates the next action based on the feedback, dynamically changing the site's UX.

[0384] "Example of form 3"

[0385] Step 1: The user visits the advertiser's website.

[0386] Step 2: The emotion engine estimates emotions from the user's behavior and reactions.

[0387] Step 3: The UX generation AI generates the next action based on the emotional information from the emotion engine.

[0388] Step 4: Dynamically change the UX based on the actions the site generates.

[0389] (Example 1)

[0390] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0391] Traditional advertiser websites generated subsequent actions based solely on user behavior data, making it difficult to provide personalized suggestions that considered user emotions. As a result, improvements in conversion rates were limited. This invention aims to improve conversion rates by combining user behavior data and emotional data to generate more accurate subsequent actions.

[0392] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0393] In this invention, the server includes means for collecting user behavior data, means for collecting user sentiment data, means for transmitting the collected behavior data and sentiment data to the server, means for cleansing and preprocessing the transmitted data, means for training an AI model using the preprocessed data, means for inferring the user's next action in real time using the trained AI model, and means for displaying the inferred action to the user. This enables personalized suggestions that combine the user's behavior data and sentiment data.

[0394] "User behavior data" refers to information about actions users perform on a website, such as clicks, page transitions, and search keywords.

[0395] "User emotion data" refers to information about the emotional state of a user, estimated from their facial expressions and voice.

[0396] A "server" is a computer system that collects, processes, and stores user behavioral and emotional data, and trains and runs AI models.

[0397] "Data cleansing" is the process of removing missing or outlier values ​​from collected data and formatting the data.

[0398] "Preprocessing" refers to the process of converting data into a format suitable for training an AI model.

[0399] An "AI model" is an artificial intelligence algorithm that infers the next action based on user behavior data and emotional data.

[0400] "Training" is the process of improving the prediction accuracy of an AI model by having it learn from a large amount of data.

[0401] "Real-time inference" is a process that instantly analyzes user behavior and emotional data to immediately generate the next action.

[0402] "Personalized suggestions" are suggestions optimized for the user, generated based on the user's individual behavioral and emotional data.

[0403] Modes for carrying out the invention

[0404] This invention relates to a system that learns from the access logs of advertiser websites and generates the next action based on user behavior data and sentiment data. This system collects user behavior data and sentiment data, sends it to a server for processing, and then provides personalized suggestions.

[0405] server

[0406] The server is a computer system for collecting, processing, and storing user behavioral and emotional data. Specifically, the server uses Apache Hadoop to distribute and process large amounts of access log data, and cleans and preprocesses the data. Next, TensorFlow is used to train an AI model (UX generation AI) to generate the next action that leads to a conversion based on the user's initial behavior. This AI model analyzes the user's behavior, such as accessing a specific product page or searching for a specific keyword, and determines which page to guide them to next and which product to recommend.

[0407] terminal

[0408] The device collects user behavior data in real time when a user visits an advertiser's website. Specifically, it uses JavaScript (registered trademark) to collect data such as user clicks, page transitions, and search keywords, and sends this data to the server. The device also has an emotion engine that estimates the user's emotions from their facial expressions and voice. For example, it uses a combination of OpenCV and TensorFlow to analyze the user's facial expressions and determine whether the user is excited or not.

[0409] User

[0410] Users visit advertiser websites, browse specific product pages, or search for specific keywords. User behavior and emotional data are collected in real time and sent to a server. Based on this data, the server's UX generation AI generates the next action and provides personalized suggestions to the user. For example, if excitement is detected while the user is browsing a product page, the server generates an action such as highlighting the "Buy Now" button.

[0411] Specific example

[0412] Specific Example 1

[0413] A user visits an advertiser's website and accesses a specific product page. The device collects this behavioral data and sends it to a server. The server uses UX generation AI to suggest related product pages that the user might be interested in next.

[0414] Specific Example 2

[0415] A user searches for specific keywords on an advertiser's website. The device collects this search data and sends it to the server. The server uses UX generation AI to recommend products related to the keywords the user searched for.

[0416] Specific example 3

[0417] As a user browses a product page, the emotion engine detects the user's level of excitement. Based on this information, the server's UX generation AI generates an action that highlights the "Buy Now" button.

[0418] Example of a prompt

[0419] "Build an AI model that generates subsequent actions based on the user's initial behavior, using access logs from advertiser websites. Design a system that collects user behavior and sentiment data in real time and provides personalized suggestions."

[0420] By explaining the system's processing from the perspectives of the server, terminal, and user, the role and specific operation of each component become clear.

[0421] The flow of the specific processing in Example 1 will be explained using Figure 17.

[0422] Step 1:

[0423] The device collects user behavior data when a user visits an advertiser's website. Specifically, it uses JavaScript (registered trademark) to obtain data such as user clicks, page transitions, and search keywords in real time. The input is user operation data, and the output is collected behavior data. This behavior data includes recording the search keyword and search time when a user searches for "smartphone."

[0424] Step 2:

[0425] The device also collects user emotion data. Specifically, it uses the camera and microphone built into the device to analyze the user's facial expressions and voice. By combining OpenCV and TensorFlow, it estimates emotions from the user's facial expressions and evaluates the intensity of those emotions through voice analysis. The input is the user's facial expressions and voice data, and the output is the analyzed emotion data. For example, if a user smiles while browsing a product page, the degree of that smile is recorded.

[0426] Step 3:

[0427] The device sends collected behavioral and emotional data to the server. The input is the collected behavioral and emotional data, and the output is the data sent to the server. For example, if a user searches for "smartphone" and smiles while doing so, that information is sent to the server.

[0428] Step 4:

[0429] The server cleanses and preprocesses the transmitted data. Specifically, it imputes missing values ​​and removes outliers, converting the data into a format suitable for training the AI ​​model. The input is the transmitted behavioral and sentiment data, and the output is the preprocessed data. For example, if the user's search keywords are incomplete, the server will perform a process to complete them.

[0430] Step 5:

[0431] The server uses pre-processed data to train a UX generation AI model using TensorFlow. The input is the pre-processed data, and the output is the trained AI model. This model generates subsequent actions that lead to conversions based on the user's initial actions. For example, if a user searches for "smartphone," the model learns to recommend "smartphone accessories" next.

[0432] Step 6:

[0433] The server uses a trained UX generation AI model to infer the user's next action in real time. The input is user behavior and emotion data, and the output is the inferred next action. Specifically, it takes user behavior and emotion data from when the user visits the site as input to determine which page to guide the user to next and which product to recommend. For example, if a user searches for "smartphone" and an excited state is detected, it will generate an action that highlights the "Buy Now" button.

[0434] Step 7:

[0435] The device displays the next action received from the server to the user. The input is the next action sent from the server, and the output is a personalized suggestion displayed to the user. Specifically, suggestions generated based on the user's behavioral and sentiment data are reflected on the web page. For example, a "Buy Now" button might be highlighted, or banners recommending related products might be displayed.

[0436] (Application Example 1)

[0437] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server," and the smart device 14 will be referred to as a "terminal."

[0438] Traditional advertising systems can suggest the next action based on user behavior data, but they cannot provide personalized suggestions that take user emotions into account, limiting their ability to improve conversion rates. Furthermore, because they do not dynamically change the site user experience in response to user emotions, it is difficult to maximize user interest and engagement.

[0439] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means. In this invention, the server includes means for training access logs of the advertiser site, means for building artificial intelligence that generates user actions leading to conversions, means for generating and proposing the next action that leads to conversions based on the initial actions of users who visit the advertiser site, means for combining an emotion engine that recognizes the user's emotions, and means for dynamically changing the site user experience based on the proposal. This makes it possible to integrate user behavior data and emotion data to generate the next action and provide personalized proposals optimized for each user.

[0440] An "advertiser site" is a website operated by a company or organization that displays advertisements.

[0441] An "access log" is data that records a user's behavior when they visit a website.

[0442] "Artificial intelligence" is a technology that allows computers to learn and reason by mimicking human intelligence.

[0443] "User action" refers to the specific operations or actions that users perform on a website.

[0444] "Initial actions" refer to the first actions or processes a user takes when visiting a website.

[0445] An "emotion engine" is a technology that estimates emotions from a user's behavior and reactions.

[0446] "Website user experience" refers to the experience and satisfaction that users feel when using a website.

[0447] "Dynamic modification" refers to changing the content and display of a website in real time in response to user behavior and emotions.

[0448] "Conversion rate" refers to the percentage of users who visit a website and actually purchase a product or use a service.

[0449] "Personalization" refers to providing optimal content and suggestions based on the individual user's characteristics and behavior.

[0450] The system for implementing this invention learns from the access logs of advertiser websites and integrates user behavior data and sentiment data to generate the next action. A specific embodiment of this system is described below.

[0451] System Configuration

[0452] hardware

[0453] Server: A high-performance server used for data collection, training, and analysis.

[0454] User device: A device used by a user, such as a smartphone, tablet, or personal computer.

[0455] software

[0456] Artificial Intelligence (AI): An AI model that learns from user behavior data and generates the next action.

[0457] Emotion engine: Software used to estimate emotions from user behavior and reactions.

[0458] Database: A database used to store user behavior data and emotional data.

[0459] API: An interface for collecting user behavior data and sending it to a server.

[0460] Data processing and data calculation

[0461] 1. Collection of user behavior data:

[0462] We collect user behavior data (e.g., access to specific product pages, searches for specific keywords, etc.) when users visit advertiser websites via API.

[0463] 2. Analysis of emotions:

[0464] An emotion engine is used to estimate a user's emotions from collected behavioral data. For example, it can detect a user's level of excitement from their mouse movements and click frequency while they are browsing a product page.

[0465] 3. Generate the next action:

[0466] Artificial intelligence is used to generate the next action based on user behavior and emotional data. For example, if a user is excited, the system might generate an action such as highlighting the "Buy Now" button.

[0467] Specific example

[0468] For example, suppose a user visits an advertiser's website on their smartphone and is viewing a specific product page. If the emotion engine detects the user's excited state, the artificial intelligence will generate an action that highlights the "Buy Now" button. This action is a personalized suggestion tailored to the user's emotions and contributes to an improved conversion rate.

[0469] Example of a prompt

[0470] When a user visits an advertiser's website and is viewing a specific product page, and the emotion engine detects the user's state of excitement, what actions will the artificial intelligence generate?

[0471] In this way, the emotion-responsive advertising assistant system integrates user behavioral data and emotional data to generate the next action and provide personalized suggestions optimized for each user.

[0472] The flow of a specific process in Application Example 1 will be explained using Figure 18.

[0473] Step 1:

[0474] A user visits an advertiser's website. The user accesses the advertiser's website using a device such as a smartphone or computer. At this time, user behavior data (e.g., access to a specific product page, searching for a specific keyword) is collected from the device. The input is the user's behavior data, and the output is the collected behavior data.

[0475] Step 2:

[0476] The device sends collected behavioral data to the server. The device sends collected behavioral data to the server using an API. The input is the collected behavioral data, and the output is the behavioral data sent to the server.

[0477] Step 3:

[0478] The server saves the behavioral data to the database. The server saves the received behavioral data to the database. The input is the behavioral data sent to the server, and the output is the behavioral data saved in the database.

[0479] Step 4:

[0480] The server uses an emotion engine to estimate the user's emotions from behavioral data. The server uses the emotion engine to analyze the user's emotions from stored behavioral data. For example, it can detect the user's level of excitement from mouse movements and click frequency while they are browsing a product page. The input is behavioral data stored in the database, and the output is estimated user emotion data.

[0481] Step 5:

[0482] The server uses artificial intelligence to generate the next action. The server uses artificial intelligence to generate the next action based on behavioral and emotional data. For example, if the user is excited, it will generate an action to highlight the "Buy Now" button. The input is behavioral and emotional data, and the output is the generated next action.

[0483] Step 6:

[0484] The server sends the next generated action to the terminal. The server sends the next generated action to the terminal. The input is the next generated action, and the output is the next action sent to the terminal.

[0485] Step 7:

[0486] The device dynamically changes the site user experience based on the next action it receives. The device dynamically changes the site's display and operation based on the next action it receives. For example, it might highlight the "Buy Now" button. The input is the next action sent to the device, and the output is the dynamically changed site user experience.

[0487] (Example 2)

[0488] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0489] Traditional advertiser websites can generate the next action based on user behavior data and dynamically change the site's user experience (UX). However, because UX optimization that takes user emotions into account is not performed, there are limitations to improving user satisfaction and conversion rates. Furthermore, there is a lack of systems to provide the optimal UX for each user, making it difficult to provide an UX that meets the needs of individual users.

[0490] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0491] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action that leads to a conversion based on the initial actions of users who visit the advertiser site, means for dynamically changing the site user experience based on the suggestion, an emotion recognition engine that recognizes the user's emotions in real time, and means for generating the next action based on feedback from the emotion recognition engine. This makes it possible to optimize the UX by considering both user behavior data and emotion data, providing the optimal UX for each user and improving conversion rates and user satisfaction.

[0492] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[0493] An "access log" is a record of user behavior when they visit a website, and includes information such as page viewing history, click history, and time spent on the site.

[0494] Artificial intelligence is a technology that allows computers to learn and reason by mimicking human intelligence, and to generate the next action based on user behavior data and emotional data.

[0495] "User experience" refers to the experience and satisfaction a user feels when using a website, and is influenced by factors such as the site's design, functionality, and content quality.

[0496] An "emotion recognition engine" is a technology that recognizes emotions in real time from the user's facial expressions, voice, text input, etc., and is used to understand the user's emotional state.

[0497] "Feedback" refers to the process of returning information to a system to determine its next action based on the information it has received, and it involves sending emotional data from the emotion recognition engine to artificial intelligence.

[0498] "Dynamic modification" means changing the website's display content and functions in real time in response to user behavior and emotions, in order to provide the optimal experience for each user.

[0499] Modes for carrying out the invention

[0500] This invention is a system that dynamically changes the user experience (UX) of an advertiser's website, generating subsequent actions based on user behavior data and emotional data. Specific embodiments of this system are described below.

[0501] Hardware and software to be used

[0502] Server: The server hosts the generative AI model and emotion recognition engine. This allows for real-time recognition of user emotions and dynamic modification of the UX.

[0503] Device: The device used by the user (PC, smartphone, tablet, etc.) receives dynamic UX changes sent from the server.

[0504] Generative AI Model: The generative AI model generates the next action based on user behavior data and emotional data.

[0505] Emotion Recognition Engine: The emotion recognition engine recognizes the user's emotions in real time and feeds that information back into the generating AI model.

[0506] Program processing

[0507] 1. User Visit: A user visits the advertiser's website. For example, the user opens a browser, enters the advertiser's URL, and accesses the site.

[0508] 2. Data Collection: The server collects user behavior data (clicks, scrolls, time spent on the site, etc.). This provides information such as which pages the user is viewing and which links they are clicking.

[0509] 3. Emotion Recognition: The emotion recognition engine recognizes the user's emotions in real time. For example, when a user is reading a product review, it can detect whether they are feeling anxious based on their facial expressions and tone of voice.

[0510] 4. Feedback: The server feeds back the emotion data recognized by the emotion recognition engine to the generating AI model. For example, it sends information such as "the user is feeling anxious" to the generating AI model.

[0511] 5. Action Generation: The generation AI model generates the next action based on the feedbacked emotional and behavioral data. For example, if a user is feeling anxious, it will generate an action to highlight information about the "peace of mind return guarantee" as an action to alleviate that anxiety.

[0512] 6. UX Changes: The server dynamically changes the site's UX based on generated actions. For example, it dynamically changes HTML and CSS to prioritize the display of specific product information.

[0513] 7. Display: The device displays the new UX sent from the server. For example, information such as "Worry-Free Returns Guarantee" is highlighted on the page the user is viewing.

[0514] Specific example

[0515] Example 1: When a user visits a site and is looking for a specific product, the generative AI model prioritizes displaying information about that product.

[0516] Example 2: When the emotion recognition engine detects anxiety while a user is reading a product review, the generative AI model highlights information about the "peace of mind return guarantee."

[0517] Example of a prompt

[0518] "Please generate an action that prioritizes displaying specific products when a user visits the site."

[0519] "When a user feels uneasy while reading a product review, generate an action to alleviate that uneasy feeling."

[0520] This system makes it possible to provide an optimal user experience for each individual user, thereby improving conversion rates and user satisfaction.

[0521] The flow of the specific processing in Example 2 will be explained using Figure 19.

[0522] Step 1:

[0523] A user visits the site.

[0524] The user opens a browser, enters the advertiser's URL, and accesses the site. The input is the user's access action, and the output is the display of the site's homepage.

[0525] Step 2:

[0526] The server collects user behavior data.

[0527] The server collects user behavior data in real time, such as clicks, scrolls, and time spent on the site. The input is user behavior data, and the output is a log of the collected behavior data. Specifically, it records information such as which pages the user viewed and which links they clicked.

[0528] Step 3:

[0529] The emotion recognition engine recognizes the user's emotions in real time.

[0530] The emotion recognition engine recognizes emotions from the user's facial expressions, voice, and text input. The input is the user's facial expression and voice data, and the output is the recognized emotion data. For example, when a user is reading a product review, the engine can detect whether they are feeling anxious based on their facial expression and tone of voice.

[0531] Step 4:

[0532] The server generates emotional data and feeds it back to the AI ​​model.

[0533] The server sends the emotion data recognized by the emotion recognition engine to the generative AI model. The input is emotion data, and the output is feedback to the generative AI model. For example, it sends information that "the user is feeling anxious" to the generative AI model.

[0534] Step 5:

[0535] The generative AI model generates the next action.

[0536] The generative AI model generates the next action based on the feedback data of emotional and behavioral data. The input is emotional and behavioral data, and the output is the generated action. For example, if the user is feeling anxious, the model will generate an action to highlight information about the "stress-free return guarantee" as an action to alleviate that anxiety.

[0537] Step 6:

[0538] The server dynamically changes the site's UX based on the actions it generates.

[0539] The server dynamically modifies the site's UX based on actions generated by a generative AI model. The input is the generated action, and the output is the modified UX. For example, it dynamically modifies HTML and CSS to prioritize the display of specific product information.

[0540] Step 7:

[0541] The device displays a new UX.

[0542] The device displays the new UX sent from the server. The input is the modified UX, and the output is the new UX displayed on the user's device. For example, the page the user is viewing will have the information "Worry-Free Returns Guarantee" highlighted.

[0543] (Application Example 2)

[0544] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0545] Traditional advertiser websites can generate the next action based on user behavior patterns and dynamically change the site's user experience, but they have not been able to provide an optimal user experience that takes user emotions into consideration. Therefore, in order to further improve user satisfaction and conversion rates, a system is needed that recognizes user emotions in real time and dynamically changes the site's displayed content based on that.

[0546] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0547] In this invention, the server includes means for learning access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, and means for generating and suggesting subsequent actions that lead to conversions based on the initial actions of users who visit advertiser sites. This makes it possible to learn user behavior patterns and emotions, generate subsequent actions based on them, and dynamically change the user experience of the site.

[0548] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[0549] An "access log" is data that records a user's behavior when they visit a website, and includes information such as the date and time of the visit, the pages viewed, and the links clicked.

[0550] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to automatically perform specific tasks.

[0551] "User experience" refers to the experience and satisfaction that users feel when using a website or application, and is influenced by factors such as ease of use, design, and functionality.

[0552] An "emotion engine" is a technology that recognizes emotions in real time from a user's facial expressions, voice, etc., and is used to analyze the user's psychological state.

[0553] "Feedback" refers to the return of information to a system to determine its next action based on the information it has received, and it is the process by which the system adjusts its operation based on the user's behavior and emotions.

[0554] "Dynamic modification" refers to a system automatically changing its display content and functions in real time according to the situation, in order to provide the optimal experience for each user.

[0555] A system for implementing this invention includes means for training an advertiser's website access logs, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action leading to a conversion based on the initial actions of a user who visits the advertiser's website, an emotion engine that recognizes the user's emotions in real time, and user experience generating artificial intelligence that generates the next action based on feedback from the emotion engine.

[0556] System program

[0557] The program for this system works as follows:

[0558] Hardware and software

[0559] Hardware: Smartphone camera

[0560] Software: OpenCV (image processing library), Keras (deep learning library), UX generation AI module

[0561] Data processing and data calculation

[0562] 1. Image Capture: The user's face is captured using the smartphone's camera. This obtains the user's facial expression data.

[0563] 2. Emotion Recognition: Face detection is performed using OpenCV, and emotions are recognized using a Keras model. Specifically, the captured face image is converted to grayscale, the face region is extracted, and input into the emotion recognition model.

[0564] 3. UX Generation: The recognized emotions are fed back to the UX generation AI to generate the next action. For example, if the user is feeling "anxious," the UX generation AI will generate an action that highlights information about the "peace of mind return guarantee."

[0565] 4. UX Update: Dynamically change the site's user experience based on generated actions. This makes it possible to provide the best possible experience for each user.

[0566] Specific example

[0567] When a user is browsing an online shopping site on their smartphone, the camera captures the user's face, and if the emotion engine detects "anxiety," the UX generation AI generates an action that highlights information about the "peace of mind return guarantee." Based on this action, the site's display content is dynamically changed to alleviate the user's anxiety.

[0568] Example of a prompt

[0569] Create a program that, when a user is browsing an online shopping site, captures their face with a camera, and if the emotion engine detects "anxiety," generates an action where the UX generation AI highlights information about the "peace of mind return guarantee."

[0570] In this way, by providing a user experience that responds to users' emotions, it is possible to improve user satisfaction and conversion rates.

[0571] The flow of a specific process in Application Example 2 will be explained using Figure 20.

[0572] Step 1:

[0573] The device uses the smartphone's camera to capture the user's face. The input is real-time video data, and the output is the captured face image. This face image is used in subsequent processing steps.

[0574] Step 2:

[0575] The device uses OpenCV to detect the facial region from a captured facial image. The input is the facial image obtained in step 1, and the output is coordinate data indicating the facial region. The facial portion is then extracted based on this coordinate data.

[0576] Step 3:

[0577] The device converts the extracted facial image to grayscale and performs preprocessing for input into the Keras model. The input is the facial region data obtained in step 2, and the output is the preprocessed facial image data. This data is then input into the emotion recognition model.

[0578] Step 4:

[0579] The device inputs pre-processed facial image data into a Keras model to recognize the user's emotions. The input is the facial image data obtained in step 3, and the output is the recognized emotion label (e.g., "anxiety," "excitement," etc.). This emotion label is used in subsequent processing steps.

[0580] Step 5:

[0581] The server feeds the recognized emotion labels back to the UX generation AI to generate the next action. The input is the emotion label obtained in step 4, and the output is the generated action (e.g., highlight the "Return Guarantee" information). This action is used to dynamically change the content displayed on the site.

[0582] Step 6:

[0583] The server dynamically modifies the site's user experience based on the actions generated. The input is the action obtained in step 5, and the output is the modified site display. This makes it possible to provide the optimal experience for each user.

[0584] (Example 3)

[0585] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0586] Traditional advertiser websites faced challenges in improving conversion rates because they couldn't adequately understand user behavior patterns and emotions. Furthermore, they lacked the ability to generate optimized actions for each user and dynamically change the site's user experience (UX). This made it difficult to provide personalized experiences tailored to users' interests and emotions.

[0587] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means.

[0588] In this invention, the server includes means for collecting access logs from advertiser sites, means for analyzing the collected access logs and learning user behavior patterns, means for generating the next action based on the learned behavior patterns, means for reflecting the generated action on the advertiser site, means for collecting user behavior and reactions in real time and estimating emotions, means for generating the next action based on the estimated emotions, and means for reflecting the generated action on the advertiser site. This makes it possible to generate optimal actions based on user behavior patterns and emotions and dynamically change the site's UX.

[0589] An "advertiser site" is a website that displays advertisements, and users can view the content of these advertisements by accessing it.

[0590] An "access log" is data that records information about when a user accesses a website, and includes information such as the IP address, date and time of access, and pages visited.

[0591] "Behavioral patterns" refer to the tendencies and patterns of a series of actions that users take on a website, including actions such as visiting or clicking on specific pages.

[0592] A "machine learning model" refers to an algorithm or mathematical model that learns patterns and rules based on data to perform predictions and classifications.

[0593] A "natural language processing model" refers to algorithms and mathematical models for understanding and generating human language, and is used for analyzing and generating text data.

[0594] An "emotion engine" is a system that estimates emotions from a user's behavior and reactions, and analyzes emotions using natural language processing models and other methods.

[0595] "UX generation AI" refers to an artificial intelligence system that generates the next action based on the user's behavior patterns and emotions in order to optimize the user experience (UX).

[0596] "Dynamic modification" refers to changing the content and display of a website in real time, including changing elements of a webpage in response to user behavior and emotions.

[0597] This invention is a system that collects access logs from advertiser websites, analyzes user behavior patterns and emotions, generates optimal actions, and dynamically changes the user experience (UX) of the site. This system is implemented using the following hardware and software.

[0598] Hardware and software to be used

[0599] 1. Server

[0600] Web server software: Use Apache, Nginx, etc., to save user access data to log files.

[0601] Database: Use a database such as MySQL or PostgreSQL to store and manage the collected access logs.

[0602] Machine learning libraries: We use programming languages ​​such as Python and R, along with machine learning libraries such as Scikit-learn and TensorFlow, to learn user behavior patterns.

[0603] Natural language processing models: We use OpenAI's GPT-4 and IBM Watson®, among others, to estimate user emotions.

[0604] 2. Terminal

[0605] Web browser: Used by users to access advertiser websites.

[0606] JavaScript (registered trademark) and HTML: Used to dynamically change the content of a website.

[0607] Specific operation of the system

[0608] 1. Collection of access logs

[0609] When a user accesses an advertiser's website, the server records information such as the user's IP address, access date and time, and visited pages in a log file.

[0610] The server periodically analyzes log files and saves them to the database.

[0611] 2. Learning behavioral patterns

[0612] The system analyzes access logs collected by the server to learn user behavior patterns. For example, it learns that users who visit a specific product page A are highly likely to also visit product page B.

[0613] The server trains a machine learning model and extracts behavioral patterns.

[0614] 3. Generate the next action

[0615] Based on the behavioral patterns the server has learned, it generates the following actions. For example, it might generate an action to recommend product page B to a user who has visited product page A.

[0616] The server-generated actions are saved to the database.

[0617] 4. Reflecting the Action

[0618] The server-generated actions are reflected on the advertiser's website. Specifically, JavaScript (registered trademark) and HTML are used to dynamically change the content of the web page.

[0619] The device displays the actions generated for the user.

[0620] 5. Estimation of emotions

[0621] The server collects user behavior and reactions in real time and estimates their emotions. For example, it collects behavioral data such as page scrolling speed and click frequency.

[0622] The server inputs this data into the emotion engine to estimate the user's emotions.

[0623] 6. Generating emotion-based actions

[0624] Based on the server's emotion engine output, the UX generation AI generates the next action. For example, if the user is showing signs of excitement, it generates an action to highlight the "Buy Now" button.

[0625] The server-generated actions are reflected on the advertiser's website.

[0626] Examples of specific cases and prompt statements

[0627] Example 1: Learning behavioral patterns and generating actions

[0628] A user visits a specific product page A.

[0629] The server collects access logs and records that a user visited product page A.

[0630] The server analyzes past access logs and learns patterns such as a high probability that a user will visit product page B after visiting product page A.

[0631] Based on this pattern, the server generates an action that recommends product page B to a user who visited product page A.

[0632] Example of a prompt

[0633] When a user visits product page A, generate an action that recommends product page B based on patterns learned from past access logs.

[0634] Example 2: Integration of an emotion engine and UX generation AI

[0635] When a user is browsing a product page, they exhibit behaviors that indicate excitement (for example, rapidly scrolling through the page).

[0636] The server inputs this action into the emotion engine to estimate the user's state of excitement.

[0637] Based on the emotion engine's output, the server uses a UX generation AI to generate an action that highlights the "Buy Now" button.

[0638] Example of a prompt

[0639] When a user exhibits behavior indicating excitement while browsing a product page, generate an action that highlights the "Buy Now" button based on the emotion engine's output.

[0640] The above describes specific embodiments of the system of the present invention. The flow of the specific processing in Example 3 will be explained with reference to Figure 21.

[0641] Step 1:

[0642] Collection of access logs

[0643] The server collects access logs from advertiser websites. Specifically, it uses web server software such as Apache or Nginx to save user access data to log files.

[0644] Input: Information such as the user's IP address, access date and time, and visited pages.

[0645] Output: Access data recorded in the log file.

[0646] Specific operation: When a user accesses an advertiser's site, the server records that access information in a log file.

[0647] Step 2:

[0648] Learning behavioral patterns

[0649] The system analyzes access logs collected by the server to learn user behavior patterns. This analysis uses programming languages ​​such as Python and R, and machine learning libraries such as Scikit-learn and TensorFlow.

[0650] Input: Access data recorded in the log file.

[0651] Output: Learned behavioral pattern model.

[0652] Specific operation: The server reads past access logs, analyzes user behavior over time, and trains a machine learning model to extract behavioral patterns.

[0653] Step 3:

[0654] Next Action Generation

[0655] Based on the behavioral patterns the server has learned, it generates the following actions. For example, it might generate an action to recommend product page B to a user who has visited a specific product page A.

[0656] Input: Trained behavioral pattern model, current user behavior data.

[0657] Output: The next action generated.

[0658] Specific operation: When a user visits product page A, the server refers to the behavioral pattern model and generates an action that recommends product page B.

[0659] Step 4:

[0660] Reflecting the action

[0661] The server-generated actions are reflected on the advertiser's website. Specifically, JavaScript (registered trademark) and HTML are used to dynamically change the content of the web page.

[0662] Input: The next action generated.

[0663] Output: A dynamically modified webpage.

[0664] Specific operation: The server generates JavaScript code to add a link to product page B on product page A, and the device displays the link to product page B to the user.

[0665] Step 5:

[0666] Estimation of emotions

[0667] The server collects user behavior and reactions in real time and estimates their emotions. This is done using tools such as Google Analytics and Hotjar.

[0668] Input: User behavior data (page scroll speed, click frequency, etc.).

[0669] Output: Estimated user sentiment.

[0670] Specific operation: While a user is browsing a product page, the server collects behavioral data such as page scrolling speed and click frequency, and inputs this data into the sentiment engine to estimate the user's emotions.

[0671] Step 6:

[0672] Generating emotion-based actions

[0673] Based on the server's emotion engine output, the UX generation AI generates the next action. For example, if the user is showing signs of excitement, it generates an action to highlight the "Buy Now" button.

[0674] Input: Estimated user sentiment, current user behavior data.

[0675] Output: The next action generated.

[0676] Specific operation: The server receives the output from the emotion engine, inputs it into the UX generation AI, and generates an action that highlights the "Buy Now" button.

[0677] Step 7:

[0678] Reflecting emotion-based actions

[0679] The server-generated actions are reflected on the advertiser's website. Specifically, JavaScript (registered trademark) and HTML are used to dynamically change the content of the web page.

[0680] Input: The next action generated.

[0681] Output: A dynamically modified webpage.

[0682] Specific operation: The server generates JavaScript code that highlights the "Buy Now" button, and the device displays the highlighted "Buy Now" button to the user.

[0683] (Application Example 3)

[0684] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0685] Traditional advertising systems could generate subsequent actions based on user behavior patterns, but they struggled to provide a personalized user experience that took user emotions into account. As a result, they couldn't display the most suitable ads based on user interests and emotions, and thus couldn't expect to improve conversion rates. Furthermore, there was a lack of means to dynamically change the user experience on the site, making it difficult to display ads optimized for each user.

[0686] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means.

[0687] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting subsequent actions that lead to conversions based on the initial actions of users who visit advertiser sites, an emotion engine that recognizes user emotions, and user experience generation artificial intelligence that generates subsequent actions based on the emotions recognized by the emotion engine. This makes it possible to display optimal advertisements based on user behavior patterns and emotions and to dynamically change the user experience of the site.

[0688] An "advertiser site" is a website designed to display advertisements and is an online platform intended for user visits.

[0689] "Access logs" are data that records a user's behavior when they visit a website, including information such as page views and clicks.

[0690] Artificial intelligence is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to analyze user behavior patterns and generate optimal actions.

[0691] "User experience" refers to the overall experience a user has when using a website or application, and includes aspects such as the ease of use of the interface and the appeal of the content.

[0692] An "emotion engine" is a technology for recognizing and analyzing a user's emotions, estimating their emotional state from their behavior and reactions.

[0693] "User experience generating artificial intelligence" is an artificial intelligence that generates the next action based on the user's emotions recognized by the emotion engine, and is a technology for providing a personalized user experience.

[0694] "Dynamic modification" means changing the content and interface of a website or application in real time in response to user behavior and emotions, thereby providing an optimized experience for each user.

[0695] The system for implementing this invention learns from the access logs of advertiser websites and generates optimal actions based on user behavior patterns and emotions. Specific embodiments of this system are described below.

[0696] System Configuration

[0697] hardware

[0698] Server: A high-performance server for data processing and running artificial intelligence models.

[0699] Device: A smartphone or computer used by the user to access the system.

[0700] Sensors: Cameras and microphones used to recognize the user's emotions.

[0701] software

[0702] Python: A programming language used for data processing and implementing artificial intelligence models.

[0703] scikit-learn: A machine learning library for learning user behavior patterns.

[0704] Custom Sentiment Engine: Software for analyzing user emotions in real time.

[0705] Custom Ad Recommendation Engine (AdRecommendationEngine): Software for generating optimal ads based on user emotions and behavioral patterns.

[0706] Data processing and calculations

[0707] Access log collection and learning

[0708] The server collects access logs when users visit advertiser sites. These access logs include page view history, click information, and time spent on the site. The collected data is analyzed using machine learning algorithms such as clustering and classification with scikit-learn.

[0709] Recognition of emotions

[0710] The device's built-in camera and microphone capture the user's facial expressions and voice tone in real time. This data is analyzed by a custom SentimentEngine to estimate the user's emotional state.

[0711] Next Action Generation

[0712] The server combines user behavior patterns learned from access logs with sentiment data recognized by the sentiment engine to generate the next ad to display. This process utilizes a custom ad recommendation engine (AdRecommendationEngine). The generated ad is personalized according to the user's emotional state and displayed at the optimal time.

[0713] Specific example

[0714] For example, if the emotion engine detects the user's state of excitement while they are browsing the page for "Product A," it will then highlight an advertisement for "Product B." In this way, it is possible to provide optimal advertisements based on the user's behavior patterns and emotions.

[0715] Example of a prompt

[0716] When a user is browsing a product page, if the sentiment engine detects the user's level of excitement, generate the next ad to display. Recommend the most relevant ad based on the user's access logs and sentiment data.

[0717] In this way, it becomes possible to provide personalized advertising based on users' behavior patterns and emotions.

[0718] The flow of the specific processing in Application Example 3 will be explained using Figure 22.

[0719] Step 1:

[0720] The server collects access logs when users visit advertiser websites. It receives data such as the user's page view history, click information, and time spent on the site as input, and stores this data in a database. The output is the collected access log data.

[0721] Step 2:

[0722] The server learns user behavior patterns using collected access log data. It receives access log data as input and applies machine learning algorithms such as clustering and classification using scikit-learn. The output is a model that illustrates user behavior patterns.

[0723] Step 3:

[0724] The device uses a camera and microphone to capture facial expressions and voice tone in order to recognize the user's emotions in real time. It receives user facial expression and voice data as input and sends this to a custom emotion engine (SentimentEngine). The output is the user's emotional state.

[0725] Step 4:

[0726] The server combines the emotion data recognized by the emotion engine with the behavioral pattern model to generate the next action. It receives user emotion data and a behavioral pattern model as input and uses a custom ad recommendation engine (AdRecommendationEngine) to generate the most suitable ad. The output is a personalized ad.

[0727] Step 5:

[0728] The device displays the generated personalized ads to the user. It receives ad data sent from the server as input and displays it on the user's screen. As output, it displays ads optimized for the user.

[0729] Step 6:

[0730] When a user takes action on an advertisement they see (click, purchase, etc.), user behavior data is collected as input and sent back to the server. New access log data is obtained as output.

[0731] Step 7:

[0732] The server updates its behavioral pattern model using newly collected access log data. It receives the new access log data as input and performs additional training on the existing model. The updated behavioral pattern model is obtained as output.

[0733] In this way, it becomes possible to provide personalized ads based on user behavior patterns and emotions, and to dynamically change the user experience on the site.

[0734] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0735] Data generation model 58 is a form of so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0736] Other examples of generative AI include Gemini® (registered trademark) (Internet search). <url: https: gemini.google.com ?hl="ja">) are some examples.

[0737] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0738] [Second Embodiment]

[0739] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0740] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0741] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0742] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0743] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0744] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0745] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0746] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0747] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0748] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0749] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0750] Next, the identification process performed by the identification processing unit 290 of the data processing device 12 will be described.

[0751] "Example of form 1"

[0752] The system of this invention constructs an AI (UX generation AI) that learns from the access logs of advertiser websites. Based on the initial actions of users who visit the advertiser website, this AI generates and proposes the next actions that will lead to a conversion. Specifically, based on the initial actions of a user when they visit the site (e.g., accessing a specific product page, searching for a specific keyword, etc.), it generates actions such as which page to guide the user to next and which products to recommend.

[0753] "Example of form 2"

[0754] The generated actions are reflected on the advertiser's website, dynamically changing the site's user experience (UX). Specifically, when a user visits the site, the displayed content changes based on the generated actions. For example, if an action recommending a specific product is generated, information about that product will be prioritized when a user visits the site. This makes it possible to provide an optimal UX for each user and improve conversion rates.

[0755] "Example of form 3"

[0756] The system of this invention learns from the access logs of advertiser websites to understand user behavior patterns and generates optimal actions based on them. Specifically, it learns user behavior patterns from past access logs and generates the next action based on those patterns. For example, if the system learns that a user who visits a particular product page is highly likely to purchase another specific product, it will generate the next action based on that pattern.

[0757] The following describes the processing flow for each example of the form.

[0758] "Example of form 1"

[0759] Step 1: Collect access logs from the advertiser's website. This includes initial user actions when they visit the site (e.g., accessing a specific product page, searching for specific keywords, etc.).

[0760] Step 2: The collected access logs are used to train the AI ​​(UX generation AI). The AI ​​learns user behavior patterns and generates the next action based on them.

[0761] Step 3: Reflect the generated actions on the advertiser's site. Specifically, when a user visits the site, the displayed content will change based on the generated actions.

[0762] "Example of form 2"

[0763] Step 1: Collect information on the initial actions of users when they visit the site.

[0764] Step 2: Based on the collected initial actions, the AI ​​(UX generation AI) generates the next action that will lead to a conversion.

[0765] Step 3: Reflect the generated actions on the advertiser's site and dynamically change the site's UX.

[0766] "Example of form 3"

[0767] Step 1: Collect access logs from the advertiser's website to understand user behavior patterns.

[0768] Step 2: Learn user behavior patterns from collected access logs and generate the next action based on those patterns.

[0769] Step 3: Reflect the generated actions on the advertiser's site and dynamically change the site's UX.

[0770] (Example 1)

[0771] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0772] Traditional advertiser websites lacked the means to effectively utilize user behavior data to improve conversion rates. In particular, it was difficult to appropriately suggest the next action based on the user's initial behavior and to dynamically change the user experience on the site. As a result, it was difficult to make suggestions that matched the user's interests and concerns, making it difficult to improve conversion rates.

[0773] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0774] In this invention, the server includes means for collecting access logs from advertiser sites, means for storing the collected access logs in a database, means for pre-processing the stored access logs, means for training a generation AI model using the pre-processed data, means for generating and proposing the next action based on the user's initial behavior, means for sending the generated next action to the user's terminal, and means for dynamically changing the user experience of the site based on the proposal. This makes it possible to effectively utilize user behavior data and improve the conversion rate.

[0775] An "advertiser site" is a website that displays advertisements and allows users to access information about products and services.

[0776] An "access log" is a record of user behavior when accessing a website, and includes information such as pages viewed, search keywords, and date and time of visit.

[0777] A "database" is a system for storing collected access logs, enabling efficient management and retrieval of data.

[0778] "Preprocessing" refers to processes such as cleaning and normalizing data to make it easier to analyze collected data.

[0779] A "generative AI model" is an artificial intelligence model that uses machine learning algorithms to learn user behavior patterns and generate the next action.

[0780] "Initial user behavior" refers to the first action a user takes when accessing a website, and includes actions such as viewing a specific product page or searching for specific keywords.

[0781] "Next action" refers to suggestions for the user's next course of action, generated based on the user's initial actions. This includes directing the user to a specific page or recommending products.

[0782] "User experience" refers to the overall experience a user has when using a website, and includes factors such as the ease of use of the site and the quality of the information provided.

[0783] "Dynamic modification" refers to changing the content and structure of a website in real time based on user behavior data.

[0784] Modes for carrying out the invention

[0785] System Overview

[0786] The system of this invention collects access logs from advertiser websites and builds a generative AI model that learns user behavior patterns based on these logs. This AI model generates and suggests the next action based on the user's initial actions. Specifically, it generates actions such as which page to guide the user to next and which products to recommend, based on the user's initial actions when visiting the site (e.g., accessing a specific product page, searching for a specific keyword, etc.).

[0787] Hardware and software to be used

[0788] The server collects, stores, and preprocesses access logs, trains the AI ​​model, and generates subsequent actions. Specifically, it uses the following hardware and software:

[0789] Hardware: High-performance server (CPU, GPU, memory, storage)

[0790] Software: Database management systems (e.g., MySQL, PostgreSQL), machine learning libraries (e.g., TensorFlow, PyTorch)

[0791] The device records user behavior data and sends it to the server. It also receives suggestions for the next action sent from the server and displays them to the user.

[0792] Hardware: User's device (e.g., smartphone, computer)

[0793] Software: Web browsers, mobile applications

[0794] Specific operation of the system

[0795] 1. Collection of access logs

[0796] The server collects real-time data on the behavior of users who visit advertiser websites. Specifically, it records information such as which pages users viewed and which keywords they searched for.

[0797] 2. Data Storage

[0798] The server stores the collected access logs in a database. The data stored includes user ID, visit date and time, viewed pages, search keywords, and more.

[0799] 3. Data preprocessing

[0800] The server preprocesses the stored data. Specifically, it performs data cleaning (imputing missing values ​​and removing outliers), data normalization (scaling), and other similar operations.

[0801] 4. Training the AI ​​model

[0802] The server uses pre-processed data to train a generative AI model. Machine learning libraries such as TensorFlow and PyTorch are used for training.

[0803] 5. Generate the next action

[0804] The server uses a pre-trained AI model to generate the next action based on the user's initial behavior. Specifically, it predicts which page the user should be directed to next and which products should be recommended.

[0805] 6. Submitting the proposal

[0806] The server then sends the generated next action to the user's device. This allows the user to be presented with appropriate pages and products.

[0807] Specific example

[0808] Suppose a user accesses an advertiser's website and views a page for a specific smartphone model. The device records this action and sends it to a server. The server stores this data in a database, preprocesses it, and then trains a generative AI model. The trained AI model determines that the user should next be directed to a page about smartphone accessories or related products and sends this suggestion to the device. The device displays the suggestion to the user, and the user views the suggested page.

[0809] Example of a prompt

[0810] "Please build an AI model that generates which page to direct users to next and which products to recommend, based on their initial user behavior data when they visit an advertiser's website and view a specific product page. Specifically, it should be a system that suggests the next action that will lead to a conversion."

[0811] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0812] Step 1:

[0813] Collection of access logs

[0814] The server collects user behavior data in real time from users who visit advertiser websites. Specifically, it records information such as which pages users viewed and which keywords they searched for. The input is user behavior data, and the output is the collected access logs.

[0815] Step 2:

[0816] Data storage

[0817] The server stores the collected access logs in a database. The data stored includes user ID, visit date and time, viewed pages, and search keywords. The input is the collected access logs, and the output is the access logs stored in the database.

[0818] Step 3:

[0819] Data preprocessing

[0820] The server preprocesses the stored data. Specifically, it performs data cleaning (imputing missing values ​​and removing outliers) and data normalization (scaling). The input is access logs stored in the database, and the output is the preprocessed data.

[0821] Step 4:

[0822] AI model training

[0823] The server uses pre-processed data to train a generative AI model. Machine learning libraries such as TensorFlow and PyTorch are used for training. The input is pre-processed data, and the output is a trained generative AI model.

[0824] Step 5:

[0825] Next Action Generation

[0826] The server uses a pre-trained AI model to generate the next action based on the user's initial behavior. Specifically, it predicts which page the user should be directed to next and which product should be recommended. The input is data on the user's initial behavior, and the output is the generated next action.

[0827] Step 6:

[0828] Submit a proposal

[0829] The server sends the generated next action to the user's device. This suggests appropriate pages and products to the user. The input is the generated next action, and the output is the suggestions sent to the user's device.

[0830] Step 7:

[0831] Receiving and displaying proposals

[0832] The terminal receives a suggestion for the next action sent from the server and displays it to the user. The input is the suggestion sent from the server, and the output is the suggestion displayed to the user.

[0833] Step 8:

[0834] User behavior

[0835] The user accepts the suggestions displayed on their device and takes the next action. For example, they might view the suggested product page or purchase the recommended product. The input is the suggestions displayed on the device, and the output is new user behavior data.

[0836] (Application Example 1)

[0837] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0838] Traditional advertiser websites struggled to appropriately suggest the next action that would lead to a conversion based on the user's initial behavior. As a result, they failed to increase user purchase intent and could not expect an improvement in conversion rates. Furthermore, they were unable to provide a site user experience optimized for each user, thus failing to improve user satisfaction.

[0839] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0840] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action that leads to a conversion based on the initial actions of a user who visits the advertiser site, means for dynamically changing the site user experience based on the suggestion, and means for analyzing the user's initial actions and suggesting in real time which product to view next and which page to proceed to. This makes it possible to appropriately suggest the next action that leads to a conversion based on the user's initial actions, and is expected to improve the conversion rate and user satisfaction.

[0841] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[0842] "Access logs" are data that records a user's behavior when they access a website, and include information such as page views and search keywords.

[0843] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to automatically perform specific tasks.

[0844] "Initial user behavior" refers to the first action a user takes when visiting a website, such as accessing a specific product page or searching for specific keywords.

[0845] "Contract completion" refers to a user purchasing goods or services on a website, signifying the completion of a transaction.

[0846] "Next action" refers to the next action a user should take on the website, suggesting actions that are highly likely to lead to a conversion.

[0847] "Website user experience" refers to the overall experience a user has when using a website, and includes factors such as ease of use and satisfaction.

[0848] "Dynamic modification" refers to changing the content and structure of a website in real time in response to user behavior and circumstances.

[0849] "Providing suggestions in real time" refers to instantly analyzing user behavior and immediately suggesting the next action based on the results.

[0850] The system for implementing this invention involves building artificial intelligence that learns from the access logs of advertiser websites and generates user actions that lead to conversions. A specific embodiment of this system is described below.

[0851] System Configuration

[0852] The system consists of the following main components:

[0853] 1. Server: Collects access logs from advertiser websites and stores them as training data.

[0854] 2. Artificial Intelligence (AI): Generates and suggests the next action based on the user's initial actions.

[0855] 3. User device: The device the user uses to access the advertiser's website (e.g., smartphone, tablet, PC).

[0856] 4. Database: Data storage for saving access logs and user behavior data.

[0857] Program processing

[0858] Data collection and learning

[0859] The server collects access logs of users who visit advertiser websites. These access logs include information such as which pages users viewed and which keywords they searched for. The collected data is stored in a database, which artificial intelligence uses as training data.

[0860] Analysis of user behavior

[0861] Artificial intelligence analyzes collected access logs and learns initial user behavior patterns. This generates a model that predicts what actions users should take to lead to conversions.

[0862] Proposed next steps

[0863] When a user accesses an advertiser's website, artificial intelligence analyzes the user's initial behavior in real time and suggests which products to view next and which pages to proceed to. These suggestions are displayed on the user's device to guide their actions.

[0864] Hardware and software to be used

[0865] Hardware: Servers, user terminals (smartphones, tablets, PCs)

[0866] Software: Python, Pandas, Scikit-learn, database management systems (e.g., MySQL)

[0867] Specific example

[0868] For example, if a user views "Product A's page" on an advertiser's website, with 5 page views and the search keyword being "special offer," the artificial intelligence will suggest that the user then view "Product B's page." This suggestion increases the user's purchase intent and contributes to improving the conversion rate.

[0869] Example of a prompt

[0870] If a user views product A's page, has 5 page views, and searches for "special offer," predict which page they will view next.

[0871] In this way, it becomes possible to appropriately suggest the next action that will lead to a sale based on the user's initial actions, and an improvement in the conversion rate and user satisfaction can be expected.

[0872] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0873] Step 1:

[0874] The server collects access logs of users who visit advertiser websites. Specifically, it records information such as which pages users viewed and which keywords they searched for. The input is user behavior data, and the output is access log data.

[0875] Step 2:

[0876] The server stores the collected access log data in a database. The input is the access log data, and the output is the log data stored in the database. Specifically, the data is stored using a database management system (e.g., MySQL).

[0877] Step 3:

[0878] The server provides access log data stored in the database to the artificial intelligence, which uses it as training data. The input is access log data obtained from the database, and the output is a trained artificial intelligence model. Specifically, the data is preprocessed using the Python Pandas library, and the model is trained using the Scikit-learn library.

[0879] Step 4:

[0880] When a user accesses an advertiser's website, the device sends initial user behavior data to the server. The input is the user's initial behavior data, and the output is the data sent to the server. Specifically, the system captures the user's behavior in real time and sends it to the server.

[0881] Step 5:

[0882] The server inputs the received initial action data into artificial intelligence and predicts the next action. The input is the user's initial action data, and the output is a suggestion for the next action. Specifically, the data is input into a trained artificial intelligence model to obtain the prediction result.

[0883] Step 6:

[0884] The server sends the predicted next action to the user's terminal. The input is the prediction result, and the output is the suggestion displayed on the user's terminal. Specifically, the server formats the prediction result and sends it to the user's terminal.

[0885] Step 7:

[0886] The user's device displays the received suggestions to the user. The input is the suggestions sent from the server, and the output is the information displayed to the user. Specifically, the suggestion content is displayed in the user interface.

[0887] In this way, it becomes possible to appropriately suggest the next action that will lead to a sale based on the user's initial actions, and an improvement in the conversion rate and user satisfaction can be expected.

[0888] (Example 2)

[0889] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0890] Traditional advertiser websites have struggled to effectively utilize user behavior data to provide an optimized user experience (UX) for individual users. As a result, they have faced challenges in achieving sufficient improvements in conversion rates. In particular, they were unable to dynamically change the site's display content based on the user's initial behavior, making it difficult to address the diverse needs of each user.

[0891] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0892] This invention includes a server that includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, and means for generating and suggesting subsequent actions that lead to conversions based on the initial actions of users who visit advertiser sites. This enables a system that includes means for collecting user behavior data and storing the generated actions in a database, and means for acquiring the generated actions when a user revisits the site and dynamically changing the displayed content. This makes it possible to provide an optimized UX for each user and improve the conversion rate.

[0893] An "advertiser site" is a website that displays advertisements and allows users to access information about products and services.

[0894] "Access logs" are data that records a user's behavior when they visit a website, and include information such as page views, clicks, and time spent on the site.

[0895] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn and reason, and is particularly used to analyze user behavior data and generate optimal actions.

[0896] "User action" refers to the specific operations or actions that users perform on a website, including actions such as clicking on a product, adding it to a cart, or making a purchase.

[0897] "User experience (UX)" refers to the overall experience and satisfaction that users feel when using a website, and is influenced by factors such as the site's ease of use, design, and the quality of information provided.

[0898] "Dynamic modification" refers to changing the content and functionality of a website in real time, in order to provide the most relevant information based on the user's actions and circumstances.

[0899] "Behavioral data" refers to data about a series of actions and behaviors performed by a user on a website, including browsing history, click patterns, and purchase history.

[0900] A "database" is a system for efficiently storing, managing, and retrieving data, and is used to store generated action and user behavior data.

[0901] "Generated actions" refer to the optimal next actions or suggestions that artificial intelligence generates by analyzing the user's behavioral data.

[0902] "Displayed content" refers to the information and content displayed to users on a website, including text, images, videos, links, etc.

[0903] This invention is a system for dynamically optimizing the user experience (UX) of advertiser websites. Specific embodiments of this system are described below.

[0904] System Overview

[0905] This system consists of three main elements: a server, a terminal, and a user. The server collects user behavior data and generates optimal actions using a generative AI model. The terminal receives the actions generated from the server when the user visits the site and dynamically changes the displayed content. The user experiences an optimized UX by visiting the site.

[0906] Hardware and software to be used

[0907] Server: Use a high-performance server machine (e.g., an AWS EC2 instance).

[0908] Database: Use a relational database such as MySQL or PostgreSQL.

[0909] Generative AI Model: We use advanced generative AI models such as OpenAI's GPT-4.

[0910] Analytics tools: Use user behavior analysis tools such as Google Analytics or Mixpanel.

[0911] Data processing and data calculation

[0912] The server collects user behavior data (e.g., browsing history, click patterns, purchase history, etc.) in real time. The collected data is stored in a database. Next, the server inputs prompt messages into a generative AI model to generate the optimal action. The generated action is stored in the database, and when the user revisits the site, the device retrieves it and dynamically changes the displayed content.

[0913] Specific example

[0914] For example, suppose a user visits an online shopping site and frequently browses the sports equipment page. In this case, the server inputs the following prompt message into the AI ​​model:

[0915] Please recommend the best product for this user.

[0916] The generative AI model analyzes the input data and generates an action to "recommend sports equipment." The server saves this action in a database, and when the user revisits the site, the device retrieves this action. Based on the retrieved action, the device displays a banner for sports equipment on the homepage.

[0917] In this way, we can provide a user-optimized UX for each individual user and improve the conversion rate.

[0918] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0919] Step 1:

[0920] A user visits the site.

[0921] Input: The user opens a browser and enters the URL of the advertiser's website.

[0922] Output: A user accesses the site, and initial behavioral data is generated.

[0923] Specific action: The user accesses an online shopping site, and the homepage is displayed.

[0924] Step 2:

[0925] The server collects user behavior data.

[0926] Input: User behavioral data such as browsing history, click patterns, and purchase history.

[0927] Output: The collected behavioral data is stored in the database.

[0928] Specific operation: The server uses analytics tools such as Google Analytics and Mixpanel to collect user behavior data in real time and store it in a database.

[0929] Step 3:

[0930] The server inputs prompt messages into the generated AI model, which then generates the optimal action.

[0931] Input: Collected behavioral data and prompt text (e.g., "Recommend the best product for this user").

[0932] Output: The optimal action obtained from the generative AI model.

[0933] Specific operation: Based on the collected behavioral data, the server inputs a prompt message, "Recommend the best product for this user," into a generating AI model (e.g., GPT-4), and generates the optimal action.

[0934] Step 4:

[0935] The server saves the generated actions to the database.

[0936] Input: The optimal action obtained from the generative AI model.

[0937] Output: Actions saved in the database.

[0938] Specific operation: The server saves the generated actions to a database such as MySQL or PostgreSQL.

[0939] Step 5:

[0940] The device retrieves actions generated from the server when a user visits.

[0941] Input: A request made by a user when they visit the site again.

[0942] Output: Generated actions retrieved from the server.

[0943] Specific operation: When the user visits the site again, the device sends a request to the server and retrieves the generated action.

[0944] Step 6:

[0945] The displayed content is dynamically changed based on the actions taken by the device.

[0946] Input: Generated action retrieved from the server.

[0947] Output: Dynamically changed display content.

[0948] Specific operation: The device dynamically changes the displayed content based on the action it receives, such as displaying a banner for sports equipment on the top page.

[0949] Step 7:

[0950] Users experience an optimized UX.

[0951] Input: Dynamically changed display content.

[0952] Output: Improved user satisfaction and increased conversion rates.

[0953] Specific behavior: Users simply visit the site to experience a personalized UX, with information on sports equipment being displayed preferentially.

[0954] (Application Example 2)

[0955] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[0956] Traditional advertiser websites struggled to provide an optimal user experience for each individual user, making it difficult to improve conversion rates. Furthermore, they lacked effective ways to utilize users' past purchase and browsing history, making it impossible to recommend appropriate products to users. This resulted in a failure to capture user interest, ultimately leading to lower conversion rates.

[0957] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0958] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action that leads to a conversion based on the initial actions of a user who visits the advertiser site, means for dynamically changing the user experience of the site based on the suggestion, and means for acquiring the user's past purchase and browsing history and dynamically changing the displayed content based on the generated actions. This makes it possible to provide an optimal user experience for each user and improve the conversion rate.

[0959] An "advertiser site" is a website used to display advertisements and is an online platform where users can visit to view and purchase products and services.

[0960] An "access log" is data that records a user's behavior when they visit a website, and includes information such as the date and time of the visit, the pages viewed, and the links clicked.

[0961] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn, reason, and make judgments, and in particular, it has the ability to analyze user behavior patterns and generate the next action.

[0962] "User experience" refers to the overall experience a user has when using a website or application, and includes elements such as ease of use, satisfaction, and convenience.

[0963] "Purchase history" refers to a record of products and services that a user has purchased in the past, including information such as the date and time of purchase, product name, and price.

[0964] "Browsing history" refers to a record of pages and products viewed by a user on a website, including information such as the date and time of viewing, pages viewed, and products viewed.

[0965] "Dynamically changing" refers to modifying the content and functionality of a website in real time in response to user behavior and circumstances, moving away from a fixed display to a more flexible one.

[0966] "Conversion rate" is an indicator that shows the percentage of users who actually purchased a product or service out of all users who visited a website, and it is an important indicator for measuring the effectiveness of marketing and sales.

[0967] A system for implementing this invention includes means for training an advertiser's website access logs, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action leading to a conversion based on the initial actions of a user visiting the advertiser's website, means for dynamically changing the user experience of the website based on the suggestion, and means for obtaining the user's past purchase and browsing history and dynamically changing the displayed content based on the generated actions.

[0968] System Configuration

[0969] The server will use the following hardware and software.

[0970] Hardware: Servers with high-performance processors, sufficient memory, and storage.

[0971] Software: Python, API request library (requests), Generative AI Model API

[0972] Data processing and data calculation

[0973] 1. Retrieving User Information: When a user visits an advertiser's site, the server retrieves the user's past purchase and browsing history. This is done using API requests.

[0974] 2. Sending prompts to the generating AI model: The server sends the acquired user information as prompts to the generating AI model. The generating AI model then generates the most suitable action for the user (such as recommended products).

[0975] 3. Dynamic changes to site content: The server dynamically changes the site's display content based on the generated actions. Specifically, it reflects recommended products in the site's content.

[0976] Specific example

[0977] For example, if a user has previously purchased "smartphone accessories" and recently viewed "smartwatches," the AI ​​model will generate an action recommending "smartwatch accessories."

[0978] Example of a prompt

[0979] User's past purchase history: Smartphone accessories, Browsing history: Smartwatches

[0980] By sending this prompt message to the AI ​​model, the system will recommend the most suitable products to the user. This allows for a personalized user experience and an improved conversion rate.

[0981] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0982] Step 1:

[0983] The server retrieves the user's past purchase and browsing history when the user visits the advertiser's website.

[0984] Input: User ID

[0985] Data processing: Use API requests to retrieve users' purchase and browsing history.

[0986] Output: User purchase history and browsing history data

[0987] Step 2:

[0988] The server sends the acquired user information as a prompt to the generated AI model.

[0989] Input: User's purchase history and browsing history data

[0990] Data processing: Convert user information into text-based prompt statements and send them to the API of the generated AI model.

[0991] Output: Recommended actions from the generated AI model (e.g., recommended products)

[0992] Step 3:

[0993] The server dynamically changes the site's display content based on the generated actions.

[0994] Input: Recommended actions from the generated AI model

[0995] Data processing: Retrieve the site's content data and update it to prioritize the display of recommended products.

[0996] Output: Updated site content data

[0997] Step 4:

[0998] When users visit a site, they view dynamically changed content.

[0999] Input: Updated site content data

[1000] Data processing: Display updated content in the user's browser.

[1001] Output: User-optimized display content

[1002] Step 5:

[1003] The server records new user behavior data in the access log.

[1004] Input: User behavior data (e.g., clicks, pages viewed)

[1005] Data processing: Add new behavioral data to the access log.

[1006] Output: Updated access log

[1007] (Example 3)

[1008] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1009] Traditional advertiser websites struggled to fully understand user behavior patterns and propose optimal actions. As a result, user conversion rates were low, and advertising effectiveness was not fully realized. Furthermore, it was difficult to provide personalized recommendations and address the individual needs of users.

[1010] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means.

[1011] In this invention, the server includes means for collecting access logs from advertiser sites, means for storing the collected access logs in a database, means for pre-processing the stored access logs, means for constructing a generative AI model that learns user behavior patterns using the pre-processed data, means for generating the next action based on the learned behavior patterns, means for transmitting the generated action to a terminal, means for the user to act based on the action displayed on the terminal, and means for recording that action again as an access log. This makes it possible to accurately grasp user behavior patterns and propose actions optimized for each individual user.

[1012] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[1013] "Access logs" are data that records a user's behavior when they visit a website, and include actions such as page views, clicks, and purchases.

[1014] A "database" is a system for storing collected access logs, and includes relational databases and NoSQL databases, among others.

[1015] "Preprocessing" refers to the process of preparing collected access logs into a format that is easy to analyze, and includes data cleaning and normalization.

[1016] A "generative AI model" is a model that uses machine learning algorithms to learn user behavior patterns and predict their next actions.

[1017] An "action" refers to the next step a user should take, such as purchasing a specific product or visiting a specific page.

[1018] A "device" refers to a device used by a user, and includes personal computers, smartphones, tablets, and other similar devices.

[1019] "Behavioral patterns" refer to the tendencies of a series of actions a user takes on a website, and are learned from past access logs.

[1020] A "suggestion" refers to the next action presented to the user by the generative AI model based on the behavioral patterns it has learned.

[1021] "Conversion rate" refers to the percentage of users who take the suggested action and achieve their objective, such as actually purchasing a product.

[1022] This invention is a system that collects access logs from advertiser websites, learns user behavior patterns, generates the optimal next action to take, and proposes it to the user. A specific embodiment of this system is described below.

[1023] Server Processing

[1024] The server collects access logs from advertiser websites. These access logs include information such as pages visited by users, links clicked, and products purchased. This data is collected using web server logs such as Apache or Nginx.

[1025] Next, the server stores the collected access logs in a database. This database uses a relational database such as MySQL or PostgreSQL. The stored data includes user IDs, visited pages, action timestamps, and more.

[1026] The saved access logs undergo preprocessing, such as data cleaning (imputing missing values ​​and removing outliers) and normalization (scaling the data). The Python Pandas library is often used for this preprocessing.

[1027] Using pre-processed data, the server builds a generative AI model. This generative AI model is built using machine learning frameworks such as TensorFlow or PyTorch. The model learns user behavior patterns from past access logs and predicts the next action to take.

[1028] Based on learned behavioral patterns, the server generates the optimal action the user should take next. For example, it might suggest products that a user is likely to purchase next after visiting a specific product page. This suggestion is made in real time using a generative AI model.

[1029] The generated actions are sent to the device in JSON format via the REST API.

[1030] Terminal processing

[1031] The device receives the optimal action sent from the server and presents it to the user. For example, this could be done by displaying it as a pop-up on a webpage or by sending a notification via email.

[1032] User processing

[1033] The user reviews the next action displayed on their device and purchases the product if necessary. The user's response is also recorded as an access log and used for future learning.

[1034] Specific example

[1035] For example, suppose a user visits the "Smartphone A" page on an advertiser's website. The server learns from past access logs and recognizes a pattern where users who visit "Smartphone A" are highly likely to purchase "Smartphone Case B". In this case, the server suggests that the user purchase "Smartphone Case B" as their next action.

[1036] Example of a prompt

[1037] "When a user visits page A on their smartphone, suggest the next product they are most likely to purchase."

[1038] In this way, a system is realized in which the server, terminal, and user work together to generate and present optimal actions to the user based on the user's behavior patterns. The flow of specific processing in Example 3 will be explained with reference to Figure 15.

[1039] Step 1:

[1040] The server collects access logs from advertiser websites. Specifically, it records user actions such as page views, clicks, and purchases within the site. Input includes user behavior data, and output is this data stored in log files.

[1041] Step 2:

[1042] The server stores the collected access logs in a database. Specifically, it uses a relational database such as MySQL or PostgreSQL. The input includes the collected access logs, and the output is structured data stored in the database.

[1043] Step 3:

[1044] The server preprocesses the stored access logs. Specifically, it performs data cleaning (imputing missing values ​​and removing outliers) and normalization (scaling the data). This process uses the Python Pandas library. The input includes the raw data stored in the database, and the output is preprocessed, clean data.

[1045] Step 4:

[1046] The server builds a generative AI model that learns user behavior patterns using preprocessed data. Specifically, it trains a neural network using machine learning frameworks such as TensorFlow or PyTorch. The input includes preprocessed data, and the output is a trained generative AI model.

[1047] Step 5:

[1048] The server generates the next action based on learned behavioral patterns. Specifically, it uses a generative AI model to predict the next action the user should take. The input includes a trained generative AI model and real-time user behavior data, and the output is the optimal next action to take.

[1049] Step 6:

[1050] The server sends the generated action to the terminal. Specifically, it sends the data in JSON format via a REST API. The input includes the generated action, and the output is the action data sent to the terminal.

[1051] Step 7:

[1052] The device presents the user with the optimal action received from the server. Specifically, this can be done by displaying it as a pop-up on a web page or by sending a notification via email. The input includes action data sent from the server, and the output is the action presented to the user.

[1053] Step 8:

[1054] The user reviews the next action displayed on their device and purchases the product if necessary. Specifically, they purchase the product on the website according to the suggested action. The input includes the action displayed on the device, and the output is the user's purchase behavior.

[1055] Step 9:

[1056] The server records the user's response again as an access log. Specifically, it saves the actions taken by the user as a new access log. The input includes user behavior data, and the output is an updated access log.

[1057] (Application Example 3)

[1058] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[1059] Traditional advertising delivery systems struggled to adequately understand user behavior patterns, making it difficult to display the right ads at the right time. As a result, they failed to capture user interest, and improvements in conversion rates were not expected. Furthermore, the lack of user-optimized ad display led to a decline in advertising effectiveness.

[1060] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means. In this invention, the server includes means for learning access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and proposing the next action that leads to a conversion based on the initial actions of a user who has visited the advertiser site, means for dynamically changing the site user experience based on the proposal, and means for learning the user's past website visit history and displaying advertisements that the user is likely to be interested in in real time. This makes it possible to grasp the user's behavior patterns in detail and display the appropriate advertisement at the optimal time.

[1061] An "advertiser site" is a website designed to display advertisements and is an online platform intended for user visits.

[1062] An "access log" is data that records a user's behavior when they visit a website, including information such as the pages visited and the time spent on each page.

[1063] Artificial intelligence is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to analyze user behavior patterns and generate the next action.

[1064] "User experience" refers to the experience and feelings a user has when using a website or application, and it affects the ease of use and satisfaction level of the site.

[1065] "Dynamic modification" means changing the content and layout of a website in real time according to user behavior and circumstances, providing an optimized display for each user.

[1066] "Website visit history" refers to a record of websites a user has visited in the past, including information such as pages visited, time spent on each site, and links clicked.

[1067] "Real-time display" means showing advertisements and content instantly based on the user's current actions and circumstances, providing information without delay.

[1068] The system for implementing this invention learns from the access logs of advertiser websites, analyzes user behavior patterns, and displays the most suitable advertisements in real time. A specific embodiment of this system is described below.

[1069] The server first collects access logs from advertiser websites and stores them in a database. These access logs include information such as the pages visited by users, the time spent on each page, and the links clicked. Next, the server uses these access logs to build an artificial intelligence model to learn user behavior patterns. This model is trained using libraries such as pandas or scikit-learn in Python.

[1070] When a user visits an advertiser's website, the server analyzes the user's initial behavior in real time and generates the optimal next action. This action is to display advertisements that the user is likely to be interested in. The server refers to the user's past website visit history and selects the most suitable advertisements based on information such as pages the user has visited in the past and the time spent on each page.

[1071] For example, if a user has previously visited a specific fashion brand's website and stayed on a particular product page for 120 seconds, the server will prioritize displaying new products and sales information from the same brand to that user. In this way, ads are optimized for each user, and an improvement in conversion rates can be expected.

[1072] As a concrete example, by inputting the following prompt into the AI ​​model, it is possible to generate advertisements based on user behavior patterns.

[1073] Example of a prompt:

[1074] Based on a user's past website visits, predict the next ads they should see. For example, if a user visited the website of "Fashion Brand A" and stayed on a specific product page for 120 seconds, then display new products and sales information from "Fashion Brand A" to that user.

[1075] This system allows for a detailed understanding of user behavior patterns and enables the display of appropriate advertisements at the optimal time. This is expected to improve advertising effectiveness and conversion rates.

[1076] The flow of the specific processing in Application Example 3 will be explained using Figure 16.

[1077] Step 1:

[1078] The server collects access logs from advertiser websites and stores them in a database. It receives information such as pages visited, time spent on the site, and links clicked as input, and stores this data in the database. Specifically, the web server collects access logs in real time and stores them in a database management system (e.g., MySQL or PostgreSQL).

[1079] Step 2:

[1080] The server uses collected access logs to build an artificial intelligence model that learns user behavior patterns. It uses access logs stored in a database as input and performs data processing such as feature extraction (e.g., page visit count, average time spent on page). The output is a trained artificial intelligence model. Specifically, it preprocesses the data using Python's pandas library and trains the model using tools like scikit-learn's RandomForestClassifier.

[1081] Step 3:

[1082] When a user visits an advertiser's website, the server analyzes the user's initial behavior in real time. It receives the user's current behavioral data (pages visited, time spent on each page, etc.) as input and feeds this into a previously built artificial intelligence model. The output generates the optimal next action to take (the advertisement to display). Specifically, it inputs real-time collected data into the model and obtains prediction results.

[1083] Step 4:

[1084] The server references the user's past website visit history and selects advertisements that are likely to interest the user. It uses the user's past visit history data as input and analyzes past behavioral patterns as data processing. The output is the selection of the most suitable advertisement. Specifically, it retrieves past visit history from a database and analyzes it using an artificial intelligence model.

[1085] Step 5:

[1086] The server displays selected advertisements to users in real time. It uses selected ad data as input and displays the advertisements on the user's screen as output. Specifically, it dynamically changes the content of the webpage and inserts the advertisements.

[1087] Step 6:

[1088] The server collects user responses after ad display as access logs and stores them in the database. It receives user response data to ads (clicks, time spent on site, etc.) as input and stores this data in the database. Specifically, the web server collects user responses in real time and stores them in the database management system.

[1089] This series of processes allows for a detailed understanding of user behavior patterns, enabling the display of appropriate advertisements at the optimal time. This is expected to improve advertising effectiveness and conversion rates.

[1090] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1091] "Example of form 1"

[1092] One embodiment of this system is a system that combines an emotion engine. This system combines an AI (UX generation AI) that learns from the access logs of advertiser sites and generates the next action that leads to a conversion based on the user's initial behavior, with an emotion engine that recognizes the user's emotions. Specifically, the emotion engine estimates the user's emotions from their actions and reactions when they visit the site, and the UX generation AI generates the next action based on those emotions. For example, if the emotion engine detects the user's excited state while they are browsing a product page, the UX generation AI takes that excited state into account and generates an action such as highlighting the "Buy Now" button. This provides a personalized UX that responds to the user's emotions and contributes to improving the conversion rate.

[1093] "Example of form 2"

[1094] Another embodiment of the present invention involves a system in which an emotion engine and a UX generation AI work together. In this system, the emotion engine recognizes the user's emotions in real time and feeds that information back to the UX generation AI. Based on this feedback, the UX generation AI generates the next action, dynamically changing the site's UX. For example, if the emotion engine detects the user's anxiety while the user is reading a product review, the UX generation AI generates an action to alleviate that anxiety, such as highlighting information about the "stress-free return guarantee." This provides a UX that responds to the user's emotions, thereby improving user satisfaction and conversion rates.

[1095] "Example of form 3"

[1096] Furthermore, another embodiment of the present invention is a system that combines an emotion engine and a UX generation AI. In this system, the emotion engine recognizes the user's emotions, and the UX generation AI generates the next action based on those emotions. Specifically, the emotion engine estimates the user's emotions from their actions and reactions when they visit the site, and the UX generation AI generates the next action based on those emotions. For example, if the emotion engine detects the user's excited state while they are browsing a product page, the UX generation AI takes that excited state into account and generates an action such as highlighting the "Buy Now" button. This provides a personalized UX that responds to the user's emotions and contributes to improving the conversion rate.

[1097] The following describes the processing flow for each example of the form.

[1098] "Example of form 1"

[1099] Step 1: The user visits the advertiser's website.

[1100] Step 2: The emotion engine estimates emotions from the user's behavior and reactions.

[1101] Step 3: The UX generation AI generates the next action based on the emotional information from the emotion engine.

[1102] Step 4: Dynamically change the UX based on the actions the site generates.

[1103] "Example of form 2"

[1104] Step 1: The user visits the advertiser's website.

[1105] Step 2: The emotion engine recognizes the user's emotions in real time.

[1106] Step 3: The emotion engine feeds back the recognized emotion information to the UX generation AI.

[1107] Step 4: The UX generation AI generates the next action based on the feedback, dynamically changing the site's UX.

[1108] "Example of form 3"

[1109] Step 1: The user visits the advertiser's website.

[1110] Step 2: The emotion engine estimates emotions from the user's behavior and reactions.

[1111] Step 3: The UX generation AI generates the next action based on the emotional information from the emotion engine.

[1112] Step 4: Dynamically change the UX based on the actions the site generates.

[1113] (Example 1)

[1114] Next, we will describe Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1115] Traditional advertiser websites generated subsequent actions based solely on user behavior data, making it difficult to provide personalized suggestions that considered user emotions. As a result, improvements in conversion rates were limited. This invention aims to improve conversion rates by combining user behavior data and emotional data to generate more accurate subsequent actions.

[1116] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1117] In this invention, the server includes means for collecting user behavior data, means for collecting user sentiment data, means for transmitting the collected behavior data and sentiment data to the server, means for cleansing and preprocessing the transmitted data, means for training an AI model using the preprocessed data, means for inferring the user's next action in real time using the trained AI model, and means for displaying the inferred action to the user. This enables personalized suggestions that combine the user's behavior data and sentiment data.

[1118] "User behavior data" refers to information about actions users perform on a website, such as clicks, page transitions, and search keywords.

[1119] "User emotion data" refers to information about the emotional state of a user, estimated from their facial expressions and voice.

[1120] A "server" is a computer system that collects, processes, and stores user behavioral and emotional data, and trains and runs AI models.

[1121] "Data cleansing" is the process of removing missing or outlier values ​​from collected data and formatting the data.

[1122] "Preprocessing" refers to the process of converting data into a format suitable for training an AI model.

[1123] An "AI model" is an artificial intelligence algorithm that infers the next action based on user behavior data and emotional data.

[1124] "Training" is the process of improving the prediction accuracy of an AI model by having it learn from a large amount of data.

[1125] "Real-time inference" is a process that instantly analyzes user behavior and emotional data to immediately generate the next action.

[1126] "Personalized suggestions" are suggestions optimized for the user, generated based on the user's individual behavioral and emotional data.

[1127] Modes for carrying out the invention

[1128] This invention relates to a system that learns from the access logs of advertiser websites and generates the next action based on user behavior data and sentiment data. This system collects user behavior data and sentiment data, sends it to a server for processing, and then provides personalized suggestions.

[1129] server

[1130] The server is a computer system for collecting, processing, and storing user behavioral and emotional data. Specifically, the server uses Apache Hadoop to distribute and process large amounts of access log data, and cleans and preprocesses the data. Next, TensorFlow is used to train an AI model (UX generation AI) to generate the next action that leads to a conversion based on the user's initial behavior. This AI model analyzes the user's behavior, such as accessing a specific product page or searching for a specific keyword, and determines which page to guide them to next and which product to recommend.

[1131] terminal

[1132] The device collects user behavior data in real time when a user visits an advertiser's website. Specifically, it uses JavaScript (registered trademark) to collect data such as user clicks, page transitions, and search keywords, and sends this data to the server. The device also has an emotion engine that estimates the user's emotions from their facial expressions and voice. For example, it uses a combination of OpenCV and TensorFlow to analyze the user's facial expressions and determine whether the user is excited or not.

[1133] User

[1134] Users visit advertiser websites, browse specific product pages, or search for specific keywords. User behavior and emotional data are collected in real time and sent to a server. Based on this data, the server's UX generation AI generates the next action and provides personalized suggestions to the user. For example, if excitement is detected while the user is browsing a product page, the server generates an action such as highlighting the "Buy Now" button.

[1135] Specific example

[1136] Specific Example 1

[1137] A user visits an advertiser's website and accesses a specific product page. The device collects this behavioral data and sends it to a server. The server uses UX generation AI to suggest related product pages that the user might be interested in next.

[1138] Specific Example 2

[1139] A user searches for specific keywords on an advertiser's website. The device collects this search data and sends it to the server. The server uses UX generation AI to recommend products related to the keywords the user searched for.

[1140] Specific example 3

[1141] As a user browses a product page, the emotion engine detects the user's level of excitement. Based on this information, the server's UX generation AI generates an action that highlights the "Buy Now" button.

[1142] Example of a prompt

[1143] "Build an AI model that generates subsequent actions based on the user's initial behavior, using access logs from advertiser websites. Design a system that collects user behavior and sentiment data in real time and provides personalized suggestions."

[1144] By explaining the system's processing from the perspectives of the server, terminal, and user, the role and specific operation of each component become clear.

[1145] The flow of the specific processing in Example 1 will be explained using Figure 17.

[1146] Step 1:

[1147] The device collects user behavior data when a user visits an advertiser's website. Specifically, it uses JavaScript (registered trademark) to obtain data such as user clicks, page transitions, and search keywords in real time. The input is user operation data, and the output is collected behavior data. This behavior data includes recording the search keyword and search time when a user searches for "smartphone."

[1148] Step 2:

[1149] The device also collects user emotion data. Specifically, it uses the camera and microphone built into the device to analyze the user's facial expressions and voice. By combining OpenCV and TensorFlow, it estimates emotions from the user's facial expressions and evaluates the intensity of those emotions through voice analysis. The input is the user's facial expressions and voice data, and the output is the analyzed emotion data. For example, if a user smiles while browsing a product page, the degree of that smile is recorded.

[1150] Step 3:

[1151] The device sends collected behavioral and emotional data to the server. The input is the collected behavioral and emotional data, and the output is the data sent to the server. For example, if a user searches for "smartphone" and smiles while doing so, that information is sent to the server.

[1152] Step 4:

[1153] The server cleanses and preprocesses the transmitted data. Specifically, it imputes missing values ​​and removes outliers, converting the data into a format suitable for training the AI ​​model. The input is the transmitted behavioral and sentiment data, and the output is the preprocessed data. For example, if the user's search keywords are incomplete, the server will perform a process to complete them.

[1154] Step 5:

[1155] The server uses pre-processed data to train a UX generation AI model using TensorFlow. The input is the pre-processed data, and the output is the trained AI model. This model generates subsequent actions that lead to conversions based on the user's initial actions. For example, if a user searches for "smartphone," the model learns to recommend "smartphone accessories" next.

[1156] Step 6:

[1157] The server uses a trained UX generation AI model to infer the user's next action in real time. The input is user behavior and emotion data, and the output is the inferred next action. Specifically, it takes user behavior and emotion data from when the user visits the site as input to determine which page to guide the user to next and which product to recommend. For example, if a user searches for "smartphone" and an excited state is detected, it will generate an action that highlights the "Buy Now" button.

[1158] Step 7:

[1159] The device displays the next action received from the server to the user. The input is the next action sent from the server, and the output is a personalized suggestion displayed to the user. Specifically, suggestions generated based on the user's behavioral and sentiment data are reflected on the web page. For example, a "Buy Now" button might be highlighted, or banners recommending related products might be displayed.

[1160] (Application Example 1)

[1161] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[1162] Traditional advertising systems can suggest the next action based on user behavior data, but they cannot provide personalized suggestions that take user emotions into account, limiting their ability to improve conversion rates. Furthermore, because they do not dynamically change the site user experience in response to user emotions, it is difficult to maximize user interest and engagement.

[1163] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means. In this invention, the server includes means for training access logs of the advertiser site, means for building artificial intelligence that generates user actions leading to conversions, means for generating and proposing the next action that leads to conversions based on the initial actions of users who visit the advertiser site, means for combining an emotion engine that recognizes the user's emotions, and means for dynamically changing the site user experience based on the proposal. This makes it possible to integrate user behavior data and emotion data to generate the next action and provide personalized proposals optimized for each user.

[1164] An "advertiser site" is a website operated by a company or organization that displays advertisements.

[1165] An "access log" is data that records a user's behavior when they visit a website.

[1166] "Artificial intelligence" is a technology that allows computers to learn and reason by mimicking human intelligence.

[1167] "User action" refers to the specific operations or actions that users perform on a website.

[1168] "Initial actions" refer to the first actions or processes a user takes when visiting a website.

[1169] An "emotion engine" is a technology that estimates emotions from a user's behavior and reactions.

[1170] "Website user experience" refers to the experience and satisfaction that users feel when using a website.

[1171] "Dynamic modification" refers to changing the content and display of a website in real time in response to user behavior and emotions.

[1172] "Conversion rate" refers to the percentage of users who visit a website and actually purchase a product or use a service.

[1173] "Personalization" refers to providing optimal content and suggestions based on the individual user's characteristics and behavior.

[1174] The system for implementing this invention learns from the access logs of advertiser websites and integrates user behavior data and sentiment data to generate the next action. A specific embodiment of this system is described below.

[1175] System Configuration

[1176] hardware

[1177] Server: A high-performance server used for data collection, training, and analysis.

[1178] User device: A device used by a user, such as a smartphone, tablet, or personal computer.

[1179] software

[1180] Artificial Intelligence (AI): An AI model that learns from user behavior data and generates the next action.

[1181] Emotion engine: Software used to estimate emotions from user behavior and reactions.

[1182] Database: A database used to store user behavior data and emotional data.

[1183] API: An interface for collecting user behavior data and sending it to a server.

[1184] Data processing and data calculation

[1185] 1. Collection of user behavior data:

[1186] We collect user behavior data (e.g., access to specific product pages, searches for specific keywords, etc.) when users visit advertiser websites via API.

[1187] 2. Analysis of emotions:

[1188] An emotion engine is used to estimate a user's emotions from collected behavioral data. For example, it can detect a user's level of excitement from their mouse movements and click frequency while they are browsing a product page.

[1189] 3. Generate the next action:

[1190] Artificial intelligence is used to generate the next action based on user behavior and emotional data. For example, if a user is excited, the system might generate an action such as highlighting the "Buy Now" button.

[1191] Specific example

[1192] For example, suppose a user visits an advertiser's website on their smartphone and is viewing a specific product page. If the emotion engine detects the user's excited state, the artificial intelligence will generate an action that highlights the "Buy Now" button. This action is a personalized suggestion tailored to the user's emotions and contributes to an improved conversion rate.

[1193] Example of a prompt

[1194] When a user visits an advertiser's website and is viewing a specific product page, and the emotion engine detects the user's state of excitement, what actions will the artificial intelligence generate?

[1195] In this way, the emotion-responsive advertising assistant system integrates user behavioral data and emotional data to generate the next action and provide personalized suggestions optimized for each user.

[1196] The flow of a specific process in Application Example 1 will be explained using Figure 18.

[1197] Step 1:

[1198] A user visits an advertiser's website. The user accesses the advertiser's website using a device such as a smartphone or computer. At this time, user behavior data (e.g., access to a specific product page, searching for a specific keyword) is collected from the device. The input is the user's behavior data, and the output is the collected behavior data.

[1199] Step 2:

[1200] The device sends collected behavioral data to the server. The device sends collected behavioral data to the server using an API. The input is the collected behavioral data, and the output is the behavioral data sent to the server.

[1201] Step 3:

[1202] The server saves the behavioral data to the database. The server saves the received behavioral data to the database. The input is the behavioral data sent to the server, and the output is the behavioral data saved in the database.

[1203] Step 4:

[1204] The server uses an emotion engine to estimate the user's emotions from behavioral data. The server uses the emotion engine to analyze the user's emotions from stored behavioral data. For example, it can detect the user's level of excitement from mouse movements and click frequency while they are browsing a product page. The input is behavioral data stored in the database, and the output is estimated user emotion data.

[1205] Step 5:

[1206] The server uses artificial intelligence to generate the next action. The server uses artificial intelligence to generate the next action based on behavioral and emotional data. For example, if the user is excited, it will generate an action to highlight the "Buy Now" button. The input is behavioral and emotional data, and the output is the generated next action.

[1207] Step 6:

[1208] The server sends the next generated action to the terminal. The server sends the next generated action to the terminal. The input is the next generated action, and the output is the next action sent to the terminal.

[1209] Step 7:

[1210] The device dynamically changes the site user experience based on the next action it receives. The device dynamically changes the site's display and operation based on the next action it receives. For example, it might highlight the "Buy Now" button. The input is the next action sent to the device, and the output is the dynamically changed site user experience.

[1211] (Example 2)

[1212] Next, we will describe Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1213] Traditional advertiser websites can generate the next action based on user behavior data and dynamically change the site's user experience (UX). However, because UX optimization that takes user emotions into account is not performed, there are limitations to improving user satisfaction and conversion rates. Furthermore, there is a lack of systems to provide the optimal UX for each user, making it difficult to provide an UX that meets the needs of individual users.

[1214] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1215] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action that leads to a conversion based on the initial actions of users who visit the advertiser site, means for dynamically changing the site user experience based on the suggestion, an emotion recognition engine that recognizes the user's emotions in real time, and means for generating the next action based on feedback from the emotion recognition engine. This makes it possible to optimize the UX by considering both user behavior data and emotion data, providing the optimal UX for each user and improving conversion rates and user satisfaction.

[1216] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[1217] An "access log" is a record of user behavior when they visit a website, and includes information such as page viewing history, click history, and time spent on the site.

[1218] Artificial intelligence is a technology that allows computers to learn and reason by mimicking human intelligence, and to generate the next action based on user behavior data and emotional data.

[1219] "User experience" refers to the experience and satisfaction a user feels when using a website, and is influenced by factors such as the site's design, functionality, and content quality.

[1220] An "emotion recognition engine" is a technology that recognizes emotions in real time from the user's facial expressions, voice, text input, etc., and is used to understand the user's emotional state.

[1221] "Feedback" refers to the process of returning information to a system to determine its next action based on the information it has received, and it involves sending emotional data from the emotion recognition engine to artificial intelligence.

[1222] "Dynamic modification" means changing the website's display content and functions in real time in response to user behavior and emotions, in order to provide the optimal experience for each user.

[1223] Modes for carrying out the invention

[1224] This invention is a system that dynamically changes the user experience (UX) of an advertiser's website, generating subsequent actions based on user behavior data and emotional data. Specific embodiments of this system are described below.

[1225] Hardware and software to be used

[1226] Server: The server hosts the generative AI model and emotion recognition engine. This allows for real-time recognition of user emotions and dynamic modification of the UX.

[1227] Device: The device used by the user (PC, smartphone, tablet, etc.) receives dynamic UX changes sent from the server.

[1228] Generative AI Model: The generative AI model generates the next action based on user behavior data and emotional data.

[1229] Emotion Recognition Engine: The emotion recognition engine recognizes the user's emotions in real time and feeds that information back into the generating AI model.

[1230] Program processing

[1231] 1. User Visit: A user visits the advertiser's website. For example, the user opens a browser, enters the advertiser's URL, and accesses the site.

[1232] 2. Data Collection: The server collects user behavior data (clicks, scrolls, time spent on the site, etc.). This provides information such as which pages the user is viewing and which links they are clicking.

[1233] 3. Emotion Recognition: The emotion recognition engine recognizes the user's emotions in real time. For example, when a user is reading a product review, it can detect whether they are feeling anxious based on their facial expressions and tone of voice.

[1234] 4. Feedback: The server feeds back the emotion data recognized by the emotion recognition engine to the generating AI model. For example, it sends information such as "the user is feeling anxious" to the generating AI model.

[1235] 5. Action Generation: The generation AI model generates the next action based on the feedbacked emotional and behavioral data. For example, if a user is feeling anxious, it will generate an action to highlight information about the "peace of mind return guarantee" as an action to alleviate that anxiety.

[1236] 6. UX Changes: The server dynamically changes the site's UX based on generated actions. For example, it dynamically changes HTML and CSS to prioritize the display of specific product information.

[1237] 7. Display: The device displays the new UX sent from the server. For example, information such as "Worry-Free Returns Guarantee" is highlighted on the page the user is viewing.

[1238] Specific example

[1239] Example 1: When a user visits a site and is looking for a specific product, the generative AI model prioritizes displaying information about that product.

[1240] Example 2: When the emotion recognition engine detects anxiety while a user is reading a product review, the generative AI model highlights information about the "peace of mind return guarantee."

[1241] Example of a prompt

[1242] "Please generate an action that prioritizes displaying specific products when a user visits the site."

[1243] "When a user feels uneasy while reading a product review, generate an action to alleviate that uneasy feeling."

[1244] This system makes it possible to provide an optimal user experience for each individual user, thereby improving conversion rates and user satisfaction.

[1245] The flow of the specific processing in Example 2 will be explained using Figure 19.

[1246] Step 1:

[1247] A user visits the site.

[1248] The user opens a browser, enters the advertiser's URL, and accesses the site. The input is the user's access action, and the output is the display of the site's homepage.

[1249] Step 2:

[1250] The server collects user behavior data.

[1251] The server collects user behavior data in real time, such as clicks, scrolls, and time spent on the site. The input is user behavior data, and the output is a log of the collected behavior data. Specifically, it records information such as which pages the user viewed and which links they clicked.

[1252] Step 3:

[1253] The emotion recognition engine recognizes the user's emotions in real time.

[1254] The emotion recognition engine recognizes emotions from the user's facial expressions, voice, and text input. The input is the user's facial expression and voice data, and the output is the recognized emotion data. For example, when a user is reading a product review, the engine can detect whether they are feeling anxious based on their facial expression and tone of voice.

[1255] Step 4:

[1256] The server generates emotional data and feeds it back to the AI ​​model.

[1257] The server sends the emotion data recognized by the emotion recognition engine to the generative AI model. The input is emotion data, and the output is feedback to the generative AI model. For example, it sends information that "the user is feeling anxious" to the generative AI model.

[1258] Step 5:

[1259] The generative AI model generates the next action.

[1260] The generative AI model generates the next action based on the feedback data of emotional and behavioral data. The input is emotional and behavioral data, and the output is the generated action. For example, if the user is feeling anxious, the model will generate an action to highlight information about the "stress-free return guarantee" as an action to alleviate that anxiety.

[1261] Step 6:

[1262] The server dynamically changes the site's UX based on the actions it generates.

[1263] The server dynamically modifies the site's UX based on actions generated by a generative AI model. The input is the generated action, and the output is the modified UX. For example, it dynamically modifies HTML and CSS to prioritize the display of specific product information.

[1264] Step 7:

[1265] The device displays a new UX.

[1266] The device displays the new UX sent from the server. The input is the modified UX, and the output is the new UX displayed on the user's device. For example, the page the user is viewing will have the information "Worry-Free Returns Guarantee" highlighted.

[1267] (Application Example 2)

[1268] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[1269] Traditional advertiser websites can generate the next action based on user behavior patterns and dynamically change the site's user experience, but they have not been able to provide an optimal user experience that takes user emotions into consideration. Therefore, in order to further improve user satisfaction and conversion rates, a system is needed that recognizes user emotions in real time and dynamically changes the site's displayed content based on that.

[1270] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1271] In this invention, the server includes means for learning access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, and means for generating and suggesting subsequent actions that lead to conversions based on the initial actions of users who visit advertiser sites. This makes it possible to learn user behavior patterns and emotions, generate subsequent actions based on them, and dynamically change the user experience of the site.

[1272] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[1273] An "access log" is data that records a user's behavior when they visit a website, and includes information such as the date and time of the visit, the pages viewed, and the links clicked.

[1274] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to automatically perform specific tasks.

[1275] "User experience" refers to the experience and satisfaction that users feel when using a website or application, and is influenced by factors such as ease of use, design, and functionality.

[1276] An "emotion engine" is a technology that recognizes emotions in real time from a user's facial expressions, voice, etc., and is used to analyze the user's psychological state.

[1277] "Feedback" refers to the return of information to a system to determine its next action based on the information it has received, and it is the process by which the system adjusts its operation based on the user's behavior and emotions.

[1278] "Dynamic modification" refers to a system automatically changing its display content and functions in real time according to the situation, in order to provide the optimal experience for each user.

[1279] A system for implementing this invention includes means for training an advertiser's website access logs, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action leading to a conversion based on the initial actions of a user who visits the advertiser's website, an emotion engine that recognizes the user's emotions in real time, and user experience generating artificial intelligence that generates the next action based on feedback from the emotion engine.

[1280] System program

[1281] The program for this system works as follows:

[1282] Hardware and software

[1283] Hardware: Smartphone camera

[1284] Software: OpenCV (image processing library), Keras (deep learning library), UX generation AI module

[1285] Data processing and data calculation

[1286] 1. Image Capture: The user's face is captured using the smartphone's camera. This obtains the user's facial expression data.

[1287] 2. Emotion Recognition: Face detection is performed using OpenCV, and emotions are recognized using a Keras model. Specifically, the captured face image is converted to grayscale, the face region is extracted, and input into the emotion recognition model.

[1288] 3. UX Generation: The recognized emotions are fed back to the UX generation AI to generate the next action. For example, if the user is feeling "anxious," the UX generation AI will generate an action that highlights information about the "peace of mind return guarantee."

[1289] 4. UX Update: Dynamically change the site's user experience based on generated actions. This makes it possible to provide the best possible experience for each user.

[1290] Specific example

[1291] When a user is browsing an online shopping site on their smartphone, the camera captures the user's face, and if the emotion engine detects "anxiety," the UX generation AI generates an action that highlights information about the "peace of mind return guarantee." Based on this action, the site's display content is dynamically changed to alleviate the user's anxiety.

[1292] Example of a prompt

[1293] Create a program that, when a user is browsing an online shopping site, captures their face with a camera, and if the emotion engine detects "anxiety," generates an action where the UX generation AI highlights information about the "peace of mind return guarantee."

[1294] In this way, by providing a user experience that responds to users' emotions, it is possible to improve user satisfaction and conversion rates.

[1295] The flow of a specific process in Application Example 2 will be explained using Figure 20.

[1296] Step 1:

[1297] The device uses the smartphone's camera to capture the user's face. The input is real-time video data, and the output is the captured face image. This face image is used in subsequent processing steps.

[1298] Step 2:

[1299] The device uses OpenCV to detect the facial region from a captured facial image. The input is the facial image obtained in step 1, and the output is coordinate data indicating the facial region. The facial portion is then extracted based on this coordinate data.

[1300] Step 3:

[1301] The device converts the extracted facial image to grayscale and performs preprocessing for input into the Keras model. The input is the facial region data obtained in step 2, and the output is the preprocessed facial image data. This data is then input into the emotion recognition model.

[1302] Step 4:

[1303] The device inputs pre-processed facial image data into a Keras model to recognize the user's emotions. The input is the facial image data obtained in step 3, and the output is the recognized emotion label (e.g., "anxiety," "excitement," etc.). This emotion label is used in subsequent processing steps.

[1304] Step 5:

[1305] The server feeds the recognized emotion labels back to the UX generation AI to generate the next action. The input is the emotion label obtained in step 4, and the output is the generated action (e.g., highlight the "Return Guarantee" information). This action is used to dynamically change the content displayed on the site.

[1306] Step 6:

[1307] The server dynamically modifies the site's user experience based on the actions generated. The input is the action obtained in step 5, and the output is the modified site display. This makes it possible to provide the optimal experience for each user.

[1308] (Example 3)

[1309] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[1310] Traditional advertiser websites faced challenges in improving conversion rates because they couldn't adequately understand user behavior patterns and emotions. Furthermore, they lacked the ability to generate optimized actions for each user and dynamically change the site's user experience (UX). This made it difficult to provide personalized experiences tailored to users' interests and emotions.

[1311] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means.

[1312] In this invention, the server includes means for collecting access logs from advertiser sites, means for analyzing the collected access logs and learning user behavior patterns, means for generating the next action based on the learned behavior patterns, means for reflecting the generated action on the advertiser site, means for collecting user behavior and reactions in real time and estimating emotions, means for generating the next action based on the estimated emotions, and means for reflecting the generated action on the advertiser site. This makes it possible to generate optimal actions based on user behavior patterns and emotions and dynamically change the site's UX.

[1313] An "advertiser site" is a website that displays advertisements, and users can view the content of these advertisements by accessing it.

[1314] An "access log" is data that records information about when a user accesses a website, and includes information such as the IP address, date and time of access, and pages visited.

[1315] "Behavioral patterns" refer to the tendencies and patterns of a series of actions that users take on a website, including actions such as visiting or clicking on specific pages.

[1316] A "machine learning model" refers to an algorithm or mathematical model that learns patterns and rules based on data to perform predictions and classifications.

[1317] A "natural language processing model" refers to algorithms and mathematical models for understanding and generating human language, and is used for analyzing and generating text data.

[1318] An "emotion engine" is a system that estimates emotions from a user's behavior and reactions, and analyzes emotions using natural language processing models and other methods.

[1319] "UX generation AI" refers to an artificial intelligence system that generates the next action based on the user's behavior patterns and emotions in order to optimize the user experience (UX).

[1320] "Dynamic modification" refers to changing the content and display of a website in real time, including changing elements of a webpage in response to user behavior and emotions.

[1321] This invention is a system that collects access logs from advertiser websites, analyzes user behavior patterns and emotions, generates optimal actions, and dynamically changes the user experience (UX) of the site. This system is implemented using the following hardware and software.

[1322] Hardware and software to be used

[1323] 1. Server

[1324] Web server software: Use Apache, Nginx, etc., to save user access data to log files.

[1325] Database: Use a database such as MySQL or PostgreSQL to store and manage the collected access logs.

[1326] Machine learning libraries: We use programming languages ​​such as Python and R, along with machine learning libraries such as Scikit-learn and TensorFlow, to learn user behavior patterns.

[1327] Natural language processing models: We use OpenAI's GPT-4 and IBM Watson, among others, to estimate user emotions.

[1328] 2. Terminal

[1329] Web browser: Used by users to access advertiser websites.

[1330] JavaScript (registered trademark) and HTML: Used to dynamically change the content of a website.

[1331] Specific operation of the system

[1332] 1. Collection of access logs

[1333] When a user accesses an advertiser's website, the server records information such as the user's IP address, access date and time, and visited pages in a log file.

[1334] The server periodically analyzes log files and saves them to the database.

[1335] 2. Learning behavioral patterns

[1336] The system analyzes access logs collected by the server to learn user behavior patterns. For example, it learns that users who visit a specific product page A are highly likely to also visit product page B.

[1337] The server trains a machine learning model and extracts behavioral patterns.

[1338] 3. Generate the next action

[1339] Based on the behavioral patterns the server has learned, it generates the following actions. For example, it might generate an action to recommend product page B to a user who has visited product page A.

[1340] The server-generated actions are saved to the database.

[1341] 4. Reflecting the Action

[1342] The server-generated actions are reflected on the advertiser's website. Specifically, JavaScript (registered trademark) and HTML are used to dynamically change the content of the web page.

[1343] The device displays the actions generated for the user.

[1344] 5. Estimation of emotions

[1345] The server collects user behavior and reactions in real time and estimates their emotions. For example, it collects behavioral data such as page scrolling speed and click frequency.

[1346] The server inputs this data into the emotion engine to estimate the user's emotions.

[1347] 6. Generating emotion-based actions

[1348] Based on the server's emotion engine output, the UX generation AI generates the next action. For example, if the user is showing signs of excitement, it generates an action to highlight the "Buy Now" button.

[1349] The server-generated actions are reflected on the advertiser's website.

[1350] Examples of specific cases and prompt statements

[1351] Example 1: Learning behavioral patterns and generating actions

[1352] A user visits a specific product page A.

[1353] The server collects access logs and records that a user visited product page A.

[1354] The server analyzes past access logs and learns patterns such as a high probability that a user will visit product page B after visiting product page A.

[1355] Based on this pattern, the server generates an action that recommends product page B to a user who visited product page A.

[1356] Example of a prompt

[1357] When a user visits product page A, generate an action that recommends product page B based on patterns learned from past access logs.

[1358] Example 2: Integration of an emotion engine and UX generation AI

[1359] When a user is browsing a product page, they exhibit behaviors that indicate excitement (for example, rapidly scrolling through the page).

[1360] The server inputs this action into the emotion engine to estimate the user's state of excitement.

[1361] Based on the emotion engine's output, the server uses a UX generation AI to generate an action that highlights the "Buy Now" button.

[1362] Example of a prompt

[1363] When a user exhibits behavior indicating excitement while browsing a product page, generate an action that highlights the "Buy Now" button based on the emotion engine's output.

[1364] The above describes specific embodiments of the system of the present invention. The flow of the specific processing in Example 3 will be explained with reference to Figure 21.

[1365] Step 1:

[1366] Collection of access logs

[1367] The server collects access logs from advertiser websites. Specifically, it uses web server software such as Apache or Nginx to save user access data to log files.

[1368] Input: Information such as the user's IP address, access date and time, and visited pages.

[1369] Output: Access data recorded in the log file.

[1370] Specific operation: When a user accesses an advertiser's site, the server records that access information in a log file.

[1371] Step 2:

[1372] Learning behavioral patterns

[1373] The system analyzes access logs collected by the server to learn user behavior patterns. This analysis uses programming languages ​​such as Python and R, and machine learning libraries such as Scikit-learn and TensorFlow.

[1374] Input: Access data recorded in the log file.

[1375] Output: Learned behavioral pattern model.

[1376] Specific operation: The server reads past access logs, analyzes user behavior over time, and trains a machine learning model to extract behavioral patterns.

[1377] Step 3:

[1378] Next Action Generation

[1379] Based on the behavioral patterns the server has learned, it generates the following actions. For example, it might generate an action to recommend product page B to a user who has visited a specific product page A.

[1380] Input: Trained behavioral pattern model, current user behavior data.

[1381] Output: The next action generated.

[1382] Specific operation: When a user visits product page A, the server refers to the behavioral pattern model and generates an action that recommends product page B.

[1383] Step 4:

[1384] Reflecting the action

[1385] The server-generated actions are reflected on the advertiser's website. Specifically, JavaScript (registered trademark) and HTML are used to dynamically change the content of the web page.

[1386] Input: The next action generated.

[1387] Output: A dynamically modified webpage.

[1388] Specific operation: The server generates JavaScript code to add a link to product page B on product page A, and the device displays the link to product page B to the user.

[1389] Step 5:

[1390] Estimation of emotions

[1391] The server collects user behavior and reactions in real time and estimates their emotions. This is done using tools such as Google Analytics and Hotjar.

[1392] Input: User behavior data (page scroll speed, click frequency, etc.).

[1393] Output: Estimated user sentiment.

[1394] Specific operation: While a user is browsing a product page, the server collects behavioral data such as page scrolling speed and click frequency, and inputs this data into the sentiment engine to estimate the user's emotions.

[1395] Step 6:

[1396] Generating emotion-based actions

[1397] Based on the server's emotion engine output, the UX generation AI generates the next action. For example, if the user is showing signs of excitement, it generates an action to highlight the "Buy Now" button.

[1398] Input: Estimated user sentiment, current user behavior data.

[1399] Output: The next action generated.

[1400] Specific operation: The server receives the output from the emotion engine, inputs it into the UX generation AI, and generates an action that highlights the "Buy Now" button.

[1401] Step 7:

[1402] Reflecting emotion-based actions

[1403] The server-generated actions are reflected on the advertiser's website. Specifically, JavaScript (registered trademark) and HTML are used to dynamically change the content of the web page.

[1404] Input: The next action generated.

[1405] Output: A dynamically modified webpage.

[1406] Specific operation: The server generates JavaScript code that highlights the "Buy Now" button, and the device displays the highlighted "Buy Now" button to the user.

[1407] (Application Example 3)

[1408] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 as the "terminal".

[1409] Traditional advertising systems could generate subsequent actions based on user behavior patterns, but they struggled to provide a personalized user experience that took user emotions into account. As a result, they couldn't display the most suitable ads based on user interests and emotions, and thus couldn't expect to improve conversion rates. Furthermore, there was a lack of means to dynamically change the user experience on the site, making it difficult to display ads optimized for each user.

[1410] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means.

[1411] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting subsequent actions that lead to conversions based on the initial actions of users who visit advertiser sites, an emotion engine that recognizes user emotions, and user experience generation artificial intelligence that generates subsequent actions based on the emotions recognized by the emotion engine. This makes it possible to display optimal advertisements based on user behavior patterns and emotions and to dynamically change the user experience of the site.

[1412] An "advertiser site" is a website designed to display advertisements and is an online platform intended for user visits.

[1413] "Access logs" are data that records a user's behavior when they visit a website, including information such as page views and clicks.

[1414] Artificial intelligence is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to analyze user behavior patterns and generate optimal actions.

[1415] "User experience" refers to the overall experience a user has when using a website or application, and includes aspects such as the ease of use of the interface and the appeal of the content.

[1416] An "emotion engine" is a technology for recognizing and analyzing a user's emotions, estimating their emotional state from their behavior and reactions.

[1417] "User experience generating artificial intelligence" is an artificial intelligence that generates the next action based on the user's emotions recognized by the emotion engine, and is a technology for providing a personalized user experience.

[1418] "Dynamic modification" means changing the content and interface of a website or application in real time in response to user behavior and emotions, thereby providing an optimized experience for each user.

[1419] The system for implementing this invention learns from the access logs of advertiser websites and generates optimal actions based on user behavior patterns and emotions. Specific embodiments of this system are described below.

[1420] System Configuration

[1421] hardware

[1422] Server: A high-performance server for data processing and running artificial intelligence models.

[1423] Device: A smartphone or computer used by the user to access the system.

[1424] Sensors: Cameras and microphones used to recognize the user's emotions.

[1425] software

[1426] Python: A programming language used for data processing and implementing artificial intelligence models.

[1427] scikit-learn: A machine learning library for learning user behavior patterns.

[1428] Custom Sentiment Engine: Software for analyzing user emotions in real time.

[1429] Custom Ad Recommendation Engine (AdRecommendationEngine): Software for generating optimal ads based on user emotions and behavioral patterns.

[1430] Data processing and calculations

[1431] Access log collection and learning

[1432] The server collects access logs when users visit advertiser sites. These access logs include page view history, click information, and time spent on the site. The collected data is analyzed using machine learning algorithms such as clustering and classification with scikit-learn.

[1433] Recognition of emotions

[1434] The device's built-in camera and microphone capture the user's facial expressions and voice tone in real time. This data is analyzed by a custom SentimentEngine to estimate the user's emotional state.

[1435] Next Action Generation

[1436] The server combines user behavior patterns learned from access logs with sentiment data recognized by the sentiment engine to generate the next ad to display. This process utilizes a custom ad recommendation engine (AdRecommendationEngine). The generated ad is personalized according to the user's emotional state and displayed at the optimal time.

[1437] Specific example

[1438] For example, if the emotion engine detects the user's state of excitement while they are browsing the page for "Product A," it will then highlight an advertisement for "Product B." In this way, it is possible to provide optimal advertisements based on the user's behavior patterns and emotions.

[1439] Example of a prompt

[1440] When a user is browsing a product page, if the sentiment engine detects the user's level of excitement, generate the next ad to display. Recommend the most relevant ad based on the user's access logs and sentiment data.

[1441] In this way, it becomes possible to provide personalized advertising based on users' behavior patterns and emotions.

[1442] The flow of the specific processing in Application Example 3 will be explained using Figure 22.

[1443] Step 1:

[1444] The server collects access logs when users visit advertiser websites. It receives data such as the user's page view history, click information, and time spent on the site as input, and stores this data in a database. The output is the collected access log data.

[1445] Step 2:

[1446] The server learns user behavior patterns using collected access log data. It receives access log data as input and applies machine learning algorithms such as clustering and classification using scikit-learn. The output is a model that illustrates user behavior patterns.

[1447] Step 3:

[1448] The device uses a camera and microphone to capture facial expressions and voice tone in order to recognize the user's emotions in real time. It receives user facial expression and voice data as input and sends this to a custom emotion engine (SentimentEngine). The output is the user's emotional state.

[1449] Step 4:

[1450] The server combines the emotion data recognized by the emotion engine with the behavioral pattern model to generate the next action. It receives user emotion data and a behavioral pattern model as input and uses a custom ad recommendation engine (AdRecommendationEngine) to generate the most suitable ad. The output is a personalized ad.

[1451] Step 5:

[1452] The device displays the generated personalized ads to the user. It receives ad data sent from the server as input and displays it on the user's screen. As output, it displays ads optimized for the user.

[1453] Step 6:

[1454] When a user takes action on an advertisement they see (click, purchase, etc.), user behavior data is collected as input and sent back to the server. New access log data is obtained as output.

[1455] Step 7:

[1456] The server updates its behavioral pattern model using newly collected access log data. It receives the new access log data as input and performs additional training on the existing model. The updated behavioral pattern model is obtained as output.

[1457] In this way, it becomes possible to provide personalized ads based on user behavior patterns and emotions, and to dynamically change the user experience on the site.

[1458] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1459] Data generation model 58 is a form of so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1460] Other examples of generative AI include Gemini® (registered trademark) (Internet search). <url: https: gemini.google.com ?hl="ja">) are some examples.

[1461] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[1462] [Third Embodiment]

[1463] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[1464] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1465] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1466] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[1467] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1468] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1469] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1470] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1471] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1472] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1473] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1474] Next, the identification process performed by the identification processing unit 290 of the data processing device 12 will be described.

[1475] "Example of form 1"

[1476] The system of this invention constructs an AI (UX generation AI) that learns from the access logs of advertiser websites. Based on the initial actions of users who visit the advertiser website, this AI generates and proposes the next actions that will lead to a conversion. Specifically, based on the initial actions of a user when they visit the site (e.g., accessing a specific product page, searching for a specific keyword, etc.), it generates actions such as which page to guide the user to next and which products to recommend.

[1477] "Example of form 2"

[1478] The generated actions are reflected on the advertiser's website, dynamically changing the site's user experience (UX). Specifically, when a user visits the site, the displayed content changes based on the generated actions. For example, if an action recommending a specific product is generated, information about that product will be prioritized when a user visits the site. This makes it possible to provide an optimal UX for each user and improve conversion rates.

[1479] "Example of form 3"

[1480] The system of this invention learns from the access logs of advertiser websites to understand user behavior patterns and generates optimal actions based on them. Specifically, it learns user behavior patterns from past access logs and generates the next action based on those patterns. For example, if the system learns that a user who visits a particular product page is highly likely to purchase another specific product, it will generate the next action based on that pattern.

[1481] The following describes the processing flow for each example of the form.

[1482] "Example of form 1"

[1483] Step 1: Collect access logs from the advertiser's website. This includes initial user actions when they visit the site (e.g., accessing a specific product page, searching for specific keywords, etc.).

[1484] Step 2: The collected access logs are used to train the AI ​​(UX generation AI). The AI ​​learns user behavior patterns and generates the next action based on them.

[1485] Step 3: Reflect the generated actions on the advertiser's site. Specifically, when a user visits the site, the displayed content will change based on the generated actions.

[1486] "Example of form 2"

[1487] Step 1: Collect information on the initial actions of users when they visit the site.

[1488] Step 2: Based on the collected initial actions, the AI ​​(UX generation AI) generates the next action that will lead to a conversion.

[1489] Step 3: Reflect the generated actions on the advertiser's site and dynamically change the site's UX.

[1490] "Example of form 3"

[1491] Step 1: Collect access logs from the advertiser's website to understand user behavior patterns.

[1492] Step 2: Learn user behavior patterns from collected access logs and generate the next action based on those patterns.

[1493] Step 3: Reflect the generated actions on the advertiser's site and dynamically change the site's UX.

[1494] (Example 1)

[1495] Next, we will describe Embodiment 1 of Embodiment Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1496] Traditional advertiser websites lacked the means to effectively utilize user behavior data to improve conversion rates. In particular, it was difficult to appropriately suggest the next action based on the user's initial behavior and to dynamically change the user experience on the site. As a result, it was difficult to make suggestions that matched the user's interests and concerns, making it difficult to improve conversion rates.

[1497] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1498] In this invention, the server includes means for collecting access logs from advertiser sites, means for storing the collected access logs in a database, means for pre-processing the stored access logs, means for training a generation AI model using the pre-processed data, means for generating and proposing the next action based on the user's initial behavior, means for sending the generated next action to the user's terminal, and means for dynamically changing the user experience of the site based on the proposal. This makes it possible to effectively utilize user behavior data and improve the conversion rate.

[1499] An "advertiser site" is a website that displays advertisements and allows users to access information about products and services.

[1500] An "access log" is a record of user behavior when accessing a website, and includes information such as pages viewed, search keywords, and date and time of visit.

[1501] A "database" is a system for storing collected access logs, enabling efficient management and retrieval of data.

[1502] "Preprocessing" refers to processes such as cleaning and normalizing data to make it easier to analyze collected data.

[1503] A "generative AI model" is an artificial intelligence model that uses machine learning algorithms to learn user behavior patterns and generate the next action.

[1504] "Initial user behavior" refers to the first action a user takes when accessing a website, and includes actions such as viewing a specific product page or searching for specific keywords.

[1505] "Next action" refers to suggestions for the user's next course of action, generated based on the user's initial actions. This includes directing the user to a specific page or recommending products.

[1506] "User experience" refers to the overall experience a user has when using a website, and includes factors such as the ease of use of the site and the quality of the information provided.

[1507] "Dynamic modification" refers to changing the content and structure of a website in real time based on user behavior data.

[1508] Modes for carrying out the invention

[1509] System Overview

[1510] The system of this invention collects access logs from advertiser websites and builds a generative AI model that learns user behavior patterns based on these logs. This AI model generates and suggests the next action based on the user's initial actions. Specifically, it generates actions such as which page to guide the user to next and which products to recommend, based on the user's initial actions when visiting the site (e.g., accessing a specific product page, searching for a specific keyword, etc.).

[1511] Hardware and software to be used

[1512] The server collects, stores, and preprocesses access logs, trains the AI ​​model, and generates subsequent actions. Specifically, it uses the following hardware and software:

[1513] Hardware: High-performance server (CPU, GPU, memory, storage)

[1514] Software: Database management systems (e.g., MySQL, PostgreSQL), machine learning libraries (e.g., TensorFlow, PyTorch)

[1515] The device records user behavior data and sends it to the server. It also receives suggestions for the next action sent from the server and displays them to the user.

[1516] Hardware: User's device (e.g., smartphone, computer)

[1517] Software: Web browsers, mobile applications

[1518] Specific operation of the system

[1519] 1. Collection of access logs

[1520] The server collects real-time data on the behavior of users who visit advertiser websites. Specifically, it records information such as which pages users viewed and which keywords they searched for.

[1521] 2. Data Storage

[1522] The server stores the collected access logs in a database. The data stored includes user ID, visit date and time, viewed pages, search keywords, and more.

[1523] 3. Data preprocessing

[1524] The server preprocesses the stored data. Specifically, it performs data cleaning (imputing missing values ​​and removing outliers), data normalization (scaling), and other similar operations.

[1525] 4. Training the AI ​​model

[1526] The server uses pre-processed data to train a generative AI model. Machine learning libraries such as TensorFlow and PyTorch are used for training.

[1527] 5. Generate the next action

[1528] The server uses a pre-trained AI model to generate the next action based on the user's initial behavior. Specifically, it predicts which page the user should be directed to next and which products should be recommended.

[1529] 6. Submitting the proposal

[1530] The server then sends the generated next action to the user's device. This allows the user to be presented with appropriate pages and products.

[1531] Specific example

[1532] Suppose a user accesses an advertiser's website and views a page for a specific smartphone model. The device records this action and sends it to a server. The server stores this data in a database, preprocesses it, and then trains a generative AI model. The trained AI model determines that the user should next be directed to a page about smartphone accessories or related products and sends this suggestion to the device. The device displays the suggestion to the user, and the user views the suggested page.

[1533] Example of a prompt

[1534] "Please build an AI model that generates which page to direct users to next and which products to recommend, based on their initial user behavior data when they visit an advertiser's website and view a specific product page. Specifically, it should be a system that suggests the next action that will lead to a conversion."

[1535] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1536] Step 1:

[1537] Collection of access logs

[1538] The server collects user behavior data in real time from users who visit advertiser websites. Specifically, it records information such as which pages users viewed and which keywords they searched for. The input is user behavior data, and the output is the collected access logs.

[1539] Step 2:

[1540] Data storage

[1541] The server stores the collected access logs in a database. The data stored includes user ID, visit date and time, viewed pages, and search keywords. The input is the collected access logs, and the output is the access logs stored in the database.

[1542] Step 3:

[1543] Data preprocessing

[1544] The server preprocesses the stored data. Specifically, it performs data cleaning (imputing missing values ​​and removing outliers) and data normalization (scaling). The input is access logs stored in the database, and the output is the preprocessed data.

[1545] Step 4:

[1546] AI model training

[1547] The server uses pre-processed data to train a generative AI model. Machine learning libraries such as TensorFlow and PyTorch are used for training. The input is pre-processed data, and the output is a trained generative AI model.

[1548] Step 5:

[1549] Next Action Generation

[1550] The server uses a pre-trained AI model to generate the next action based on the user's initial behavior. Specifically, it predicts which page the user should be directed to next and which product should be recommended. The input is data on the user's initial behavior, and the output is the generated next action.

[1551] Step 6:

[1552] Submit a proposal

[1553] The server sends the generated next action to the user's device. This suggests appropriate pages and products to the user. The input is the generated next action, and the output is the suggestions sent to the user's device.

[1554] Step 7:

[1555] Receiving and displaying proposals

[1556] The terminal receives a suggestion for the next action sent from the server and displays it to the user. The input is the suggestion sent from the server, and the output is the suggestion displayed to the user.

[1557] Step 8:

[1558] User behavior

[1559] The user accepts the suggestions displayed on their device and takes the next action. For example, they might view the suggested product page or purchase the recommended product. The input is the suggestions displayed on the device, and the output is new user behavior data.

[1560] (Application Example 1)

[1561] Next, we will describe Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1562] Traditional advertiser websites struggled to appropriately suggest the next action that would lead to a conversion based on the user's initial behavior. As a result, they failed to increase user purchase intent and could not expect an improvement in conversion rates. Furthermore, they were unable to provide a site user experience optimized for each user, thus failing to improve user satisfaction.

[1563] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1564] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action that leads to a conversion based on the initial actions of a user who visits the advertiser site, means for dynamically changing the site user experience based on the suggestion, and means for analyzing the user's initial actions and suggesting in real time which product to view next and which page to proceed to. This makes it possible to appropriately suggest the next action that leads to a conversion based on the user's initial actions, and is expected to improve the conversion rate and user satisfaction.

[1565] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[1566] "Access logs" are data that records a user's behavior when they access a website, and include information such as page views and search keywords.

[1567] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to automatically perform specific tasks.

[1568] "Initial user behavior" refers to the first action a user takes when visiting a website, such as accessing a specific product page or searching for specific keywords.

[1569] "Contract completion" refers to a user purchasing goods or services on a website, signifying the completion of a transaction.

[1570] "Next action" refers to the next action a user should take on the website, suggesting actions that are highly likely to lead to a conversion.

[1571] "Website user experience" refers to the overall experience a user has when using a website, and includes factors such as ease of use and satisfaction.

[1572] "Dynamic modification" refers to changing the content and structure of a website in real time in response to user behavior and circumstances.

[1573] "Providing suggestions in real time" refers to instantly analyzing user behavior and immediately suggesting the next action based on the results.

[1574] The system for implementing this invention involves building artificial intelligence that learns from the access logs of advertiser websites and generates user actions that lead to conversions. A specific embodiment of this system is described below.

[1575] System Configuration

[1576] The system consists of the following main components:

[1577] 1. Server: Collects access logs from advertiser websites and stores them as training data.

[1578] 2. Artificial Intelligence (AI): Generates and suggests the next action based on the user's initial actions.

[1579] 3. User device: The device the user uses to access the advertiser's website (e.g., smartphone, tablet, PC).

[1580] 4. Database: Data storage for saving access logs and user behavior data.

[1581] Program processing

[1582] Data collection and learning

[1583] The server collects access logs of users who visit advertiser websites. These access logs include information such as which pages users viewed and which keywords they searched for. The collected data is stored in a database, which artificial intelligence uses as training data.

[1584] Analysis of user behavior

[1585] Artificial intelligence analyzes collected access logs and learns initial user behavior patterns. This generates a model that predicts what actions users should take to lead to conversions.

[1586] Proposed next steps

[1587] When a user accesses an advertiser's website, artificial intelligence analyzes the user's initial behavior in real time and suggests which products to view next and which pages to proceed to. These suggestions are displayed on the user's device to guide their actions.

[1588] Hardware and software to be used

[1589] Hardware: Servers, user terminals (smartphones, tablets, PCs)

[1590] Software: Python, Pandas, Scikit-learn, database management systems (e.g., MySQL)

[1591] Specific example

[1592] For example, if a user views "Product A's page" on an advertiser's website, with 5 page views and the search keyword being "special offer," the artificial intelligence will suggest that the user then view "Product B's page." This suggestion increases the user's purchase intent and contributes to improving the conversion rate.

[1593] Example of a prompt

[1594] If a user views product A's page, has 5 page views, and searches for "special offer," predict which page they will view next.

[1595] In this way, it becomes possible to appropriately suggest the next action that will lead to a sale based on the user's initial actions, and an improvement in the conversion rate and user satisfaction can be expected.

[1596] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1597] Step 1:

[1598] The server collects access logs of users who visit advertiser websites. Specifically, it records information such as which pages users viewed and which keywords they searched for. The input is user behavior data, and the output is access log data.

[1599] Step 2:

[1600] The server stores the collected access log data in a database. The input is the access log data, and the output is the log data stored in the database. Specifically, the data is stored using a database management system (e.g., MySQL).

[1601] Step 3:

[1602] The server provides access log data stored in the database to the artificial intelligence, which uses it as training data. The input is access log data obtained from the database, and the output is a trained artificial intelligence model. Specifically, the data is preprocessed using the Python Pandas library, and the model is trained using the Scikit-learn library.

[1603] Step 4:

[1604] When a user accesses an advertiser's website, the device sends initial user behavior data to the server. The input is the user's initial behavior data, and the output is the data sent to the server. Specifically, the system captures the user's behavior in real time and sends it to the server.

[1605] Step 5:

[1606] The server inputs the received initial action data into artificial intelligence and predicts the next action. The input is the user's initial action data, and the output is a suggestion for the next action. Specifically, the data is input into a trained artificial intelligence model to obtain the prediction result.

[1607] Step 6:

[1608] The server sends the predicted next action to the user's terminal. The input is the prediction result, and the output is the suggestion displayed on the user's terminal. Specifically, the server formats the prediction result and sends it to the user's terminal.

[1609] Step 7:

[1610] The user's device displays the received suggestions to the user. The input is the suggestions sent from the server, and the output is the information displayed to the user. Specifically, the suggestion content is displayed in the user interface.

[1611] In this way, it becomes possible to appropriately suggest the next action that will lead to a sale based on the user's initial actions, and an improvement in the conversion rate and user satisfaction can be expected.

[1612] (Example 2)

[1613] Next, we will describe Example 2 of the morphological example. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1614] Traditional advertiser websites have struggled to effectively utilize user behavior data to provide an optimized user experience (UX) for individual users. As a result, they have faced challenges in achieving sufficient improvements in conversion rates. In particular, they were unable to dynamically change the site's display content based on the user's initial behavior, making it difficult to address the diverse needs of each user.

[1615] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1616] This invention includes a server that includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, and means for generating and suggesting subsequent actions that lead to conversions based on the initial actions of users who visit advertiser sites. This enables a system that includes means for collecting user behavior data and storing the generated actions in a database, and means for acquiring the generated actions when a user revisits the site and dynamically changing the displayed content. This makes it possible to provide an optimized UX for each user and improve the conversion rate.

[1617] An "advertiser site" is a website that displays advertisements and allows users to access information about products and services.

[1618] "Access logs" are data that records a user's behavior when they visit a website, and include information such as page views, clicks, and time spent on the site.

[1619] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn and reason, and is particularly used to analyze user behavior data and generate optimal actions.

[1620] "User action" refers to the specific operations or actions that users perform on a website, including actions such as clicking on a product, adding it to a cart, or making a purchase.

[1621] "User experience (UX)" refers to the overall experience and satisfaction that users feel when using a website, and is influenced by factors such as the site's ease of use, design, and the quality of information provided.

[1622] "Dynamic modification" refers to changing the content and functionality of a website in real time, in order to provide the most relevant information based on the user's actions and circumstances.

[1623] "Behavioral data" refers to data about a series of actions and behaviors performed by a user on a website, including browsing history, click patterns, and purchase history.

[1624] A "database" is a system for efficiently storing, managing, and retrieving data, and is used to store generated action and user behavior data.

[1625] "Generated actions" refer to the optimal next actions or suggestions that artificial intelligence generates by analyzing the user's behavioral data.

[1626] "Displayed content" refers to the information and content displayed to users on a website, including text, images, videos, links, etc.

[1627] This invention is a system for dynamically optimizing the user experience (UX) of advertiser websites. Specific embodiments of this system are described below.

[1628] System Overview

[1629] This system consists of three main elements: a server, a terminal, and a user. The server collects user behavior data and generates optimal actions using a generative AI model. The terminal receives the actions generated from the server when the user visits the site and dynamically changes the displayed content. The user experiences an optimized UX by visiting the site.

[1630] Hardware and software to be used

[1631] Server: Use a high-performance server machine (e.g., an AWS EC2 instance).

[1632] Database: Use a relational database such as MySQL or PostgreSQL.

[1633] Generative AI Model: We use advanced generative AI models such as OpenAI's GPT-4.

[1634] Analytics tools: Use user behavior analysis tools such as Google Analytics or Mixpanel.

[1635] Data processing and data calculation

[1636] The server collects user behavior data (e.g., browsing history, click patterns, purchase history, etc.) in real time. The collected data is stored in a database. Next, the server inputs prompt messages into a generative AI model to generate the optimal action. The generated action is stored in the database, and when the user revisits the site, the device retrieves it and dynamically changes the displayed content.

[1637] Specific example

[1638] For example, suppose a user visits an online shopping site and frequently browses the sports equipment page. In this case, the server inputs the following prompt message into the AI ​​model:

[1639] Please recommend the best product for this user.

[1640] The generative AI model analyzes the input data and generates an action to "recommend sports equipment." The server saves this action in a database, and when the user revisits the site, the device retrieves this action. Based on the retrieved action, the device displays a banner for sports equipment on the homepage.

[1641] In this way, we can provide a user-optimized UX for each individual user and improve the conversion rate.

[1642] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1643] Step 1:

[1644] A user visits the site.

[1645] Input: The user opens a browser and enters the URL of the advertiser's website.

[1646] Output: A user accesses the site, and initial behavioral data is generated.

[1647] Specific action: The user accesses an online shopping site, and the homepage is displayed.

[1648] Step 2:

[1649] The server collects user behavior data.

[1650] Input: User behavioral data such as browsing history, click patterns, and purchase history.

[1651] Output: The collected behavioral data is stored in the database.

[1652] Specific operation: The server uses analytics tools such as Google Analytics and Mixpanel to collect user behavior data in real time and store it in a database.

[1653] Step 3:

[1654] The server inputs prompt messages into the generated AI model, which then generates the optimal action.

[1655] Input: Collected behavioral data and prompt text (e.g., "Recommend the best product for this user").

[1656] Output: The optimal action obtained from the generative AI model.

[1657] Specific operation: Based on the collected behavioral data, the server inputs a prompt message, "Recommend the best product for this user," into a generating AI model (e.g., GPT-4), and generates the optimal action.

[1658] Step 4:

[1659] The server saves the generated actions to the database.

[1660] Input: The optimal action obtained from the generative AI model.

[1661] Output: Actions saved in the database.

[1662] Specific operation: The server saves the generated actions to a database such as MySQL or PostgreSQL.

[1663] Step 5:

[1664] The device retrieves actions generated from the server when a user visits.

[1665] Input: A request made by a user when they visit the site again.

[1666] Output: Generated actions retrieved from the server.

[1667] Specific operation: When the user visits the site again, the device sends a request to the server and retrieves the generated action.

[1668] Step 6:

[1669] The displayed content is dynamically changed based on the actions taken by the device.

[1670] Input: Generated action retrieved from the server.

[1671] Output: Dynamically changed display content.

[1672] Specific operation: The device dynamically changes the displayed content based on the action it receives, such as displaying a banner for sports equipment on the top page.

[1673] Step 7:

[1674] Users experience an optimized UX.

[1675] Input: Dynamically changed display content.

[1676] Output: Improved user satisfaction and increased conversion rates.

[1677] Specific behavior: Users simply visit the site to experience a personalized UX, with information on sports equipment being displayed preferentially.

[1678] (Application Example 2)

[1679] Next, we will describe application example 2 of form example 2. In the following description, the data processing device 12 will be referred to as a "server," and the headset-type terminal 314 will be referred to as a "terminal."

[1680] Traditional advertiser websites struggled to provide an optimal user experience for each individual user, making it difficult to improve conversion rates. Furthermore, they lacked effective ways to utilize users' past purchase and browsing history, making it impossible to recommend appropriate products to users. This resulted in a failure to capture user interest, ultimately leading to lower conversion rates.

[1681] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1682] In this invention, the server includes means for training access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action that leads to a conversion based on the initial actions of a user who visits the advertiser site, means for dynamically changing the user experience of the site based on the suggestion, and means for acquiring the user's past purchase and browsing history and dynamically changing the displayed content based on the generated actions. This makes it possible to provide an optimal user experience for each user and improve the conversion rate.

[1683] An "advertiser site" is a website used to display advertisements and is an online platform where users can visit to view and purchase products and services.

[1684] An "access log" is data that records a user's behavior when they visit a website, and includes information such as the date and time of the visit, the pages viewed, and the links clicked.

[1685] "Artificial intelligence" is a technology in which computer systems imitate human intelligence to learn, reason, and make judgments, and in particular, it has the ability to analyze user behavior patterns and generate the next action.

[1686] "User experience" refers to the overall experience a user has when using a website or application, and includes elements such as ease of use, satisfaction, and convenience.

[1687] "Purchase history" refers to a record of products and services that a user has purchased in the past, including information such as the date and time of purchase, product name, and price.

[1688] "Browsing history" refers to a record of pages and products viewed by a user on a website, including information such as the date and time of viewing, pages viewed, and products viewed.

[1689] "Dynamically changing" refers to modifying the content and functionality of a website in real time in response to user behavior and circumstances, moving away from a fixed display to a more flexible one.

[1690] "Conversion rate" is an indicator that shows the percentage of users who actually purchased a product or service out of all users who visited a website, and it is an important indicator for measuring the effectiveness of marketing and sales.

[1691] A system for implementing this invention includes means for training an advertiser's website access logs, means for building artificial intelligence that generates user actions leading to conversions, means for generating and suggesting the next action leading to a conversion based on the initial actions of a user visiting the advertiser's website, means for dynamically changing the user experience of the website based on the suggestion, and means for obtaining the user's past purchase and browsing history and dynamically changing the displayed content based on the generated actions.

[1692] System Configuration

[1693] The server will use the following hardware and software.

[1694] Hardware: Servers with high-performance processors, sufficient memory, and storage.

[1695] Software: Python, API request library (requests), Generative AI Model API

[1696] Data processing and data calculation

[1697] 1. Retrieving User Information: When a user visits an advertiser's site, the server retrieves the user's past purchase and browsing history. This is done using API requests.

[1698] 2. Sending prompts to the generating AI model: The server sends the acquired user information as prompts to the generating AI model. The generating AI model then generates the most suitable action for the user (such as recommended products).

[1699] 3. Dynamic changes to site content: The server dynamically changes the site's display content based on the generated actions. Specifically, it reflects recommended products in the site's content.

[1700] Specific example

[1701] For example, if a user has previously purchased "smartphone accessories" and recently viewed "smartwatches," the AI ​​model will generate an action recommending "smartwatch accessories."

[1702] Example of a prompt

[1703] User's past purchase history: Smartphone accessories, Browsing history: Smartwatches

[1704] By sending this prompt message to the AI ​​model, the system will recommend the most suitable products to the user. This allows for a personalized user experience and an improved conversion rate.

[1705] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1706] Step 1:

[1707] The server retrieves the user's past purchase and browsing history when the user visits the advertiser's website.

[1708] Input: User ID

[1709] Data processing: Use API requests to retrieve users' purchase and browsing history.

[1710] Output: User purchase history and browsing history data

[1711] Step 2:

[1712] The server sends the acquired user information as a prompt to the generated AI model.

[1713] Input: User's purchase history and browsing history data

[1714] Data processing: Convert user information into text-based prompt statements and send them to the API of the generated AI model.

[1715] Output: Recommended actions from the generated AI model (e.g., recommended products)

[1716] Step 3:

[1717] The server dynamically changes the site's display content based on the generated actions.

[1718] Input: Recommended actions from the generated AI model

[1719] Data processing: Retrieve the site's content data and update it to prioritize the display of recommended products.

[1720] Output: Updated site content data

[1721] Step 4:

[1722] When users visit a site, they view dynamically changed content.

[1723] Input: Updated site content data

[1724] Data processing: Display updated content in the user's browser.

[1725] Output: User-optimized display content

[1726] Step 5:

[1727] The server records new user behavior data in the access log.

[1728] Input: User behavior data (e.g., clicks, pages viewed)

[1729] Data processing: Add new behavioral data to the access log.

[1730] Output: Updated access log

[1731] (Example 3)

[1732] Next, we will describe Embodiment 3 of Embodiment Example 3. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1733] Traditional advertiser websites struggled to fully understand user behavior patterns and propose optimal actions. As a result, user conversion rates were low, and advertising effectiveness was not fully realized. Furthermore, it was difficult to provide personalized recommendations and address the individual needs of users.

[1734] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 3 is realized by the following means.

[1735] In this invention, the server includes means for collecting access logs from advertiser sites, means for storing the collected access logs in a database, means for pre-processing the stored access logs, means for constructing a generative AI model that learns user behavior patterns using the pre-processed data, means for generating the next action based on the learned behavior patterns, means for transmitting the generated action to a terminal, means for the user to act based on the action displayed on the terminal, and means for recording that action again as an access log. This makes it possible to accurately grasp user behavior patterns and propose actions optimized for each individual user.

[1736] An "advertiser site" is a website that displays advertisements and allows users to visit and obtain information about products and services.

[1737] "Access logs" are data that records a user's behavior when they visit a website, and include actions such as page views, clicks, and purchases.

[1738] A "database" is a system for storing collected access logs, and includes relational databases and NoSQL databases, among others.

[1739] "Preprocessing" refers to the process of preparing collected access logs into a format that is easy to analyze, and includes data cleaning and normalization.

[1740] A "generative AI model" is a model that uses machine learning algorithms to learn user behavior patterns and predict their next actions.

[1741] An "action" refers to the next step a user should take, such as purchasing a specific product or visiting a specific page.

[1742] A "device" refers to a device used by a user, and includes personal computers, smartphones, tablets, and other similar devices.

[1743] "Behavioral patterns" refer to the tendencies of a series of actions a user takes on a website, and are learned from past access logs.

[1744] A "suggestion" refers to the next action presented to the user by the generative AI model based on the behavioral patterns it has learned.

[1745] "Conversion rate" refers to the percentage of users who take the suggested action and achieve their objective, such as actually purchasing a product.

[1746] This invention is a system that collects access logs from advertiser websites, learns user behavior patterns, generates the optimal next action to take, and proposes it to the user. A specific embodiment of this system is described below.

[1747] Server Processing

[1748] The server collects access logs from advertiser websites. These access logs include information such as pages visited by users, links clicked, and products purchased. This data is collected using web server logs such as Apache or Nginx.

[1749] Next, the server stores the collected access logs in a database. This database uses a relational database such as MySQL or PostgreSQL. The stored data includes user IDs, visited pages, action timestamps, and more.

[1750] The saved access logs undergo preprocessing, such as data cleaning (imputing missing values ​​and removing outliers) and normalization (scaling the data). The Python Pandas library is often used for this process.

[1751] Using pre-processed data, the server builds a generative AI model. This generative AI model is built using machine learning frameworks such as TensorFlow or PyTorch. The model learns user behavior patterns from past access logs and predicts the next action to take.

[1752] Based on learned behavioral patterns, the server generates the optimal action the user should take next. For example, it might suggest products that a user is likely to purchase next after visiting a specific product page. This suggestion is made in real time using a generative AI model.

[1753] The generated actions are sent to the device in JSON format via the REST API.

[1754] Terminal processing

[1755] The device receives the optimal action sent from the server and presents it to the user. For example, this could be done by displaying it as a pop-up on a webpage or by sending a notification via email.

[1756] User processing

[1757] The user reviews the next action displayed on their device and purchases the product if necessary. The user's response is also recorded as an access log and used for future learning.

[1758] Specific example

[1759] For example, suppose a user visits the "Smartphone A" page on an advertiser's website. The server learns from past access logs and recognizes a pattern where users who visit "Smartphone A" are highly likely to purchase "Smartphone Case B". In this case, the server suggests that the user purchase "Smartphone Case B" as their next action.

[1760] Example of a prompt

[1761] "When a user visits page A on their smartphone, suggest the next product they are most likely to purchase."

[1762] In this way, a system is realized in which the server, terminal, and user work together to generate and present optimal actions to the user based on the user's behavior patterns. The flow of specific processing in Example 3 will be explained with reference to Figure 15.

[1763] Step 1:

[1764] The server collects access logs from advertiser websites. Specifically, it records user actions such as page views, clicks, and purchases within the site. Input includes user behavior data, and output is this data stored in log files.

[1765] Step 2:

[1766] The server stores the collected access logs in a database. Specifically, it uses a relational database such as MySQL or PostgreSQL. The input includes the collected access logs, and the output is structured data stored in the database.

[1767] Step 3:

[1768] The server preprocesses the stored access logs. Specifically, it performs data cleaning (imputing missing values ​​and removing outliers) and normalization (scaling the data). This process uses the Python Pandas library. The input includes the raw data stored in the database, and the output is preprocessed, clean data.

[1769] Step 4:

[1770] The server builds a generative AI model that learns user behavior patterns using preprocessed data. Specifically, it trains a neural network using machine learning frameworks such as TensorFlow or PyTorch. The input includes preprocessed data, and the output is a trained generative AI model.

[1771] Step 5:

[1772] The server generates the next action based on learned behavioral patterns. Specifically, it uses a generative AI model to predict the next action the user should take. The input includes a trained generative AI model and real-time user behavior data, and the output is the optimal next action to take.

[1773] Step 6:

[1774] The server sends the generated action to the terminal. Specifically, it sends the data in JSON format via a REST API. The input includes the generated action, and the output is the action data sent to the terminal.

[1775] Step 7:

[1776] The device presents the user with the optimal action received from the server. Specifically, this can be done by displaying it as a pop-up on a web page or by sending a notification via email. The input includes action data sent from the server, and the output is the action presented to the user.

[1777] Step 8:

[1778] The user reviews the next action displayed on their device and purchases the product if necessary. Specifically, they purchase the product on the website according to the suggested action. The input includes the action displayed on the device, and the output is the user's purchase behavior.

[1779] Step 9:

[1780] The server records the user's response again as an access log. Specifically, it saves the actions taken by the user as a new access log. The input includes user behavior data, and the output is an updated access log.

[1781] (Application Example 3)

[1782] Next, we will describe application example 3 of form example 3. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1783] Traditional advertising delivery systems struggled to adequately understand user behavior patterns, making it difficult to display the right ads at the right time. As a result, they failed to capture user interest, and improvements in conversion rates were not expected. Furthermore, the lack of user-optimized ad display led to a decline in advertising effectiveness.

[1784] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means. In this invention, the server includes means for learning access logs from advertiser sites, means for building artificial intelligence that generates user actions leading to conversions, means for generating and proposing the next action that leads to a conversion based on the initial actions of a user who has visited the advertiser site, means for dynamically changing the site user experience based on the proposal, and means for learning the user's past website visit history and displaying advertisements that the user is likely to be interested in in real time. This makes it possible to grasp the user's behavior patterns in detail and display the appropriate advertisement at the optimal time.

[1785] An "advertiser site" is a website designed to display advertisements and is an online platform intended for user visits.

[1786] An "access log" is data that records a user's behavior when they visit a website, including information such as the pages visited and the time spent on each page.

[1787] Artificial intelligence is a technology in which computer systems imitate human intelligence to learn and reason, and have the ability to analyze user behavior patterns and generate the next action.

[1788] "User experience" refers to the experience and feelings a user has when using a website or application, and it affects the ease of use and satisfaction level of the site.

[1789] "Dynamic modification" means changing the content and layout of a website in real time according to user behavior and circumstances, providing an optimized display for each user.

[1790] "Website visit history" refers to a record of websites a user has visited in the past, including information such as pages visited, time spent on each site, and links clicked.

[1791] "Real-time display" means showing advertisements and content instantly based on the user's current actions and circumstances, providing information without delay.

[1792] The system for implementing this invention learns from the access logs of advertiser websites, analyzes user behavior patterns, and displays the most suitable advertisements in real time. A specific embodiment of this system is described below.

[1793] The server first collects access logs from advertiser websites and stores them in a database. These access logs include information such as the pages visited by users, the time spent on each page, and the links clicked. Next, the server uses these access logs to build an artificial intelligence model to learn user behavior patterns. This model is trained using libraries such as pandas or scikit-learn in Python.

[1794] When a user visits an advertiser's website, the server analyzes the user's initial behavior in real time and generates the optimal next action. This action is to display advertisements that the user is likely to be interested in. The server refers to the user's past website visit history and selects the most suitable advertisements based on information such as pages the user has visited in the past and the time spent on each page.

[1795] For example, if a user has previously visited a specific fashion brand's website and stayed on a particular product page for 120 seconds, the server will prioritize displaying new products and sales information from the same brand to that user. In this way, ads are optimized for each user, and an improvement in conversion rates can be expected.

[1796] As a concrete example, by inputting the following prompt into the AI ​​model, it is possible to generate advertisements based on user behavior patterns.

[1797] Example of a prompt:

[1798] Based on a user's past website visits, predict the next ads they should see. For example, if a user visited the website of "Fashion Brand A" and stayed on a specific product page for 120 seconds, then display new products and sales information from "Fashion Brand A" to that user.

[1799] This system allows for a detailed understanding of user behavior patterns and enables the display of appropriate advertisements at the optimal time. This is expected to improve advertising effectiveness and conversion rates.

[1800] The flow of the specific processing in Application Example 3 will be explained using Figure 16.

[1801] Step 1:

[1802] The server collects access logs from advertiser websites and stores them in a database. It receives information such as pages visited, time spent on the site, and links clicked as input, and stores this data in the database. Specifically, the web server collects access logs in real time and stores them in a database management system (e.g., MySQL or PostgreSQL).

[1803] Step 2:

[1804] The server uses collected access logs to build an artificial intelligence model that learns user behavior patterns. It uses access logs stored in a database as input and performs data processing such as feature extraction (e.g., page visit count, average time spent on page). The output is a trained artificial intelligence model. Specifically, it preprocesses the data using Python's pandas library and trains the model using tools like scikit-learn's RandomForestClassifier.

[1805] Step 3:

[1806] When a user visits an advertiser's website, the server analyzes the user's initial behavior in real time. It receives the user's current behavioral data (pages visited, time spent on each page, etc.) as input and feeds this into a previously built artificial intelligence model. The output generates the optimal next action to take (the advertisement to display). Specifically, it inputs real-time collected data into the model and obtains prediction results.

[1807] Step 4:

[1808] The server references the user's past website visit history and selects advertisements that are likely to interest the user. It uses the user's past visit history data as input and analyzes past behavioral patterns as data processing. The output is the selection of the most suitable advertisement. Specifically, it retrieves past visit history from a database and analyzes it using an artificial intelligence model.

[1809] Step 5:

[1810] The server displays selected advertisements to users in real time. It uses selected ad data as input and displays the advertisements on the user's screen as output. Specifically, it dynamically changes the content of the webpage and inserts the advertisements.

[1811] Step 6: ...

Claims

1. Methods for collecting access logs from advertiser websites, A means for storing the collected access logs in a database, Means for performing preprocessing to clean and normalize the stored access log data, A means for constructing a generative AI model that uses the pre-processed data to learn user behavior patterns and generate user actions that lead to conversions, A means for collecting the actions and reactions of the user who accessed the advertiser site in real time and for estimating the user's emotions using an emotion engine, A means for generating and proposing subsequent actions that lead to a sale, including which page to guide the user to or which product to recommend, based on the estimated emotion and the initial actions of the user who accessed the advertiser site, including accessing a specific product page or searching for a specific keyword, using the trained generative AI model, even if the initial actions are the same, the actions will differ depending on the estimated emotion. The system includes means for dynamically changing the user experience of the advertiser site based on the proposed subsequent actions, The generation AI model learns user behavior patterns to generate user actions that lead to conversions, based on access logs stored in the database that include the actions of users who have made a purchase. system.

2. The user experience is optimized for each user. The system according to claim 1.

Images