system

Abstract

Provide a system. A method for collecting user behavior data; means for analyzing the collected behavioral data to identify user behavior patterns; means for generating a document optimized for each user based on the identified behavioral patterns; means for transmitting the generated document to a user; a feedback means for collecting user response data and improving the accuracy of said analysis means; A system including:

Description

[Technical Field]

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

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, 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 documents] [Patent documents]

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

[0004] Conventional notifications and advertisements from official LINE accounts send uniform messages to many users, which do not necessarily match the interests of individual users. As a result, there is a problem that message open rates and click rates tend to be low. Furthermore, there is a lack of a mechanism for optimizing messages based on individual user responses, making it difficult to maximize marketing effectiveness. The purpose of this invention is to solve these problems and provide a system that generates and sends messages optimized for individual users. [Means for solving the problem]

[0005] The present invention is a system including a means for collecting user behavioral data, a means for analyzing the collected behavioral data and identifying the user's behavioral patterns, a means for generating a document optimized for each user based on the identified behavioral patterns, a means for sending the generated document to the user, and a feedback means for collecting user reaction data and improving the accuracy of the analysis means.

[0006] Specifically, user behavior data, including message open rates, link click rates, and past communication content, is collected, and behavioral patterns are analyzed using a machine learning algorithm. Based on this, individually optimized messages are generated and sent to users at the appropriate time. User reaction data is collected again as feedback, and the machine learning model is retrained to improve the accuracy of the next message optimization. This method makes it possible to provide messages that more effectively match users' interests and concerns.

[0007] "Behavioral Data" refers to a record of a series of actions taken by a user on an official LINE account, such as the user's message open rate, link click rate, and the content of past communications.

[0008] "Means of collection" refers to the functions and processes that link with the official LINE account to obtain user behavioral data and store it in a database.

[0009] "Means for analysis" refers to a set of algorithms or systems for processing collected behavioral data and identifying user behavior patterns. Specifically, this includes the use of machine learning algorithms.

[0010] A "behavioral pattern" is a collection of data that indicates the tendencies and characteristics of behavior that a user takes under specific conditions.

[0011] "Optimized documents" refers to messages and notifications that are customized based on a user's behavioral patterns, making the messages more interesting and likely to elicit a response.

[0012] "Generation means" refers to software or a system that automatically creates documents optimized for each user based on the results of analyzing behavioral patterns.

[0013] "Means of sending" refers to the function for delivering the generated optimized message to users via the LINE API, etc.

[0014] "Response data" is a record of actions users take in response to messages sent to them (e.g., opening messages, clicking links, responding, etc.).

[0015] A "feedback method" is a process or system for collecting user response data and reanalyzing it to improve the accuracy of the next message generation.

[0016] "Machine learning algorithms" are statistical methods for recognizing data patterns and creating predictive models to identify user behavior patterns and generate optimized messages. [Brief explanation of the drawings]

[0017] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5]FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0018] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0019] First, the terms used in the following description will be explained.

[0020] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0021] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0022] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0023] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple 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), etc.

[0024] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0025] [First embodiment]

[0026] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0027] 1, a 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.

[0028] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0029] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0030] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the 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.

[0031] 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 of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0032] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0033] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0034] 2, in the data processing device 12, a specific process 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" according to the technology of the present 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 process 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.

[0035] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0036] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the 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 process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0037] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. 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."

[0038] The present invention relates to a system for collecting and analyzing user behavior data for the administrator of an official LINE account, and generating and sending messages optimized for each user.

[0039] 1. Data Collection

[0040] User:Activity Log

[0041] A user takes an action such as opening a message, clicking, or responding to the official LINE account.

[0042] Server: Collecting data logs

[0043] The server connects with the official LINE account and collects user behavior data in real time, specifically using technologies such as the LINE API to collect information such as message open rates, link click rates, and past communication content.

[0044] Server: Save to database

[0045] The collected behavioral data is stored in a database, including user ID, message ID, behavior type (open, click, response), timestamp, etc.

[0046] 2. Data analysis

[0047] Server: Data preprocessing

[0048] Cleanse the stored behavioral data, fill in missing values, and remove outliers.

[0049] Server:Feature Extraction

[0050] Extract features from user behavior patterns. For example, extract data such as message open rates, click rates, and the most responsive times for the past week.

[0051] Server: Applying machine learning models

[0052] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict user behavior patterns, allowing us to understand the characteristics of each user.

[0053] 3. Individual optimization

[0054] Server: Prepare message template

[0055] It provides multiple pre-made message templates, such as new product introductions, sales announcements, newsletters, etc.

[0056] Server:Customization

[0057] Customize templates based on each user's behavioral patterns, for example, inserting specific keywords or product categories for each user.

[0058] Server: Message Creation

[0059] Automatically generate personalized messages for each user based on customized templates.

[0060] 4. Sending a message

[0061] Server: Decide when to send

[0062] Based on the analysis results, the optimal sending time for each user is determined. If, based on past data, User A is judged to be most responsive at 9:00 AM, the message will be sent at that time.

[0063] Server: Send message

[0064] Using the LINE API, etc., the generated optimized message is sent to the user at the determined time.

[0065] 5. Monitoring the reaction

[0066] Users: Message Reception and Actions

[0067] Users receive the messages sent and take action such as opening, clicking, or responding.

[0068] Server: Real-time monitoring

[0069] The server monitors users' reactions to the sent messages in real time and collects reaction data.

[0070] 6. Feedback

[0071] Server: Reaction data storage

[0072] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[0073] Server: Retrain the model

[0074] Based on the new reaction data collected, the machine learning model is retrained to improve the accuracy of the next message optimization.

[0075] Specific examples

[0076] Example 1: New product introduction

[0077] Data collection

[0078] Behavioral data is collected showing that User A has shown interest in "electronic devices" in the past.

[0079] Data analysis

[0080] It is analyzed that user A is often interested in messages containing the keyword "new product announcement."

[0081] Individual optimization

[0082] Generate a message for User A saying, "Check it out now! The latest smartphones have been released and we have special discount codes for you!"

[0083] Sending a message

[0084] Send the message at 9:00 AM. Based on past data, this time is the most responsive time for User A.

[0085] Reaction monitoring

[0086] User A opens the message and clicks the link to view product details.

[0087] feedback

[0088] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[0089] In this way, the present invention maximizes the marketing effectiveness of official LINE by generating and sending optimal messages based on each user's behavioral patterns.

[0090] The processing flow will be explained below.

[0091] Step 1: Data collection

[0092] 1. User: Activity Record

[0093] A user opens a message from the official LINE account, clicks a link, or responds to a message.

[0094] 2. Server: Collecting data logs

[0095] The server collects user behavior data in real time through the LINE API, including message open rates, link click rates, and past communication content.

[0096] 3. Server: Save to database

[0097] The collected behavioral data is organized in a database, and the user ID, message ID, behavior type (open, click, response), timestamp, etc. are recorded.

[0098] Step 2: Data analysis

[0099] 1. Server: Data preprocessing

[0100] Cleanse the stored behavioral data, fill in missing values, and remove outliers.

[0101] 2. Server: Feature Extraction

[0102] Extract features from user behavior data. For example, aggregate message open rates, click rates, and the most responsive times for the past week.

[0103] 3. Server: Applying machine learning models

[0104] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict user behavior patterns.

[0105] Step 3: Individual optimization

[0106] 1. Server: Prepare message template

[0107] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[0108] 2. Server: Customization

[0109] Customize templates individually based on each user's behavioral patterns, for example by inserting specific keywords or product categories to optimize message content.

[0110] 3. Server: Message Creation

[0111] Generate personalized messages for each user based on customized templates.

[0112] Step 4: Send the message

[0113] 1. Server: Decide when to send

[0114] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[0115] 2. Server: Send message

[0116] Use the LINE API to send the generated optimized message to the user at the determined time.

[0117] Step 5: Monitoring the reaction

[0118] 1. Users: Receiving messages and taking action

[0119] Users open the messages they receive and click on links within them or reply to them.

[0120] 2. Server: Real-time monitoring

[0121] User responses to sent messages (e.g., open rates, click rates, response content) are monitored in real time and response data is collected.

[0122] Step 6: Feedback

[0123] 1. Server: Reaction data storage

[0124] The collected reaction data is stored in a database and accumulated as material for the next analysis.

[0125] 2. Server: Retrain the model

[0126] Retrain your machine learning model based on new response data to improve the accuracy of your next message optimization.

[0127] In this way, through each processing step, the system can optimize messages from the official LINE account for each individual user, allowing for more effective communication of information.

[0128] Example 1

[0129] Next, a description will be given of 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."

[0130] Conventional messaging systems have had the problem of being unable to deliver effective marketing due to the difficulty in sending messages optimized for each user. Furthermore, they lacked a feedback mechanism for effectively analyzing user behavior data and utilizing the results in the next messaging. Therefore, a system that can maximize the effectiveness of marketing is needed.

[0131] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0132] In this invention, the server includes means for collecting user behavioral data, means for cleansing the collected behavioral data, completing missing values ​​and removing outliers, means for performing feature extraction to extract features from user behavior patterns, means for applying a machine learning algorithm to classify and predict user behavioral patterns, means for preparing multiple message templates, customizing the templates based on each user's behavioral pattern, and generating an optimal message, means for determining an optimal sending time and sending the generated message to the user, means for collecting user reaction data to the sent message and saving it as material for subsequent analysis and optimization, and feedback means for retraining a machine learning model based on the collected reaction data and improving the accuracy of message optimization, thereby enabling the generation and sending of optimal messages based on each user's behavioral pattern.

[0133] "User behavior data" is data recorded when a user takes a specific action, such as opening a message, clicking a link, or responding.

[0134] "Data cleansing" is the process of imputing missing values ​​and removing outliers from collected data.

[0135] "Feature extraction" is the process of extracting features from user behavior patterns, which makes analysis and prediction easier.

[0136] A "machine learning algorithm" is an algorithm that learns patterns and regularities from data and makes future predictions and classifications.

[0137] A "message template" is a pre-defined framework for message format and content, allowing for quick message generation and customization.

[0138] The "optimal sending time" is the time at which a message is sent that is most likely to be responsive to each user, based on collected data and analysis results.

[0139] "Response data" is a record of the actions users take in response to messages they receive, and is used as material for future analysis and optimization.

[0140] "Feedback measures" are processes that retrain machine learning models based on collected response data to improve the accuracy of the next message optimization.

[0141] This invention is a system for administrators of official messaging accounts that collects and analyzes user behavior data, and generates and sends individually optimized messages. The aim is to effectively utilize user behavior data and maximize the marketing effectiveness of messages.

[0142] This system is mainly composed of three elements: a server, a terminal, and a user. Below, we will explain the flow of data processing and data calculation using the specific hardware and software of each element.

[0143] Data collection

[0144] User:Activity Log

[0145] Users take actions such as opening, clicking, or responding to messages on official messaging accounts, and these actions act as triggers for the system.

[0146] Server: Collecting data logs

[0147] The server obtains user behavior data in real time via a messaging API (e.g., the LINE API), including message open rates, link click rates, and past communication content.

[0148] Server: Save to database

[0149] The collected behavioral data is stored in a database along with information such as user ID, message ID, behavior type (open, click, response), and timestamp.

[0150] Data analysis

[0151] Server: Data preprocessing

[0152] Stored behavioral data is cleansed, missing values ​​are filled, and outliers are removed, ensuring data accuracy and reliability.

[0153] Server:Feature Extraction

[0154] Extract features from user behavior patterns, such as message open rates and click rates over the past week, and the time of day when responses are most likely to occur.

[0155] Server: Applying machine learning models

[0156] Based on the extracted features, machine learning algorithms (e.g., clustering, regression analysis, etc.) are applied to classify and predict user behavior patterns, allowing for a detailed understanding of each user's characteristics.

[0157] Individual optimization

[0158] Server: Prepare message template

[0159] The server uses a number of pre-defined message templates (e.g., new product introductions, sales announcements, newsletters, etc.).

[0160] Server:Customization

[0161] The template is customized based on each user's behavioral patterns, for example, inserting relevant keywords for a user interested in "electronic devices."

[0162] Server: Message Creation

[0163] It automatically generates personalized messages for each user based on customized templates, using a generative AI model and providing specific prompts.

[0164] Sending a message

[0165] Server: Decide when to send

[0166] Based on the analysis results, the optimal sending time for each user is determined. For example, if past data indicates that User A responds best at 9:00 AM, then that time is set.

[0167] Server: Send message

[0168] The server uses the API to send the generated optimization message to the user at the determined timing.

[0169] Reaction monitoring

[0170] Users: Message Reception and Actions

[0171] Users receive the messages sent and take action such as opening, clicking, or responding.

[0172] Server: Real-time monitoring

[0173] The server monitors users' reactions to the sent messages in real time and stores the reaction data in a database.

[0174] feedback

[0175] Server: Reaction data storage

[0176] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[0177] Server: Retrain the model

[0178] Based on the new reaction data collected, the machine learning model is retrained to improve the accuracy of message optimization.

[0179] An example of a specific prompt might be, "Generate a message for a new product launch taking into account user behavioral data. Use keywords that have generated positive responses in past messages, and include content that will increase open rates."

[0180] In this way, the present invention is a system that generates and sends optimal messages based on each user's behavioral patterns, maximizing the marketing effectiveness of official messaging accounts.

[0181] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0182] Step 1:

[0183] Data collection

[0184] User:Activity Log

[0185] Users open messages, click links, and send responses to official messaging accounts, and these actions trigger data collection.

[0186] Input: User actions (opens, clicks, responses)

[0187] Output: When an action occurs, the data is collected immediately.

[0188] Server: Collecting data logs

[0189] The server obtains user behavior data in real time via a messaging API (e.g., LINE API).

[0190] Input: Behavioral data captured by the messaging API

[0191] Output: Raw behavioral data (open rates, click rates, communication content, etc.)

[0192] Server: Save to database

[0193] The server stores the collected behavioral data in a database, which includes the user ID, message ID, behavior type, and timestamp.

[0194] Input: Raw behavioral data

[0195] Output: Behavioral data stored in a database

[0196] Step 2:

[0197] Data Preprocessing

[0198] Server: Data Cleansing

[0199] The server cleanses the stored data, imputing missing values ​​and removing outliers.

[0200] Input: Stored behavioral data

[0201] Output: Cleansed data

[0202] Specific behavior: For example, gracefully handling time zone inconsistencies and extreme values.

[0203] Step 3:

[0204] Data analysis

[0205] Server:Feature Extraction

[0206] The server extracts features from user behavior patterns, such as message open rates and click rates over the past week, and the time periods with the highest response rates.

[0207] Input: Cleansed data

[0208] Output: Extracted features

[0209] Specific operations: Analyze user behavior patterns using time series analysis and statistical methods.

[0210] Server: Applying machine learning models

[0211] The server applies a machine learning algorithm based on the extracted features to classify and predict user behavior patterns.

[0212] Input: Feature data

[0213] Output: Classified user groups and predicted behavior patterns

[0214] Specific behavior: Uses clustering and regression analysis algorithms.

[0215] Step 4:

[0216] Individual optimization

[0217] Server: Prepare message template

[0218] The server uses a number of pre-defined message templates.

[0219] Input: Template list

[0220] Output: Available templates

[0221] Specific actions: Select and update templates regularly.

[0222] Server:Customization

[0223] The server customizes the template based on each user's behavioral patterns, for example, by inserting keywords related to "electronic devices" to match the user's interests.

[0224] Input: Template, Categorized User Groups

[0225] Output: Customized template

[0226] Specific behavior: Modify specific parts of the template to suit the user's interests.

[0227] Server: Message Creation

[0228] The server automatically generates messages optimized for each user based on customized templates.

[0229] Input:Customized template

[0230] Output: Generated optimization messages

[0231] What it does: Uses a generative AI model to generate an appropriate message based on the prompt.

[0232] Step 5:

[0233] Sending a message

[0234] Server: Decide when to send

[0235] The server determines the optimal transmission time for each user from the analysis results.

[0236] Input: User's optimal transmission time data

[0237] Output: Transmission schedule

[0238] Specific behavior: Optimize sending times based on past behavioral data.

[0239] Server: Send message

[0240] The server sends the optimization message generated using the API to the user at the determined timing.

[0241] Input: Generated message, send schedule

[0242] Output: Message sent

[0243] Specific operation: The actual message sending is carried out using the API.

[0244] Step 6:

[0245] Reaction monitoring

[0246] Users: Message Reception and Actions

[0247] The user opens the received message, clicks on the link, and sends a response.

[0248] Input: Optimized message

[0249] Output: User response data (behavior)

[0250] What it does: Collect user feedback in real time.

[0251] Server: Real-time monitoring

[0252] The server monitors users' reactions to the sent messages in real time and stores the reaction data in a database.

[0253] Input: User response data

[0254] Output: Saved reaction data

[0255] Specific actions: Monitor using real-time data analysis tools.

[0256] Step 7:

[0257] feedback

[0258] Server: Reaction data storage

[0259] The server stores the user's reaction data in a database and accumulates it as material for the next analysis.

[0260] Input: User response data

[0261] Output: Saved reaction data

[0262] Specific operation: The saving process is carried out using a database management system.

[0263] Server: Retrain the model

[0264] The server retrains the machine learning model based on the new reaction data collected, improving the accuracy of the next message optimization.

[0265] Input: New reaction data

[0266] Output: A retrained machine learning model

[0267] What it does: Continually improve the model using adaptive learning techniques.

[0268] This makes it possible to generate and send optimal messages based on each user's behavioral patterns, maximizing marketing effectiveness.

[0269] (Application example 1)

[0270] Next, a description will be given of Application 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."

[0271] Conventional user behavior data collection and analysis systems have been unable to achieve effective marketing due to insufficient optimization of messages for individual users. In particular, generic messages are often sent without sufficient consideration of the different behavioral patterns of each user, limiting the effectiveness of improving user response rates and purchasing intent. Furthermore, the system lacks a mechanism for efficiently feeding back response data to messages sent and improving the accuracy of the model.

[0272] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0273] In this invention, the server includes means for collecting user behavioral data, means for analyzing the collected behavioral data and identifying user behavior patterns, means for generating a document optimized for each user based on the identified behavioral pattern, means for sending the generated document to the user, means for collecting user reaction data and providing feedback to improve the accuracy of the analysis means, means for using a machine learning algorithm to perform clustering based on the collected behavioral data, and means for selecting a message template optimized for each user cluster and generating a customized message, thereby enabling effective message optimization based on user behavior patterns.

[0274] "User behavior data" refers to records of various interactions users have with the system, including, for example, message open rates, link click rates, and the content of past communications.

[0275] "Analysis Tools" refers to the functions and algorithms used to process and analyze collected behavioral data and identify user behavior patterns.

[0276] "Behavioral patterns" refer to models or tendencies that indicate how a user will behave based on the user's past behavioral data.

[0277] "Optimized documents" refers to messages or documents that are customized to best suit a user based on the user's behavioral patterns.

[0278] "Feedback means" refers to a data collection and processing mechanism for improving the analytical accuracy of the system and the performance of the machine learning model based on user response data.

[0279] "Clustering" refers to a technique that uses machine learning algorithms to classify users into groups with similar behavioral patterns.

[0280] A "machine learning algorithm" is a part of a computer program that analyzes large amounts of data and automatically learns patterns and predictive models.

[0281] A "template" is a pre-created message template for a specific purpose, which serves as the basis for customization.

[0282] "Customization" refers to tailoring documents and messages to specific user behavior patterns.

[0283] The present invention relates to a system for generating and sending optimized messages based on user behavior data. The system is implemented in the following steps:

[0284] First, the server uses the LINE API to collect user behavior data, including message open rates, link click rates, and past communication content. The collected data is sent to the server in real time and stored in a database.

[0285] Once the data is collected, the server pre-processes it, cleansing it to fill in missing values ​​and remove outliers, then performs feature extraction to extract parameters that characterize user behavior patterns.

[0286] The server then uses machine learning algorithms, specifically clustering algorithms, to classify users into clusters with similar behavioral patterns, leveraging Python's scikit-learn library and other machine learning tools.

[0287] After clustering is complete, the server applies pre-prepared template messages to each cluster. Each template is customized to fit the behavioral patterns of a specific user. The customized message is then sent to the user in the most effective way. For example, user A might receive a message like, "Hello, would you like to check out our latest products?"

[0288] After the message is sent, user reaction data is collected again and sent to the server. This reaction data is stored in a database and used as a feedback mechanism to improve the accuracy of the analysis method. The new data is used to retrain the machine learning model, improving the accuracy of the next message optimization.

[0289] As a concrete example, suppose there is data showing that User A has clicked on messages related to beauty products in the past. In this case, the machine learning model will classify User A into a cluster interested in beauty products and use a template message for that cluster to send a customized message such as "Don't miss out on our great beauty product campaigns!"

[0290] Examples of prompts include:

[0291] Generate customized product introduction messages according to the cluster. Cluster 1: For users interested in beauty products

[0292] Don't miss out on our great beauty deals!

[0293] Cluster 2: For users interested in home appliances

[0294] The latest home appliances have arrived! Check them out now.

[0295] In this way, the present invention maximizes marketing effectiveness by utilizing user behavior data to generate and send messages optimized for individual users.

[0296] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0297] Step 1:

[0298] The server collects user behavior data via the LINE API, including message open rates, link click rates, and past communication content. The collected data is sent to the server in real time and stored in a database.

[0299] Input: User behavior data obtained via the LINE API

[0300] Output: Behavioral data stored in a database

[0301] Specific operation: The server sends a request to the LINE API and executes an SQL query to save the retrieved data to the database.

[0302] Step 2:

[0303] The server pre-processes the stored behavioral data, performing data cleansing, filling in missing values, and removing outliers.

[0304] Input: Raw data stored in a database

[0305] Output: Cleansed behavioral data

[0306] Specific operation: The server reads data from the database and uses the Pandas library to impute missing values ​​and remove outliers.

[0307] Step 3:

[0308] The server performs feature extraction from the preprocessed data, extracting features such as the open rate, click rate, and best response time for each user.

[0309] Input: Cleansed behavioral data

[0310] Output: A dataset with extracted features

[0311] What it does: The server groups the data by user ID, calculates certain statistics, and generates new features, such as how many times a message was opened or clicked each day.

[0312] Step 4:

[0313] The server performs clustering using a machine learning algorithm to classify users into clusters with similar behavioral patterns.

[0314] Input: Feature-extracted dataset

[0315] Output: User data with assigned clusters

[0316] What it does: The server uses the scikit-learn library to run the KMeans algorithm to classify users into clusters, and each user is assigned a cluster ID.

[0317] Step 5:

[0318] The server selects the best template message for each cluster and generates a customized message.

[0319] Input: User data with assigned clusters, and message template

[0320] Output: Customized message

[0321] What happens: The server selects a template for each cluster and customizes the message by inserting user-specific information.

[0322] Step 6:

[0323] The server sends the generated customized message to the user via the LINE API.

[0324] Input: Customized Message

[0325] Output: Message sent to the user's terminal

[0326] Specific operation: The server sends a POST request to the LINE API and sends the generated message to the user at the appropriate time, for example, during the time period when the user is most likely to respond.

[0327] Step 7:

[0328] The server again collects user reaction data to the sent messages and stores it in a database.

[0329] Input: User interaction data (message opens, clicks, responses, etc.)

[0330] Output: Reaction data stored in a database

[0331] Specific operation: The server retrieves the user's reaction data from the LINE API and executes an SQL query to save it in the database.

[0332] Step 8:

[0333] The server retrains the machine learning model based on the new response data to improve the accuracy of the next message optimization.

[0334] Input: New reaction data stored in the database

[0335] Output: An updated machine learning model

[0336] Specific operation: The server extracts features from new reaction data and retrains the machine learning algorithm to improve the accuracy of the model.

[0337] Through these steps, the present invention achieves effective message optimization based on user behavior data.

[0338] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0339] The present invention relates to a system for collecting and analyzing behavioral and emotional data of users for the administrator of an official LINE account, and generating and sending messages optimized for each user.

[0340] 1. Data Collection

[0341] User:Activity Log

[0342] A user takes an action such as opening a message, clicking, or responding to the official LINE account.

[0343] Server: Data logging and emotion data collection

[0344] The server connects with the official LINE account to obtain user behavioral and emotional data in real time. Specifically, it uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also uses an emotion engine to collect user emotional data through text, voice, and image analysis.

[0345] Server: Save to database

[0346] The collected behavioral and emotional data is organized in a database, and the user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc. are recorded.

[0347] 2. Data analysis

[0348] Server: Data preprocessing

[0349] Cleanse stored behavioral and emotional data to fill in missing values ​​and remove outliers.

[0350] Server:Feature Extraction

[0351] Extract features from user behavioral and emotional data. For example, aggregate message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week.

[0352] Server: Applying machine learning models

[0353] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict the user's behavioral and emotional patterns.

[0354] 3. Individual optimization

[0355] Server: Prepare message template

[0356] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[0357] Server:Customization

[0358] Templates are individually customized based on each user's behavioral and emotional patterns. For example, message content can be optimized by inserting specific keywords or product categories. The tone and content of messages can also be adjusted according to the user's emotional state.

[0359] Server: Message Creation

[0360] Automatically generate messages optimized for each user based on customized templates.

[0361] 4. Sending a message

[0362] Server: Decide when to send

[0363] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[0364] Server: Send message

[0365] Use the LINE API to send the generated optimized message to the user at the determined time.

[0366] 5. Monitoring the reaction

[0367] Users: Message Reception and Actions

[0368] Users open the messages they receive and click on links within them or reply to them.

[0369] Server: Real-time monitoring

[0370] The server monitors user responses to sent messages (e.g., open rate, click rate, response content) in real time and collects response data.

[0371] 6. Feedback

[0372] Server: Reaction data storage

[0373] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[0374] Server: Retrain the model

[0375] Retrain your machine learning model based on new reaction data collected to improve the accuracy of your next message optimization.

[0376] Specific examples

[0377] Example 1: New product introduction

[0378] Data collection

[0379] The system collects behavioral data showing that User A has shown interest in "electronic devices" in the past. The emotion engine also identifies a positive emotional state based on User A's recent messages and voice.

[0380] Data analysis

[0381] It is analyzed that User A is often interested in messages containing the keyword "new product launch." Also, since User A's emotional state is positive, it is determined that a direct sales message would be effective.

[0382] Individual optimization

[0383] Generate a positive message for User A: "Check it out now! The latest smartphones are on sale now, and we've got a special discount code for you!"

[0384] Sending a message

[0385] Send the message at 9:00 AM. Based on past data, this time is the most responsive time for User A.

[0386] Reaction monitoring

[0387] User A opens the message and clicks the link to view product details.

[0388] feedback

[0389] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[0390] In this way, the present invention maximizes the marketing effectiveness of official LINE by combining and analyzing user behavioral data and emotional data to generate and send optimal messages according to individual needs and emotional states.

[0391] The processing flow will be explained below.

[0392] Step 1: Data collection

[0393] 1. User: Activity Record

[0394] A user opens a message from the official LINE account, clicks a link, or responds to a message.

[0395] 2. Server: Collecting data logs

[0396] The server obtains user behavior data in real time through the LINE API, including message open rates, link click rates, and past communication content.

[0397] 3. Server: Emotion data collection

[0398] The emotion engine is used to collect user emotional data through text, voice, and image analysis, for example, to identify the user's emotional state from the content of their message or their tone of voice.

[0399] 4. Server: Save to database

[0400] The collected behavioral and emotional data is organized and stored in a database, including user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc.

[0401] Step 2: Data analysis

[0402] 1. Server: Data preprocessing

[0403] Cleanse stored behavioral and emotional data to fill in missing values ​​and remove outliers.

[0404] 2. Server: Feature Extraction

[0405] Extract features from user behavioral and emotional data. For example, aggregate message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week.

[0406] 3. Server: Applying machine learning models

[0407] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict the user's behavioral and emotional patterns.

[0408] Step 3: Individual optimization

[0409] 1. Server: Prepare message template

[0410] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[0411] 2. Server: Customization

[0412] Templates are individually customized based on each user's behavioral and emotional patterns. For example, message content can be optimized by inserting specific keywords or product categories. The tone and content of messages can also be adjusted according to the user's emotional state.

[0413] 3. Server: Message Creation

[0414] Automatically generate messages optimized for each user based on customized templates.

[0415] Step 4: Send the message

[0416] 1. Server: Decide when to send

[0417] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[0418] 2. Server: Send message

[0419] Use the LINE API to send the generated optimized message to the user at the determined time.

[0420] Step 5: Monitoring the reaction

[0421] 1. Users: Receiving messages and taking action

[0422] Users open the messages they receive and click on links within them or reply to them.

[0423] 2. Server: Real-time monitoring

[0424] User responses to sent messages (e.g., open rates, click rates, response content) are monitored in real time and response data is collected.

[0425] Step 6: Feedback

[0426] 1. Server: Reaction data storage

[0427] The collected reaction data is stored in a database and accumulated as material for the next analysis.

[0428] 2. Server: Retrain the model

[0429] Retrain your machine learning model based on new response data to improve the accuracy of your next message optimization.

[0430] Example 2

[0431] Next, a description will be given of Example 2. 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."

[0432] Conventional message sending systems only considered user behavioral data and did not perform individual optimization using emotional data, making it difficult to maximize the effectiveness of marketing messages. Furthermore, there was also the problem of not being able to quickly reflect user reactions because the timing of sending messages was not optimized and real-time feedback was not adequately provided.

[0433] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting user behavioral data and emotional data, means for analyzing the collected behavioral data and emotional data and identifying the user's behavioral and emotional patterns, means for generating a document optimized for each user based on the identified behavioral and emotional patterns, means for sending the generated document to the user, and means for collecting user reaction data in real time and providing feedback to improve the accuracy of the analysis means. This makes it possible to generate and send individually optimized messages based on the user's behavioral and emotional data, thereby maximizing marketing effectiveness.

[0434] "User behavioral data" refers to information related to actions taken by users on official LINE accounts, such as opening messages, clicking links, and responding to messages.

[0435] "User emotional data" is information collected using text analysis, speech analysis, image analysis, etc. to identify a user's emotional state.

[0436] A "behavioral pattern" refers to the regularity or tendency of behavior that is analyzed based on data on a series of actions taken by a user.

[0437] "Emotional patterns" refer to changes and trends in emotions that are analyzed based on the user's emotional data.

[0438] "Optimized documents" are messages that are individually tailored based on a user's behavioral and emotional patterns and are generated to maximize the user's interest and response.

[0439] A "machine learning algorithm" is a computational method for learning patterns from data and making predictions or classifications based on the results.

[0440] "Clustering" is a type of machine learning technique that divides data into several groups.

[0441] "Regression analysis" is a machine learning technique that is used to model the relationship between predictor variables and response variables and make predictions.

[0442] A "feedback means" is a method or mechanism for collecting user reaction data and using that data to improve the accuracy of the analysis means and generation means.

[0443] "Optimizing sending timing" aims to send messages at the time when users are most likely to respond, and determines the timing of sending based on past behavioral data.

[0444] This invention relates to a system for official LINE administrators that collects and analyzes user behavioral and emotional data, and generates and sends messages optimized for each user.

[0445] 1. Data Collection

[0446] User behavior record

[0447] Users open messages sent to official LINE accounts, click links, and respond - all of this behavioral data is collected in real time.

[0448] Server data logging and emotion data collection

[0449] The server uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. Additionally, it collects user emotion data through text analysis, voice analysis, and image analysis using an emotion engine (e.g., IBM Watson® or Microsoft® Azure®).

[0450] Server database storage

[0451] The server organizes the collected behavioral and emotional data into a database, recording the user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc.

[0452] 2. Data analysis

[0453] Server data preprocessing

[0454] The server cleanses the stored behavioral and emotional data, filling in missing values ​​and removing outliers, thereby ensuring data quality.

[0455] Server Feature Extraction

[0456] The server extracts features from the behavioral and emotional data, such as message open rates, click rates, the most responsive times, and patterns of emotional change over the past week.

[0457] Applying machine learning models to servers

[0458] Based on the extracted features, the server uses machine learning algorithms (e.g., K-means clustering and linear regression) to classify and predict the user's behavioral and emotional patterns.

[0459] 3. Individual optimization

[0460] Preparing the server message template

[0461] The server prepares a number of pre-defined message templates (e.g., new product introductions, sales announcements, newsletters).

[0462] Server Customization

[0463] The server customizes the templates individually for each user based on their behavioral and emotional patterns, inserting specific keywords and product categories and adjusting the tone and content of the message depending on the user's emotional state.

[0464] Server Message Generation

[0465] The server automatically generates messages optimized for each user based on customized templates.

[0466] 4. Sending a message

[0467] Server transmission timing decision

[0468] The server then uses the analysis to determine the best time to send the message. For example, if historical data indicates that a particular user is most likely to respond at 9 a.m., it will schedule the message to be sent at that time.

[0469] Server message sending

[0470] The server uses the LINE API to send the generated optimized message to the user at the determined timing.

[0471] 5. Monitoring the reaction

[0472] User message receipt and actions

[0473] The user opens the received message, clicks on a link in the message, or makes a response.

[0474] Real-time server monitoring

[0475] The server monitors user responses to sent messages (e.g., open rate, click rate, response content) in real time and collects response data.

[0476] 6. Feedback

[0477] Saving reaction data on the server

[0478] The server stores the user's reaction data in a database and accumulates it as material for the next analysis.

[0479] Retraining the model on the server

[0480] The server retrains the machine learning model based on the new reaction data collected, improving the accuracy of the next message optimization.

[0481] Specific examples

[0482] Example 1: New product introduction

[0483] Behavioral data showing that User A has shown interest in "electronic devices" in the past is collected. In addition, analysis by an emotion engine (e.g., IBM Watson) identifies a positive emotional state based on User A's recent messages and voice.

[0484] Data analysis

[0485] It is analyzed that User A is often interested in messages containing the keyword "new product launch." Also, since User A's emotional state is positive, it is determined that a direct sales message would be effective.

[0486] Individual optimization

[0487] Generate a positive message for User A: "Check it out now! The latest smartphones are on sale now, and we've got a special discount code for you!"

[0488] Sending a message

[0489] Based on past data, it is determined that User A responds best at 9:00 AM, so the message is sent at this time.

[0490] Reaction monitoring

[0491] User A opens the message and clicks the link to view product details.

[0492] feedback

[0493] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[0494] Prompt Sentence Examples

[0495] Here are some example prompts to enter into a generative AI model:

[0496] "User A has shown interest in 'electronic devices' in the past, and analysis of the emotion engine reveals that he is in a positive emotional state. Please create a message about a new product launch that will interest User A."

[0497] Using this prompt, the generative AI model generates a message optimized for User A.

[0498] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0499] Step 1: Recording user behavior

[0500] A user opens a message from an official LINE account, clicks a link, and responds. This behavioral data is sent to the LINE server (input).

[0501] Step 2: Collecting server data logs and emotion data

[0502] The server works with the official LINE app to call the LINE API and collect (output) user behavior data in real time. Specifically, behavioral data acquired includes message open rates, link click rates, and past communication content. Additionally, an emotion engine (e.g., IBM Watson) is used to collect user emotion data through text, voice, and image analysis.

[0503] Step 3: Save the database on the server

[0504] The server stores the collected behavioral and emotional data in a database (output). Specifically, it records information such as user ID, message ID, behavior type (open, click, response), emotional state, and timestamp.

[0505] Step 4: Data Preprocessing on the Server

[0506] The server cleanses the stored behavioral and emotional data (data processing). Specifically, it complements missing values ​​and removes outliers, preparing the data for analysis (output).

[0507] Step 5: Server Feature Extraction

[0508] The server extracts features from the cleansed data (data processing). Specifically, it aggregates information such as message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week (output).

[0509] Step 6: Applying the machine learning model to the server

[0510] The server analyzes and classifies the user's behavioral and emotional patterns using machine learning algorithms (K-means clustering and linear regression) based on the extracted features (data calculation and output). For example, User A is classified into the "high reaction / positive group."

[0511] Step 7: Prepare the server message template

[0512] The server prepares message templates, including templates for new product introductions, sales announcements, newsletters, etc. (Output)

[0513] Step 8: Customizing the Server

[0514] The server customizes templates based on each user's behavioral and emotional patterns, inserting specific keywords and product categories and adjusting the tone and content of messages depending on their emotional state (data processing, output).

[0515] Step 9: Server Message Generation

[0516] The server automatically generates (outputs) a message optimized for each user based on the customized template. For example, a message such as "Check it out now! The latest smartphone has been released. We also have a special discount code!" is generated.

[0517] Step 10: Deciding when the server should send

[0518] The server determines the optimal sending time based on the user's past data (data calculation, output). For example, if user A is determined to be most responsive at 9:00 AM, the server will schedule the message to be sent at 9:00 AM.

[0519] Step 11: Server Sends Message

[0520] The server uses the LINE API to send the generated optimized message to the user at the determined timing (output).

[0521] Step 12: User receives message and takes action

[0522] The user opens the received message and takes action such as clicking a link (input).

[0523] Step 13: Real-time Server Monitoring

[0524] The server monitors users' reactions to the sent messages in real time and collects (outputs) reaction data, such as whether the message was opened, whether a link was clicked, and the content of the response.

[0525] Step 14: Save the server's reaction data

[0526] The server saves the user's reaction data in a database (output), which is then stored as material for the next analysis.

[0527] Step 15: Retraining the model on the server

[0528] The server retrains the machine learning model based on the new reaction data to improve the accuracy of the next message optimization (data calculation, output). Specifically, it learns new behavioral and emotional patterns and improves the accuracy of the model.

[0529] (Application example 2)

[0530] Next, a description will be given of Application Example 2. 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."

[0531] Conventional advertising delivery systems optimize messages based solely on user behavioral data, which means they are unable to generate effective messages that reflect the user's emotional state. Furthermore, relying solely on user behavioral data makes it difficult to accurately reflect the user's real-time interests and emotional changes, limiting the effectiveness of advertising.

[0532] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting user behavioral data and emotional data, means for analyzing the collected behavioral data and emotional data and identifying the user's behavioral patterns and emotional patterns, means for generating documents optimized for each user based on the identified behavioral patterns and emotional patterns, means for sending the generated documents to the user, and means for collecting user reaction data and providing feedback to improve the accuracy of the analysis means. This makes it possible to integrate the user's behavioral data and emotional data and generate and deliver individually optimized advertising messages.

[0533] "User behavioral data" refers to data generated when a user behaves online, such as message open rates, link click rates, and the content of past communications.

[0534] "Emotional data" is data that indicates a user's emotional state collected through text analysis, voice analysis, and image analysis.

[0535] "Behavioral patterns" refer to specific behavioral tendencies and habits that are analyzed based on user behavior data.

[0536] An "emotion pattern" indicates the tendency and changes in the user's unique emotions, which are analyzed based on the emotion data.

[0537] "Optimized documents" are messages and notifications that are individually customized and created in the most effective way based on the user's behavioral and emotional patterns.

[0538] The "feedback means" is a means for collecting user response data and reflecting it in the system to improve the accuracy of the analysis means.

[0539] The present invention is an advertising distribution system that collects, analyzes, and optimizes user behavioral and emotional data. This system is implemented using a server, user terminals (smartphones, etc.), and a network. Specifically, the invention is implemented with the following configuration.

[0540] 1. Data Collection

[0541] The server collects user behavioral and emotional data in real time. Specifically, it uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also uses an emotion engine to obtain user emotional data through text, voice, and image analysis. This data is then stored in a database.

[0542] 2. Data analysis

[0543] The server cleanses the collected behavioral and emotional data, fills in missing values, removes outliers, and extracts features through feature extraction. It then uses machine learning models (e.g., clustering and regression analysis) to classify and predict user behavioral and emotional patterns.

[0544] 3. Individual optimization

[0545] The server uses multiple pre-prepared message templates and customizes them individually based on each user's behavioral and emotional patterns, adjusting the tone and content of the message according to the user's emotional state to generate individually optimized documents.

[0546] 4. Sending a message

[0547] The server determines the optimal sending time based on the analysis results. For example, if past data indicates that a particular user is most likely to respond at 9:00 a.m., it sets the message to be sent at that time. The server then uses the LINE API to send the generated optimized message to the user.

[0548] 5. Response monitoring and feedback

[0549] The server monitors user responses to sent messages (open rates, click rates, response content) in real time and stores the response data in a database. This data is used to retrain the machine learning model and improve the accuracy of the next message optimization.

[0550] Specific examples

[0551] When User A shows interest in a certain product, the server collects their behavioral and emotional data in real time. If User A sends a message saying, "I love this product!", the emotional data indicates a positive reaction. Based on this information, the server generates a customized message for User A saying, "Check it out now! The latest smartphone has been released. We also have a special discount code!" and sends it to User A at 9:00 AM.

[0552] An example of a prompt used by a generative AI model is, "If user A expresses positive feelings about a particular product, generate a message containing the latest information and special discounts related to that product."

[0553] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0554] Step 1:

[0555] It is a means of collecting message open rates, link click rates, and past communication content from users' devices using the LINE API. The input data is user behavioral data and emotional data, and the output is the collected data. In particular, text analysis analyzes the content of users' messages and generates emotional data such as positive or negative to determine their emotional state. In addition, an emotion engine is used to perform voice and image analysis to collect user emotional data.

[0556] Step 2:

[0557] The server stores the collected behavioral and emotional data in a database. This database records user IDs, message IDs, behavior types (open, click, response), emotional states, timestamps, etc. The input data is the collected user behavioral and emotional data, and the output is structured data organized in the database. Specific operations include classifying and structuring the information.

[0558] Step 3:

[0559] The server performs data preprocessing, which includes data cleansing, missing value imputation, and outlier removal. The input data is raw data stored in a database, and the output is clean data. In particular, Pandas is used to clean the data.

[0560] Step 4:

[0561] The server performs feature extraction to extract features from user behavioral and emotional data. The input data is preprocessed clean data, and the output is extracted features. Specific operations include compiling message open rates and click rates over the past week, the most responsive times, and emotional change patterns.

[0562] Step 5:

[0563] The server uses a machine learning model to classify and predict the user's behavioral and emotional patterns based on the extracted features. The input data are the features, and the output is the user's behavioral and emotional patterns. Specific operations include pattern classification using clustering and regression analysis.

[0564] Step 6:

[0565] The server uses multiple pre-prepared message templates to generate documents optimized for each user. The input data are classified behavioral and emotional patterns, and the output is an optimized message. Specific operations include template customization and keyword insertion.

[0566] Step 7:

[0567] The server sends the generated optimized message to the user. The message is sent using the LINE API. The input data is the optimized message and the sending timing, and the output is the sent message. Specific operations include determining the sending time and delivering the message.

[0568] Step 8:

[0569] The server monitors user responses in real time and collects response data. The input data is the sent message and the user's response, and the output is the collected response data. Specific operations include monitoring the open rate, click rate, and response.

[0570] Step 9:

[0571] The server stores the collected reaction data in a database and provides a feedback mechanism for retraining the machine learning model. The input data is the collected reaction data, and the output is an updated machine learning model. Specific operations include learning new data and improving the accuracy of the model.

[0572] 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 a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the 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.

[0573] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0574] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0575] [Second embodiment]

[0576] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0577] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0578] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0579] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0580] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0581] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0582] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0583] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0584] The specific processing program 56 is an example of a "program" according to the technology of the present 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.

[0585] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0586] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. 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 process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0587] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. 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."

[0588] The present invention relates to a system for collecting and analyzing user behavior data for the administrator of an official LINE account, and generating and sending messages optimized for each user.

[0589] 1. Data Collection

[0590] User:Activity Log

[0591] A user takes an action such as opening a message, clicking, or responding to the official LINE account.

[0592] Server: Collecting data logs

[0593] The server connects with the official LINE account and collects user behavior data in real time, specifically using technologies such as the LINE API to collect information such as message open rates, link click rates, and past communication content.

[0594] Server: Save to database

[0595] The collected behavioral data is stored in a database, including user ID, message ID, behavior type (open, click, response), timestamp, etc.

[0596] 2. Data analysis

[0597] Server: Data preprocessing

[0598] Cleanse the stored behavioral data, fill in missing values, and remove outliers.

[0599] Server:Feature Extraction

[0600] Extract features from user behavior patterns. For example, extract data such as message open rates, click rates, and the most responsive times for the past week.

[0601] Server: Applying machine learning models

[0602] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict user behavior patterns, allowing us to understand the characteristics of each user.

[0603] 3. Individual optimization

[0604] Server: Prepare message template

[0605] It provides multiple pre-made message templates, such as new product introductions, sales announcements, newsletters, etc.

[0606] Server:Customization

[0607] Customize templates based on each user's behavioral patterns, for example, inserting specific keywords or product categories for each user.

[0608] Server: Message Creation

[0609] Automatically generate personalized messages for each user based on customized templates.

[0610] 4. Sending a message

[0611] Server: Decide when to send

[0612] Based on the analysis results, the optimal sending time for each user is determined. If, based on past data, User A is judged to be most responsive at 9:00 AM, the message will be sent at that time.

[0613] Server: Send message

[0614] Using the LINE API, etc., the generated optimized message is sent to the user at the determined time.

[0615] 5. Monitoring the reaction

[0616] Users: Message Reception and Actions

[0617] Users receive the messages sent and take action such as opening, clicking, or responding.

[0618] Server: Real-time monitoring

[0619] The server monitors users' reactions to the sent messages in real time and collects reaction data.

[0620] 6. Feedback

[0621] Server: Reaction data storage

[0622] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[0623] Server: Retrain the model

[0624] Based on the new reaction data collected, the machine learning model is retrained to improve the accuracy of the next message optimization.

[0625] Specific examples

[0626] Example 1: New product introduction

[0627] Data collection

[0628] Behavioral data is collected showing that User A has shown interest in "electronic devices" in the past.

[0629] Data analysis

[0630] It is analyzed that user A is often interested in messages containing the keyword "new product announcement."

[0631] Individual optimization

[0632] Generate a message for User A saying, "Check it out now! The latest smartphones have been released and we have special discount codes for you!"

[0633] Sending a message

[0634] Send the message at 9:00 AM. Based on past data, this time is the most responsive time for User A.

[0635] Reaction monitoring

[0636] User A opens the message and clicks the link to view product details.

[0637] feedback

[0638] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[0639] In this way, the present invention maximizes the marketing effectiveness of official LINE by generating and sending optimal messages based on each user's behavioral patterns.

[0640] The processing flow will be explained below.

[0641] Step 1: Data collection

[0642] 1. User: Activity Record

[0643] A user opens a message from the official LINE account, clicks a link, or responds to a message.

[0644] 2. Server: Collecting data logs

[0645] The server collects user behavior data in real time through the LINE API, including message open rates, link click rates, and past communication content.

[0646] 3. Server: Save to database

[0647] The collected behavioral data is organized in a database, and the user ID, message ID, behavior type (open, click, response), timestamp, etc. are recorded.

[0648] Step 2: Data analysis

[0649] 1. Server: Data preprocessing

[0650] Cleanse the stored behavioral data, fill in missing values, and remove outliers.

[0651] 2. Server: Feature Extraction

[0652] Extract features from user behavior data. For example, aggregate message open rates, click rates, and the most responsive times for the past week.

[0653] 3. Server: Applying machine learning models

[0654] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict user behavior patterns.

[0655] Step 3: Individual optimization

[0656] 1. Server: Prepare message template

[0657] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[0658] 2. Server: Customization

[0659] Customize templates individually based on each user's behavioral patterns, for example by inserting specific keywords or product categories to optimize message content.

[0660] 3. Server: Message Creation

[0661] Generate personalized messages for each user based on customized templates.

[0662] Step 4: Send the message

[0663] 1. Server: Decide when to send

[0664] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[0665] 2. Server: Send message

[0666] Use the LINE API to send the generated optimized message to the user at the determined time.

[0667] Step 5: Monitoring the reaction

[0668] 1. Users: Receiving messages and taking action

[0669] Users open the messages they receive and click on links within them or reply to them.

[0670] 2. Server: Real-time monitoring

[0671] User responses to sent messages (e.g., open rates, click rates, response content) are monitored in real time and response data is collected.

[0672] Step 6: Feedback

[0673] 1. Server: Reaction data storage

[0674] The collected reaction data is stored in a database and accumulated as material for the next analysis.

[0675] 2. Server: Retrain the model

[0676] Retrain your machine learning model based on new response data to improve the accuracy of your next message optimization.

[0677] In this way, through each processing step, the system can optimize messages from the official LINE account for each individual user, allowing for more effective communication of information.

[0678] Example 1

[0679] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0680] Conventional messaging systems have had the problem of being unable to deliver effective marketing due to the difficulty in sending messages optimized for each user. Furthermore, they lacked a feedback mechanism for effectively analyzing user behavior data and utilizing the results in the next messaging. Therefore, a system that can maximize the effectiveness of marketing is needed.

[0681] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0682] In this invention, the server includes means for collecting user behavioral data, means for cleansing the collected behavioral data, completing missing values ​​and removing outliers, means for performing feature extraction to extract features from user behavior patterns, means for applying a machine learning algorithm to classify and predict user behavioral patterns, means for preparing multiple message templates, customizing the templates based on each user's behavioral pattern, and generating an optimal message, means for determining an optimal sending time and sending the generated message to the user, means for collecting user reaction data to the sent message and saving it as material for subsequent analysis and optimization, and feedback means for retraining a machine learning model based on the collected reaction data and improving the accuracy of message optimization, thereby enabling the generation and sending of optimal messages based on each user's behavioral pattern.

[0683] "User behavior data" is data recorded when a user takes a specific action, such as opening a message, clicking a link, or responding.

[0684] "Data cleansing" is the process of imputing missing values ​​and removing outliers from collected data.

[0685] "Feature extraction" is the process of extracting features from user behavior patterns, which makes analysis and prediction easier.

[0686] A "machine learning algorithm" is an algorithm that learns patterns and regularities from data and makes future predictions and classifications.

[0687] A "message template" is a pre-defined framework for message format and content, allowing for quick message generation and customization.

[0688] The "optimal sending time" is the time at which a message is sent that is most likely to be responsive to each user, based on collected data and analysis results.

[0689] "Response data" is a record of the actions users take in response to messages they receive, and is used as material for future analysis and optimization.

[0690] "Feedback measures" are processes that retrain machine learning models based on collected response data to improve the accuracy of the next message optimization.

[0691] This invention is a system for administrators of official messaging accounts that collects and analyzes user behavior data, and generates and sends individually optimized messages. The aim is to effectively utilize user behavior data and maximize the marketing effectiveness of messages.

[0692] This system is mainly composed of three elements: a server, a terminal, and a user. Below, we will explain the flow of data processing and data calculation using the specific hardware and software of each element.

[0693] Data collection

[0694] User:Activity Log

[0695] Users take actions such as opening, clicking, or responding to messages on official messaging accounts, and these actions act as triggers for the system.

[0696] Server: Collecting data logs

[0697] The server obtains user behavior data in real time via a messaging API (e.g., the LINE API), including message open rates, link click rates, and past communication content.

[0698] Server: Save to database

[0699] The collected behavioral data is stored in a database along with information such as user ID, message ID, behavior type (open, click, response), and timestamp.

[0700] Data analysis

[0701] Server: Data preprocessing

[0702] Stored behavioral data is cleansed, missing values ​​are filled, and outliers are removed, ensuring data accuracy and reliability.

[0703] Server:Feature Extraction

[0704] Extract features from user behavior patterns, such as message open rates and click rates over the past week, and the time of day when responses are most likely to occur.

[0705] Server: Applying machine learning models

[0706] Based on the extracted features, machine learning algorithms (e.g., clustering, regression analysis, etc.) are applied to classify and predict user behavior patterns, allowing for a detailed understanding of each user's characteristics.

[0707] Individual optimization

[0708] Server: Prepare message template

[0709] The server uses a number of pre-defined message templates (e.g., new product introductions, sales announcements, newsletters, etc.).

[0710] Server:Customization

[0711] The template is customized based on each user's behavioral patterns, for example, inserting relevant keywords for a user interested in "electronic devices."

[0712] Server: Message Creation

[0713] It automatically generates personalized messages for each user based on customized templates, using a generative AI model and providing specific prompts.

[0714] Sending a message

[0715] Server: Decide when to send

[0716] Based on the analysis results, the optimal sending time for each user is determined. For example, if past data indicates that User A responds best at 9:00 AM, then that time is set.

[0717] Server: Send message

[0718] The server uses the API to send the generated optimization message to the user at the determined timing.

[0719] Reaction monitoring

[0720] Users: Message Reception and Actions

[0721] Users receive the messages sent and take action such as opening, clicking, or responding.

[0722] Server: Real-time monitoring

[0723] The server monitors users' reactions to the sent messages in real time and stores the reaction data in a database.

[0724] feedback

[0725] Server: Reaction data storage

[0726] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[0727] Server: Retrain the model

[0728] Based on the new reaction data collected, the machine learning model is retrained to improve the accuracy of message optimization.

[0729] An example of a specific prompt might be, "Generate a message for a new product launch taking into account user behavioral data. Use keywords that have generated positive responses in past messages, and include content that will increase open rates."

[0730] In this way, the present invention is a system that generates and sends optimal messages based on each user's behavioral patterns, maximizing the marketing effectiveness of official messaging accounts.

[0731] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0732] Step 1:

[0733] Data collection

[0734] User:Activity Log

[0735] Users open messages, click links, and send responses to official messaging accounts, and these actions trigger data collection.

[0736] Input: User actions (opens, clicks, responses)

[0737] Output: When an action occurs, the data is collected immediately.

[0738] Server: Collecting data logs

[0739] The server obtains user behavior data in real time via a messaging API (e.g., LINE API).

[0740] Input: Behavioral data captured by the messaging API

[0741] Output: Raw behavioral data (open rates, click rates, communication content, etc.)

[0742] Server: Save to database

[0743] The server stores the collected behavioral data in a database, which includes the user ID, message ID, behavior type, and timestamp.

[0744] Input: Raw behavioral data

[0745] Output: Behavioral data stored in a database

[0746] Step 2:

[0747] Data Preprocessing

[0748] Server: Data Cleansing

[0749] The server cleanses the stored data, imputing missing values ​​and removing outliers.

[0750] Input: Stored behavioral data

[0751] Output: Cleansed data

[0752] Specific behavior: For example, gracefully handling time zone inconsistencies and extreme values.

[0753] Step 3:

[0754] Data analysis

[0755] Server:Feature Extraction

[0756] The server extracts features from user behavior patterns, such as message open rates and click rates over the past week, and the time periods with the highest response rates.

[0757] Input: Cleansed data

[0758] Output: Extracted features

[0759] Specific operations: Analyze user behavior patterns using time series analysis and statistical methods.

[0760] Server: Applying machine learning models

[0761] The server applies a machine learning algorithm based on the extracted features to classify and predict user behavior patterns.

[0762] Input: Feature data

[0763] Output: Classified user groups and predicted behavior patterns

[0764] Specific behavior: Uses clustering and regression analysis algorithms.

[0765] Step 4:

[0766] Individual optimization

[0767] Server: Prepare message template

[0768] The server uses a number of pre-defined message templates.

[0769] Input: Template list

[0770] Output: Available templates

[0771] Specific actions: Select and update templates regularly.

[0772] Server:Customization

[0773] The server customizes the template based on each user's behavioral patterns, for example, by inserting keywords related to "electronic devices" to match the user's interests.

[0774] Input: Template, Categorized User Groups

[0775] Output: Customized template

[0776] Specific behavior: Modify specific parts of the template to suit the user's interests.

[0777] Server: Message Creation

[0778] The server automatically generates messages optimized for each user based on customized templates.

[0779] Input:Customized template

[0780] Output: Generated optimization messages

[0781] What it does: Uses a generative AI model to generate an appropriate message based on the prompt.

[0782] Step 5:

[0783] Sending a message

[0784] Server: Decide when to send

[0785] The server determines the optimal transmission time for each user from the analysis results.

[0786] Input: User's optimal transmission time data

[0787] Output: Transmission schedule

[0788] Specific behavior: Optimize sending times based on past behavioral data.

[0789] Server: Send message

[0790] The server sends the optimization message generated using the API to the user at the determined timing.

[0791] Input: Generated message, send schedule

[0792] Output: Message sent

[0793] Specific operation: The actual message sending is carried out using the API.

[0794] Step 6:

[0795] Reaction monitoring

[0796] Users: Message Reception and Actions

[0797] The user opens the received message, clicks on the link, and sends a response.

[0798] Input: Optimized message

[0799] Output: User response data (behavior)

[0800] What it does: Collect user feedback in real time.

[0801] Server: Real-time monitoring

[0802] The server monitors users' reactions to the sent messages in real time and stores the reaction data in a database.

[0803] Input: User response data

[0804] Output: Saved reaction data

[0805] Specific actions: Monitor using real-time data analysis tools.

[0806] Step 7:

[0807] feedback

[0808] Server: Reaction data storage

[0809] The server stores the user's reaction data in a database and accumulates it as material for the next analysis.

[0810] Input: User response data

[0811] Output: Saved reaction data

[0812] Specific operation: The saving process is carried out using a database management system.

[0813] Server: Retrain the model

[0814] The server retrains the machine learning model based on the new reaction data collected, improving the accuracy of the next message optimization.

[0815] Input: New reaction data

[0816] Output: A retrained machine learning model

[0817] What it does: Continually improve the model using adaptive learning techniques.

[0818] This makes it possible to generate and send optimal messages based on each user's behavioral patterns, maximizing marketing effectiveness.

[0819] (Application example 1)

[0820] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0821] Conventional user behavior data collection and analysis systems have been unable to achieve effective marketing due to insufficient optimization of messages for individual users. In particular, generic messages are often sent without sufficient consideration of the different behavioral patterns of each user, limiting the effectiveness of improving user response rates and purchasing intent. Furthermore, the system lacks a mechanism for efficiently feeding back response data to messages sent and improving the accuracy of the model.

[0822] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0823] In this invention, the server includes means for collecting user behavioral data, means for analyzing the collected behavioral data and identifying user behavior patterns, means for generating a document optimized for each user based on the identified behavioral pattern, means for sending the generated document to the user, means for collecting user reaction data and providing feedback to improve the accuracy of the analysis means, means for using a machine learning algorithm to perform clustering based on the collected behavioral data, and means for selecting a message template optimized for each user cluster and generating a customized message, thereby enabling effective message optimization based on user behavior patterns.

[0824] "User behavior data" refers to records of various interactions users have with the system, including, for example, message open rates, link click rates, and the content of past communications.

[0825] "Analysis Tools" refers to the functions and algorithms used to process and analyze collected behavioral data and identify user behavior patterns.

[0826] "Behavioral patterns" refer to models or tendencies that indicate how a user will behave based on the user's past behavioral data.

[0827] "Optimized documents" refers to messages or documents that are customized to best suit a user based on the user's behavioral patterns.

[0828] "Feedback means" refers to a data collection and processing mechanism for improving the analytical accuracy of the system and the performance of the machine learning model based on user response data.

[0829] "Clustering" refers to a technique that uses machine learning algorithms to classify users into groups with similar behavioral patterns.

[0830] A "machine learning algorithm" is a part of a computer program that analyzes large amounts of data and automatically learns patterns and predictive models.

[0831] A "template" is a pre-created message template for a specific purpose, which serves as the basis for customization.

[0832] "Customization" refers to tailoring documents and messages to specific user behavior patterns.

[0833] The present invention relates to a system for generating and sending optimized messages based on user behavior data. The system is implemented in the following steps:

[0834] First, the server uses the LINE API to collect user behavior data, including message open rates, link click rates, and past communication content. The collected data is sent to the server in real time and stored in a database.

[0835] Once the data is collected, the server pre-processes it, cleansing it to fill in missing values ​​and remove outliers, then performs feature extraction to extract parameters that characterize user behavior patterns.

[0836] The server then uses machine learning algorithms, specifically clustering algorithms, to classify users into clusters with similar behavioral patterns, leveraging Python's scikit-learn library and other machine learning tools.

[0837] After clustering is complete, the server applies pre-prepared template messages to each cluster. Each template is customized to fit the behavioral patterns of a specific user. The customized message is then sent to the user in the most effective way. For example, user A might receive a message like, "Hello, would you like to check out our latest products?"

[0838] After the message is sent, user reaction data is collected again and sent to the server. This reaction data is stored in a database and used as a feedback mechanism to improve the accuracy of the analysis method. The new data is used to retrain the machine learning model, improving the accuracy of the next message optimization.

[0839] As a concrete example, suppose there is data showing that User A has clicked on messages related to beauty products in the past. In this case, the machine learning model will classify User A into a cluster interested in beauty products and use a template message for that cluster to send a customized message such as "Don't miss out on our great beauty product campaigns!"

[0840] Examples of prompts include:

[0841] Generate customized product introduction messages according to the cluster. Cluster 1: For users interested in beauty products

[0842] Don't miss out on our great beauty deals!

[0843] Cluster 2: For users interested in home appliances

[0844] The latest home appliances have arrived! Check them out now.

[0845] In this way, the present invention maximizes marketing effectiveness by utilizing user behavior data to generate and send messages optimized for individual users.

[0846] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0847] Step 1:

[0848] The server collects user behavior data via the LINE API, including message open rates, link click rates, and past communication content. The collected data is sent to the server in real time and stored in a database.

[0849] Input: User behavior data obtained via the LINE API

[0850] Output: Behavioral data stored in a database

[0851] Specific operation: The server sends a request to the LINE API and executes an SQL query to save the retrieved data to the database.

[0852] Step 2:

[0853] The server pre-processes the stored behavioral data, performing data cleansing, filling in missing values, and removing outliers.

[0854] Input: Raw data stored in a database

[0855] Output: Cleansed behavioral data

[0856] Specific operation: The server reads data from the database and uses the Pandas library to impute missing values ​​and remove outliers.

[0857] Step 3:

[0858] The server performs feature extraction from the preprocessed data, extracting features such as the open rate, click rate, and best response time for each user.

[0859] Input: Cleansed behavioral data

[0860] Output: A dataset with extracted features

[0861] What it does: The server groups the data by user ID, calculates certain statistics, and generates new features, such as how many times a message was opened or clicked each day.

[0862] Step 4:

[0863] The server performs clustering using a machine learning algorithm to classify users into clusters with similar behavioral patterns.

[0864] Input: Feature-extracted dataset

[0865] Output: User data with assigned clusters

[0866] What it does: The server uses the scikit-learn library to run the KMeans algorithm to classify users into clusters, and each user is assigned a cluster ID.

[0867] Step 5:

[0868] The server selects the best template message for each cluster and generates a customized message.

[0869] Input: User data with assigned clusters, and message template

[0870] Output: Customized message

[0871] What happens: The server selects a template for each cluster and customizes the message by inserting user-specific information.

[0872] Step 6:

[0873] The server sends the generated customized message to the user via the LINE API.

[0874] Input: Customized Message

[0875] Output: Message sent to the user's terminal

[0876] Specific operation: The server sends a POST request to the LINE API and sends the generated message to the user at the appropriate time, for example, during the time period when the user is most likely to respond.

[0877] Step 7:

[0878] The server again collects user reaction data to the sent messages and stores it in a database.

[0879] Input: User interaction data (message opens, clicks, responses, etc.)

[0880] Output: Reaction data stored in a database

[0881] Specific operation: The server retrieves the user's reaction data from the LINE API and executes an SQL query to save it in the database.

[0882] Step 8:

[0883] The server retrains the machine learning model based on the new response data to improve the accuracy of the next message optimization.

[0884] Input: New reaction data stored in the database

[0885] Output: An updated machine learning model

[0886] Specific operation: The server extracts features from new reaction data and retrains the machine learning algorithm to improve the accuracy of the model.

[0887] Through these steps, the present invention achieves effective message optimization based on user behavior data.

[0888] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0889] The present invention relates to a system for collecting and analyzing behavioral and emotional data of users for the administrator of an official LINE account, and generating and sending messages optimized for each user.

[0890] 1. Data Collection

[0891] User:Activity Log

[0892] A user takes an action such as opening a message, clicking, or responding to the official LINE account.

[0893] Server: Data logging and emotion data collection

[0894] The server connects with the official LINE account to obtain user behavioral and emotional data in real time. Specifically, it uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also uses an emotion engine to collect user emotional data through text, voice, and image analysis.

[0895] Server: Save to database

[0896] The collected behavioral and emotional data is organized in a database, and the user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc. are recorded.

[0897] 2. Data analysis

[0898] Server: Data preprocessing

[0899] Cleanse stored behavioral and emotional data to fill in missing values ​​and remove outliers.

[0900] Server:Feature Extraction

[0901] Extract features from user behavioral and emotional data. For example, aggregate message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week.

[0902] Server: Applying machine learning models

[0903] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict the user's behavioral and emotional patterns.

[0904] 3. Individual optimization

[0905] Server: Prepare message template

[0906] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[0907] Server:Customization

[0908] Templates are individually customized based on each user's behavioral and emotional patterns. For example, message content can be optimized by inserting specific keywords or product categories. The tone and content of messages can also be adjusted according to the user's emotional state.

[0909] Server: Message Creation

[0910] Automatically generate messages optimized for each user based on customized templates.

[0911] 4. Sending a message

[0912] Server: Decide when to send

[0913] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[0914] Server: Send message

[0915] Use the LINE API to send the generated optimized message to the user at the determined time.

[0916] 5. Monitoring the reaction

[0917] Users: Message Reception and Actions

[0918] Users open the messages they receive and click on links within them or reply to them.

[0919] Server: Real-time monitoring

[0920] The server monitors user responses to sent messages (e.g., open rate, click rate, response content) in real time and collects response data.

[0921] 6. Feedback

[0922] Server: Reaction data storage

[0923] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[0924] Server: Retrain the model

[0925] Retrain your machine learning model based on new reaction data collected to improve the accuracy of your next message optimization.

[0926] Specific examples

[0927] Example 1: New product introduction

[0928] Data collection

[0929] The system collects behavioral data showing that User A has shown interest in "electronic devices" in the past. The emotion engine also identifies a positive emotional state based on User A's recent messages and voice.

[0930] Data analysis

[0931] It is analyzed that User A is often interested in messages containing the keyword "new product launch." Also, since User A's emotional state is positive, it is determined that a direct sales message would be effective.

[0932] Individual optimization

[0933] Generate a positive message for User A: "Check it out now! The latest smartphones are on sale now, and we've got a special discount code for you!"

[0934] Sending a message

[0935] Send the message at 9:00 AM. Based on past data, this time is the most responsive time for User A.

[0936] Reaction monitoring

[0937] User A opens the message and clicks the link to view product details.

[0938] feedback

[0939] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[0940] In this way, the present invention maximizes the marketing effectiveness of official LINE by combining and analyzing user behavioral data and emotional data to generate and send optimal messages according to individual needs and emotional states.

[0941] The processing flow will be explained below.

[0942] Step 1: Data collection

[0943] 1. User: Activity Record

[0944] A user opens a message from the official LINE account, clicks a link, or responds to a message.

[0945] 2. Server: Collecting data logs

[0946] The server obtains user behavior data in real time through the LINE API, including message open rates, link click rates, and past communication content.

[0947] 3. Server: Emotion data collection

[0948] The emotion engine is used to collect user emotional data through text, voice, and image analysis, for example, to identify the user's emotional state from the content of their message or their tone of voice.

[0949] 4. Server: Save to database

[0950] The collected behavioral and emotional data is organized and stored in a database, including user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc.

[0951] Step 2: Data analysis

[0952] 1. Server: Data preprocessing

[0953] Cleanse stored behavioral and emotional data to fill in missing values ​​and remove outliers.

[0954] 2. Server: Feature Extraction

[0955] Extract features from user behavioral and emotional data. For example, aggregate message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week.

[0956] 3. Server: Applying machine learning models

[0957] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict the user's behavioral and emotional patterns.

[0958] Step 3: Individual optimization

[0959] 1. Server: Prepare message template

[0960] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[0961] 2. Server: Customization

[0962] Templates are individually customized based on each user's behavioral and emotional patterns. For example, message content can be optimized by inserting specific keywords or product categories. The tone and content of messages can also be adjusted according to the user's emotional state.

[0963] 3. Server: Message Creation

[0964] Automatically generate messages optimized for each user based on customized templates.

[0965] Step 4: Send the message

[0966] 1. Server: Decide when to send

[0967] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[0968] 2. Server: Send message

[0969] Use the LINE API to send the generated optimized message to the user at the determined time.

[0970] Step 5: Monitoring the reaction

[0971] 1. Users: Receiving messages and taking action

[0972] Users open the messages they receive and click on links within them or reply to them.

[0973] 2. Server: Real-time monitoring

[0974] User responses to sent messages (e.g., open rates, click rates, response content) are monitored in real time and response data is collected.

[0975] Step 6: Feedback

[0976] 1. Server: Reaction data storage

[0977] The collected reaction data is stored in a database and accumulated as material for the next analysis.

[0978] 2. Server: Retrain the model

[0979] Retrain your machine learning model based on new response data to improve the accuracy of your next message optimization.

[0980] Example 2

[0981] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0982] Conventional message sending systems only considered user behavioral data and did not perform individual optimization using emotional data, making it difficult to maximize the effectiveness of marketing messages. Furthermore, there was also the problem of not being able to quickly reflect user reactions because the timing of sending messages was not optimized and real-time feedback was not adequately provided.

[0983] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting user behavioral data and emotional data, means for analyzing the collected behavioral data and emotional data and identifying the user's behavioral and emotional patterns, means for generating a document optimized for each user based on the identified behavioral and emotional patterns, means for sending the generated document to the user, and means for collecting user reaction data in real time and providing feedback to improve the accuracy of the analysis means. This makes it possible to generate and send individually optimized messages based on the user's behavioral and emotional data, thereby maximizing marketing effectiveness.

[0984] "User behavioral data" refers to information related to actions taken by users on official LINE accounts, such as opening messages, clicking links, and responding to messages.

[0985] "User emotional data" is information collected using text analysis, speech analysis, image analysis, etc. to identify a user's emotional state.

[0986] A "behavioral pattern" refers to the regularity or tendency of behavior that is analyzed based on data on a series of actions taken by a user.

[0987] "Emotional patterns" refer to changes and trends in emotions that are analyzed based on the user's emotional data.

[0988] "Optimized documents" are messages that are individually tailored based on a user's behavioral and emotional patterns and are generated to maximize the user's interest and response.

[0989] A "machine learning algorithm" is a computational method for learning patterns from data and making predictions or classifications based on the results.

[0990] "Clustering" is a type of machine learning technique that divides data into several groups.

[0991] "Regression analysis" is a machine learning technique that is used to model the relationship between predictor variables and response variables and make predictions.

[0992] A "feedback means" is a method or mechanism for collecting user reaction data and using that data to improve the accuracy of the analysis means and generation means.

[0993] "Optimizing sending timing" aims to send messages at the time when users are most likely to respond, and determines the timing of sending based on past behavioral data.

[0994] This invention relates to a system for official LINE administrators that collects and analyzes user behavioral and emotional data, and generates and sends messages optimized for each user.

[0995] 1. Data Collection

[0996] User behavior record

[0997] Users open messages sent to official LINE accounts, click links, and respond - all of this behavioral data is collected in real time.

[0998] Server data logging and emotion data collection

[0999] The server uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also collects user emotional data through text, voice, and image analysis by emotion engines (such as IBM Watson and Microsoft Azure).

[1000] Server database storage

[1001] The server organizes the collected behavioral and emotional data into a database, recording the user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc.

[1002] 2. Data analysis

[1003] Server data preprocessing

[1004] The server cleanses the stored behavioral and emotional data, filling in missing values ​​and removing outliers, thereby ensuring data quality.

[1005] Server Feature Extraction

[1006] The server extracts features from the behavioral and emotional data, such as message open rates, click rates, the most responsive times, and patterns of emotional change over the past week.

[1007] Applying machine learning models to servers

[1008] Based on the extracted features, the server uses machine learning algorithms (e.g., K-means clustering and linear regression) to classify and predict the user's behavioral and emotional patterns.

[1009] 3. Individual optimization

[1010] Preparing the server message template

[1011] The server prepares a number of pre-defined message templates (e.g., new product introductions, sales announcements, newsletters).

[1012] Server Customization

[1013] The server customizes the templates individually for each user based on their behavioral and emotional patterns, inserting specific keywords and product categories and adjusting the tone and content of the message depending on the user's emotional state.

[1014] Server Message Generation

[1015] The server automatically generates messages optimized for each user based on customized templates.

[1016] 4. Sending a message

[1017] Server transmission timing decision

[1018] The server then uses the analysis to determine the best time to send the message. For example, if historical data indicates that a particular user is most likely to respond at 9 a.m., it will schedule the message to be sent at that time.

[1019] Server message sending

[1020] The server uses the LINE API to send the generated optimized message to the user at the determined timing.

[1021] 5. Monitoring the reaction

[1022] User message receipt and actions

[1023] The user opens the received message, clicks on a link in the message, or makes a response.

[1024] Real-time server monitoring

[1025] The server monitors user responses to sent messages (e.g., open rate, click rate, response content) in real time and collects response data.

[1026] 6. Feedback

[1027] Saving reaction data on the server

[1028] The server stores the user's reaction data in a database and accumulates it as material for the next analysis.

[1029] Retraining the model on the server

[1030] The server retrains the machine learning model based on the new reaction data collected, improving the accuracy of the next message optimization.

[1031] Specific examples

[1032] Example 1: New product introduction

[1033] Behavioral data showing that User A has shown interest in "electronic devices" in the past is collected. In addition, analysis by an emotion engine (e.g., IBM Watson) identifies a positive emotional state based on User A's recent messages and voice.

[1034] Data analysis

[1035] It is analyzed that User A is often interested in messages containing the keyword "new product launch." Also, since User A's emotional state is positive, it is determined that a direct sales message would be effective.

[1036] Individual optimization

[1037] Generate a positive message for User A: "Check it out now! The latest smartphones are on sale now, and we've got a special discount code for you!"

[1038] Sending a message

[1039] Based on past data, it is determined that User A responds best at 9:00 AM, so the message is sent at this time.

[1040] Reaction monitoring

[1041] User A opens the message and clicks the link to view product details.

[1042] feedback

[1043] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[1044] Prompt Sentence Examples

[1045] Here are some example prompts to enter into a generative AI model:

[1046] "User A has shown interest in 'electronic devices' in the past, and analysis of the emotion engine reveals that he is in a positive emotional state. Please create a message about a new product launch that will interest User A."

[1047] Using this prompt, the generative AI model generates a message optimized for User A.

[1048] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1049] Step 1: Recording user behavior

[1050] A user opens a message from an official LINE account, clicks a link, and responds. This behavioral data is sent to the LINE server (input).

[1051] Step 2: Collecting server data logs and emotion data

[1052] The server works with the official LINE app to call the LINE API and collect (output) user behavior data in real time. Specifically, behavioral data acquired includes message open rates, link click rates, and past communication content. Additionally, an emotion engine (e.g., IBM Watson) is used to collect user emotion data through text, voice, and image analysis.

[1053] Step 3: Save the database on the server

[1054] The server stores the collected behavioral and emotional data in a database (output). Specifically, it records information such as user ID, message ID, behavior type (open, click, response), emotional state, and timestamp.

[1055] Step 4: Data Preprocessing on the Server

[1056] The server cleanses the stored behavioral and emotional data (data processing). Specifically, it complements missing values ​​and removes outliers, preparing the data for analysis (output).

[1057] Step 5: Server Feature Extraction

[1058] The server extracts features from the cleansed data (data processing). Specifically, it aggregates information such as message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week (output).

[1059] Step 6: Applying the machine learning model to the server

[1060] The server analyzes and classifies the user's behavioral and emotional patterns using machine learning algorithms (K-means clustering and linear regression) based on the extracted features (data calculation and output). For example, User A is classified into the "high reaction / positive group."

[1061] Step 7: Prepare the server message template

[1062] The server prepares message templates, including templates for new product introductions, sales announcements, newsletters, etc. (Output)

[1063] Step 8: Customizing the Server

[1064] The server customizes templates based on each user's behavioral and emotional patterns, inserting specific keywords and product categories and adjusting the tone and content of messages depending on their emotional state (data processing, output).

[1065] Step 9: Server Message Generation

[1066] The server automatically generates (outputs) a message optimized for each user based on the customized template. For example, a message such as "Check it out now! The latest smartphone has been released. We also have a special discount code!" is generated.

[1067] Step 10: Deciding when the server should send

[1068] The server determines the optimal sending time based on the user's past data (data calculation, output). For example, if user A is determined to be most responsive at 9:00 AM, the server will schedule the message to be sent at 9:00 AM.

[1069] Step 11: Server Sends Message

[1070] The server uses the LINE API to send the generated optimized message to the user at the determined timing (output).

[1071] Step 12: User receives message and takes action

[1072] The user opens the received message and takes action such as clicking a link (input).

[1073] Step 13: Real-time Server Monitoring

[1074] The server monitors users' reactions to the sent messages in real time and collects (outputs) reaction data, such as whether the message was opened, whether a link was clicked, and the content of the response.

[1075] Step 14: Save the server's reaction data

[1076] The server saves the user's reaction data in a database (output), which is then stored as material for the next analysis.

[1077] Step 15: Retraining the model on the server

[1078] The server retrains the machine learning model based on the new reaction data to improve the accuracy of the next message optimization (data calculation, output). Specifically, it learns new behavioral and emotional patterns and improves the accuracy of the model.

[1079] (Application example 2)

[1080] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1081] Conventional advertising delivery systems optimize messages based solely on user behavioral data, which means they are unable to generate effective messages that reflect the user's emotional state. Furthermore, relying solely on user behavioral data makes it difficult to accurately reflect the user's real-time interests and emotional changes, limiting the effectiveness of advertising.

[1082] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting user behavioral data and emotional data, means for analyzing the collected behavioral data and emotional data and identifying the user's behavioral patterns and emotional patterns, means for generating documents optimized for each user based on the identified behavioral patterns and emotional patterns, means for sending the generated documents to the user, and means for collecting user reaction data and providing feedback to improve the accuracy of the analysis means. This makes it possible to integrate the user's behavioral data and emotional data and generate and deliver individually optimized advertising messages.

[1083] "User behavioral data" refers to data generated when a user behaves online, such as message open rates, link click rates, and the content of past communications.

[1084] "Emotional data" is data that indicates a user's emotional state collected through text analysis, voice analysis, and image analysis.

[1085] "Behavioral patterns" refer to specific behavioral tendencies and habits that are analyzed based on user behavior data.

[1086] An "emotion pattern" indicates the tendency and changes in the user's unique emotions, which are analyzed based on the emotion data.

[1087] "Optimized documents" are messages and notifications that are individually customized and created in the most effective way based on the user's behavioral and emotional patterns.

[1088] The "feedback means" is a means for collecting user response data and reflecting it in the system to improve the accuracy of the analysis means.

[1089] The present invention is an advertising distribution system that collects, analyzes, and optimizes user behavioral and emotional data. This system is implemented using a server, user terminals (smartphones, etc.), and a network. Specifically, the invention is implemented with the following configuration.

[1090] 1. Data Collection

[1091] The server collects user behavioral and emotional data in real time. Specifically, it uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also uses an emotion engine to obtain user emotional data through text, voice, and image analysis. This data is then stored in a database.

[1092] 2. Data analysis

[1093] The server cleanses the collected behavioral and emotional data, fills in missing values, removes outliers, and extracts features through feature extraction. It then uses machine learning models (e.g., clustering and regression analysis) to classify and predict user behavioral and emotional patterns.

[1094] 3. Individual optimization

[1095] The server uses multiple pre-prepared message templates and customizes them individually based on each user's behavioral and emotional patterns, adjusting the tone and content of the message according to the user's emotional state to generate individually optimized documents.

[1096] 4. Sending a message

[1097] The server determines the optimal sending time based on the analysis results. For example, if past data indicates that a particular user is most likely to respond at 9:00 a.m., it sets the message to be sent at that time. The server then uses the LINE API to send the generated optimized message to the user.

[1098] 5. Response monitoring and feedback

[1099] The server monitors user responses to sent messages (open rates, click rates, response content) in real time and stores the response data in a database. This data is used to retrain the machine learning model and improve the accuracy of the next message optimization.

[1100] Specific examples

[1101] When User A shows interest in a certain product, the server collects their behavioral and emotional data in real time. If User A sends a message saying, "I love this product!", the emotional data indicates a positive reaction. Based on this information, the server generates a customized message for User A saying, "Check it out now! The latest smartphone has been released. We also have a special discount code!" and sends it to User A at 9:00 AM.

[1102] An example of a prompt used by a generative AI model is, "If user A expresses positive feelings about a particular product, generate a message containing the latest information and special discounts related to that product."

[1103] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1104] Step 1:

[1105] It is a means of collecting message open rates, link click rates, and past communication content from users' devices using the LINE API. The input data is user behavioral data and emotional data, and the output is the collected data. In particular, text analysis analyzes the content of users' messages and generates emotional data such as positive or negative to determine their emotional state. In addition, an emotion engine is used to perform voice and image analysis to collect user emotional data.

[1106] Step 2:

[1107] The server stores the collected behavioral and emotional data in a database. This database records user IDs, message IDs, behavior types (open, click, response), emotional states, timestamps, etc. The input data is the collected user behavioral and emotional data, and the output is structured data organized in the database. Specific operations include classifying and structuring the information.

[1108] Step 3:

[1109] The server performs data preprocessing, which includes data cleansing, missing value imputation, and outlier removal. The input data is raw data stored in a database, and the output is clean data. In particular, Pandas is used to clean the data.

[1110] Step 4:

[1111] The server performs feature extraction to extract features from user behavioral and emotional data. The input data is preprocessed clean data, and the output is extracted features. Specific operations include compiling message open rates and click rates over the past week, the most responsive times, and emotional change patterns.

[1112] Step 5:

[1113] The server uses a machine learning model to classify and predict the user's behavioral and emotional patterns based on the extracted features. The input data are the features, and the output is the user's behavioral and emotional patterns. Specific operations include pattern classification using clustering and regression analysis.

[1114] Step 6:

[1115] The server uses multiple pre-prepared message templates to generate documents optimized for each user. The input data are classified behavioral and emotional patterns, and the output is an optimized message. Specific operations include template customization and keyword insertion.

[1116] Step 7:

[1117] The server sends the generated optimized message to the user. The message is sent using the LINE API. The input data is the optimized message and the sending timing, and the output is the sent message. Specific operations include determining the sending time and delivering the message.

[1118] Step 8:

[1119] The server monitors user responses in real time and collects response data. The input data is the sent message and the user's response, and the output is the collected response data. Specific operations include monitoring the open rate, click rate, and response.

[1120] Step 9:

[1121] The server stores the collected reaction data in a database and provides a feedback mechanism for retraining the machine learning model. The input data is the collected reaction data, and the output is an updated machine learning model. Specific operations include learning new data and improving the accuracy of the model.

[1122] 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 a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1123] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1124] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[1125] [Third embodiment]

[1126] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1127] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[1128] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1129] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[1130] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1131] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1132] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1133] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1134] The specific processing program 56 is an example of a "program" according to the technology of the present 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.

[1135] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1136] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. 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 process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1137] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. 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."

[1138] The present invention relates to a system for collecting and analyzing user behavior data for the administrator of an official LINE account, and generating and sending messages optimized for each user.

[1139] 1. Data Collection

[1140] User:Activity Log

[1141] A user takes an action such as opening a message, clicking, or responding to the official LINE account.

[1142] Server: Collecting data logs

[1143] The server connects with the official LINE account and collects user behavior data in real time, specifically using technologies such as the LINE API to collect information such as message open rates, link click rates, and past communication content.

[1144] Server: Save to database

[1145] The collected behavioral data is stored in a database, including user ID, message ID, behavior type (open, click, response), timestamp, etc.

[1146] 2. Data analysis

[1147] Server: Data preprocessing

[1148] Cleanse the stored behavioral data, fill in missing values, and remove outliers.

[1149] Server:Feature Extraction

[1150] Extract features from user behavior patterns. For example, extract data such as message open rates, click rates, and the most responsive times for the past week.

[1151] Server: Applying machine learning models

[1152] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict user behavior patterns, allowing us to understand the characteristics of each user.

[1153] 3. Individual optimization

[1154] Server: Prepare message template

[1155] It provides multiple pre-made message templates, such as new product introductions, sales announcements, newsletters, etc.

[1156] Server:Customization

[1157] Customize templates based on each user's behavioral patterns, for example, inserting specific keywords or product categories for each user.

[1158] Server: Message Creation

[1159] Automatically generate personalized messages for each user based on customized templates.

[1160] 4. Sending a message

[1161] Server: Decide when to send

[1162] Based on the analysis results, the optimal sending time for each user is determined. If, based on past data, User A is judged to be most responsive at 9:00 AM, the message will be sent at that time.

[1163] Server: Send message

[1164] Using the LINE API, etc., the generated optimized message is sent to the user at the determined time.

[1165] 5. Monitoring the reaction

[1166] Users: Message Reception and Actions

[1167] Users receive the messages sent and take action such as opening, clicking, or responding.

[1168] Server: Real-time monitoring

[1169] The server monitors users' reactions to the sent messages in real time and collects reaction data.

[1170] 6. Feedback

[1171] Server: Reaction data storage

[1172] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[1173] Server: Retrain the model

[1174] Based on the new reaction data collected, the machine learning model is retrained to improve the accuracy of the next message optimization.

[1175] Specific examples

[1176] Example 1: New product introduction

[1177] Data collection

[1178] Behavioral data is collected showing that User A has shown interest in "electronic devices" in the past.

[1179] Data analysis

[1180] It is analyzed that user A is often interested in messages containing the keyword "new product announcement."

[1181] Individual optimization

[1182] Generate a message for User A saying, "Check it out now! The latest smartphones have been released and we have special discount codes for you!"

[1183] Sending a message

[1184] Send the message at 9:00 AM. Based on past data, this time is the most responsive time for User A.

[1185] Reaction monitoring

[1186] User A opens the message and clicks the link to view product details.

[1187] feedback

[1188] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[1189] In this way, the present invention maximizes the marketing effectiveness of official LINE by generating and sending optimal messages based on each user's behavioral patterns.

[1190] The processing flow will be explained below.

[1191] Step 1: Data collection

[1192] 1. User: Activity Record

[1193] A user opens a message from the official LINE account, clicks a link, or responds to a message.

[1194] 2. Server: Collecting data logs

[1195] The server collects user behavior data in real time through the LINE API, including message open rates, link click rates, and past communication content.

[1196] 3. Server: Save to database

[1197] The collected behavioral data is organized in a database, and the user ID, message ID, behavior type (open, click, response), timestamp, etc. are recorded.

[1198] Step 2: Data analysis

[1199] 1. Server: Data preprocessing

[1200] Cleanse the stored behavioral data, fill in missing values, and remove outliers.

[1201] 2. Server: Feature Extraction

[1202] Extract features from user behavior data. For example, aggregate message open rates, click rates, and the most responsive times for the past week.

[1203] 3. Server: Applying machine learning models

[1204] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict user behavior patterns.

[1205] Step 3: Individual optimization

[1206] 1. Server: Prepare message template

[1207] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[1208] 2. Server: Customization

[1209] Customize templates individually based on each user's behavioral patterns, for example by inserting specific keywords or product categories to optimize message content.

[1210] 3. Server: Message Creation

[1211] Generate personalized messages for each user based on customized templates.

[1212] Step 4: Send the message

[1213] 1. Server: Decide when to send

[1214] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[1215] 2. Server: Send message

[1216] Use the LINE API to send the generated optimized message to the user at the determined time.

[1217] Step 5: Monitoring the reaction

[1218] 1. Users: Receiving messages and taking action

[1219] Users open the messages they receive and click on links within them or reply to them.

[1220] 2. Server: Real-time monitoring

[1221] User responses to sent messages (e.g., open rates, click rates, response content) are monitored in real time and response data is collected.

[1222] Step 6: Feedback

[1223] 1. Server: Reaction data storage

[1224] The collected reaction data is stored in a database and accumulated as material for the next analysis.

[1225] 2. Server: Retrain the model

[1226] Retrain your machine learning model based on new response data to improve the accuracy of your next message optimization.

[1227] In this way, through each processing step, the system can optimize messages from the official LINE account for each individual user, allowing for more effective communication of information.

[1228] Example 1

[1229] Next, a description will be given of Example 1. 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."

[1230] Conventional messaging systems have had the problem of being unable to deliver effective marketing due to the difficulty in sending messages optimized for each user. Furthermore, they lacked a feedback mechanism for effectively analyzing user behavior data and utilizing the results in the next messaging. Therefore, a system that can maximize the effectiveness of marketing is needed.

[1231] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1232] In this invention, the server includes means for collecting user behavioral data, means for cleansing the collected behavioral data, completing missing values ​​and removing outliers, means for performing feature extraction to extract features from user behavior patterns, means for applying a machine learning algorithm to classify and predict user behavioral patterns, means for preparing multiple message templates, customizing the templates based on each user's behavioral pattern, and generating an optimal message, means for determining an optimal sending time and sending the generated message to the user, means for collecting user reaction data to the sent message and saving it as material for subsequent analysis and optimization, and feedback means for retraining a machine learning model based on the collected reaction data and improving the accuracy of message optimization, thereby enabling the generation and sending of optimal messages based on each user's behavioral pattern.

[1233] "User behavior data" is data recorded when a user takes a specific action, such as opening a message, clicking a link, or responding.

[1234] "Data cleansing" is the process of imputing missing values ​​and removing outliers from collected data.

[1235] "Feature extraction" is the process of extracting features from user behavior patterns, which makes analysis and prediction easier.

[1236] A "machine learning algorithm" is an algorithm that learns patterns and regularities from data and makes future predictions and classifications.

[1237] A "message template" is a pre-defined framework for message format and content, allowing for quick message generation and customization.

[1238] The "optimal sending time" is the time at which a message is sent that is most likely to be responsive to each user, based on collected data and analysis results.

[1239] "Response data" is a record of the actions users take in response to messages they receive, and is used as material for future analysis and optimization.

[1240] "Feedback measures" are processes that retrain machine learning models based on collected response data to improve the accuracy of the next message optimization.

[1241] This invention is a system for administrators of official messaging accounts that collects and analyzes user behavior data, and generates and sends individually optimized messages. The aim is to effectively utilize user behavior data and maximize the marketing effectiveness of messages.

[1242] This system is mainly composed of three elements: a server, a terminal, and a user. Below, we will explain the flow of data processing and data calculation using the specific hardware and software of each element.

[1243] Data collection

[1244] User:Activity Log

[1245] Users take actions such as opening, clicking, or responding to messages on official messaging accounts, and these actions act as triggers for the system.

[1246] Server: Collecting data logs

[1247] The server obtains user behavior data in real time via a messaging API (e.g., the LINE API), including message open rates, link click rates, and past communication content.

[1248] Server: Save to database

[1249] The collected behavioral data is stored in a database along with information such as user ID, message ID, behavior type (open, click, response), and timestamp.

[1250] Data analysis

[1251] Server: Data preprocessing

[1252] Stored behavioral data is cleansed, missing values ​​are filled, and outliers are removed, ensuring data accuracy and reliability.

[1253] Server:Feature Extraction

[1254] Extract features from user behavior patterns, such as message open rates and click rates over the past week, and the time of day when responses are most likely to occur.

[1255] Server: Applying machine learning models

[1256] Based on the extracted features, machine learning algorithms (e.g., clustering, regression analysis, etc.) are applied to classify and predict user behavior patterns, allowing for a detailed understanding of each user's characteristics.

[1257] Individual optimization

[1258] Server: Prepare message template

[1259] The server uses a number of pre-defined message templates (e.g., new product introductions, sales announcements, newsletters, etc.).

[1260] Server:Customization

[1261] The template is customized based on each user's behavioral patterns, for example, inserting relevant keywords for a user interested in "electronic devices."

[1262] Server: Message Creation

[1263] It automatically generates personalized messages for each user based on customized templates, using a generative AI model and providing specific prompts.

[1264] Sending a message

[1265] Server: Decide when to send

[1266] Based on the analysis results, the optimal sending time for each user is determined. For example, if past data indicates that User A responds best at 9:00 AM, then that time is set.

[1267] Server: Send message

[1268] The server uses the API to send the generated optimization message to the user at the determined timing.

[1269] Reaction monitoring

[1270] Users: Message Reception and Actions

[1271] Users receive the messages sent and take action such as opening, clicking, or responding.

[1272] Server: Real-time monitoring

[1273] The server monitors users' reactions to the sent messages in real time and stores the reaction data in a database.

[1274] feedback

[1275] Server: Reaction data storage

[1276] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[1277] Server: Retrain the model

[1278] Based on the new reaction data collected, the machine learning model is retrained to improve the accuracy of message optimization.

[1279] An example of a specific prompt might be, "Generate a message for a new product launch taking into account user behavioral data. Use keywords that have generated positive responses in past messages, and include content that will increase open rates."

[1280] In this way, the present invention is a system that generates and sends optimal messages based on each user's behavioral patterns, maximizing the marketing effectiveness of official messaging accounts.

[1281] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1282] Step 1:

[1283] Data collection

[1284] User:Activity Log

[1285] Users open messages, click links, and send responses to official messaging accounts, and these actions trigger data collection.

[1286] Input: User actions (opens, clicks, responses)

[1287] Output: When an action occurs, the data is collected immediately.

[1288] Server: Collecting data logs

[1289] The server obtains user behavior data in real time via a messaging API (e.g., LINE API).

[1290] Input: Behavioral data captured by the messaging API

[1291] Output: Raw behavioral data (open rates, click rates, communication content, etc.)

[1292] Server: Save to database

[1293] The server stores the collected behavioral data in a database, which includes the user ID, message ID, behavior type, and timestamp.

[1294] Input: Raw behavioral data

[1295] Output: Behavioral data stored in a database

[1296] Step 2:

[1297] Data Preprocessing

[1298] Server: Data Cleansing

[1299] The server cleanses the stored data, imputing missing values ​​and removing outliers.

[1300] Input: Stored behavioral data

[1301] Output: Cleansed data

[1302] Specific behavior: For example, gracefully handling time zone inconsistencies and extreme values.

[1303] Step 3:

[1304] Data analysis

[1305] Server:Feature Extraction

[1306] The server extracts features from user behavior patterns, such as message open rates and click rates over the past week, and the time periods with the highest response rates.

[1307] Input: Cleansed data

[1308] Output: Extracted features

[1309] Specific operations: Analyze user behavior patterns using time series analysis and statistical methods.

[1310] Server: Applying machine learning models

[1311] The server applies a machine learning algorithm based on the extracted features to classify and predict user behavior patterns.

[1312] Input: Feature data

[1313] Output: Classified user groups and predicted behavior patterns

[1314] Specific behavior: Uses clustering and regression analysis algorithms.

[1315] Step 4:

[1316] Individual optimization

[1317] Server: Prepare message template

[1318] The server uses a number of pre-defined message templates.

[1319] Input: Template list

[1320] Output: Available templates

[1321] Specific actions: Select and update templates regularly.

[1322] Server:Customization

[1323] The server customizes the template based on each user's behavioral patterns, for example, by inserting keywords related to "electronic devices" to match the user's interests.

[1324] Input: Template, Categorized User Groups

[1325] Output: Customized template

[1326] Specific behavior: Modify specific parts of the template to suit the user's interests.

[1327] Server: Message Creation

[1328] The server automatically generates messages optimized for each user based on customized templates.

[1329] Input:Customized template

[1330] Output: Generated optimization messages

[1331] What it does: Uses a generative AI model to generate an appropriate message based on the prompt.

[1332] Step 5:

[1333] Sending a message

[1334] Server: Decide when to send

[1335] The server determines the optimal transmission time for each user from the analysis results.

[1336] Input: User's optimal transmission time data

[1337] Output: Transmission schedule

[1338] Specific behavior: Optimize sending times based on past behavioral data.

[1339] Server: Send message

[1340] The server sends the optimization message generated using the API to the user at the determined timing.

[1341] Input: Generated message, send schedule

[1342] Output: Message sent

[1343] Specific operation: The actual message sending is carried out using the API.

[1344] Step 6:

[1345] Reaction monitoring

[1346] Users: Message Reception and Actions

[1347] The user opens the received message, clicks on the link, and sends a response.

[1348] Input: Optimized message

[1349] Output: User response data (behavior)

[1350] What it does: Collect user feedback in real time.

[1351] Server: Real-time monitoring

[1352] The server monitors users' reactions to the sent messages in real time and stores the reaction data in a database.

[1353] Input: User response data

[1354] Output: Saved reaction data

[1355] Specific actions: Monitor using real-time data analysis tools.

[1356] Step 7:

[1357] feedback

[1358] Server: Reaction data storage

[1359] The server stores the user's reaction data in a database and accumulates it as material for the next analysis.

[1360] Input: User response data

[1361] Output: Saved reaction data

[1362] Specific operation: The saving process is carried out using a database management system.

[1363] Server: Retrain the model

[1364] The server retrains the machine learning model based on the new reaction data collected, improving the accuracy of the next message optimization.

[1365] Input: New reaction data

[1366] Output: A retrained machine learning model

[1367] What it does: Continually improve the model using adaptive learning techniques.

[1368] This makes it possible to generate and send optimal messages based on each user's behavioral patterns, maximizing marketing effectiveness.

[1369] (Application example 1)

[1370] Next, a description will be given of Application Example 1. 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."

[1371] Conventional user behavior data collection and analysis systems have been unable to achieve effective marketing due to insufficient optimization of messages for individual users. In particular, generic messages are often sent without sufficient consideration of the different behavioral patterns of each user, limiting the effectiveness of improving user response rates and purchasing intent. Furthermore, the system lacks a mechanism for efficiently feeding back response data to messages sent and improving the accuracy of the model.

[1372] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1373] In this invention, the server includes means for collecting user behavioral data, means for analyzing the collected behavioral data and identifying user behavior patterns, means for generating a document optimized for each user based on the identified behavioral pattern, means for sending the generated document to the user, means for collecting user reaction data and providing feedback to improve the accuracy of the analysis means, means for using a machine learning algorithm to perform clustering based on the collected behavioral data, and means for selecting a message template optimized for each user cluster and generating a customized message, thereby enabling effective message optimization based on user behavior patterns.

[1374] "User behavior data" refers to records of various interactions users have with the system, including, for example, message open rates, link click rates, and the content of past communications.

[1375] "Analysis Tools" refers to the functions and algorithms used to process and analyze collected behavioral data and identify user behavior patterns.

[1376] "Behavioral patterns" refer to models or tendencies that indicate how a user will behave based on the user's past behavioral data.

[1377] "Optimized documents" refers to messages or documents that are customized to best suit a user based on the user's behavioral patterns.

[1378] "Feedback means" refers to a data collection and processing mechanism for improving the analytical accuracy of the system and the performance of the machine learning model based on user response data.

[1379] "Clustering" refers to a technique that uses machine learning algorithms to classify users into groups with similar behavioral patterns.

[1380] A "machine learning algorithm" is a part of a computer program that analyzes large amounts of data and automatically learns patterns and predictive models.

[1381] A "template" is a pre-created message template for a specific purpose, which serves as the basis for customization.

[1382] "Customization" refers to tailoring documents and messages to specific user behavior patterns.

[1383] The present invention relates to a system for generating and sending optimized messages based on user behavior data. The system is implemented in the following steps:

[1384] First, the server uses the LINE API to collect user behavior data, including message open rates, link click rates, and past communication content. The collected data is sent to the server in real time and stored in a database.

[1385] Once the data is collected, the server pre-processes it, cleansing it to fill in missing values ​​and remove outliers, then performs feature extraction to extract parameters that characterize user behavior patterns.

[1386] The server then uses machine learning algorithms, specifically clustering algorithms, to classify users into clusters with similar behavioral patterns, leveraging Python's scikit-learn library and other machine learning tools.

[1387] After clustering is complete, the server applies pre-prepared template messages to each cluster. Each template is customized to fit the behavioral patterns of a specific user. The customized message is then sent to the user in the most effective way. For example, user A might receive a message like, "Hello, would you like to check out our latest products?"

[1388] After the message is sent, user reaction data is collected again and sent to the server. This reaction data is stored in a database and used as a feedback mechanism to improve the accuracy of the analysis method. The new data is used to retrain the machine learning model, improving the accuracy of the next message optimization.

[1389] As a concrete example, suppose there is data showing that User A has clicked on messages related to beauty products in the past. In this case, the machine learning model will classify User A into a cluster interested in beauty products and use a template message for that cluster to send a customized message such as "Don't miss out on our great beauty product campaigns!"

[1390] Examples of prompts include:

[1391] Generate customized product introduction messages according to the cluster. Cluster 1: For users interested in beauty products

[1392] Don't miss out on our great beauty deals!

[1393] Cluster 2: For users interested in home appliances

[1394] The latest home appliances have arrived! Check them out now.

[1395] In this way, the present invention maximizes marketing effectiveness by utilizing user behavior data to generate and send messages optimized for individual users.

[1396] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1397] Step 1:

[1398] The server collects user behavior data via the LINE API, including message open rates, link click rates, and past communication content. The collected data is sent to the server in real time and stored in a database.

[1399] Input: User behavior data obtained via the LINE API

[1400] Output: Behavioral data stored in a database

[1401] Specific operation: The server sends a request to the LINE API and executes an SQL query to save the retrieved data to the database.

[1402] Step 2:

[1403] The server pre-processes the stored behavioral data, performing data cleansing, filling in missing values, and removing outliers.

[1404] Input: Raw data stored in a database

[1405] Output: Cleansed behavioral data

[1406] Specific operation: The server reads data from the database and uses the Pandas library to impute missing values ​​and remove outliers.

[1407] Step 3:

[1408] The server performs feature extraction from the preprocessed data, extracting features such as the open rate, click rate, and best response time for each user.

[1409] Input: Cleansed behavioral data

[1410] Output: A dataset with extracted features

[1411] What it does: The server groups the data by user ID, calculates certain statistics, and generates new features, such as how many times a message was opened or clicked each day.

[1412] Step 4:

[1413] The server performs clustering using a machine learning algorithm to classify users into clusters with similar behavioral patterns.

[1414] Input: Feature-extracted dataset

[1415] Output: User data with assigned clusters

[1416] What it does: The server uses the scikit-learn library to run the KMeans algorithm to classify users into clusters, and each user is assigned a cluster ID.

[1417] Step 5:

[1418] The server selects the best template message for each cluster and generates a customized message.

[1419] Input: User data with assigned clusters, and message template

[1420] Output: Customized message

[1421] What happens: The server selects a template for each cluster and customizes the message by inserting user-specific information.

[1422] Step 6:

[1423] The server sends the generated customized message to the user via the LINE API.

[1424] Input: Customized Message

[1425] Output: Message sent to the user's terminal

[1426] Specific operation: The server sends a POST request to the LINE API and sends the generated message to the user at the appropriate time, for example, during the time period when the user is most likely to respond.

[1427] Step 7:

[1428] The server again collects user reaction data to the sent messages and stores it in a database.

[1429] Input: User interaction data (message opens, clicks, responses, etc.)

[1430] Output: Reaction data stored in a database

[1431] Specific operation: The server retrieves the user's reaction data from the LINE API and executes an SQL query to save it in the database.

[1432] Step 8:

[1433] The server retrains the machine learning model based on the new response data to improve the accuracy of the next message optimization.

[1434] Input: New reaction data stored in the database

[1435] Output: An updated machine learning model

[1436] Specific operation: The server extracts features from new reaction data and retrains the machine learning algorithm to improve the accuracy of the model.

[1437] Through these steps, the present invention achieves effective message optimization based on user behavior data.

[1438] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1439] The present invention relates to a system for collecting and analyzing behavioral and emotional data of users for the administrator of an official LINE account, and generating and sending messages optimized for each user.

[1440] 1. Data Collection

[1441] User:Activity Log

[1442] A user takes an action such as opening a message, clicking, or responding to the official LINE account.

[1443] Server: Data logging and emotion data collection

[1444] The server connects with the official LINE account to obtain user behavioral and emotional data in real time. Specifically, it uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also uses an emotion engine to collect user emotional data through text, voice, and image analysis.

[1445] Server: Save to database

[1446] The collected behavioral and emotional data is organized in a database, and the user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc. are recorded.

[1447] 2. Data analysis

[1448] Server: Data preprocessing

[1449] Cleanse stored behavioral and emotional data to fill in missing values ​​and remove outliers.

[1450] Server:Feature Extraction

[1451] Extract features from user behavioral and emotional data. For example, aggregate message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week.

[1452] Server: Applying machine learning models

[1453] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict the user's behavioral and emotional patterns.

[1454] 3. Individual optimization

[1455] Server: Prepare message template

[1456] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[1457] Server:Customization

[1458] Templates are individually customized based on each user's behavioral and emotional patterns. For example, message content can be optimized by inserting specific keywords or product categories. The tone and content of messages can also be adjusted according to the user's emotional state.

[1459] Server: Message Creation

[1460] Automatically generate messages optimized for each user based on customized templates.

[1461] 4. Sending a message

[1462] Server: Decide when to send

[1463] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[1464] Server: Send message

[1465] Use the LINE API to send the generated optimized message to the user at the determined time.

[1466] 5. Monitoring the reaction

[1467] Users: Message Reception and Actions

[1468] Users open the messages they receive and click on links within them or reply to them.

[1469] Server: Real-time monitoring

[1470] The server monitors user responses to sent messages (e.g., open rate, click rate, response content) in real time and collects response data.

[1471] 6. Feedback

[1472] Server: Reaction data storage

[1473] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[1474] Server: Retrain the model

[1475] Retrain your machine learning model based on new reaction data collected to improve the accuracy of your next message optimization.

[1476] Specific examples

[1477] Example 1: New product introduction

[1478] Data collection

[1479] The system collects behavioral data showing that User A has shown interest in "electronic devices" in the past. The emotion engine also identifies a positive emotional state based on User A's recent messages and voice.

[1480] Data analysis

[1481] It is analyzed that User A is often interested in messages containing the keyword "new product launch." Also, since User A's emotional state is positive, it is determined that a direct sales message would be effective.

[1482] Individual optimization

[1483] Generate a positive message for User A: "Check it out now! The latest smartphones are on sale now, and we've got a special discount code for you!"

[1484] Sending a message

[1485] Send the message at 9:00 AM. Based on past data, this time is the most responsive time for User A.

[1486] Reaction monitoring

[1487] User A opens the message and clicks the link to view product details.

[1488] feedback

[1489] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[1490] In this way, the present invention maximizes the marketing effectiveness of official LINE by combining and analyzing user behavioral data and emotional data to generate and send optimal messages according to individual needs and emotional states.

[1491] The processing flow will be explained below.

[1492] Step 1: Data collection

[1493] 1. User: Activity Record

[1494] A user opens a message from the official LINE account, clicks a link, or responds to a message.

[1495] 2. Server: Collecting data logs

[1496] The server obtains user behavior data in real time through the LINE API, including message open rates, link click rates, and past communication content.

[1497] 3. Server: Emotion data collection

[1498] The emotion engine is used to collect user emotional data through text, voice, and image analysis, for example, to identify the user's emotional state from the content of their message or their tone of voice.

[1499] 4. Server: Save to database

[1500] The collected behavioral and emotional data is organized and stored in a database, including user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc.

[1501] Step 2: Data analysis

[1502] 1. Server: Data preprocessing

[1503] Cleanse stored behavioral and emotional data to fill in missing values ​​and remove outliers.

[1504] 2. Server: Feature Extraction

[1505] Extract features from user behavioral and emotional data. For example, aggregate message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week.

[1506] 3. Server: Applying machine learning models

[1507] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict the user's behavioral and emotional patterns.

[1508] Step 3: Individual optimization

[1509] 1. Server: Prepare message template

[1510] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[1511] 2. Server: Customization

[1512] Templates are individually customized based on each user's behavioral and emotional patterns. For example, message content can be optimized by inserting specific keywords or product categories. The tone and content of messages can also be adjusted according to the user's emotional state.

[1513] 3. Server: Message Creation

[1514] Automatically generate messages optimized for each user based on customized templates.

[1515] Step 4: Send the message

[1516] 1. Server: Decide when to send

[1517] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[1518] 2. Server: Send message

[1519] Use the LINE API to send the generated optimized message to the user at the determined time.

[1520] Step 5: Monitoring the reaction

[1521] 1. Users: Receiving messages and taking action

[1522] Users open the messages they receive and click on links within them or reply to them.

[1523] 2. Server: Real-time monitoring

[1524] User responses to sent messages (e.g., open rates, click rates, response content) are monitored in real time and response data is collected.

[1525] Step 6: Feedback

[1526] 1. Server: Reaction data storage

[1527] The collected reaction data is stored in a database and accumulated as material for the next analysis.

[1528] 2. Server: Retrain the model

[1529] Retrain your machine learning model based on new response data to improve the accuracy of your next message optimization.

[1530] Example 2

[1531] Next, a description will be given of 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."

[1532] Conventional message sending systems only considered user behavioral data and did not perform individual optimization using emotional data, making it difficult to maximize the effectiveness of marketing messages. Furthermore, there was also the problem of not being able to quickly reflect user reactions because the timing of sending messages was not optimized and real-time feedback was not adequately provided.

[1533] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting user behavioral data and emotional data, means for analyzing the collected behavioral data and emotional data and identifying the user's behavioral and emotional patterns, means for generating a document optimized for each user based on the identified behavioral and emotional patterns, means for sending the generated document to the user, and means for collecting user reaction data in real time and providing feedback to improve the accuracy of the analysis means. This makes it possible to generate and send individually optimized messages based on the user's behavioral and emotional data, thereby maximizing marketing effectiveness.

[1534] "User behavioral data" refers to information related to actions taken by users on official LINE accounts, such as opening messages, clicking links, and responding to messages.

[1535] "User emotional data" is information collected using text analysis, speech analysis, image analysis, etc. to identify a user's emotional state.

[1536] A "behavioral pattern" refers to the regularity or tendency of behavior that is analyzed based on data on a series of actions taken by a user.

[1537] "Emotional patterns" refer to changes and trends in emotions that are analyzed based on the user's emotional data.

[1538] "Optimized documents" are messages that are individually tailored based on a user's behavioral and emotional patterns and are generated to maximize the user's interest and response.

[1539] A "machine learning algorithm" is a computational method for learning patterns from data and making predictions or classifications based on the results.

[1540] "Clustering" is a type of machine learning technique that divides data into several groups.

[1541] "Regression analysis" is a machine learning technique that is used to model the relationship between predictor variables and response variables and make predictions.

[1542] A "feedback means" is a method or mechanism for collecting user reaction data and using that data to improve the accuracy of the analysis means and generation means.

[1543] "Optimizing sending timing" aims to send messages at the time when users are most likely to respond, and determines the timing of sending based on past behavioral data.

[1544] This invention relates to a system for official LINE administrators that collects and analyzes user behavioral and emotional data, and generates and sends messages optimized for each user.

[1545] 1. Data Collection

[1546] User behavior record

[1547] Users open messages sent to official LINE accounts, click links, and respond - all of this behavioral data is collected in real time.

[1548] Server data logging and emotion data collection

[1549] The server uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also collects user emotional data through text, voice, and image analysis by emotion engines (such as IBM Watson and Microsoft Azure).

[1550] Server database storage

[1551] The server organizes the collected behavioral and emotional data into a database, recording the user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc.

[1552] 2. Data analysis

[1553] Server data preprocessing

[1554] The server cleanses the stored behavioral and emotional data, filling in missing values ​​and removing outliers, thereby ensuring data quality.

[1555] Server Feature Extraction

[1556] The server extracts features from the behavioral and emotional data, such as message open rates, click rates, the most responsive times, and patterns of emotional change over the past week.

[1557] Applying machine learning models to servers

[1558] Based on the extracted features, the server uses machine learning algorithms (e.g., K-means clustering and linear regression) to classify and predict the user's behavioral and emotional patterns.

[1559] 3. Individual optimization

[1560] Preparing the server message template

[1561] The server prepares a number of pre-defined message templates (e.g., new product introductions, sales announcements, newsletters).

[1562] Server Customization

[1563] The server customizes the templates individually for each user based on their behavioral and emotional patterns, inserting specific keywords and product categories and adjusting the tone and content of the message depending on the user's emotional state.

[1564] Server Message Generation

[1565] The server automatically generates messages optimized for each user based on customized templates.

[1566] 4. Sending a message

[1567] Server transmission timing decision

[1568] The server then uses the analysis to determine the best time to send the message. For example, if historical data indicates that a particular user is most likely to respond at 9 a.m., it will schedule the message to be sent at that time.

[1569] Server message sending

[1570] The server uses the LINE API to send the generated optimized message to the user at the determined timing.

[1571] 5. Monitoring the reaction

[1572] User message receipt and actions

[1573] The user opens the received message, clicks on a link in the message, or makes a response.

[1574] Real-time server monitoring

[1575] The server monitors user responses to sent messages (e.g., open rate, click rate, response content) in real time and collects response data.

[1576] 6. Feedback

[1577] Saving reaction data on the server

[1578] The server stores the user's reaction data in a database and accumulates it as material for the next analysis.

[1579] Retraining the model on the server

[1580] The server retrains the machine learning model based on the new reaction data collected, improving the accuracy of the next message optimization.

[1581] Specific examples

[1582] Example 1: New product introduction

[1583] Behavioral data showing that User A has shown interest in "electronic devices" in the past is collected. In addition, analysis by an emotion engine (e.g., IBM Watson) identifies a positive emotional state based on User A's recent messages and voice.

[1584] Data analysis

[1585] It is analyzed that User A is often interested in messages containing the keyword "new product launch." Also, since User A's emotional state is positive, it is determined that a direct sales message would be effective.

[1586] Individual optimization

[1587] Generate a positive message for User A: "Check it out now! The latest smartphones are on sale now, and we've got a special discount code for you!"

[1588] Sending a message

[1589] Based on past data, it is determined that User A responds best at 9:00 AM, so the message is sent at this time.

[1590] Reaction monitoring

[1591] User A opens the message and clicks the link to view product details.

[1592] feedback

[1593] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[1594] Prompt Sentence Examples

[1595] Here are some example prompts to enter into a generative AI model:

[1596] "User A has shown interest in 'electronic devices' in the past, and analysis of the emotion engine reveals that he is in a positive emotional state. Please create a message about a new product launch that will interest User A."

[1597] Using this prompt, the generative AI model generates a message optimized for User A.

[1598] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1599] Step 1: Recording user behavior

[1600] A user opens a message from an official LINE account, clicks a link, and responds. This behavioral data is sent to the LINE server (input).

[1601] Step 2: Collecting server data logs and emotion data

[1602] The server works with the official LINE app to call the LINE API and collect (output) user behavior data in real time. Specifically, behavioral data acquired includes message open rates, link click rates, and past communication content. Additionally, an emotion engine (e.g., IBM Watson) is used to collect user emotion data through text, voice, and image analysis.

[1603] Step 3: Save the database on the server

[1604] The server stores the collected behavioral and emotional data in a database (output). Specifically, it records information such as user ID, message ID, behavior type (open, click, response), emotional state, and timestamp.

[1605] Step 4: Data Preprocessing on the Server

[1606] The server cleanses the stored behavioral and emotional data (data processing). Specifically, it complements missing values ​​and removes outliers, preparing the data for analysis (output).

[1607] Step 5: Server Feature Extraction

[1608] The server extracts features from the cleansed data (data processing). Specifically, it aggregates information such as message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week (output).

[1609] Step 6: Applying the machine learning model to the server

[1610] The server analyzes and classifies the user's behavioral and emotional patterns using machine learning algorithms (K-means clustering and linear regression) based on the extracted features (data calculation and output). For example, User A is classified into the "high reaction / positive group."

[1611] Step 7: Prepare the server message template

[1612] The server prepares message templates, including templates for new product introductions, sales announcements, newsletters, etc. (Output)

[1613] Step 8: Customizing the Server

[1614] The server customizes templates based on each user's behavioral and emotional patterns, inserting specific keywords and product categories and adjusting the tone and content of messages depending on their emotional state (data processing, output).

[1615] Step 9: Server Message Generation

[1616] The server automatically generates (outputs) a message optimized for each user based on the customized template. For example, a message such as "Check it out now! The latest smartphone has been released. We also have a special discount code!" is generated.

[1617] Step 10: Deciding when the server should send

[1618] The server determines the optimal sending time based on the user's past data (data calculation, output). For example, if user A is determined to be most responsive at 9:00 AM, the server will schedule the message to be sent at 9:00 AM.

[1619] Step 11: Server Sends Message

[1620] The server uses the LINE API to send the generated optimized message to the user at the determined timing (output).

[1621] Step 12: User receives message and takes action

[1622] The user opens the received message and takes action such as clicking a link (input).

[1623] Step 13: Real-time Server Monitoring

[1624] The server monitors users' reactions to the sent messages in real time and collects (outputs) reaction data, such as whether the message was opened, whether a link was clicked, and the content of the response.

[1625] Step 14: Save the server's reaction data

[1626] The server saves the user's reaction data in a database (output), which is then stored as material for the next analysis.

[1627] Step 15: Retraining the model on the server

[1628] The server retrains the machine learning model based on the new reaction data to improve the accuracy of the next message optimization (data calculation, output). Specifically, it learns new behavioral and emotional patterns and improves the accuracy of the model.

[1629] (Application example 2)

[1630] Next, a description will be given of Application 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."

[1631] Conventional advertising delivery systems optimize messages based solely on user behavioral data, which means they are unable to generate effective messages that reflect the user's emotional state. Furthermore, relying solely on user behavioral data makes it difficult to accurately reflect the user's real-time interests and emotional changes, limiting the effectiveness of advertising.

[1632] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting user behavioral data and emotional data, means for analyzing the collected behavioral data and emotional data and identifying the user's behavioral patterns and emotional patterns, means for generating documents optimized for each user based on the identified behavioral patterns and emotional patterns, means for sending the generated documents to the user, and means for collecting user reaction data and providing feedback to improve the accuracy of the analysis means. This makes it possible to integrate the user's behavioral data and emotional data and generate and deliver individually optimized advertising messages.

[1633] "User behavioral data" refers to data generated when a user behaves online, such as message open rates, link click rates, and the content of past communications.

[1634] "Emotional data" is data that indicates a user's emotional state collected through text analysis, voice analysis, and image analysis.

[1635] "Behavioral patterns" refer to specific behavioral tendencies and habits that are analyzed based on user behavior data.

[1636] An "emotion pattern" indicates the tendency and changes in the user's unique emotions, which are analyzed based on the emotion data.

[1637] "Optimized documents" are messages and notifications that are individually customized and created in the most effective way based on the user's behavioral and emotional patterns.

[1638] The "feedback means" is a means for collecting user response data and reflecting it in the system to improve the accuracy of the analysis means.

[1639] The present invention is an advertising distribution system that collects, analyzes, and optimizes user behavioral and emotional data. This system is implemented using a server, user terminals (smartphones, etc.), and a network. Specifically, the invention is implemented with the following configuration.

[1640] 1. Data Collection

[1641] The server collects user behavioral and emotional data in real time. Specifically, it uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also uses an emotion engine to obtain user emotional data through text, voice, and image analysis. This data is then stored in a database.

[1642] 2. Data analysis

[1643] The server cleanses the collected behavioral and emotional data, fills in missing values, removes outliers, and extracts features through feature extraction. It then uses machine learning models (e.g., clustering and regression analysis) to classify and predict user behavioral and emotional patterns.

[1644] 3. Individual optimization

[1645] The server uses multiple pre-prepared message templates and customizes them individually based on each user's behavioral and emotional patterns, adjusting the tone and content of the message according to the user's emotional state to generate individually optimized documents.

[1646] 4. Sending a message

[1647] The server determines the optimal sending time based on the analysis results. For example, if past data indicates that a particular user is most likely to respond at 9:00 a.m., it sets the message to be sent at that time. The server then uses the LINE API to send the generated optimized message to the user.

[1648] 5. Response monitoring and feedback

[1649] The server monitors user responses to sent messages (open rates, click rates, response content) in real time and stores the response data in a database. This data is used to retrain the machine learning model and improve the accuracy of the next message optimization.

[1650] Specific examples

[1651] When User A shows interest in a certain product, the server collects their behavioral and emotional data in real time. If User A sends a message saying, "I love this product!", the emotional data indicates a positive reaction. Based on this information, the server generates a customized message for User A saying, "Check it out now! The latest smartphone has been released. We also have a special discount code!" and sends it to User A at 9:00 AM.

[1652] An example of a prompt used by a generative AI model is, "If user A expresses positive feelings about a particular product, generate a message containing the latest information and special discounts related to that product."

[1653] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1654] Step 1:

[1655] It is a means of collecting message open rates, link click rates, and past communication content from users' devices using the LINE API. The input data is user behavioral data and emotional data, and the output is the collected data. In particular, text analysis analyzes the content of users' messages and generates emotional data such as positive or negative to determine their emotional state. In addition, an emotion engine is used to perform voice and image analysis to collect user emotional data.

[1656] Step 2:

[1657] The server stores the collected behavioral and emotional data in a database. This database records user IDs, message IDs, behavior types (open, click, response), emotional states, timestamps, etc. The input data is the collected user behavioral and emotional data, and the output is structured data organized in the database. Specific operations include classifying and structuring the information.

[1658] Step 3:

[1659] The server performs data preprocessing, which includes data cleansing, missing value imputation, and outlier removal. The input data is raw data stored in a database, and the output is clean data. In particular, Pandas is used to clean the data.

[1660] Step 4:

[1661] The server performs feature extraction to extract features from user behavioral and emotional data. The input data is preprocessed clean data, and the output is extracted features. Specific operations include compiling message open rates and click rates over the past week, the most responsive times, and emotional change patterns.

[1662] Step 5:

[1663] The server uses a machine learning model to classify and predict the user's behavioral and emotional patterns based on the extracted features. The input data are the features, and the output is the user's behavioral and emotional patterns. Specific operations include pattern classification using clustering and regression analysis.

[1664] Step 6:

[1665] The server uses multiple pre-prepared message templates to generate documents optimized for each user. The input data are classified behavioral and emotional patterns, and the output is an optimized message. Specific operations include template customization and keyword insertion.

[1666] Step 7:

[1667] The server sends the generated optimized message to the user. The message is sent using the LINE API. The input data is the optimized message and the sending timing, and the output is the sent message. Specific operations include determining the sending time and delivering the message.

[1668] Step 8:

[1669] The server monitors user responses in real time and collects response data. The input data is the sent message and the user's response, and the output is the collected response data. Specific operations include monitoring the open rate, click rate, and response.

[1670] Step 9:

[1671] The server stores the collected reaction data in a database and provides a feedback mechanism for retraining the machine learning model. The input data is the collected reaction data, and the output is an updated machine learning model. Specific operations include learning new data and improving the accuracy of the model.

[1672] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1673] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1674] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1675] [Fourth embodiment]

[1676] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1677] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1678] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1679] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1680] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1681] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1682] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1683] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1684] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1685] The specific processing program 56 is an example of a "program" according to the technology of the present 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.

[1686] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1687] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. 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 process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1688] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1689] The present invention relates to a system for collecting and analyzing user behavior data for the administrator of an official LINE account, and generating and sending messages optimized for each user.

[1690] 1. Data Collection

[1691] User:Activity Log

[1692] A user takes an action such as opening a message, clicking, or responding to the official LINE account.

[1693] Server: Collecting data logs

[1694] The server connects with the official LINE account and collects user behavior data in real time, specifically using technologies such as the LINE API to collect information such as message open rates, link click rates, and past communication content.

[1695] Server: Save to database

[1696] The collected behavioral data is stored in a database, including user ID, message ID, behavior type (open, click, response), timestamp, etc.

[1697] 2. Data analysis

[1698] Server: Data preprocessing

[1699] Cleanse the stored behavioral data, fill in missing values, and remove outliers.

[1700] Server:Feature Extraction

[1701] Extract features from user behavior patterns. For example, extract data such as message open rates, click rates, and the most responsive times for the past week.

[1702] Server: Applying machine learning models

[1703] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict user behavior patterns, allowing us to understand the characteristics of each user.

[1704] 3. Individual optimization

[1705] Server: Prepare message template

[1706] It provides multiple pre-made message templates, such as new product introductions, sales announcements, newsletters, etc.

[1707] Server:Customization

[1708] Customize templates based on each user's behavioral patterns, for example, inserting specific keywords or product categories for each user.

[1709] Server: Message Creation

[1710] Automatically generate personalized messages for each user based on customized templates.

[1711] 4. Sending a message

[1712] Server: Decide when to send

[1713] Based on the analysis results, the optimal sending time for each user is determined. If, based on past data, User A is judged to be most responsive at 9:00 AM, the message will be sent at that time.

[1714] Server: Send message

[1715] Using the LINE API, etc., the generated optimized message is sent to the user at the determined time.

[1716] 5. Monitoring the reaction

[1717] Users: Message Reception and Actions

[1718] Users receive the messages sent and take action such as opening, clicking, or responding.

[1719] Server: Real-time monitoring

[1720] The server monitors users' reactions to the sent messages in real time and collects reaction data.

[1721] 6. Feedback

[1722] Server: Reaction data storage

[1723] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[1724] Server: Retrain the model

[1725] Based on the new reaction data collected, the machine learning model is retrained to improve the accuracy of the next message optimization.

[1726] Specific examples

[1727] Example 1: New product introduction

[1728] Data collection

[1729] Behavioral data is collected showing that User A has shown interest in "electronic devices" in the past.

[1730] Data analysis

[1731] It is analyzed that user A is often interested in messages containing the keyword "new product announcement."

[1732] Individual optimization

[1733] Generate a message for User A saying, "Check it out now! The latest smartphones have been released and we have special discount codes for you!"

[1734] Sending a message

[1735] Send the message at 9:00 AM. Based on past data, this time is the most responsive time for User A.

[1736] Reaction monitoring

[1737] User A opens the message and clicks the link to view product details.

[1738] feedback

[1739] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[1740] In this way, the present invention maximizes the marketing effectiveness of official LINE by generating and sending optimal messages based on each user's behavioral patterns.

[1741] The processing flow will be explained below.

[1742] Step 1: Data collection

[1743] 1. User: Activity Record

[1744] A user opens a message from the official LINE account, clicks a link, or responds to a message.

[1745] 2. Server: Collecting data logs

[1746] The server collects user behavior data in real time through the LINE API, including message open rates, link click rates, and past communication content.

[1747] 3. Server: Save to database

[1748] The collected behavioral data is organized in a database, and the user ID, message ID, behavior type (open, click, response), timestamp, etc. are recorded.

[1749] Step 2: Data analysis

[1750] 1. Server: Data preprocessing

[1751] Cleanse the stored behavioral data, fill in missing values, and remove outliers.

[1752] 2. Server: Feature Extraction

[1753] Extract features from user behavior data. For example, aggregate message open rates, click rates, and the most responsive times for the past week.

[1754] 3. Server: Applying machine learning models

[1755] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict user behavior patterns.

[1756] Step 3: Individual optimization

[1757] 1. Server: Prepare message template

[1758] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[1759] 2. Server: Customization

[1760] Customize templates individually based on each user's behavioral patterns, for example by inserting specific keywords or product categories to optimize message content.

[1761] 3. Server: Message Creation

[1762] Generate personalized messages for each user based on customized templates.

[1763] Step 4: Send the message

[1764] 1. Server: Decide when to send

[1765] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[1766] 2. Server: Send message

[1767] Use the LINE API to send the generated optimized message to the user at the determined time.

[1768] Step 5: Monitoring the reaction

[1769] 1. Users: Receiving messages and taking action

[1770] Users open the messages they receive and click on links within them or reply to them.

[1771] 2. Server: Real-time monitoring

[1772] User responses to sent messages (e.g., open rates, click rates, response content) are monitored in real time and response data is collected.

[1773] Step 6: Feedback

[1774] 1. Server: Reaction data storage

[1775] The collected reaction data is stored in a database and accumulated as material for the next analysis.

[1776] 2. Server: Retrain the model

[1777] Retrain your machine learning model based on new response data to improve the accuracy of your next message optimization.

[1778] In this way, through each processing step, the system can optimize messages from the official LINE account for each individual user, allowing for more effective communication of information.

[1779] Example 1

[1780] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1781] Conventional messaging systems have had the problem of being unable to deliver effective marketing due to the difficulty in sending messages optimized for each user. Furthermore, they lacked a feedback mechanism for effectively analyzing user behavior data and utilizing the results in the next messaging. Therefore, a system that can maximize the effectiveness of marketing is needed.

[1782] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1783] In this invention, the server includes means for collecting user behavioral data, means for cleansing the collected behavioral data, completing missing values ​​and removing outliers, means for performing feature extraction to extract features from user behavior patterns, means for applying a machine learning algorithm to classify and predict user behavioral patterns, means for preparing multiple message templates, customizing the templates based on each user's behavioral pattern, and generating an optimal message, means for determining an optimal sending time and sending the generated message to the user, means for collecting user reaction data to the sent message and saving it as material for subsequent analysis and optimization, and feedback means for retraining a machine learning model based on the collected reaction data and improving the accuracy of message optimization, thereby enabling the generation and sending of optimal messages based on each user's behavioral pattern.

[1784] "User behavior data" is data recorded when a user takes a specific action, such as opening a message, clicking a link, or responding.

[1785] "Data cleansing" is the process of imputing missing values ​​and removing outliers from collected data.

[1786] "Feature extraction" is the process of extracting features from user behavior patterns, which makes analysis and prediction easier.

[1787] A "machine learning algorithm" is an algorithm that learns patterns and regularities from data and makes future predictions and classifications.

[1788] A "message template" is a pre-defined framework for message format and content, allowing for quick message generation and customization.

[1789] The "optimal sending time" is the time at which a message is sent that is most likely to be responsive to each user, based on collected data and analysis results.

[1790] "Response data" is a record of the actions users take in response to messages they receive, and is used as material for future analysis and optimization.

[1791] "Feedback measures" are processes that retrain machine learning models based on collected response data to improve the accuracy of the next message optimization.

[1792] This invention is a system for administrators of official messaging accounts that collects and analyzes user behavior data, and generates and sends individually optimized messages. The aim is to effectively utilize user behavior data and maximize the marketing effectiveness of messages.

[1793] This system is mainly composed of three elements: a server, a terminal, and a user. Below, we will explain the flow of data processing and data calculation using the specific hardware and software of each element.

[1794] Data collection

[1795] User:Activity Log

[1796] Users take actions such as opening, clicking, or responding to messages on official messaging accounts, and these actions act as triggers for the system.

[1797] Server: Collecting data logs

[1798] The server obtains user behavior data in real time via a messaging API (e.g., the LINE API), including message open rates, link click rates, and past communication content.

[1799] Server: Save to database

[1800] The collected behavioral data is stored in a database along with information such as user ID, message ID, behavior type (open, click, response), and timestamp.

[1801] Data analysis

[1802] Server: Data preprocessing

[1803] Stored behavioral data is cleansed, missing values ​​are filled, and outliers are removed, ensuring data accuracy and reliability.

[1804] Server:Feature Extraction

[1805] Extract features from user behavior patterns, such as message open rates and click rates over the past week, and the time of day when responses are most likely to occur.

[1806] Server: Applying machine learning models

[1807] Based on the extracted features, machine learning algorithms (e.g., clustering, regression analysis, etc.) are applied to classify and predict user behavior patterns, allowing for a detailed understanding of each user's characteristics.

[1808] Individual optimization

[1809] Server: Prepare message template

[1810] The server uses a number of pre-defined message templates (e.g., new product introductions, sales announcements, newsletters, etc.).

[1811] Server:Customization

[1812] The template is customized based on each user's behavioral patterns, for example, inserting relevant keywords for a user interested in "electronic devices."

[1813] Server: Message Creation

[1814] It automatically generates personalized messages for each user based on customized templates, using a generative AI model and providing specific prompts.

[1815] Sending a message

[1816] Server: Decide when to send

[1817] Based on the analysis results, the optimal sending time for each user is determined. For example, if past data indicates that User A responds best at 9:00 AM, then that time is set.

[1818] Server: Send message

[1819] The server uses the API to send the generated optimization message to the user at the determined timing.

[1820] Reaction monitoring

[1821] Users: Message Reception and Actions

[1822] Users receive the messages sent and take action such as opening, clicking, or responding.

[1823] Server: Real-time monitoring

[1824] The server monitors users' reactions to the sent messages in real time and stores the reaction data in a database.

[1825] feedback

[1826] Server: Reaction data storage

[1827] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[1828] Server: Retrain the model

[1829] Based on the new reaction data collected, the machine learning model is retrained to improve the accuracy of message optimization.

[1830] An example of a specific prompt might be, "Generate a message for a new product launch taking into account user behavioral data. Use keywords that have generated positive responses in past messages, and include content that will increase open rates."

[1831] In this way, the present invention is a system that generates and sends optimal messages based on each user's behavioral patterns, maximizing the marketing effectiveness of official messaging accounts.

[1832] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1833] Step 1:

[1834] Data collection

[1835] User:Activity Log

[1836] Users open messages, click links, and send responses to official messaging accounts, and these actions trigger data collection.

[1837] Input: User actions (opens, clicks, responses)

[1838] Output: When an action occurs, the data is collected immediately.

[1839] Server: Collecting data logs

[1840] The server obtains user behavior data in real time via a messaging API (e.g., LINE API).

[1841] Input: Behavioral data captured by the messaging API

[1842] Output: Raw behavioral data (open rates, click rates, communication content, etc.)

[1843] Server: Save to database

[1844] The server stores the collected behavioral data in a database, which includes the user ID, message ID, behavior type, and timestamp.

[1845] Input: Raw behavioral data

[1846] Output: Behavioral data stored in a database

[1847] Step 2:

[1848] Data Preprocessing

[1849] Server: Data Cleansing

[1850] The server cleanses the stored data, imputing missing values ​​and removing outliers.

[1851] Input: Stored behavioral data

[1852] Output: Cleansed data

[1853] Specific behavior: For example, gracefully handling time zone inconsistencies and extreme values.

[1854] Step 3:

[1855] Data analysis

[1856] Server:Feature Extraction

[1857] The server extracts features from user behavior patterns, such as message open rates and click rates over the past week, and the time periods with the highest response rates.

[1858] Input: Cleansed data

[1859] Output: Extracted features

[1860] Specific operations: Analyze user behavior patterns using time series analysis and statistical methods.

[1861] Server: Applying machine learning models

[1862] The server applies a machine learning algorithm based on the extracted features to classify and predict user behavior patterns.

[1863] Input: Feature data

[1864] Output: Classified user groups and predicted behavior patterns

[1865] Specific behavior: Uses clustering and regression analysis algorithms.

[1866] Step 4:

[1867] Individual optimization

[1868] Server: Prepare message template

[1869] The server uses a number of pre-defined message templates.

[1870] Input: Template list

[1871] Output: Available templates

[1872] Specific actions: Select and update templates regularly.

[1873] Server:Customization

[1874] The server customizes the template based on each user's behavioral patterns, for example, by inserting keywords related to "electronic devices" to match the user's interests.

[1875] Input: Template, Categorized User Groups

[1876] Output: Customized template

[1877] Specific behavior: Modify specific parts of the template to suit the user's interests.

[1878] Server: Message Creation

[1879] The server automatically generates messages optimized for each user based on customized templates.

[1880] Input:Customized template

[1881] Output: Generated optimization messages

[1882] What it does: Uses a generative AI model to generate an appropriate message based on the prompt.

[1883] Step 5:

[1884] Sending a message

[1885] Server: Decide when to send

[1886] The server determines the optimal transmission time for each user from the analysis results.

[1887] Input: User's optimal transmission time data

[1888] Output: Transmission schedule

[1889] Specific behavior: Optimize sending times based on past behavioral data.

[1890] Server: Send message

[1891] The server sends the optimization message generated using the API to the user at the determined timing.

[1892] Input: Generated message, send schedule

[1893] Output: Message sent

[1894] Specific operation: The actual message sending is carried out using the API.

[1895] Step 6:

[1896] Reaction monitoring

[1897] Users: Message Reception and Actions

[1898] The user opens the received message, clicks on the link, and sends a response.

[1899] Input: Optimized message

[1900] Output: User response data (behavior)

[1901] What it does: Collect user feedback in real time.

[1902] Server: Real-time monitoring

[1903] The server monitors users' reactions to the sent messages in real time and stores the reaction data in a database.

[1904] Input: User response data

[1905] Output: Saved reaction data

[1906] Specific actions: Monitor using real-time data analysis tools.

[1907] Step 7:

[1908] feedback

[1909] Server: Reaction data storage

[1910] The server stores the user's reaction data in a database and accumulates it as material for the next analysis.

[1911] Input: User response data

[1912] Output: Saved reaction data

[1913] Specific operation: The saving process is carried out using a database management system.

[1914] Server: Retrain the model

[1915] The server retrains the machine learning model based on the new reaction data collected, improving the accuracy of the next message optimization.

[1916] Input: New reaction data

[1917] Output: A retrained machine learning model

[1918] What it does: Continually improve the model using adaptive learning techniques.

[1919] This makes it possible to generate and send optimal messages based on each user's behavioral patterns, maximizing marketing effectiveness.

[1920] (Application example 1)

[1921] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1922] Conventional user behavior data collection and analysis systems have been unable to achieve effective marketing due to insufficient optimization of messages for individual users. In particular, generic messages are often sent without sufficient consideration of the different behavioral patterns of each user, limiting the effectiveness of improving user response rates and purchasing intent. Furthermore, the system lacks a mechanism for efficiently feeding back response data to messages sent and improving the accuracy of the model.

[1923] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1924] In this invention, the server includes means for collecting user behavioral data, means for analyzing the collected behavioral data and identifying user behavior patterns, means for generating a document optimized for each user based on the identified behavioral pattern, means for sending the generated document to the user, means for collecting user reaction data and providing feedback to improve the accuracy of the analysis means, means for using a machine learning algorithm to perform clustering based on the collected behavioral data, and means for selecting a message template optimized for each user cluster and generating a customized message, thereby enabling effective message optimization based on user behavior patterns.

[1925] "User behavior data" refers to records of various interactions users have with the system, including, for example, message open rates, link click rates, and the content of past communications.

[1926] "Analysis Tools" refers to the functions and algorithms used to process and analyze collected behavioral data and identify user behavior patterns.

[1927] "Behavioral patterns" refer to models or tendencies that indicate how a user will behave based on the user's past behavioral data.

[1928] "Optimized documents" refers to messages or documents that are customized to best suit a user based on the user's behavioral patterns.

[1929] "Feedback means" refers to a data collection and processing mechanism for improving the analytical accuracy of the system and the performance of the machine learning model based on user response data.

[1930] "Clustering" refers to a technique that uses machine learning algorithms to classify users into groups with similar behavioral patterns.

[1931] A "machine learning algorithm" is a part of a computer program that analyzes large amounts of data and automatically learns patterns and predictive models.

[1932] A "template" is a pre-created message template for a specific purpose, which serves as the basis for customization.

[1933] "Customization" refers to tailoring documents and messages to specific user behavior patterns.

[1934] The present invention relates to a system for generating and sending optimized messages based on user behavior data. The system is implemented in the following steps:

[1935] First, the server uses the LINE API to collect user behavior data, including message open rates, link click rates, and past communication content. The collected data is sent to the server in real time and stored in a database.

[1936] Once the data is collected, the server pre-processes it, cleansing it to fill in missing values ​​and remove outliers, then performs feature extraction to extract parameters that characterize user behavior patterns.

[1937] The server then uses machine learning algorithms, specifically clustering algorithms, to classify users into clusters with similar behavioral patterns, leveraging Python's scikit-learn library and other machine learning tools.

[1938] After clustering is complete, the server applies pre-prepared template messages to each cluster. Each template is customized to fit the behavioral patterns of a specific user. The customized message is then sent to the user in the most effective way. For example, user A might receive a message like, "Hello, would you like to check out our latest products?"

[1939] After the message is sent, user reaction data is collected again and sent to the server. This reaction data is stored in a database and used as a feedback mechanism to improve the accuracy of the analysis method. The new data is used to retrain the machine learning model, improving the accuracy of the next message optimization.

[1940] As a concrete example, suppose there is data showing that User A has clicked on messages related to beauty products in the past. In this case, the machine learning model will classify User A into a cluster interested in beauty products and use a template message for that cluster to send a customized message such as "Don't miss out on our great beauty product campaigns!"

[1941] Examples of prompts include:

[1942] Generate customized product introduction messages according to the cluster. Cluster 1: For users interested in beauty products

[1943] Don't miss out on our great beauty deals!

[1944] Cluster 2: For users interested in home appliances

[1945] The latest home appliances have arrived! Check them out now.

[1946] In this way, the present invention maximizes marketing effectiveness by utilizing user behavior data to generate and send messages optimized for individual users.

[1947] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1948] Step 1:

[1949] The server collects user behavior data via the LINE API, including message open rates, link click rates, and past communication content. The collected data is sent to the server in real time and stored in a database.

[1950] Input: User behavior data obtained via the LINE API

[1951] Output: Behavioral data stored in a database

[1952] Specific operation: The server sends a request to the LINE API and executes an SQL query to save the retrieved data to the database.

[1953] Step 2:

[1954] The server pre-processes the stored behavioral data, performing data cleansing, filling in missing values, and removing outliers.

[1955] Input: Raw data stored in a database

[1956] Output: Cleansed behavioral data

[1957] Specific operation: The server reads data from the database and uses the Pandas library to impute missing values ​​and remove outliers.

[1958] Step 3:

[1959] The server performs feature extraction from the preprocessed data, extracting features such as the open rate, click rate, and best response time for each user.

[1960] Input: Cleansed behavioral data

[1961] Output: A dataset with extracted features

[1962] What it does: The server groups the data by user ID, calculates certain statistics, and generates new features, such as how many times a message was opened or clicked each day.

[1963] Step 4:

[1964] The server performs clustering using a machine learning algorithm to classify users into clusters with similar behavioral patterns.

[1965] Input: Feature-extracted dataset

[1966] Output: User data with assigned clusters

[1967] What it does: The server uses the scikit-learn library to run the KMeans algorithm to classify users into clusters, and each user is assigned a cluster ID.

[1968] Step 5:

[1969] The server selects the best template message for each cluster and generates a customized message.

[1970] Input: User data with assigned clusters, and message template

[1971] Output: Customized message

[1972] What happens: The server selects a template for each cluster and customizes the message by inserting user-specific information.

[1973] Step 6:

[1974] The server sends the generated customized message to the user via the LINE API.

[1975] Input: Customized Message

[1976] Output: Message sent to the user's terminal

[1977] Specific operation: The server sends a POST request to the LINE API and sends the generated message to the user at the appropriate time, for example, during the time period when the user is most likely to respond.

[1978] Step 7:

[1979] The server again collects user reaction data to the sent messages and stores it in a database.

[1980] Input: User interaction data (message opens, clicks, responses, etc.)

[1981] Output: Reaction data stored in a database

[1982] Specific operation: The server retrieves the user's reaction data from the LINE API and executes an SQL query to save it in the database.

[1983] Step 8:

[1984] The server retrains the machine learning model based on the new response data to improve the accuracy of the next message optimization.

[1985] Input: New reaction data stored in the database

[1986] Output: An updated machine learning model

[1987] Specific operation: The server extracts features from new reaction data and retrains the machine learning algorithm to improve the accuracy of the model.

[1988] Through these steps, the present invention achieves effective message optimization based on user behavior data.

[1989] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1990] The present invention relates to a system for collecting and analyzing behavioral and emotional data of users for the administrator of an official LINE account, and generating and sending messages optimized for each user.

[1991] 1. Data Collection

[1992] User:Activity Log

[1993] A user takes an action such as opening a message, clicking, or responding to the official LINE account.

[1994] Server: Data logging and emotion data collection

[1995] The server connects with the official LINE account to obtain user behavioral and emotional data in real time. Specifically, it uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also uses an emotion engine to collect user emotional data through text, voice, and image analysis.

[1996] Server: Save to database

[1997] The collected behavioral and emotional data is organized in a database, and the user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc. are recorded.

[1998] 2. Data analysis

[1999] Server: Data preprocessing

[2000] Cleanse stored behavioral and emotional data to fill in missing values ​​and remove outliers.

[2001] Server:Feature Extraction

[2002] Extract features from user behavioral and emotional data. For example, aggregate message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week.

[2003] Server: Applying machine learning models

[2004] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict the user's behavioral and emotional patterns.

[2005] 3. Individual optimization

[2006] Server: Prepare message template

[2007] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[2008] Server:Customization

[2009] Templates are individually customized based on each user's behavioral and emotional patterns. For example, message content can be optimized by inserting specific keywords or product categories. The tone and content of messages can also be adjusted according to the user's emotional state.

[2010] Server: Message Creation

[2011] Automatically generate messages optimized for each user based on customized templates.

[2012] 4. Sending a message

[2013] Server: Decide when to send

[2014] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[2015] Server: Send message

[2016] Use the LINE API to send the generated optimized message to the user at the determined time.

[2017] 5. Monitoring the reaction

[2018] Users: Message Reception and Actions

[2019] Users open the messages they receive and click on links within them or reply to them.

[2020] Server: Real-time monitoring

[2021] The server monitors user responses to sent messages (e.g., open rate, click rate, response content) in real time and collects response data.

[2022] 6. Feedback

[2023] Server: Reaction data storage

[2024] The user's reaction data is saved in a database and accumulated as material for the next analysis.

[2025] Server: Retrain the model

[2026] Retrain your machine learning model based on new reaction data collected to improve the accuracy of your next message optimization.

[2027] Specific examples

[2028] Example 1: New product introduction

[2029] Data collection

[2030] The system collects behavioral data showing that User A has shown interest in "electronic devices" in the past. The emotion engine also identifies a positive emotional state based on User A's recent messages and voice.

[2031] Data analysis

[2032] It is analyzed that User A is often interested in messages containing the keyword "new product launch." Also, since User A's emotional state is positive, it is determined that a direct sales message would be effective.

[2033] Individual optimization

[2034] Generate a positive message for User A: "Check it out now! The latest smartphones are on sale now, and we've got a special discount code for you!"

[2035] Sending a message

[2036] Send the message at 9:00 AM. Based on past data, this time is the most responsive time for User A.

[2037] Reaction monitoring

[2038] User A opens the message and clicks the link to view product details.

[2039] feedback

[2040] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[2041] In this way, the present invention maximizes the marketing effectiveness of official LINE by combining and analyzing user behavioral data and emotional data to generate and send optimal messages according to individual needs and emotional states.

[2042] The processing flow will be explained below.

[2043] Step 1: Data collection

[2044] 1. User: Activity Record

[2045] A user opens a message from the official LINE account, clicks a link, or responds to a message.

[2046] 2. Server: Collecting data logs

[2047] The server obtains user behavior data in real time through the LINE API, including message open rates, link click rates, and past communication content.

[2048] 3. Server: Emotion data collection

[2049] The emotion engine is used to collect user emotional data through text, voice, and image analysis, for example, to identify the user's emotional state from the content of their message or their tone of voice.

[2050] 4. Server: Save to database

[2051] The collected behavioral and emotional data is organized and stored in a database, including user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc.

[2052] Step 2: Data analysis

[2053] 1. Server: Data preprocessing

[2054] Cleanse stored behavioral and emotional data to fill in missing values ​​and remove outliers.

[2055] 2. Server: Feature Extraction

[2056] Extract features from user behavioral and emotional data. For example, aggregate message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week.

[2057] 3. Server: Applying machine learning models

[2058] Based on the extracted features, machine learning algorithms (such as clustering and regression analysis) are used to classify and predict the user's behavioral and emotional patterns.

[2059] Step 3: Individual optimization

[2060] 1. Server: Prepare message template

[2061] Prepare multiple pre-made message templates (e.g., new product introductions, sales announcements, newsletters).

[2062] 2. Server: Customization

[2063] Templates are individually customized based on each user's behavioral and emotional patterns. For example, message content can be optimized by inserting specific keywords or product categories. The tone and content of messages can also be adjusted according to the user's emotional state.

[2064] 3. Server: Message Creation

[2065] Automatically generate messages optimized for each user based on customized templates.

[2066] Step 4: Send the message

[2067] 1. Server: Decide when to send

[2068] Based on the analysis results, the optimal sending time is determined. For example, if past data indicates that User A is most likely to respond at 9:00 AM, the message is scheduled to be sent at that time.

[2069] 2. Server: Send message

[2070] Use the LINE API to send the generated optimized message to the user at the determined time.

[2071] Step 5: Monitoring the reaction

[2072] 1. Users: Receiving messages and taking action

[2073] Users open the messages they receive and click on links within them or reply to them.

[2074] 2. Server: Real-time monitoring

[2075] User responses to sent messages (e.g., open rates, click rates, response content) are monitored in real time and response data is collected.

[2076] Step 6: Feedback

[2077] 1. Server: Reaction data storage

[2078] The collected reaction data is stored in a database and accumulated as material for the next analysis.

[2079] 2. Server: Retrain the model

[2080] Retrain your machine learning model based on new response data to improve the accuracy of your next message optimization.

[2081] Example 2

[2082] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[2083] Conventional message sending systems only considered user behavioral data and did not perform individual optimization using emotional data, making it difficult to maximize the effectiveness of marketing messages. Furthermore, there was also the problem of not being able to quickly reflect user reactions because the timing of sending messages was not optimized and real-time feedback was not adequately provided.

[2084] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for collecting user behavioral data and emotional data, means for analyzing the collected behavioral data and emotional data and identifying the user's behavioral and emotional patterns, means for generating a document optimized for each user based on the identified behavioral and emotional patterns, means for sending the generated document to the user, and means for collecting user reaction data in real time and providing feedback to improve the accuracy of the analysis means. This makes it possible to generate and send individually optimized messages based on the user's behavioral and emotional data, thereby maximizing marketing effectiveness.

[2085] "User behavioral data" refers to information related to actions taken by users on official LINE accounts, such as opening messages, clicking links, and responding to messages.

[2086] "User emotional data" is information collected using text analysis, speech analysis, image analysis, etc. to identify a user's emotional state.

[2087] A "behavioral pattern" refers to the regularity or tendency of behavior that is analyzed based on data on a series of actions taken by a user.

[2088] "Emotional patterns" refer to changes and trends in emotions that are analyzed based on the user's emotional data.

[2089] "Optimized documents" are messages that are individually tailored based on a user's behavioral and emotional patterns and are generated to maximize the user's interest and response.

[2090] A "machine learning algorithm" is a computational method for learning patterns from data and making predictions or classifications based on the results.

[2091] "Clustering" is a type of machine learning technique that divides data into several groups.

[2092] "Regression analysis" is a machine learning technique that is used to model the relationship between predictor variables and response variables and make predictions.

[2093] A "feedback means" is a method or mechanism for collecting user reaction data and using that data to improve the accuracy of the analysis means and generation means.

[2094] "Optimizing sending timing" aims to send messages at the time when users are most likely to respond, and determines the timing of sending based on past behavioral data.

[2095] This invention relates to a system for official LINE administrators that collects and analyzes user behavioral and emotional data, and generates and sends messages optimized for each user.

[2096] 1. Data Collection

[2097] User behavior record

[2098] Users open messages sent to official LINE accounts, click links, and respond - all of this behavioral data is collected in real time.

[2099] Server data logging and emotion data collection

[2100] The server uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also collects user emotional data through text, voice, and image analysis by emotion engines (such as IBM Watson and Microsoft Azure).

[2101] Server database storage

[2102] The server organizes the collected behavioral and emotional data into a database, recording the user ID, message ID, behavior type (open, click, response), emotional state, timestamp, etc.

[2103] 2. Data analysis

[2104] Server data preprocessing

[2105] The server cleanses the stored behavioral and emotional data, filling in missing values ​​and removing outliers, thereby ensuring data quality.

[2106] Server Feature Extraction

[2107] The server extracts features from the behavioral and emotional data, such as message open rates, click rates, the most responsive times, and patterns of emotional change over the past week.

[2108] Applying machine learning models to servers

[2109] Based on the extracted features, the server uses machine learning algorithms (e.g., K-means clustering and linear regression) to classify and predict the user's behavioral and emotional patterns.

[2110] 3. Individual optimization

[2111] Preparing the server message template

[2112] The server prepares a number of pre-defined message templates (e.g., new product introductions, sales announcements, newsletters).

[2113] Server Customization

[2114] The server customizes the templates individually for each user based on their behavioral and emotional patterns, inserting specific keywords and product categories and adjusting the tone and content of the message depending on the user's emotional state.

[2115] Server Message Generation

[2116] The server automatically generates messages optimized for each user based on customized templates.

[2117] 4. Sending a message

[2118] Server transmission timing decision

[2119] The server then uses the analysis to determine the best time to send the message. For example, if historical data indicates that a particular user is most likely to respond at 9 a.m., it will schedule the message to be sent at that time.

[2120] Server message sending

[2121] The server uses the LINE API to send the generated optimized message to the user at the determined timing.

[2122] 5. Monitoring the reaction

[2123] User message receipt and actions

[2124] The user opens the received message, clicks on a link in the message, or makes a response.

[2125] Real-time server monitoring

[2126] The server monitors user responses to sent messages (e.g., open rate, click rate, response content) in real time and collects response data.

[2127] 6. Feedback

[2128] Saving reaction data on the server

[2129] The server stores the user's reaction data in a database and accumulates it as material for the next analysis.

[2130] Retraining the model on the server

[2131] The server retrains the machine learning model based on the new reaction data collected, improving the accuracy of the next message optimization.

[2132] Specific examples

[2133] Example 1: New product introduction

[2134] Behavioral data showing that User A has shown interest in "electronic devices" in the past is collected. In addition, analysis by an emotion engine (e.g., IBM Watson) identifies a positive emotional state based on User A's recent messages and voice.

[2135] Data analysis

[2136] It is analyzed that User A is often interested in messages containing the keyword "new product launch." Also, since User A's emotional state is positive, it is determined that a direct sales message would be effective.

[2137] Individual optimization

[2138] Generate a positive message for User A: "Check it out now! The latest smartphones are on sale now, and we've got a special discount code for you!"

[2139] Sending a message

[2140] Based on past data, it is determined that User A responds best at 9:00 AM, so the message is sent at this time.

[2141] Reaction monitoring

[2142] User A opens the message and clicks the link to view product details.

[2143] feedback

[2144] User A's reaction data will be collected again to help improve the accuracy of the next new product announcement.

[2145] Prompt Sentence Examples

[2146] Here are some example prompts to enter into a generative AI model:

[2147] "User A has shown interest in 'electronic devices' in the past, and analysis of the emotion engine reveals that he is in a positive emotional state. Please create a message about a new product launch that will interest User A."

[2148] Using this prompt, the generative AI model generates a message optimized for User A.

[2149] The flow of the identification process in the second embodiment will be described with reference to FIG.

[2150] Step 1: Recording user behavior

[2151] A user opens a message from an official LINE account, clicks a link, and responds. This behavioral data is sent to the LINE server (input).

[2152] Step 2: Collecting server data logs and emotion data

[2153] The server works with the official LINE app to call the LINE API and collect (output) user behavior data in real time. Specifically, behavioral data acquired includes message open rates, link click rates, and past communication content. Additionally, an emotion engine (e.g., IBM Watson) is used to collect user emotion data through text, voice, and image analysis.

[2154] Step 3: Save the database on the server

[2155] The server stores the collected behavioral and emotional data in a database (output). Specifically, it records information such as user ID, message ID, behavior type (open, click, response), emotional state, and timestamp.

[2156] Step 4: Data Preprocessing on the Server

[2157] The server cleanses the stored behavioral and emotional data (data processing). Specifically, it complements missing values ​​and removes outliers, preparing the data for analysis (output).

[2158] Step 5: Server Feature Extraction

[2159] The server extracts features from the cleansed data (data processing). Specifically, it aggregates information such as message open rates, click rates, the most responsive time periods, and patterns of emotional change over the past week (output).

[2160] Step 6: Applying the machine learning model to the server

[2161] The server analyzes and classifies the user's behavioral and emotional patterns using machine learning algorithms (K-means clustering and linear regression) based on the extracted features (data calculation and output). For example, User A is classified into the "high reaction / positive group."

[2162] Step 7: Prepare the server message template

[2163] The server prepares message templates, including templates for new product introductions, sales announcements, newsletters, etc. (Output)

[2164] Step 8: Customizing the Server

[2165] The server customizes templates based on each user's behavioral and emotional patterns, inserting specific keywords and product categories and adjusting the tone and content of messages depending on their emotional state (data processing, output).

[2166] Step 9: Server Message Generation

[2167] The server automatically generates (outputs) a message optimized for each user based on the customized template. For example, a message such as "Check it out now! The latest smartphone has been released. We also have a special discount code!" is generated.

[2168] Step 10: Deciding when the server should send

[2169] The server determines the optimal sending time based on the user's past data (data calculation, output). For example, if user A is determined to be most responsive at 9:00 AM, the server will schedule the message to be sent at 9:00 AM.

[2170] Step 11: Server Sends Message

[2171] The server uses the LINE API to send the generated optimized message to the user at the determined timing (output).

[2172] Step 12: User receives message and takes action

[2173] The user opens the received message and takes action such as clicking a link (input).

[2174] Step 13: Real-time Server Monitoring

[2175] The server monitors users' reactions to the sent messages in real time and collects (outputs) reaction data, such as whether the message was opened, whether a link was clicked, and the content of the response.

[2176] Step 14: Save the server's reaction data

[2177] The server saves the user's reaction data in a database (output), which is then stored as material for the next analysis.

[2178] Step 15: Retraining the model on the server

[2179] The server retrains the machine learning model based on the new reaction data to improve the accuracy of the next message optimization (data calculation, output). Specifically, it learns new behavioral and emotional patterns and improves the accuracy of the model.

[2180] (Application example 2)

[2181] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[2182] Conventional advertising delivery systems optimize messages based solely on user behavioral data, which means they are unable to generate effective messages that reflect the user's emotional state. Furthermore, relying solely on user behavioral data makes it difficult to accurately reflect the user's real-time interests and emotional changes, limiting the effectiveness of advertising.

[2183] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for collecting user behavioral data and emotional data, means for analyzing the collected behavioral data and emotional data and identifying the user's behavioral patterns and emotional patterns, means for generating documents optimized for each user based on the identified behavioral patterns and emotional patterns, means for sending the generated documents to the user, and means for collecting user reaction data and providing feedback to improve the accuracy of the analysis means. This makes it possible to integrate the user's behavioral data and emotional data and generate and deliver individually optimized advertising messages.

[2184] "User behavioral data" refers to data generated when a user behaves online, such as message open rates, link click rates, and the content of past communications.

[2185] "Emotional data" is data that indicates a user's emotional state collected through text analysis, voice analysis, and image analysis.

[2186] "Behavioral patterns" refer to specific behavioral tendencies and habits that are analyzed based on user behavior data.

[2187] An "emotion pattern" indicates the tendency and changes in the user's unique emotions, which are analyzed based on the emotion data.

[2188] "Optimized documents" are messages and notifications that are individually customized and created in the most effective way based on the user's behavioral and emotional patterns.

[2189] The "feedback means" is a means for collecting user response data and reflecting it in the system to improve the accuracy of the analysis means.

[2190] The present invention is an advertising distribution system that collects, analyzes, and optimizes user behavioral and emotional data. This system is implemented using a server, user terminals (smartphones, etc.), and a network. Specifically, the invention is implemented with the following configuration.

[2191] 1. Data Collection

[2192] The server collects user behavioral and emotional data in real time. Specifically, it uses the LINE API to collect behavioral data such as message open rates, link click rates, and past communication content. It also uses an emotion engine to obtain user emotional data through text, voice, and image analysis. This data is then stored in a database.

[2193] 2. Data analysis

[2194] The server cleanses the collected behavioral and emotional data, fills in missing values, removes outliers, and extracts features through feature extraction. It then uses machine learning models (e.g., clustering and regression analysis) to classify and predict user behavioral and emotional patterns.

[2195] 3. Individual optimization

[2196] The server uses multiple pre-prepared message templates and customizes them individually based on each user's behavioral and emotional patterns, adjusting the tone and content of the message according to the user's emotional state to generate individually optimized documents.

[2197] 4. Sending a message

[2198] The server determines the optimal sending time based on the analysis results. For example, if past data indicates that a particular user is most likely to respond at 9:00 a.m., it sets the message to be sent at that time. The server then uses the LINE API to send the generated optimized message to the user.

[2199] 5. Response monitoring and feedback

[2200] The server monitors user responses to sent messages (open rates, click rates, response content) in real time and stores the response data in a database. This data is used to retrain the machine learning model and improve the accuracy of the next message optimization.

[2201] Specific examples

[2202] When User A shows interest in a certain product, the server collects their behavioral and emotional data in real time. If User A sends a message saying, "I love this product!", the emotional data indicates a positive reaction. Based on this information, the server generates a customized message for User A saying, "Check it out now! The latest smartphone has been released. We also have a special discount code!" and sends it to User A at 9:00 AM.

[2203] An example of a prompt used by a generative AI model is, "If user A expresses positive feelings about a particular product, generate a message containing the latest information and special discounts related to that product."

[2204] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2205] Step 1:

[2206] It is a means of collecting message open rates, link click rates, and past communication content from users' devices using the LINE API. The input data is user behavioral data and emotional data, and the output is the collected data. In particular, text analysis analyzes the content of users' messages and generates emotional data such as positive or negative to determine their emotional state. In addition, an emotion engine is used to perform voice and image analysis to collect user emotional data.

[2207] Step 2:

[2208] The server stores the collected behavioral and emotional data in a database. This database records user IDs, message IDs, behavior types (open, click, response), emotional states, timestamps, etc. The input data is the collected user behavioral and emotional data, and the output is structured data organized in the database. Specific operations include classifying and structuring the information.

[2209] Step 3:

[2210] The server performs data preprocessing, which includes data cleansing, missing value imputation, and outlier removal. The input data is raw data stored in a database, and the output is clean data. In particular, Pandas is used to clean the data.

[2211] Step 4:

[2212] The server performs feature extraction to extract features from user behavioral and emotional data. The input data is preprocessed clean data, and the output is extracted features. Specific operations include compiling message open rates and click rates over the past week, the most responsive times, and emotional change patterns.

[2213] Step 5:

[2214] The server uses a machine learning model to classify and predict the user's behavioral and emotional patterns based on the extracted features. The input data are the features, and the output is the user's behavioral and emotional patterns. Specific operations include pattern classification using clustering and regression analysis.

[2215] Step 6:

[2216] The server uses multiple pre-prepared message templates to generate documents optimized for each user. The input data are classified behavioral and emotional patterns, and the output is an optimized message. Specific operations include template customization and keyword insertion.

[2217] Step 7:

[2218] The server sends the generated optimized message to the user. The message is sent using the LINE API. The input data is the optimized message and the sending timing, and the output is the sent message. Specific operations include determining the sending time and delivering the message.

[2219] Step 8:

[2220] The server monitors user responses in real time and collects response data. The input data is the sent message and the user's response, and the output is the collected response data. Specific operations include monitoring the open rate, click rate, and response.

[2221] Step 9:

[2222] The server stores the collected reaction data in a database and provides a feedback mechanism for retraining the machine learning model. The input data is the collected reaction data, and the output is an updated machine learning model. Specific operations include learning new data and improving the accuracy of the model.

[2223] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[2224] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[2225] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[2226] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[2227] FIG. 9 illustrates an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and behaviors arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[2228] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[2229] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[2230] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[2231] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[2232] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[2233] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[2234] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[2235] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[2236] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[2237] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[2238] The hardware resource for executing a specific process can be any of the following types of processors: A processor, for example, is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[2239] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[2240] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[2241] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[2242] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[2243] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[2244] The following is further disclosed regarding the above embodiment.

[2245] (Claim 1)

[2246] a means for collecting user behavior data;

[2247] means for analyzing the collected behavioral data to identify user behavior patterns;

[2248] means for generating a document optimized for each user based on the identified behavioral patterns;

[2249] means for transmitting the generated document to a user;

[2250] a feedback means for collecting user response data and improving the accuracy of said analysis means;

[2251] A system including:

[2252] (Claim 2)

[2253] 2. The system of claim 1, wherein the user behavior data includes message open rates, link click rates, and past communication content.

[2254] (Claim 3)

[2255] 10. The system of claim 1, wherein the step of generating a document optimized for each user is performed using a machine learning algorithm.

[2256] "Example 1"

[2257] (Claim 1)

[2258] a means for collecting user behavior data;

[2259] A means for cleansing the collected behavioral data to impute missing values ​​and remove outliers;

[2260] A means for performing feature extraction to extract features from user behavior patterns;

[2261] A means for applying machine learning algorithms to classify and predict user behavior patterns;

[2262] A means for preparing a plurality of message templates, customizing the templates based on the behavioral patterns of each user, and generating an optimal message;

[2263] means for determining an optimal transmission time and transmitting the generated message to the user;

[2264] A means for collecting user response data to the sent messages and storing the data as material for next analysis and optimization;

[2265] A feedback mechanism to retrain the machine learning model based on the collected reaction data and improve the accuracy of message optimization.

[2266] A system including:

[2267] (Claim 2)

[2268] 2. The system of claim 1, wherein the user behavior data includes message open rates, link click rates, and past communication content.

[2269] (Claim 3)

[2270] 10. The system of claim 1, wherein the step of generating a message optimized for each user is performed using a generative AI model.

[2271] "Application Example 1"

[2272] (Claim 1)

[2273] a means for collecting user behavior data;

[2274] means for analyzing the collected behavioral data to identify user behavior patterns;

[2275] means for generating a document optimized for each user based on the identified behavioral patterns;

[2276] means for transmitting the generated document to a user;

[2277] a feedback means for collecting user response data and improving the accuracy of said analysis means;

[2278] a means for using a machine learning algorithm to perform clustering based on the collected behavioral data;

[2279] a means for selecting an optimized message template for each cluster of users and generating a customized message;

[2280] A system including:

[2281] (Claim 2)

[2282] 2. The system of claim 1, wherein the user behavior data includes message open rates, link click rates, and past communication content.

[2283] (Claim 3)

[2284] 10. The system of claim 1, wherein the step of generating a document optimized for each user is performed using a machine learning algorithm.

[2285] "Example 2: Combining Emotion Engines"

[2286] (Claim 1)

[2287] means for collecting user behavioral and emotional data;

[2288] means for analyzing the collected behavioral and emotional data to identify user behavioral and emotional patterns;

[2289] means for generating a document optimized for the user based on the identified behavioral and emotional patterns;

[2290] means for transmitting the generated document to a user;

[2291] a feedback means for collecting user reaction data in real time and improving the accuracy of the analysis means;

[2292] A system including:

[2293] (Claim 2)

[2294] 2. The system of claim 1, wherein the user behavior data includes message open rates, link click rates, and past communication content.

[2295] (Claim 3)

[2296] 2. The system of claim 1, wherein the user's emotional data is collected through text analysis, speech analysis, and image analysis.

[2297] (Claim 4)

[2298] 10. The system of claim 1, wherein the step of generating a document optimized for each user is performed using a machine learning algorithm.

[2299] (Claim 5)

[2300] 5. The system of claim 4, wherein the machine learning algorithms include clustering and regression analysis.

[2301] (Claim 6)

[2302] 10. The system of claim 1, wherein the document generation adjusts the tone and content of a message based on a user's behavioral and emotional patterns.

[2303] (Claim 7)

[2304] 10. The system of claim 1, wherein the timing of sending the generated document is optimized based on past data analysis.

[2305] "Application example 2 when combining emotion engines"

[2306] (Claim 1)

[2307] means for collecting user behavioral and emotional data;

[2308] means for analyzing the collected behavioral and emotional data to identify user behavioral and emotional patterns;

[2309] means for generating a document optimized for the user based on the identified behavioral and emotional patterns;

[2310] means for transmitting the generated document to a user;

[2311] a feedback means for collecting user response data and improving the accuracy of said analysis means;

[2312] A system including:

[2313] (Claim 2)

[2314] 2. The system of claim 1, wherein the user behavioral and emotional data includes message open rates, link click rates, past communication content, and text analysis, voice analysis, and image analysis.

[2315] (Claim 3)

[2316] 10. The system of claim 1, wherein the step of generating a document optimized for each user is performed using a machine learning algorithm. [Explanation of symbols]

[2317] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. a means for collecting user behavior data; means for analyzing the collected behavioral data to identify user behavior patterns; means for generating a document optimized for each user based on the identified behavioral patterns; means for transmitting the generated document to a user; a feedback means for collecting user response data and improving the accuracy of said analysis means; A system including:

2. 2. The system of claim 1, wherein the user behavior data includes message open rates, link click rates, and past communication content.

3. 10. The system of claim 1, wherein the step of generating a document optimized for each user is performed using a machine learning algorithm.

Images