system

Abstract

Provide a system. The method comprises: collecting data from a plurality of data sources; a means of formatting and cleaning the collected data; means for training an artificial intelligence model using the reformulated and cleaned data; means for customizing the artificial intelligence model based on user preferences; A means to distribute and sell customized artificial intelligence models; and 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 AI systems have difficulty in efficiently and comprehensively processing data because they do not integrate a series of processes, such as data collection from specific data sources, data processing, model training, customization, and distribution. In addition, they lack a means to customize AI models based on individual user preferences, making it impossible to achieve highly accurate personalization. [Means for solving the problem]

[0005] The present invention solves the above-mentioned problems by providing a system that includes a means for collecting data from multiple data sources, a means for shaping and cleaning the collected data, a means for training an AI model using the shaping and cleaning data, a means for customizing the AI ​​model based on user preferences, and a means for distributing and selling the customized AI model, which enables efficient and integrated data processing and achieves advanced personalization.

[0006] "Data Source" means the source or platform from which data is collected.

[0007] A "means for collecting data" is a method or device for collecting data from multiple data sources.

[0008] A "data formatting means" is a method or device for converting collected data into a regular format.

[0009] A "data cleaning means" is a method or device for removing noise and defects from collected data to obtain clean data.

[0010] An "artificial intelligence model" is an algorithm or mathematical model that uses techniques such as machine learning and neural networks to perform specific tasks.

[0011] A "means for training an artificial intelligence model" is a method or apparatus for using collected and shaped data to train an artificial intelligence model and improve its performance.

[0012] "Customization means" means a method or device for adjusting or modifying an artificial intelligence model based on a user's individual preferences and requirements.

[0013] A "distribution and sales means" is a method or device for providing a customized artificial intelligence model to the market or to specific clients and commercially exploiting it.

[0014] "User preferences" refers to information about the preferences, interests, and behavioral patterns of a particular user. [Brief explanation of the drawings]

[0015] [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

[0016] 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.

[0017] First, the terms used in the following description will be explained.

[0018] 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).

[0019] 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.

[0020] 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.

[0021] 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.

[0022] 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."

[0023] [First embodiment]

[0024] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0025] 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.

[0026] 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).

[0027] 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.

[0028] 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.

[0029] 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.

[0030] 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.

[0031] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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."

[0036] This invention is a system that collects data from multiple data sources, formats and cleans the data, trains an artificial intelligence (AI) model, and provides customized AI models to markets or specific clients. This system executes each step centrally on a server, achieving highly accurate personalization based on the individual preferences of users.

[0037] Feature Overview

[0038] The system includes the following main features:

[0039] 1. Data collection

[0040] 2. Data formatting and cleaning

[0041] 3. Training the AI ​​model

[0042] 4. Customization

[0043] 5. Distribution and Sales

[0044] Data collection

[0045] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this data collection process, the server provides authentication information to each data source and makes the necessary API calls. For example, the server collects user browsing history from web servers and usage history from mobile apps.

[0046] As a specific example, a server collects the user's smartphone app usage history as a data source, and periodically sends the data to the server, where it is stored in an integrated database.

[0047] Data Shaping and Cleaning

[0048] Since the collected data comes in various formats, the server converts it into a unified format. It also cleans missing or outlier values ​​if they are included. For example, if data associated with a specific user ID is missing, it performs the appropriate imputation process.

[0049] As a specific example, if there are missing values ​​in a user's website browsing history collected by the server, a cleaned dataset is created by filling in the missing values ​​with previous and subsequent data.

[0050] Training an AI model

[0051] Using the cleaned data, the server trains an AI model, which improves its ability to predict user behavior and provide optimal content to users, for example by using machine learning algorithms to find patterns in the training data.

[0052] As a specific example, a server trains a model based on a user's past browsing history to predict which website they are likely to visit next.

[0053] Customization

[0054] It is difficult to apply a trained AI model to all users as is, so the server customizes the AI ​​model based on the preferences of each individual user, providing a personalized service optimized for each user.

[0055] As a specific example, a server generates an AI model that preferentially recommends specific content to a particular user based on the type of content the user prefers.

[0056] Distribution and Sales

[0057] The final customized AI models are then distributed and sold to the market or to specific clients. The server commercializes these models by uploading them to the cloud or by deploying them directly to client servers.

[0058] As a specific example, a server provides services by introducing a content recommendation system customized for a company and deploying an AI model directly on the company's server.

[0059] summary

[0060] This invention is a system that integrates data collection, formatting, cleaning, training, customization, and sales. The server efficiently executes this series of processes, realizing highly personalized services based on user preferences. This system provides high-performance AI models based on information aggregated from multiple data sources, creating significant value for users.

[0061] The processing flow will be explained below.

[0062] Detailed description of the process

[0063] Data collection and formatting process

[0064] Step 1:

[0065] The server sets authentication information for multiple data sources such as web servers, mobile apps, and social media platforms, and prepares for data collection. Specifically, by setting API keys and secret keys, data can be securely collected from external data sources. For example, the web server calls an API to obtain user visit history.

[0066] Step 2:

[0067] The server sends data collection requests from each data source. Specifically, it collects user behavior data through APIs and stores it in an integrated database. During this process, it analyzes data in JSON and XML formats to extract the necessary information. For example, it collects data from mobile apps about user usage time and feature usage.

[0068] Step 3:

[0069] The server normalizes the collected data and formats it into a unified format. Specifically, it unifies different field names and data types and stores the data according to a common schema. For example, it can combine browsing history from a web server and usage history from a mobile app based on the same user ID.

[0070] Data Cleaning Process

[0071] Step 4:

[0072] The server cleans the formatted data. Specifically, it detects missing values ​​and outliers and performs appropriate imputation or deletion. For example, if the number of visits is abnormally high, that data is removed. If there are missing values, they are imputed with the mean or median.

[0073] AI model training process

[0074] Step 5:

[0075] The server splits the cleaned and shaped data into training data and test data by randomly shuffling the entire data and dividing it into 80% training data and 20% test data. This split allows for the evaluation of the model's performance.

[0076] Step 6:

[0077] The server uses the training data to train the AI ​​model. Specifically, it applies machine learning algorithms (e.g., random forests, neural networks) to learn features from the data. During the training process, it adjusts hyperparameters such as the learning rate and the number of epochs.

[0078] Step 7:

[0079] The server evaluates the trained model using test data. Specifically, it calculates metrics such as prediction accuracy, recall, and F-measure to verify the model's performance. Based on these results, the server fine-tunes the model to achieve optimal performance.

[0080] Customization and Distribution Process

[0081] Step 8:

[0082] The server collects user preference data and customizes the trained model by adjusting the model parameters based on the user's past behavioral data and explicit preferences (e.g., favorite genres, frequently used features).

[0083] Step 9:

[0084] The server deploys the customized AI model to a cloud server or a client server. Specifically, it exports the model and uploads it to the deployment environment using a secure communication protocol. After deployment, it checks and tests its operation in the production environment.

[0085] Step 10:

[0086] The server distributes and sells customized AI models to markets or specific clients, specifically by processing purchases of the models through marketplaces or sales platforms and providing buyers with license keys and installation instructions.

[0087] In this way, a system is realized that collects, cleans, shapes, trains, customizes, and distributes data from multiple sources, enabling highly accurate personalization based on user preferences.

[0088] Example 1

[0089] 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."

[0090] Existing AI systems require a lot of time and effort to effectively collect, shape, and clean data from multiple disparate sources. It is also difficult to create highly accurate models based on the cleaned data and customize them to suit individual user preferences. Furthermore, efficiently distributing customized models and making them commercially available has been a challenge.

[0091] 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.

[0092] In this invention, the server includes means for collecting information from multiple sources, means for unifying and purifying the collected information, means for training an AI model using the unified and purified information, means for periodically transmitting information from terminals and storing it in an integrated database, means for identifying and completing missing values ​​and outliers in the purification process, and means for predicting the site a user is likely to visit next based on the user's past behavior history. This enables efficient data collection from multiple data sources, shaping and cleaning, the creation and customization of highly accurate AI models based on user preferences, and the efficient distribution and commercialization of the models.

[0093] "Source" refers to the systems or platforms that provide data, such as web servers, mobile applications, and social networking services.

[0094] "Means of collection" refers to the technical means of providing API calls and credentials to retrieve data from multiple sources.

[0095] "Harmonization means" refers to procedures and methods for converting data of different formats into a unified format.

[0096] "Cleaning methods" refers to technical means for shaping and cleaning data and filling in missing or outlier values.

[0097] "Training methods" refers to machine learning algorithms and techniques used to create artificial intelligence models using collected and cleaned data.

[0098] "Means for periodically transmitting information" refers to the settings and technical means for periodically transmitting information from the terminal to the server and storing it in a database.

[0099] "Means for identifying and imputing missing and outliers" refers to technology for detecting missing and outliers in data and imputing them in an appropriate manner.

[0100] "Methods for predicting which site a user is likely to visit next based on their behavioral history" refers to AI models and algorithms that analyze a user's past behavioral history and predict which website they are likely to visit next.

[0101] "Artificial intelligence model" refers to a mathematical model that uses machine learning algorithms to analyze data and perform specific tasks.

[0102] The present invention is a system for collecting information from multiple sources, unifying and purifying the information, training an artificial intelligence (AI) model, and providing customized AI models to the market or specific clients. Detailed embodiments of the present invention are described below.

[0103] 1. Data collection

[0104] The server collects data from multiple sources, such as web servers, mobile applications, social networking services, etc. It retrieves data from each source by making an API call and providing the necessary authentication information.

[0105] Examples:

[0106] The server uses the Google (registered trademark) Analytics API to obtain the website browsing history of a specific user.

[0107] The server uses the Firebase Analytics API to obtain mobile app usage data.

[0108] The server performs authentication using OAuth to obtain data from the SNS platform.

[0109] 2. Data formatting and cleaning

[0110] The collected data often comes in different formats, so the server unifies them and also detects missing and outlier values ​​and imputes them in an appropriate way.

[0111] Specific behavior:

[0112] The server uses the Python pandas library to unify data in different formats.

[0113] The server uses the mean or mode to impute missing values.

[0114] The server detects outliers and imputes them using a rational imputation method.

[0115] 3. Training the AI ​​model

[0116] Using the harmonized and cleansed data, the server trains AI models, improving their ability to predict user behavior and deliver optimal content.

[0117] Specific behavior:

[0118] The server builds and trains machine learning models using the Scikit-learn library.

[0119] The server generates a predictive model based on past behavioral data and evaluates its performance.

[0120] 4. Customization

[0121] The trained AI model is then customized based on each user's individual preferences, providing each user with the most personalized service possible.

[0122] Examples:

[0123] The server acquires the user's preference information (movies, music, sports, etc.) from a database.

[0124] The server fine-tunes the AI ​​model for each user segment to provide optimal recommendation results.

[0125] 5. Distribution and Sales

[0126] The final customized AI models are then distributed and sold to the market or to specific clients. The server uploads these models to the cloud or deploys them directly to the client's server.

[0127] Specific behavior:

[0128] The server uploads the completed AI model to an S3 bucket on Amazon Web Services (AWS (registered trademark)).

[0129] The server connects to a specific client server via SSH and deploys the AI ​​model.

[0130] Prompt Sentence Examples

[0131] Below are some example prompts to input to a generative AI model:

[0132] "Based on my browsing history over the past month, what website would you recommend next?"

[0133] This system allows for efficient data collection from multiple sources, shaping and cleaning, creating highly accurate AI models, providing personalized services based on user preferences, and efficiently distributing the models to markets and clients.

[0134] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0135] Step 1:

[0136] Data collection

[0137] Input: Data from multiple sources (web servers, mobile applications, social networking services).

[0138] Processing: The server retrieves data by making API calls to each source and providing the necessary authentication information: Google Analytics API to retrieve website browsing history, Firebase Analytics API to collect mobile app usage data, and OAuth authentication to access social media platforms.

[0139] Output: Raw data stored on the server.

[0140] Specific behavior:

[0141] The server uses the Google Analytics API to obtain the user's website browsing history.

[0142] The server uses the Firebase Analytics API to obtain mobile app usage data.

[0143] The server performs OAuth authentication to obtain the social networking data.

[0144] Step 2:

[0145] Data Shaping and Cleaning

[0146] Input: Raw data stored on the server.

[0147] Processing: The server converts the collected data into a unified format, and also detects missing and outlier values ​​and imputes them in a rational way.

[0148] Output: Unified and cleaned data.

[0149] Specific behavior:

[0150] The server uses the Python pandas library to unify data in different formats.

[0151] The server uses the mean or mode to impute missing values.

[0152] The server detects outliers and interpolates them by inferring reasonable values ​​from the surrounding data.

[0153] Step 3:

[0154] Training artificial intelligence models

[0155] Input: Unified and cleaned data.

[0156] Processing: The server uses the formatted data to train an AI model, specifically by building a machine learning model using the Scikit-learn library and applying the training data to generate a predictive model.

[0157] Output: The trained AI model.

[0158] Specific behavior:

[0159] The server builds machine learning models using the Scikit-learn library.

[0160] The server applies past behavioral data as training data to generate a predictive model.

[0161] The server evaluates the performance of the generated model and adjusts the hyperparameters as needed.

[0162] Step 4:

[0163] Customization

[0164] Input: Trained AI model, user preference information.

[0165] Processing: The server customizes the AI ​​model based on the user's individual preferences, thereby providing personalized services optimized for each individual user.

[0166] Output: A customized AI model.

[0167] Specific behavior:

[0168] The server acquires the user's preference information (e.g., movies, music, sports) from a database.

[0169] The server customizes the AI ​​model and recommends the best content for a specific user.

[0170] Step 5:

[0171] Distribution and Sales

[0172] Input: A customized AI model.

[0173] Processing: The server distributes and sells the AI ​​model to a marketplace or specific clients. The model is uploaded to the cloud or deployed directly to the client's server.

[0174] Output: AI model deployed to market or client.

[0175] Specific behavior:

[0176] The server uploads the completed AI model to an AWS S3 bucket.

[0177] The server connects to a specific client server via SSH and deploys the AI ​​model.

[0178] (Application example 1)

[0179] 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."

[0180] Current content distribution services lack sophisticated personalization based on each user's preferences and viewing history, making it difficult to recommend content that matches a user's interests. Furthermore, in addition to primary data processing, they lack integrated training of the AI ​​models required for cleaning and formatting collected data, as well as customization. Therefore, there is a need to provide more appropriate content to users.

[0181] 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.

[0182] In this invention, the server includes means for collecting data from multiple data sources, means for formatting and cleaning the collected data, and means for training an artificial intelligence model using the formatted and cleaned data, thereby enabling the provision of personalized services based on user preferences and enabling smart device applications that individually recommend content based on viewing history.

[0183] "Multiple data sources" refers to different types of information sources, such as web servers, mobile apps, and social media platforms.

[0184] "Means of collecting data" refers to the methods and systems that obtain information from each data source using API calls and authentication information.

[0185] "Shaping and cleaning methods" refers to the processes and techniques used to fill in missing values ​​in collected data and convert it into a unified format.

[0186] "Artificial intelligence model" refers to a model that uses machine learning algorithms to learn patterns from large amounts of data and make predictions and recommendations.

[0187] "Training" refers to the process of applying machine learning algorithms to improve the performance of an artificial intelligence model using collected and cleaned data.

[0188] "Preference-based customization" refers to the process of adjusting and individually optimizing artificial intelligence models based on each user's past behavioral data and interests.

[0189] "Distribution and Sale Means" refers to methods and systems for providing trained and customized artificial intelligence models to markets and clients.

[0190] "Viewing history" refers to a record of content that a user has viewed or listened to in the past.

[0191] "Personalized content recommendation" refers to the process of individually recommending specific content to each user based on their preferences and viewing history.

[0192] "Smart device applications" refers to applications that run on smartphones, smart glasses, and other digital devices.

[0193] This invention is a system that collects data from multiple data sources, shapes and cleans it, and then trains a customized artificial intelligence (AI) model to provide personalized content recommendations for a specific user. This system is particularly intended for use in smart device applications.

[0194] Data collection

[0195] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this process, the server uses authentication information to make the necessary API calls. The collected data includes user viewing history, clicks, browsing history, etc. For example, smartphone app usage history and social media activity can be included as data sources.

[0196] Data Shaping and Cleaning

[0197] The collected data is formatted into a unified format by the server, and missing values ​​and outliers are cleaned. For example, if data on the viewing history linked to a specific user ID is missing, it is supplemented from previous and subsequent data. The data format is also unified, formatted into a data frame, and then converted into a format that can be applied to machine learning algorithms.

[0198] Training an AI model

[0199] The server uses the cleaned data to train an AI model. During this process, machine learning algorithms are applied to build a model based on the user's behavioral patterns and viewing history. For example, algorithms such as Random Forest Classifier are used to create a model that predicts the content that a user is likely to watch next based on their viewing history.

[0200] Customization

[0201] Since it is difficult to apply a trained AI model to all users in its initial state, it is customized based on the preference information of each individual user. The server generates recommendations optimized for each user based on the content type preferred by that particular user. For example, based on the "video1, video2" that a user has previously watched, the server will recommend content that the user is likely to watch next.

[0202] Distribution and Sales

[0203] The final customized AI models are distributed and sold to the market or specific clients. The server uploads these models to the cloud or deploys them directly to client servers for commercialization.

[0204] Examples of concrete examples and prompts

[0205] As a specific example, a history of the user's past viewing of "video1, video2" is collected, and based on that, it is determined which content to recommend next.

[0206] Example prompt sentence:

[0207] Based on the past viewing history "video1, video2", please recommend the next video this user might watch.

[0208] This system makes it possible to provide highly accurate personalized services based on users' viewing history and preferences. The server collects and cleans data, and trains and customizes AI models to provide individually optimized content recommendations to smart device applications.

[0209] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0210] Step 1:

[0211] The server collects data from web servers, mobile apps, and social media platforms. This involves making API calls and using authentication credentials to retrieve the required data (view history, clicks, etc.) from each platform. The collected data is temporarily stored in local storage or a cloud database. The input here is the authentication credentials and API endpoints, and the output is the collected raw data.

[0212] Step 2:

[0213] The server formats and cleans the collected raw data. This includes filling in missing values, correcting outliers, and standardizing data formats. Specifically, it converts the data into a data frame using libraries such as Pandas, and fills in missing values ​​from the surrounding data. The input here is the collected raw data, and the output is a cleaned dataset.

[0214] Step 3:

[0215] The server uses the cleaned data to train an AI model. This process involves applying a machine learning algorithm (e.g., RandomForestClassifier) ​​and optimizing the model's parameters. Specifically, the data is divided into training data and test data, and a model is built using the training data. The input here is the cleaned dataset, and the output is a trained AI model.

[0216] Step 4:

[0217] The server customizes the trained AI model based on the user's preferences. This involves adjusting the model's predictions based on the user's past viewing history and click count. The server then uses the embodied user data to generate personalized recommendations. The input here is the user's preference data, and the output is the customized model's recommendations.

[0218] Step 5:

[0219] The server provides the customized AI model and its recommendation results to the smart device application. This process involves uploading it to the cloud or deploying it directly to the client's server. The user also receives personalized content recommendations through their smartphone or smart glasses. The input here is the customized AI model and the recommendation results, and the output is the display of recommended content on the user's smart device.

[0220] In this way, by linking each step, a highly personalized content recommendation system is realized.

[0221] 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.

[0222] This invention is a system that collects data from multiple data sources, formats and cleans the data to train an artificial intelligence (AI) model, and then combines it with an emotion engine that recognizes user preferences and emotions to provide customized AI models to the market or specific clients. This system executes each step centrally on a server, achieving highly accurate personalization based on the individual preferences and emotions of users.

[0223] Feature Overview

[0224] The system includes the following main features:

[0225] 1. Data collection

[0226] 2. Data formatting and cleaning

[0227] 3. Training the AI ​​model

[0228] 4. Customization

[0229] 5. Emotion recognition

[0230] 6. Distribution and Sales

[0231] Data collection

[0232] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this data collection process, the server provides authentication information to each data source and makes the necessary API calls. For example, it collects user visit history from web servers and usage history from mobile apps.

[0233] As a specific example, a server collects the user's smartphone app usage history as a data source, and periodically sends the data to the server, where it is stored in an integrated database.

[0234] Data Shaping and Cleaning

[0235] Since the collected data comes in various formats, the server converts it into a unified format. It also cleans missing or outlier values ​​if they are included. For example, if data associated with a specific user ID is missing, it performs the appropriate imputation process.

[0236] As a specific example, if there are missing values ​​in a user's website browsing history collected by the server, a cleaned dataset is created by filling in the missing values ​​with previous and subsequent data.

[0237] Training an AI model

[0238] Using the cleaned data, the server trains an AI model, which improves its ability to predict user behavior and provide optimal content to users, for example by using machine learning algorithms to find patterns in the training data.

[0239] As a specific example, a server trains a model based on a user's past browsing history to predict which website they are likely to visit next.

[0240] emotion recognition

[0241] The server incorporates an emotion engine that recognizes emotions using the user's voice, facial expressions, and text data as additional data sources, enabling data collection based on the user's emotional state. For example, speech recognition technology can be used to analyze the user's speech to determine the user's emotional state.

[0242] As a specific example, a server analyzes voice data obtained from an app installed on a user's smartphone and determines whether the user is happy or sad.

[0243] Customizing AI models

[0244] It is difficult to apply a trained AI model to all users as is, so the server customizes the AI ​​model based on the user's preference data and recognized emotional data, thereby providing personalized services optimized for each user.

[0245] For example, a server could generate an AI model that prioritizes recommending certain content to a particular user based on the type of content the user prefers, or prioritizes providing different content when the user is in a particular emotional state.

[0246] Distribution and Sales

[0247] The final customized AI models are then distributed and sold to the market or to specific clients. The server commercializes these models by uploading them to the cloud or by deploying them directly to client servers.

[0248] As a specific example, a server provides services by introducing a content recommendation system customized for a company and deploying an AI model directly on the company's server.

[0249] summary

[0250] This invention is a system that integrates data collection, formatting, cleaning, AI model training, emotion recognition, customization, and distribution. The server efficiently executes this series of processes, enabling highly personalized services based on user preferences and emotions. This system provides high-performance AI models based on information aggregated from multiple data sources, creating significant value for users.

[0251] The processing flow will be explained below.

[0252] Detailed description of the process

[0253] The process of data collection, formatting, cleaning, training AI models, emotion recognition, customizing AI models, distribution and sales

[0254] Step 1:

[0255] The server sets authentication information for multiple data sources such as web servers, mobile apps, and social media platforms, and prepares for data collection. Specifically, it sets API keys and secret keys to enable secure data collection from external data sources. For example, it calls an API to obtain user visit history from a web server.

[0256] Step 2:

[0257] The server sends data collection requests from each data source. Specifically, it collects user behavior data through APIs and stores it in an integrated database. During this process, it analyzes data in JSON and XML formats to extract the necessary information. For example, it collects data from mobile apps about user usage time and feature usage.

[0258] Step 3:

[0259] The server normalizes the collected data and formats it into a unified format. Specifically, it unifies different field names and data types and stores the data according to a common schema. For example, it can combine browsing history from a web server and usage history from a mobile app based on the same user ID.

[0260] Step 4:

[0261] The server cleans the formatted data. Specifically, it detects missing values ​​and outliers and performs appropriate imputation or deletion. For example, if the number of visits is abnormally high, that data is removed. If there are missing values, they are imputed with the mean or median.

[0262] Step 5:

[0263] The server splits the cleaned and shaped data into training data and test data by randomly shuffling the entire data and dividing it into 80% training data and 20% test data. This split allows for the evaluation of the model's performance.

[0264] Step 6:

[0265] The server uses the training data to train the AI ​​model. Specifically, it applies machine learning algorithms (e.g., random forests, neural networks) to learn features from the data. During the training process, it adjusts hyperparameters such as the learning rate and the number of epochs.

[0266] Step 7:

[0267] The server evaluates the trained model using test data. Specifically, it calculates metrics such as prediction accuracy, recall, and F-measure to verify the model's performance. Based on these results, the server fine-tunes the model to achieve optimal performance.

[0268] Step 8:

[0269] The server runs an emotion engine that recognizes emotions based on the user's voice, facial expression, and text data, and collects additional data. Specifically, it uses voice recognition, facial recognition, and text analysis technologies to identify the user's emotions and stores that information in an integrated database. For example, it can determine the user's emotional state, such as joy, anger, or sadness, from voice data.

[0270] Step 9:

[0271] The server then incorporates the recognized user emotion data into the training and customization process. Specifically, the emotion data is input as features into the AI ​​model, and a model is generated to make predictions and recommendations that take the user's emotions into account.

[0272] Step 10:

[0273] The server uses the training and emotion data to generate a customized AI model. Specifically, it combines the user's preference data and emotion data to generate an AI model optimized for each user.

[0274] Step 11:

[0275] The server deploys the customized AI model to a cloud server or a client server. Specifically, it exports the model and uploads it to the deployment environment using a secure communication protocol. After deployment, it checks and tests its operation in the production environment.

[0276] Step 12:

[0277] The server distributes and sells customized AI models to markets or specific clients, specifically by processing purchases of the models through marketplaces or sales platforms and providing buyers with license keys and installation instructions.

[0278] This provides a system that collects data from multiple sources, cleans, shapes, trains, recognizes emotions, customizes, and distributes them, enabling highly accurate personalization based on user preferences and emotions.

[0279] Example 2

[0280] 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."

[0281] In modern society, user needs are becoming more diverse, and there is a demand for personalized services based on individual preferences and emotions. However, conventional systems have difficulty collecting and formatting data from a variety of data sources, quickly and accurately grasping user preferences and emotions, and customizing AI models. This has led to issues such as a decline in the quality of service provided to users and a lack of improvement in satisfaction.

[0282] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0283] In this invention, the server includes means for collecting data from multiple information sources, means for shaping and cleaning the collected data, means for training an AI model using the shaped and cleaned data, means for customizing the trained AI model based on a user's preferences and emotions, means for distributing and selling the customized AI model, and means for recognizing the user's emotional state using voice, facial expression, and text data, thereby enabling the provision of highly accurate personalized services that meet the diverse needs of users.

[0284] "Sources" are the locations or systems from which data is collected, specifically web servers, mobile device apps, and social networking services.

[0285] "Data transformation" is the process of converting collected data into a consistent format or structure.

[0286] "Data cleaning" is the process of detecting missing or outliers from collected data and correcting or supplementing them in an appropriate manner.

[0287] An "artificial intelligence model" is a model trained using machine learning algorithms and used to predict user behavior and provide optimal content.

[0288] "Training" is the process of using collected data to teach an artificial intelligence model patterns.

[0289] "User preferences" refer to the type of content or service that a particular user prefers.

[0290] "Emotion recognition" is the process of analyzing a user's voice, facial expressions, and text data to determine their emotional state.

[0291] "Customization" is the process of adjusting a trained artificial intelligence model based on a user's preferences and emotions, optimizing it to provide services tailored to individual needs.

[0292] "Distribution" is the act of offering a customized artificial intelligence model to the market or to specific clients.

[0293] "Sales" is the process of commercially trading and obtaining payment for customized artificial intelligence models.

[0294] This invention is a system that collects data from multiple sources, formats and cleans the data to train an artificial intelligence (AI) model, and then combines it with an emotion engine that recognizes the user's preferences and emotions to provide a customized AI model. This system executes each step centrally on a server, achieving highly accurate personalization based on the user's individual preferences and emotions.

[0295] Data collection

[0296] The server collects data from web servers, mobile communication device apps, and social networking services. During this data collection process, the server provides the necessary authentication information to each information source and makes API calls. For example, the server obtains a user's smartphone app usage history and sends that data to the server to store in an integrated database. This system targets data such as user behavior history, posted content, comments, and usage time.

[0297] Data Shaping and Cleaning

[0298] The server formats the collected data into a unified format and cleans up missing and outlier values ​​within the data. For example, it converts website browsing history collected in different formats into a consistent format. If there is missing data, it complements it by referencing the preceding and following data. This results in a formatted and cleaned dataset, which is then used to train the AI ​​model, as described below.

[0299] Training an AI model

[0300] The server uses the cleaned data to train an AI model. In this process, machine learning algorithms are used to detect patterns in the collected data and generate a model that can predict behavior and recommend optimal content. For example, the server trains a model that predicts which website a user is likely to visit next based on their past website browsing history. This training process typically uses machine learning libraries such as Python's Scikit-learn or TENSORFLOW (registered trademark).

[0301] emotion recognition

[0302] The server analyzes voice, facial expressions, and text data to recognize the user's emotional state. For example, it uses voice recognition technology to analyze the user's emotional state (happiness, sadness, anger, etc.) from their speech. Analysis tools such as OpenCV and Google Cloud Speech-to-Text are used for this. As a specific example, it can analyze voice data obtained from an app installed on the user's smartphone to determine whether the user is happy or angry.

[0303] Customizing AI models

[0304] The server customizes the trained AI model based on the user's preferences and emotional data. This process allows for services optimized for individual users. For example, the server generates a model that preferentially recommends specific content to a particular user based on the type of content the user prefers. It is also possible to provide different content depending on the user's emotional state.

[0305] Distribution and Sales

[0306] The server provides customized AI models to the market or specific clients. Distribution methods include uploading the model to the cloud or deploying it directly to the client's server. For example, a content recommendation system customized for a company can be deployed to the client's server and provided as a service.

[0307] Example prompt statement

[0308] Examples of prompts to input to a generative AI model include:

[0309] "Predict what site a user might visit next based on their past website browsing data."

[0310] This invention is a system that efficiently collects, formats, cleans, trains AI models, recognizes emotions, customizes models, and distributes data, and provides highly personalized services based on user preferences and emotions, thereby creating high value for users.

[0311] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0312] Step 1:

[0313] Data collection

[0314] The server collects data from web servers, mobile communication device apps, and social networking services. In this process, the server provides authentication information to each information source and calls APIs to retrieve data. For example, the server obtains a user's smartphone app usage history and periodically sends that data to the server. The input data is the user's behavioral history data, and the output is the history data stored in an integrated database. Specifically, the server periodically schedules and executes data collection jobs.

[0315] Step 2:

[0316] Data Shaping and Cleaning

[0317] The server converts the collected data into a consistent format and cleans missing and outlier values. For example, it converts website browsing history in different formats into a common format. The input data is the original collected data, and the output is the converted and cleaned data. Specifically, it performs data format conversion processing and applies a missing value imputation algorithm to impute data.

[0318] Step 3:

[0319] Training an AI model

[0320] The server trains an AI model based on the cleaned data. It uses machine learning algorithms to detect patterns and generate a predictive model of user behavior. The input data is the cleaned and formatted dataset, and the output is the trained AI model. Specifically, it uses Python's Scikit-learn and TensorFlow to execute the model training process.

[0321] Step 4:

[0322] emotion recognition

[0323] The server analyzes voice, facial expressions, and text data to recognize the user's emotional state. For example, it uses voice recognition technology to analyze the user's speech and determine their emotional state. The input data is voice, facial expressions, and text data, and the output is the recognized emotional state. Specifically, it uses tools such as OpenCV and Google Cloud Speech-to-Text to analyze the voice and classify emotions.

[0324] Step 5:

[0325] Customizing AI models

[0326] The server customizes the trained AI model based on the user's preference and emotional data, thereby providing services optimized for each individual user. The input data is the user's preference and emotional data, and the output is a customized AI model. Specifically, the server fine-tunes the model's weights and parameters to generate an optimal model for each user.

[0327] Step 6:

[0328] Distribution and Sales

[0329] The server distributes and sells the customized AI model to the market or to specific clients. Distribution methods include uploading the model to the cloud or deploying it to the client's server. The input data is the customized AI model, and the output is the distributed model and a commercial contract. Specifically, the model is uploaded using a cloud storage service and provided to clients via an API.

[0330] (Application example 2)

[0331] 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."

[0332] In today's online shopping environment, users face difficulties in finding the right products for them from the vast amount of information available. Furthermore, online shopping sites lack the ability to appropriately recommend products based on users' individual preferences and emotional state. As a result, users' purchasing motivation declines and site sales do not increase. It is necessary to solve these problems and provide users with a highly personalized experience.

[0333] The identification processing 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 data from multiple data sources, means for formatting and cleaning the collected data, means for training an AI model using the formatted and cleaned data, means for customizing the AI ​​model based on user preferences and emotions, means for converting voice data into text and recognizing user emotions from the text data, and means for distributing and selling the customized AI model. This makes it possible to recommend products based on the user's individual preferences and emotional state, significantly improving the quality of the online shopping experience, increasing user satisfaction, and increasing sales.

[0334] A "data source" is a source of information for collecting information related to user behavior, preferences, emotions, etc.

[0335] "Formatting" is the process of converting collected data into a unified format and making it suitable for analysis.

[0336] "Cleaning" is a process for removing missing values ​​and outliers from collected data to maintain data quality.

[0337] An "artificial intelligence model" is an algorithm or mathematical model that performs a specific task by analyzing data and learning patterns.

[0338] "Training" is the process of using large amounts of data to train an artificial intelligence model and improve its accuracy.

[0339] "User preferences" refer to the tendency of a particular user to prefer certain products, services, or information.

[0340] "Customization" refers to adjusting a system or model to provide services and content optimized for individual users.

[0341] "Voice data" is a recording of the voice spoken by the user, and is data before it is converted into text information.

[0342] "Text data" refers to data in which voice data or other types of information are expressed as textual information.

[0343] "Emotion recognition" is a technology that analyzes data such as voice and text to determine a user's emotional state.

[0344] "Distribution" refers to providing a customized artificial intelligence model to a user or client.

[0345] "Sale" refers to the commercial trading of customized artificial intelligence models.

[0346] The present invention is a system that analyzes user behavior and emotions to provide personalized product recommendations through the steps of data collection, formatting and cleaning, AI model training, emotion recognition, customization, distribution, and sales.

[0347] Data collection

[0348] The server collects data from multiple sources, such as web servers, mobile apps, and social network platforms. This data includes users' browsing history, purchase history, voice data, and text data. For example, it can collect website visit history, mobile app usage history, and social networking platform posts.

[0349] Data Shaping and Cleaning

[0350] Since the collected data has different formats, the server converts them into a unified format. In addition, it detects and cleans missing and outlier values. Specifically, it complements missing data from the surrounding information and removes outliers.

[0351] Training an AI model

[0352] The server uses the cleaned and formatted data to train an AI model. It uses machine learning algorithms to learn user behavior patterns and build a model that predicts what users will want to buy next. For example, it uses a random forest classifier to predict future behavior based on users' browsing and purchase history.

[0353] emotion recognition

[0354] The server converts the voice data into text and recognizes the user's emotions from the text data. Using voice recognition technology, the server converts the user's speech into text and analyzes the emotions from that text. This makes it possible to determine the emotional state of the user. For example, the category of products the user is interested in can be determined from the text "My recent favorite is gadgets."

[0355] Customizing AI models

[0356] The trained AI model is customized based on the user's preferences and emotions, allowing it to provide optimal product recommendations for each user. For example, if a user is interested in gadgets, new gadget products will be recommended to that user first.

[0357] Distribution and Sales

[0358] Customized AI models are distributed and sold to markets or specific clients by the server. They are commercialized by being uploaded to the cloud or deployed directly to the client's server. For example, if a company provides a customized product recommendation system for a company, the AI ​​model is deployed on the company's server.

[0359] Specific prompt examples

[0360] "User text input: "My current favorite thing is gadgets.""

[0361] "Audio data: 'sample_audio.wav'"

[0362] As described above, the present invention provides a concrete means for implementing personalized product recommendations based on user behavior and emotions.

[0363] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0364] Step 1:

[0365] The server collects data from multiple data sources: user visit history from web servers, usage history from mobile apps, and user comments from social media platforms. These input data are sent to the server and stored in a centralized database. The output is an integrated raw data set.

[0366] Step 2:

[0367] The server formats the collected raw data and converts it into a unified format. In this process, data in different formats is unified. Next, a cleaning process is performed to detect missing values ​​and outliers and to impute or remove them. The input is the raw data collected in step 1, and the output is a clean dataset that has been formatted and cleaned.

[0368] Step 3:

[0369] The server uses the cleaned and sanitized data to train an AI model. This process uses a machine learning algorithm (e.g., a random forest classifier) ​​to learn user behavior patterns. The input is the clean dataset created in step 2, and the output is the trained AI model.

[0370] Step 4:

[0371] The server converts the voice data into text. Specifically, it uses voice recognition technology to analyze the user's speech and obtain it as text. The text data is then analyzed to recognize the user's emotions. The input is the voice data, and the output is the text data and its emotional evaluation.

[0372] Step 5:

[0373] The server customizes the trained AI model based on the user's preferences and emotions, thereby building a product recommendation model optimized for the specific user. The input is the AI ​​model from step 3 and the emotion evaluation from step 4, and the output is the customized AI model.

[0374] Step 6:

[0375] The server distributes and sells the customized AI model by uploading it to the cloud or by deploying it directly to the client's server. The input is the customized AI model created in step 5, and the output is an instance of the distributed AI model.

[0376] Specific operation example

[0377] "User text input: "My current favorite thing is gadgets.""

[0378] "Audio data: 'sample_audio.wav'"

[0379] In this specific example, if a user inputs the text "My recent favorite is gadgets" and provides audio data 'sample_audio.wav', the server can recommend personalized gadgets to this user through the processes of steps 1 to 6.

[0380] 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.

[0381] 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.

[0382] 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.

[0383] [Second embodiment]

[0384] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0385] 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.

[0386] 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).

[0387] 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.

[0388] 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.

[0389] 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).

[0390] 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. 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.

[0391] 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.

[0392] 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.

[0393] 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.

[0394] 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.

[0395] 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."

[0396] This invention is a system that collects data from multiple data sources, formats and cleans the data, trains an artificial intelligence (AI) model, and provides customized AI models to markets or specific clients. This system executes each step centrally on a server, achieving highly accurate personalization based on the individual preferences of users.

[0397] Feature Overview

[0398] The system includes the following main features:

[0399] 1. Data collection

[0400] 2. Data formatting and cleaning

[0401] 3. Training the AI ​​model

[0402] 4. Customization

[0403] 5. Distribution and Sales

[0404] Data collection

[0405] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this data collection process, the server provides authentication information to each data source and makes the necessary API calls. For example, the server collects user browsing history from web servers and usage history from mobile apps.

[0406] As a specific example, a server collects the user's smartphone app usage history as a data source, and periodically sends the data to the server, where it is stored in an integrated database.

[0407] Data Shaping and Cleaning

[0408] Since the collected data comes in various formats, the server converts it into a unified format. It also cleans missing or outlier values ​​if they are included. For example, if data associated with a specific user ID is missing, it performs the appropriate imputation process.

[0409] As a specific example, if there are missing values ​​in a user's website browsing history collected by the server, a cleaned dataset is created by filling in the missing values ​​with previous and subsequent data.

[0410] Training an AI model

[0411] Using the cleaned data, the server trains an AI model, which improves its ability to predict user behavior and provide optimal content to users, for example by using machine learning algorithms to find patterns in the training data.

[0412] As a specific example, a server trains a model based on a user's past browsing history to predict which website they are likely to visit next.

[0413] Customization

[0414] It is difficult to apply a trained AI model to all users as is, so the server customizes the AI ​​model based on the preferences of each individual user, providing a personalized service optimized for each user.

[0415] As a specific example, a server generates an AI model that preferentially recommends specific content to a particular user based on the type of content the user prefers.

[0416] Distribution and Sales

[0417] The final customized AI models are then distributed and sold to the market or to specific clients. The server commercializes these models by uploading them to the cloud or by deploying them directly to client servers.

[0418] As a specific example, a server provides services by introducing a content recommendation system customized for a company and deploying an AI model directly on the company's server.

[0419] summary

[0420] This invention is a system that integrates data collection, formatting, cleaning, training, customization, and sales. The server efficiently executes this series of processes, realizing highly personalized services based on user preferences. This system provides high-performance AI models based on information aggregated from multiple data sources, creating significant value for users.

[0421] The processing flow will be explained below.

[0422] Detailed description of the process

[0423] Data collection and formatting process

[0424] Step 1:

[0425] The server sets authentication information for multiple data sources such as web servers, mobile apps, and social media platforms, and prepares for data collection. Specifically, by setting API keys and secret keys, data can be securely collected from external data sources. For example, the web server calls an API to obtain user visit history.

[0426] Step 2:

[0427] The server sends data collection requests from each data source. Specifically, it collects user behavior data through APIs and stores it in an integrated database. During this process, it analyzes data in JSON and XML formats to extract the necessary information. For example, it collects data from mobile apps about user usage time and feature usage.

[0428] Step 3:

[0429] The server normalizes the collected data and formats it into a unified format. Specifically, it unifies different field names and data types and stores the data according to a common schema. For example, it can combine browsing history from a web server and usage history from a mobile app based on the same user ID.

[0430] Data Cleaning Process

[0431] Step 4:

[0432] The server cleans the formatted data. Specifically, it detects missing values ​​and outliers and performs appropriate imputation or deletion. For example, if the number of visits is abnormally high, that data is removed. If there are missing values, they are imputed with the mean or median.

[0433] AI model training process

[0434] Step 5:

[0435] The server splits the cleaned and shaped data into training data and test data by randomly shuffling the entire data and dividing it into 80% training data and 20% test data. This split allows for the evaluation of the model's performance.

[0436] Step 6:

[0437] The server uses the training data to train the AI ​​model. Specifically, it applies machine learning algorithms (e.g., random forests, neural networks) to learn features from the data. During the training process, it adjusts hyperparameters such as the learning rate and the number of epochs.

[0438] Step 7:

[0439] The server evaluates the trained model using test data. Specifically, it calculates metrics such as prediction accuracy, recall, and F-measure to verify the model's performance. Based on these results, the server fine-tunes the model to achieve optimal performance.

[0440] Customization and Distribution Process

[0441] Step 8:

[0442] The server collects user preference data and customizes the trained model by adjusting the model parameters based on the user's past behavioral data and explicit preferences (e.g., favorite genres, frequently used features).

[0443] Step 9:

[0444] The server deploys the customized AI model to a cloud server or a client server. Specifically, it exports the model and uploads it to the deployment environment using a secure communication protocol. After deployment, it checks and tests its operation in the production environment.

[0445] Step 10:

[0446] The server distributes and sells customized AI models to markets or specific clients, specifically by processing purchases of the models through marketplaces or sales platforms and providing buyers with license keys and installation instructions.

[0447] In this way, a system is realized that collects, cleans, shapes, trains, customizes, and distributes data from multiple sources, enabling highly accurate personalization based on user preferences.

[0448] Example 1

[0449] 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."

[0450] Existing AI systems require a lot of time and effort to effectively collect, shape, and clean data from multiple disparate sources. It is also difficult to create highly accurate models based on the cleaned data and customize them to suit individual user preferences. Furthermore, efficiently distributing customized models and making them commercially available has been a challenge.

[0451] 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.

[0452] In this invention, the server includes means for collecting information from multiple sources, means for unifying and purifying the collected information, means for training an AI model using the unified and purified information, means for periodically transmitting information from terminals and storing it in an integrated database, means for identifying and completing missing values ​​and outliers in the purification process, and means for predicting the site a user is likely to visit next based on the user's past behavior history. This enables efficient data collection from multiple data sources, shaping and cleaning, the creation and customization of highly accurate AI models based on user preferences, and the efficient distribution and commercialization of the models.

[0453] "Source" refers to the systems or platforms that provide data, such as web servers, mobile applications, and social networking services.

[0454] "Means of collection" refers to the technical means of providing API calls and credentials to retrieve data from multiple sources.

[0455] "Harmonization means" refers to procedures and methods for converting data of different formats into a unified format.

[0456] "Cleaning methods" refers to technical means for shaping and cleaning data and filling in missing or outlier values.

[0457] "Training methods" refers to machine learning algorithms and techniques used to create artificial intelligence models using collected and cleaned data.

[0458] "Means for periodically transmitting information" refers to the settings and technical means for periodically transmitting information from the terminal to the server and storing it in a database.

[0459] "Means for identifying and imputing missing and outliers" refers to technology for detecting missing and outliers in data and imputing them in an appropriate manner.

[0460] "Methods for predicting which site a user is likely to visit next based on their behavioral history" refers to AI models and algorithms that analyze a user's past behavioral history and predict which website they are likely to visit next.

[0461] "Artificial intelligence model" refers to a mathematical model that uses machine learning algorithms to analyze data and perform specific tasks.

[0462] The present invention is a system for collecting information from multiple sources, unifying and purifying the information, training an artificial intelligence (AI) model, and providing customized AI models to the market or specific clients. Detailed embodiments of the present invention are described below.

[0463] 1. Data collection

[0464] The server collects data from multiple sources, such as web servers, mobile applications, social networking services, etc. It retrieves data from each source by making an API call and providing the necessary authentication information.

[0465] Examples:

[0466] The server uses the Google Analytics API to obtain the website browsing history of a specific user.

[0467] The server uses the Firebase Analytics API to obtain mobile app usage data.

[0468] The server performs authentication using OAuth to obtain data from the SNS platform.

[0469] 2. Data formatting and cleaning

[0470] The collected data often comes in different formats, so the server unifies them and also detects missing and outlier values ​​and imputes them in an appropriate way.

[0471] Specific behavior:

[0472] The server uses the Python pandas library to unify data in different formats.

[0473] The server uses the mean or mode to impute missing values.

[0474] The server detects outliers and imputes them using a rational imputation method.

[0475] 3. Training the AI ​​model

[0476] Using the harmonized and cleansed data, the server trains AI models, improving their ability to predict user behavior and deliver optimal content.

[0477] Specific behavior:

[0478] The server builds and trains machine learning models using the Scikit-learn library.

[0479] The server generates a predictive model based on past behavioral data and evaluates its performance.

[0480] 4. Customization

[0481] The trained AI model is then customized based on each user's individual preferences, providing each user with the most personalized service possible.

[0482] Examples:

[0483] The server acquires the user's preference information (movies, music, sports, etc.) from a database.

[0484] The server fine-tunes the AI ​​model for each user segment to provide optimal recommendation results.

[0485] 5. Distribution and Sales

[0486] The final customized AI models are then distributed and sold to the market or to specific clients. The server uploads these models to the cloud or deploys them directly to the client's server.

[0487] Specific behavior:

[0488] The server uploads the completed AI model to an S3 bucket on Amazon Web Services (AWS).

[0489] The server connects to a specific client server via SSH and deploys the AI ​​model.

[0490] Prompt Sentence Examples

[0491] Below are some example prompts to input to a generative AI model:

[0492] "Based on my browsing history over the past month, what website would you recommend next?"

[0493] This system allows for efficient data collection from multiple sources, shaping and cleaning, creating highly accurate AI models, providing personalized services based on user preferences, and efficiently distributing the models to markets and clients.

[0494] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0495] Step 1:

[0496] Data collection

[0497] Input: Data from multiple sources (web servers, mobile applications, social networking services).

[0498] Processing: The server retrieves data by making API calls to each source and providing the necessary authentication information: Google Analytics API to retrieve website browsing history, Firebase Analytics API to collect mobile app usage data, and OAuth authentication to access social media platforms.

[0499] Output: Raw data stored on the server.

[0500] Specific behavior:

[0501] The server uses the Google Analytics API to obtain the user's website browsing history.

[0502] The server uses the Firebase Analytics API to obtain mobile app usage data.

[0503] The server performs OAuth authentication to obtain the social networking data.

[0504] Step 2:

[0505] Data Shaping and Cleaning

[0506] Input: Raw data stored on the server.

[0507] Processing: The server converts the collected data into a unified format, and also detects missing and outlier values ​​and imputes them in a rational way.

[0508] Output: Unified and cleaned data.

[0509] Specific behavior:

[0510] The server uses the Python pandas library to unify data in different formats.

[0511] The server uses the mean or mode to impute missing values.

[0512] The server detects outliers and interpolates them by inferring reasonable values ​​from the surrounding data.

[0513] Step 3:

[0514] Training artificial intelligence models

[0515] Input: Unified and cleaned data.

[0516] Processing: The server uses the formatted data to train an AI model, specifically by building a machine learning model using the Scikit-learn library and applying the training data to generate a predictive model.

[0517] Output: The trained AI model.

[0518] Specific behavior:

[0519] The server builds machine learning models using the Scikit-learn library.

[0520] The server applies past behavioral data as training data to generate a predictive model.

[0521] The server evaluates the performance of the generated model and adjusts the hyperparameters as needed.

[0522] Step 4:

[0523] Customization

[0524] Input: Trained AI model, user preference information.

[0525] Processing: The server customizes the AI ​​model based on the user's individual preferences, thereby providing personalized services optimized for each individual user.

[0526] Output: A customized AI model.

[0527] Specific behavior:

[0528] The server acquires the user's preference information (e.g., movies, music, sports) from a database.

[0529] The server customizes the AI ​​model and recommends the best content for a specific user.

[0530] Step 5:

[0531] Distribution and Sales

[0532] Input: A customized AI model.

[0533] Processing: The server distributes and sells the AI ​​model to a marketplace or specific clients. The model is uploaded to the cloud or deployed directly to the client's server.

[0534] Output: AI model deployed to market or client.

[0535] Specific behavior:

[0536] The server uploads the completed AI model to an AWS S3 bucket.

[0537] The server connects to a specific client server via SSH and deploys the AI ​​model.

[0538] (Application example 1)

[0539] 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."

[0540] Current content distribution services lack sophisticated personalization based on each user's preferences and viewing history, making it difficult to recommend content that matches a user's interests. Furthermore, in addition to primary data processing, they lack integrated training of the AI ​​models required for cleaning and formatting collected data, as well as customization. Therefore, there is a need to provide more appropriate content to users.

[0541] 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.

[0542] In this invention, the server includes means for collecting data from multiple data sources, means for formatting and cleaning the collected data, and means for training an artificial intelligence model using the formatted and cleaned data, thereby enabling the provision of personalized services based on user preferences and enabling smart device applications that individually recommend content based on viewing history.

[0543] "Multiple data sources" refers to different types of information sources, such as web servers, mobile apps, and social media platforms.

[0544] "Means of collecting data" refers to the methods and systems that obtain information from each data source using API calls and authentication information.

[0545] "Shaping and cleaning methods" refers to the processes and techniques used to fill in missing values ​​in collected data and convert it into a unified format.

[0546] "Artificial intelligence model" refers to a model that uses machine learning algorithms to learn patterns from large amounts of data and make predictions and recommendations.

[0547] "Training" refers to the process of applying machine learning algorithms to improve the performance of an artificial intelligence model using collected and cleaned data.

[0548] "Preference-based customization" refers to the process of adjusting and individually optimizing artificial intelligence models based on each user's past behavioral data and interests.

[0549] "Distribution and Sale Means" refers to methods and systems for providing trained and customized artificial intelligence models to markets and clients.

[0550] "Viewing history" refers to a record of content that a user has viewed or listened to in the past.

[0551] "Personalized content recommendation" refers to the process of individually recommending specific content to each user based on their preferences and viewing history.

[0552] "Smart device applications" refers to applications that run on smartphones, smart glasses, and other digital devices.

[0553] This invention is a system that collects data from multiple data sources, shapes and cleans it, and then trains a customized artificial intelligence (AI) model to provide personalized content recommendations for a specific user. This system is particularly intended for use in smart device applications.

[0554] Data collection

[0555] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this process, the server uses authentication information to make the necessary API calls. The collected data includes user viewing history, clicks, browsing history, etc. For example, smartphone app usage history and social media activity can be included as data sources.

[0556] Data Shaping and Cleaning

[0557] The collected data is formatted into a unified format by the server, and missing values ​​and outliers are cleaned. For example, if data on the viewing history linked to a specific user ID is missing, it is supplemented from previous and subsequent data. The data format is also unified, formatted into a data frame, and then converted into a format that can be applied to machine learning algorithms.

[0558] Training an AI model

[0559] The server uses the cleaned data to train an AI model. During this process, machine learning algorithms are applied to build a model based on the user's behavioral patterns and viewing history. For example, algorithms such as Random Forest Classifier are used to create a model that predicts the content that a user is likely to watch next based on their viewing history.

[0560] Customization

[0561] Since it is difficult to apply a trained AI model to all users in its initial state, it is customized based on the preference information of each individual user. The server generates recommendations optimized for each user based on the content type preferred by that particular user. For example, based on the "video1, video2" that a user has previously watched, the server will recommend content that the user is likely to watch next.

[0562] Distribution and Sales

[0563] The final customized AI models are distributed and sold to the market or specific clients. The server uploads these models to the cloud or deploys them directly to client servers for commercialization.

[0564] Examples of concrete examples and prompts

[0565] As a specific example, a history of the user's past viewing of "video1, video2" is collected, and based on that, it is determined which content to recommend next.

[0566] Example prompt sentence:

[0567] Based on the past viewing history "video1, video2", please recommend the next video this user might watch.

[0568] This system makes it possible to provide highly accurate personalized services based on users' viewing history and preferences. The server collects and cleans data, and trains and customizes AI models to provide individually optimized content recommendations to smart device applications.

[0569] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0570] Step 1:

[0571] The server collects data from web servers, mobile apps, and social media platforms. This involves making API calls and using authentication credentials to retrieve the required data (view history, clicks, etc.) from each platform. The collected data is temporarily stored in local storage or a cloud database. The input here is the authentication credentials and API endpoints, and the output is the collected raw data.

[0572] Step 2:

[0573] The server formats and cleans the collected raw data. This includes filling in missing values, correcting outliers, and standardizing data formats. Specifically, it converts the data into a data frame using libraries such as Pandas, and fills in missing values ​​from the surrounding data. The input here is the collected raw data, and the output is a cleaned dataset.

[0574] Step 3:

[0575] The server uses the cleaned data to train an AI model. This process involves applying a machine learning algorithm (e.g., RandomForestClassifier) ​​and optimizing the model's parameters. Specifically, the data is divided into training data and test data, and a model is built using the training data. The input here is the cleaned dataset, and the output is a trained AI model.

[0576] Step 4:

[0577] The server customizes the trained AI model based on the user's preferences. This involves adjusting the model's predictions based on the user's past viewing history and click count. The server then uses the embodied user data to generate personalized recommendations. The input here is the user's preference data, and the output is the customized model's recommendations.

[0578] Step 5:

[0579] The server provides the customized AI model and its recommendation results to the smart device application. This process involves uploading it to the cloud or deploying it directly to the client's server. The user also receives personalized content recommendations through their smartphone or smart glasses. The input here is the customized AI model and the recommendation results, and the output is the display of recommended content on the user's smart device.

[0580] In this way, by linking each step, a highly personalized content recommendation system is realized.

[0581] 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.

[0582] This invention is a system that collects data from multiple data sources, formats and cleans the data to train an artificial intelligence (AI) model, and then combines it with an emotion engine that recognizes user preferences and emotions to provide customized AI models to the market or specific clients. This system executes each step centrally on a server, achieving highly accurate personalization based on the individual preferences and emotions of users.

[0583] Feature Overview

[0584] The system includes the following main features:

[0585] 1. Data collection

[0586] 2. Data formatting and cleaning

[0587] 3. Training the AI ​​model

[0588] 4. Customization

[0589] 5. Emotion recognition

[0590] 6. Distribution and Sales

[0591] Data collection

[0592] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this data collection process, the server provides authentication information to each data source and makes the necessary API calls. For example, it collects user visit history from web servers and usage history from mobile apps.

[0593] As a specific example, a server collects the user's smartphone app usage history as a data source, and periodically sends the data to the server, where it is stored in an integrated database.

[0594] Data Shaping and Cleaning

[0595] Since the collected data comes in various formats, the server converts it into a unified format. It also cleans missing or outlier values ​​if they are included. For example, if data associated with a specific user ID is missing, it performs the appropriate imputation process.

[0596] As a specific example, if there are missing values ​​in a user's website browsing history collected by the server, a cleaned dataset is created by filling in the missing values ​​with previous and subsequent data.

[0597] Training an AI model

[0598] Using the cleaned data, the server trains an AI model, which improves its ability to predict user behavior and provide optimal content to users, for example by using machine learning algorithms to find patterns in the training data.

[0599] As a specific example, a server trains a model based on a user's past browsing history to predict which website they are likely to visit next.

[0600] emotion recognition

[0601] The server incorporates an emotion engine that recognizes emotions using the user's voice, facial expressions, and text data as additional data sources, enabling data collection based on the user's emotional state. For example, speech recognition technology can be used to analyze the user's speech to determine the user's emotional state.

[0602] As a specific example, a server analyzes voice data obtained from an app installed on a user's smartphone and determines whether the user is happy or sad.

[0603] Customizing AI models

[0604] It is difficult to apply a trained AI model to all users as is, so the server customizes the AI ​​model based on the user's preference data and recognized emotional data, thereby providing personalized services optimized for each user.

[0605] For example, a server could generate an AI model that prioritizes recommending certain content to a particular user based on the type of content the user prefers, or prioritizes providing different content when the user is in a particular emotional state.

[0606] Distribution and Sales

[0607] The final customized AI models are then distributed and sold to the market or to specific clients. The server commercializes these models by uploading them to the cloud or by deploying them directly to client servers.

[0608] As a specific example, a server provides services by introducing a content recommendation system customized for a company and deploying an AI model directly on the company's server.

[0609] summary

[0610] This invention is a system that integrates data collection, formatting, cleaning, AI model training, emotion recognition, customization, and distribution. The server efficiently executes this series of processes, enabling highly personalized services based on user preferences and emotions. This system provides high-performance AI models based on information aggregated from multiple data sources, creating significant value for users.

[0611] The processing flow will be explained below.

[0612] Detailed description of the process

[0613] The process of data collection, formatting, cleaning, training AI models, emotion recognition, customizing AI models, distribution and sales

[0614] Step 1:

[0615] The server sets authentication information for multiple data sources such as web servers, mobile apps, and social media platforms, and prepares for data collection. Specifically, it sets API keys and secret keys to enable secure data collection from external data sources. For example, it calls an API to obtain user visit history from a web server.

[0616] Step 2:

[0617] The server sends data collection requests from each data source. Specifically, it collects user behavior data through APIs and stores it in an integrated database. During this process, it analyzes data in JSON and XML formats to extract the necessary information. For example, it collects data from mobile apps about user usage time and feature usage.

[0618] Step 3:

[0619] The server normalizes the collected data and formats it into a unified format. Specifically, it unifies different field names and data types and stores the data according to a common schema. For example, it can combine browsing history from a web server and usage history from a mobile app based on the same user ID.

[0620] Step 4:

[0621] The server cleans the formatted data. Specifically, it detects missing values ​​and outliers and performs appropriate imputation or deletion. For example, if the number of visits is abnormally high, that data is removed. If there are missing values, they are imputed with the mean or median.

[0622] Step 5:

[0623] The server splits the cleaned and shaped data into training data and test data by randomly shuffling the entire data and dividing it into 80% training data and 20% test data. This split allows for the evaluation of the model's performance.

[0624] Step 6:

[0625] The server uses the training data to train the AI ​​model. Specifically, it applies machine learning algorithms (e.g., random forests, neural networks) to learn features from the data. During the training process, it adjusts hyperparameters such as the learning rate and the number of epochs.

[0626] Step 7:

[0627] The server evaluates the trained model using test data. Specifically, it calculates metrics such as prediction accuracy, recall, and F-measure to verify the model's performance. Based on these results, the server fine-tunes the model to achieve optimal performance.

[0628] Step 8:

[0629] The server runs an emotion engine that recognizes emotions based on the user's voice, facial expression, and text data, and collects additional data. Specifically, it uses voice recognition, facial recognition, and text analysis technologies to identify the user's emotions and stores that information in an integrated database. For example, it can determine the user's emotional state, such as joy, anger, or sadness, from voice data.

[0630] Step 9:

[0631] The server then incorporates the recognized user emotion data into the training and customization process. Specifically, the emotion data is input as features into the AI ​​model, and a model is generated to make predictions and recommendations that take the user's emotions into account.

[0632] Step 10:

[0633] The server uses the training and emotion data to generate a customized AI model. Specifically, it combines the user's preference data and emotion data to generate an AI model optimized for each user.

[0634] Step 11:

[0635] The server deploys the customized AI model to a cloud server or a client server. Specifically, it exports the model and uploads it to the deployment environment using a secure communication protocol. After deployment, it checks and tests its operation in the production environment.

[0636] Step 12:

[0637] The server distributes and sells customized AI models to markets or specific clients, specifically by processing purchases of the models through marketplaces or sales platforms and providing buyers with license keys and installation instructions.

[0638] This provides a system that collects data from multiple sources, cleans, shapes, trains, recognizes emotions, customizes, and distributes them, enabling highly accurate personalization based on user preferences and emotions.

[0639] Example 2

[0640] 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."

[0641] In modern society, user needs are becoming more diverse, and there is a demand for personalized services based on individual preferences and emotions. However, conventional systems have difficulty collecting and formatting data from a variety of data sources, quickly and accurately grasping user preferences and emotions, and customizing AI models. This has led to issues such as a decline in the quality of service provided to users and a lack of improvement in satisfaction.

[0642] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0643] In this invention, the server includes means for collecting data from multiple information sources, means for shaping and cleaning the collected data, means for training an AI model using the shaped and cleaned data, means for customizing the trained AI model based on a user's preferences and emotions, means for distributing and selling the customized AI model, and means for recognizing the user's emotional state using voice, facial expression, and text data, thereby enabling the provision of highly accurate personalized services that meet the diverse needs of users.

[0644] "Sources" are the locations or systems from which data is collected, specifically web servers, mobile device apps, and social networking services.

[0645] "Data transformation" is the process of converting collected data into a consistent format or structure.

[0646] "Data cleaning" is the process of detecting missing or outliers from collected data and correcting or supplementing them in an appropriate manner.

[0647] An "artificial intelligence model" is a model trained using machine learning algorithms and used to predict user behavior and provide optimal content.

[0648] "Training" is the process of using collected data to teach an artificial intelligence model patterns.

[0649] "User preferences" refer to the type of content or service that a particular user prefers.

[0650] "Emotion recognition" is the process of analyzing a user's voice, facial expressions, and text data to determine their emotional state.

[0651] "Customization" is the process of adjusting a trained artificial intelligence model based on a user's preferences and emotions, optimizing it to provide services tailored to individual needs.

[0652] "Distribution" is the act of offering a customized artificial intelligence model to the market or to specific clients.

[0653] "Sales" is the process of commercially trading and obtaining payment for customized artificial intelligence models.

[0654] This invention is a system that collects data from multiple sources, formats and cleans the data to train an artificial intelligence (AI) model, and then combines it with an emotion engine that recognizes the user's preferences and emotions to provide a customized AI model. This system executes each step centrally on a server, achieving highly accurate personalization based on the user's individual preferences and emotions.

[0655] Data collection

[0656] The server collects data from web servers, mobile communication device apps, and social networking services. During this data collection process, the server provides the necessary authentication information to each information source and makes API calls. For example, the server obtains a user's smartphone app usage history and sends that data to the server to store in an integrated database. This system targets data such as user behavior history, posted content, comments, and usage time.

[0657] Data Shaping and Cleaning

[0658] The server formats the collected data into a unified format and cleans up missing and outlier values ​​within the data. For example, it converts website browsing history collected in different formats into a consistent format. If there is missing data, it complements it by referencing the preceding and following data. This results in a formatted and cleaned dataset, which is then used to train the AI ​​model, as described below.

[0659] Training an AI model

[0660] The server uses the cleaned data to train an AI model. This process uses machine learning algorithms to detect patterns in the collected data and generate a model that can predict behavior and recommend optimal content. For example, the server trains a model that predicts which website a user is likely to visit next based on their past website browsing history. This training process is typically performed using machine learning libraries such as Python's Scikit-learn and TensorFlow.

[0661] emotion recognition

[0662] The server analyzes voice, facial expressions, and text data to recognize the user's emotional state. For example, it uses voice recognition technology to analyze the user's emotional state (happiness, sadness, anger, etc.) from their speech. Analysis tools such as OpenCV and Google Cloud Speech-to-Text are used for this. As a specific example, it can analyze voice data obtained from an app installed on the user's smartphone to determine whether the user is happy or angry.

[0663] Customizing AI models

[0664] The server customizes the trained AI model based on the user's preferences and emotional data. This process allows for services optimized for individual users. For example, the server generates a model that preferentially recommends specific content to a particular user based on the type of content the user prefers. It is also possible to provide different content depending on the user's emotional state.

[0665] Distribution and Sales

[0666] The server provides customized AI models to the market or specific clients. Distribution methods include uploading the model to the cloud or deploying it directly to the client's server. For example, a content recommendation system customized for a company can be deployed to the client's server and provided as a service.

[0667] Example prompt statement

[0668] Examples of prompts to input to a generative AI model include:

[0669] "Predict what site a user might visit next based on their past website browsing data."

[0670] This invention is a system that efficiently collects, formats, cleans, trains AI models, recognizes emotions, customizes models, and distributes data, and provides highly personalized services based on user preferences and emotions, thereby creating high value for users.

[0671] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0672] Step 1:

[0673] Data collection

[0674] The server collects data from web servers, mobile communication device apps, and social networking services. In this process, the server provides authentication information to each information source and calls APIs to retrieve data. For example, the server obtains a user's smartphone app usage history and periodically sends that data to the server. The input data is the user's behavioral history data, and the output is the history data stored in an integrated database. Specifically, the server periodically schedules and executes data collection jobs.

[0675] Step 2:

[0676] Data Shaping and Cleaning

[0677] The server converts the collected data into a consistent format and cleans missing and outlier values. For example, it converts website browsing history in different formats into a common format. The input data is the original collected data, and the output is the converted and cleaned data. Specifically, it performs data format conversion processing and applies a missing value imputation algorithm to impute data.

[0678] Step 3:

[0679] Training an AI model

[0680] The server trains an AI model based on the cleaned data. It uses machine learning algorithms to detect patterns and generate a predictive model of user behavior. The input data is the cleaned and formatted dataset, and the output is the trained AI model. Specifically, it uses Python's Scikit-learn and TensorFlow to execute the model training process.

[0681] Step 4:

[0682] emotion recognition

[0683] The server analyzes voice, facial expressions, and text data to recognize the user's emotional state. For example, it uses voice recognition technology to analyze the user's speech and determine their emotional state. The input data is voice, facial expressions, and text data, and the output is the recognized emotional state. Specifically, it uses tools such as OpenCV and Google Cloud Speech-to-Text to analyze the voice and classify emotions.

[0684] Step 5:

[0685] Customizing AI models

[0686] The server customizes the trained AI model based on the user's preference and emotional data, thereby providing services optimized for each individual user. The input data is the user's preference and emotional data, and the output is a customized AI model. Specifically, the server fine-tunes the model's weights and parameters to generate an optimal model for each user.

[0687] Step 6:

[0688] Distribution and Sales

[0689] The server distributes and sells the customized AI model to the market or to specific clients. Distribution methods include uploading the model to the cloud or deploying it to the client's server. The input data is the customized AI model, and the output is the distributed model and a commercial contract. Specifically, the model is uploaded using a cloud storage service and provided to clients via an API.

[0690] (Application example 2)

[0691] 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."

[0692] In today's online shopping environment, users face difficulties in finding the right products for them from the vast amount of information available. Furthermore, online shopping sites lack the ability to appropriately recommend products based on users' individual preferences and emotional state. As a result, users' purchasing motivation declines and site sales do not increase. It is necessary to solve these problems and provide users with a highly personalized experience.

[0693] The identification processing 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 data from multiple data sources, means for formatting and cleaning the collected data, means for training an AI model using the formatted and cleaned data, means for customizing the AI ​​model based on user preferences and emotions, means for converting voice data into text and recognizing user emotions from the text data, and means for distributing and selling the customized AI model. This makes it possible to recommend products based on the user's individual preferences and emotional state, significantly improving the quality of the online shopping experience, increasing user satisfaction, and increasing sales.

[0694] A "data source" is a source of information for collecting information related to user behavior, preferences, emotions, etc.

[0695] "Formatting" is the process of converting collected data into a unified format and making it suitable for analysis.

[0696] "Cleaning" is a process for removing missing values ​​and outliers from collected data to maintain data quality.

[0697] An "artificial intelligence model" is an algorithm or mathematical model that performs a specific task by analyzing data and learning patterns.

[0698] "Training" is the process of using large amounts of data to train an artificial intelligence model and improve its accuracy.

[0699] "User preferences" refer to the tendency of a particular user to prefer certain products, services, or information.

[0700] "Customization" refers to adjusting a system or model to provide services and content optimized for individual users.

[0701] "Voice data" is a recording of the voice spoken by the user, and is data before it is converted into text information.

[0702] "Text data" refers to data in which voice data or other types of information are expressed as textual information.

[0703] "Emotion recognition" is a technology that analyzes data such as voice and text to determine a user's emotional state.

[0704] "Distribution" refers to providing a customized artificial intelligence model to a user or client.

[0705] "Sale" refers to the commercial trading of customized artificial intelligence models.

[0706] The present invention is a system that analyzes user behavior and emotions to provide personalized product recommendations through the steps of data collection, formatting and cleaning, AI model training, emotion recognition, customization, distribution, and sales.

[0707] Data collection

[0708] The server collects data from multiple sources, such as web servers, mobile apps, and social network platforms. This data includes users' browsing history, purchase history, voice data, and text data. For example, it can collect website visit history, mobile app usage history, and social networking platform posts.

[0709] Data Shaping and Cleaning

[0710] Since the collected data has different formats, the server converts them into a unified format. In addition, it detects and cleans missing and outlier values. Specifically, it complements missing data from the surrounding information and removes outliers.

[0711] Training an AI model

[0712] The server uses the cleaned and formatted data to train an AI model. It uses machine learning algorithms to learn user behavior patterns and build a model that predicts what users will want to buy next. For example, it uses a random forest classifier to predict future behavior based on users' browsing and purchase history.

[0713] emotion recognition

[0714] The server converts the voice data into text and recognizes the user's emotions from the text data. Using voice recognition technology, the server converts the user's speech into text and analyzes the emotions from that text. This makes it possible to determine the emotional state of the user. For example, the category of products the user is interested in can be determined from the text "My recent favorite is gadgets."

[0715] Customizing AI models

[0716] The trained AI model is customized based on the user's preferences and emotions, allowing it to provide optimal product recommendations for each user. For example, if a user is interested in gadgets, new gadget products will be recommended to that user first.

[0717] Distribution and Sales

[0718] Customized AI models are distributed and sold to markets or specific clients by the server. They are commercialized by being uploaded to the cloud or deployed directly to the client's server. For example, if a company provides a customized product recommendation system for a company, the AI ​​model is deployed on the company's server.

[0719] Specific prompt examples

[0720] "User text input: "My current favorite thing is gadgets.""

[0721] "Audio data: 'sample_audio.wav'"

[0722] As described above, the present invention provides a concrete means for implementing personalized product recommendations based on user behavior and emotions.

[0723] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0724] Step 1:

[0725] The server collects data from multiple data sources: user visit history from web servers, usage history from mobile apps, and user comments from social media platforms. These input data are sent to the server and stored in a centralized database. The output is an integrated raw data set.

[0726] Step 2:

[0727] The server formats the collected raw data and converts it into a unified format. In this process, data in different formats is unified. Next, a cleaning process is performed to detect missing values ​​and outliers and to impute or remove them. The input is the raw data collected in step 1, and the output is a clean dataset that has been formatted and cleaned.

[0728] Step 3:

[0729] The server uses the cleaned and sanitized data to train an AI model. This process uses a machine learning algorithm (e.g., a random forest classifier) ​​to learn user behavior patterns. The input is the clean dataset created in step 2, and the output is the trained AI model.

[0730] Step 4:

[0731] The server converts the voice data into text. Specifically, it uses voice recognition technology to analyze the user's speech and obtain it as text. The text data is then analyzed to recognize the user's emotions. The input is the voice data, and the output is the text data and its emotional evaluation.

[0732] Step 5:

[0733] The server customizes the trained AI model based on the user's preferences and emotions, thereby building a product recommendation model optimized for the specific user. The input is the AI ​​model from step 3 and the emotion evaluation from step 4, and the output is the customized AI model.

[0734] Step 6:

[0735] The server distributes and sells the customized AI model by uploading it to the cloud or by deploying it directly to the client's server. The input is the customized AI model created in step 5, and the output is an instance of the distributed AI model.

[0736] Specific operation example

[0737] "User text input: "My current favorite thing is gadgets.""

[0738] "Audio data: 'sample_audio.wav'"

[0739] In this specific example, if a user inputs the text "My recent favorite is gadgets" and provides audio data 'sample_audio.wav', the server can recommend personalized gadgets to this user through the processes of steps 1 to 6.

[0740] 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.

[0741] 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.

[0742] 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.

[0743] [Third embodiment]

[0744] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0745] 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.

[0746] 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).

[0747] 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.

[0748] 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.

[0749] 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).

[0750] 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. 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.

[0751] 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.

[0752] 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.

[0753] 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.

[0754] 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.

[0755] 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."

[0756] This invention is a system that collects data from multiple data sources, formats and cleans the data, trains an artificial intelligence (AI) model, and provides customized AI models to markets or specific clients. This system executes each step centrally on a server, achieving highly accurate personalization based on the individual preferences of users.

[0757] Feature Overview

[0758] The system includes the following main features:

[0759] 1. Data collection

[0760] 2. Data formatting and cleaning

[0761] 3. Training the AI ​​model

[0762] 4. Customization

[0763] 5. Distribution and Sales

[0764] Data collection

[0765] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this data collection process, the server provides authentication information to each data source and makes the necessary API calls. For example, the server collects user browsing history from web servers and usage history from mobile apps.

[0766] As a specific example, a server collects the user's smartphone app usage history as a data source, and periodically sends the data to the server, where it is stored in an integrated database.

[0767] Data Shaping and Cleaning

[0768] Since the collected data comes in various formats, the server converts it into a unified format. It also cleans missing or outlier values ​​if they are included. For example, if data associated with a specific user ID is missing, it performs the appropriate imputation process.

[0769] As a specific example, if there are missing values ​​in a user's website browsing history collected by the server, a cleaned dataset is created by filling in the missing values ​​with previous and subsequent data.

[0770] Training an AI model

[0771] Using the cleaned data, the server trains an AI model, which improves its ability to predict user behavior and provide optimal content to users, for example by using machine learning algorithms to find patterns in the training data.

[0772] As a specific example, a server trains a model based on a user's past browsing history to predict which website they are likely to visit next.

[0773] Customization

[0774] It is difficult to apply a trained AI model to all users as is, so the server customizes the AI ​​model based on the preferences of each individual user, providing a personalized service optimized for each user.

[0775] As a specific example, a server generates an AI model that preferentially recommends specific content to a particular user based on the type of content the user prefers.

[0776] Distribution and Sales

[0777] The final customized AI models are then distributed and sold to the market or to specific clients. The server commercializes these models by uploading them to the cloud or by deploying them directly to client servers.

[0778] As a specific example, a server provides services by introducing a content recommendation system customized for a company and deploying an AI model directly on the company's server.

[0779] summary

[0780] This invention is a system that integrates data collection, formatting, cleaning, training, customization, and sales. The server efficiently executes this series of processes, realizing highly personalized services based on user preferences. This system provides high-performance AI models based on information aggregated from multiple data sources, creating significant value for users.

[0781] The processing flow will be explained below.

[0782] Detailed description of the process

[0783] Data collection and formatting process

[0784] Step 1:

[0785] The server sets authentication information for multiple data sources such as web servers, mobile apps, and social media platforms, and prepares for data collection. Specifically, by setting API keys and secret keys, data can be securely collected from external data sources. For example, the web server calls an API to obtain user visit history.

[0786] Step 2:

[0787] The server sends data collection requests from each data source. Specifically, it collects user behavior data through APIs and stores it in an integrated database. During this process, it analyzes data in JSON and XML formats to extract the necessary information. For example, it collects data from mobile apps about user usage time and feature usage.

[0788] Step 3:

[0789] The server normalizes the collected data and formats it into a unified format. Specifically, it unifies different field names and data types and stores the data according to a common schema. For example, it can combine browsing history from a web server and usage history from a mobile app based on the same user ID.

[0790] Data Cleaning Process

[0791] Step 4:

[0792] The server cleans the formatted data. Specifically, it detects missing values ​​and outliers and performs appropriate imputation or deletion. For example, if the number of visits is abnormally high, that data is removed. If there are missing values, they are imputed with the mean or median.

[0793] AI model training process

[0794] Step 5:

[0795] The server splits the cleaned and shaped data into training data and test data by randomly shuffling the entire data and dividing it into 80% training data and 20% test data. This split allows for the evaluation of the model's performance.

[0796] Step 6:

[0797] The server uses the training data to train the AI ​​model. Specifically, it applies machine learning algorithms (e.g., random forests, neural networks) to learn features from the data. During the training process, it adjusts hyperparameters such as the learning rate and the number of epochs.

[0798] Step 7:

[0799] The server evaluates the trained model using test data. Specifically, it calculates metrics such as prediction accuracy, recall, and F-measure to verify the model's performance. Based on these results, the server fine-tunes the model to achieve optimal performance.

[0800] Customization and Distribution Process

[0801] Step 8:

[0802] The server collects user preference data and customizes the trained model by adjusting the model parameters based on the user's past behavioral data and explicit preferences (e.g., favorite genres, frequently used features).

[0803] Step 9:

[0804] The server deploys the customized AI model to a cloud server or a client server. Specifically, it exports the model and uploads it to the deployment environment using a secure communication protocol. After deployment, it checks and tests its operation in the production environment.

[0805] Step 10:

[0806] The server distributes and sells customized AI models to markets or specific clients, specifically by processing purchases of the models through marketplaces or sales platforms and providing buyers with license keys and installation instructions.

[0807] In this way, a system is realized that collects, cleans, shapes, trains, customizes, and distributes data from multiple sources, enabling highly accurate personalization based on user preferences.

[0808] Example 1

[0809] 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."

[0810] Existing AI systems require a lot of time and effort to effectively collect, shape, and clean data from multiple disparate sources. It is also difficult to create highly accurate models based on the cleaned data and customize them to suit individual user preferences. Furthermore, efficiently distributing customized models and making them commercially available has been a challenge.

[0811] 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.

[0812] In this invention, the server includes means for collecting information from multiple sources, means for unifying and purifying the collected information, means for training an AI model using the unified and purified information, means for periodically transmitting information from terminals and storing it in an integrated database, means for identifying and completing missing values ​​and outliers in the purification process, and means for predicting the site a user is likely to visit next based on the user's past behavior history. This enables efficient data collection from multiple data sources, shaping and cleaning, the creation and customization of highly accurate AI models based on user preferences, and the efficient distribution and commercialization of the models.

[0813] "Source" refers to the systems or platforms that provide data, such as web servers, mobile applications, and social networking services.

[0814] "Means of collection" refers to the technical means of providing API calls and credentials to retrieve data from multiple sources.

[0815] "Harmonization means" refers to procedures and methods for converting data of different formats into a unified format.

[0816] "Cleaning methods" refers to technical means for shaping and cleaning data and filling in missing or outlier values.

[0817] "Training methods" refers to machine learning algorithms and techniques used to create artificial intelligence models using collected and cleaned data.

[0818] "Means for periodically transmitting information" refers to the settings and technical means for periodically transmitting information from the terminal to the server and storing it in a database.

[0819] "Means for identifying and imputing missing and outliers" refers to technology for detecting missing and outliers in data and imputing them in an appropriate manner.

[0820] "Methods for predicting which site a user is likely to visit next based on their behavioral history" refers to AI models and algorithms that analyze a user's past behavioral history and predict which website they are likely to visit next.

[0821] "Artificial intelligence model" refers to a mathematical model that uses machine learning algorithms to analyze data and perform specific tasks.

[0822] The present invention is a system for collecting information from multiple sources, unifying and purifying the information, training an artificial intelligence (AI) model, and providing customized AI models to the market or specific clients. Detailed embodiments of the present invention are described below.

[0823] 1. Data collection

[0824] The server collects data from multiple sources, such as web servers, mobile applications, social networking services, etc. It retrieves data from each source by making an API call and providing the necessary authentication information.

[0825] Examples:

[0826] The server uses the Google Analytics API to obtain the website browsing history of a specific user.

[0827] The server uses the Firebase Analytics API to obtain mobile app usage data.

[0828] The server performs authentication using OAuth to obtain data from the SNS platform.

[0829] 2. Data formatting and cleaning

[0830] The collected data often comes in different formats, so the server unifies them and also detects missing and outlier values ​​and imputes them in an appropriate way.

[0831] Specific behavior:

[0832] The server uses the Python pandas library to unify data in different formats.

[0833] The server uses the mean or mode to impute missing values.

[0834] The server detects outliers and imputes them using a rational imputation method.

[0835] 3. Training the AI ​​model

[0836] Using the harmonized and cleansed data, the server trains AI models, improving their ability to predict user behavior and deliver optimal content.

[0837] Specific behavior:

[0838] The server builds and trains machine learning models using the Scikit-learn library.

[0839] The server generates a predictive model based on past behavioral data and evaluates its performance.

[0840] 4. Customization

[0841] The trained AI model is then customized based on each user's individual preferences, providing each user with the most personalized service possible.

[0842] Examples:

[0843] The server acquires the user's preference information (movies, music, sports, etc.) from a database.

[0844] The server fine-tunes the AI ​​model for each user segment to provide optimal recommendation results.

[0845] 5. Distribution and Sales

[0846] The final customized AI models are then distributed and sold to the market or to specific clients. The server uploads these models to the cloud or deploys them directly to the client's server.

[0847] Specific behavior:

[0848] The server uploads the completed AI model to an S3 bucket on Amazon Web Services (AWS).

[0849] The server connects to a specific client server via SSH and deploys the AI ​​model.

[0850] Prompt Sentence Examples

[0851] Below are some example prompts to input to a generative AI model:

[0852] "Based on my browsing history over the past month, what website would you recommend next?"

[0853] This system allows for efficient data collection from multiple sources, shaping and cleaning, creating highly accurate AI models, providing personalized services based on user preferences, and efficiently distributing the models to markets and clients.

[0854] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0855] Step 1:

[0856] Data collection

[0857] Input: Data from multiple sources (web servers, mobile applications, social networking services).

[0858] Processing: The server retrieves data by making API calls to each source and providing the necessary authentication information: Google Analytics API to retrieve website browsing history, Firebase Analytics API to collect mobile app usage data, and OAuth authentication to access social media platforms.

[0859] Output: Raw data stored on the server.

[0860] Specific behavior:

[0861] The server uses the Google Analytics API to obtain the user's website browsing history.

[0862] The server uses the Firebase Analytics API to obtain mobile app usage data.

[0863] The server performs OAuth authentication to obtain the social networking data.

[0864] Step 2:

[0865] Data Shaping and Cleaning

[0866] Input: Raw data stored on the server.

[0867] Processing: The server converts the collected data into a unified format, and also detects missing and outlier values ​​and imputes them in a rational way.

[0868] Output: Unified and cleaned data.

[0869] Specific behavior:

[0870] The server uses the Python pandas library to unify data in different formats.

[0871] The server uses the mean or mode to impute missing values.

[0872] The server detects outliers and interpolates them by inferring reasonable values ​​from the surrounding data.

[0873] Step 3:

[0874] Training artificial intelligence models

[0875] Input: Unified and cleaned data.

[0876] Processing: The server uses the formatted data to train an AI model, specifically by building a machine learning model using the Scikit-learn library and applying the training data to generate a predictive model.

[0877] Output: The trained AI model.

[0878] Specific behavior:

[0879] The server builds machine learning models using the Scikit-learn library.

[0880] The server applies past behavioral data as training data to generate a predictive model.

[0881] The server evaluates the performance of the generated model and adjusts the hyperparameters as needed.

[0882] Step 4:

[0883] Customization

[0884] Input: Trained AI model, user preference information.

[0885] Processing: The server customizes the AI ​​model based on the user's individual preferences, thereby providing personalized services optimized for each individual user.

[0886] Output: A customized AI model.

[0887] Specific behavior:

[0888] The server acquires the user's preference information (e.g., movies, music, sports) from a database.

[0889] The server customizes the AI ​​model and recommends the best content for a specific user.

[0890] Step 5:

[0891] Distribution and Sales

[0892] Input: A customized AI model.

[0893] Processing: The server distributes and sells the AI ​​model to a marketplace or specific clients. The model is uploaded to the cloud or deployed directly to the client's server.

[0894] Output: AI model deployed to market or client.

[0895] Specific behavior:

[0896] The server uploads the completed AI model to an AWS S3 bucket.

[0897] The server connects to a specific client server via SSH and deploys the AI ​​model.

[0898] (Application example 1)

[0899] 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."

[0900] Current content distribution services lack sophisticated personalization based on each user's preferences and viewing history, making it difficult to recommend content that matches a user's interests. Furthermore, in addition to primary data processing, they lack integrated training of the AI ​​models required for cleaning and formatting collected data, as well as customization. Therefore, there is a need to provide more appropriate content to users.

[0901] 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.

[0902] In this invention, the server includes means for collecting data from multiple data sources, means for formatting and cleaning the collected data, and means for training an artificial intelligence model using the formatted and cleaned data, thereby enabling the provision of personalized services based on user preferences and enabling smart device applications that individually recommend content based on viewing history.

[0903] "Multiple data sources" refers to different types of information sources, such as web servers, mobile apps, and social media platforms.

[0904] "Means of collecting data" refers to the methods and systems that obtain information from each data source using API calls and authentication information.

[0905] "Shaping and cleaning methods" refers to the processes and techniques used to fill in missing values ​​in collected data and convert it into a unified format.

[0906] "Artificial intelligence model" refers to a model that uses machine learning algorithms to learn patterns from large amounts of data and make predictions and recommendations.

[0907] "Training" refers to the process of applying machine learning algorithms to improve the performance of an artificial intelligence model using collected and cleaned data.

[0908] "Preference-based customization" refers to the process of adjusting and individually optimizing artificial intelligence models based on each user's past behavioral data and interests.

[0909] "Distribution and Sale Means" refers to methods and systems for providing trained and customized artificial intelligence models to markets and clients.

[0910] "Viewing history" refers to a record of content that a user has viewed or listened to in the past.

[0911] "Personalized content recommendation" refers to the process of individually recommending specific content to each user based on their preferences and viewing history.

[0912] "Smart device applications" refers to applications that run on smartphones, smart glasses, and other digital devices.

[0913] This invention is a system that collects data from multiple data sources, shapes and cleans it, and then trains a customized artificial intelligence (AI) model to provide personalized content recommendations for a specific user. This system is particularly intended for use in smart device applications.

[0914] Data collection

[0915] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this process, the server uses authentication information to make the necessary API calls. The collected data includes user viewing history, clicks, browsing history, etc. For example, smartphone app usage history and social media activity can be included as data sources.

[0916] Data Shaping and Cleaning

[0917] The collected data is formatted into a unified format by the server, and missing values ​​and outliers are cleaned. For example, if data on the viewing history linked to a specific user ID is missing, it is supplemented from previous and subsequent data. The data format is also unified, formatted into a data frame, and then converted into a format that can be applied to machine learning algorithms.

[0918] Training an AI model

[0919] The server uses the cleaned data to train an AI model. During this process, machine learning algorithms are applied to build a model based on the user's behavioral patterns and viewing history. For example, algorithms such as Random Forest Classifier are used to create a model that predicts the content that a user is likely to watch next based on their viewing history.

[0920] Customization

[0921] Since it is difficult to apply a trained AI model to all users in its initial state, it is customized based on the preference information of each individual user. The server generates recommendations optimized for each user based on the content type preferred by that particular user. For example, based on the "video1, video2" that a user has previously watched, the server will recommend content that the user is likely to watch next.

[0922] Distribution and Sales

[0923] The final customized AI models are distributed and sold to the market or specific clients. The server uploads these models to the cloud or deploys them directly to client servers for commercialization.

[0924] Examples of concrete examples and prompts

[0925] As a specific example, a history of the user's past viewing of "video1, video2" is collected, and based on that, it is determined which content to recommend next.

[0926] Example prompt sentence:

[0927] Based on the past viewing history "video1, video2", please recommend the next video this user might watch.

[0928] This system makes it possible to provide highly accurate personalized services based on users' viewing history and preferences. The server collects and cleans data, and trains and customizes AI models to provide individually optimized content recommendations to smart device applications.

[0929] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0930] Step 1:

[0931] The server collects data from web servers, mobile apps, and social media platforms. This involves making API calls and using authentication credentials to retrieve the required data (view history, clicks, etc.) from each platform. The collected data is temporarily stored in local storage or a cloud database. The input here is the authentication credentials and API endpoints, and the output is the collected raw data.

[0932] Step 2:

[0933] The server formats and cleans the collected raw data. This includes filling in missing values, correcting outliers, and standardizing data formats. Specifically, it converts the data into a data frame using libraries such as Pandas, and fills in missing values ​​from the surrounding data. The input here is the collected raw data, and the output is a cleaned dataset.

[0934] Step 3:

[0935] The server uses the cleaned data to train an AI model. This process involves applying a machine learning algorithm (e.g., RandomForestClassifier) ​​and optimizing the model's parameters. Specifically, the data is divided into training data and test data, and a model is built using the training data. The input here is the cleaned dataset, and the output is a trained AI model.

[0936] Step 4:

[0937] The server customizes the trained AI model based on the user's preferences. This involves adjusting the model's predictions based on the user's past viewing history and click count. The server then uses the embodied user data to generate personalized recommendations. The input here is the user's preference data, and the output is the customized model's recommendations.

[0938] Step 5:

[0939] The server provides the customized AI model and its recommendation results to the smart device application. This process involves uploading it to the cloud or deploying it directly to the client's server. The user also receives personalized content recommendations through their smartphone or smart glasses. The input here is the customized AI model and the recommendation results, and the output is the display of recommended content on the user's smart device.

[0940] In this way, by linking each step, a highly personalized content recommendation system is realized.

[0941] 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.

[0942] This invention is a system that collects data from multiple data sources, formats and cleans the data to train an artificial intelligence (AI) model, and then combines it with an emotion engine that recognizes user preferences and emotions to provide customized AI models to the market or specific clients. This system executes each step centrally on a server, achieving highly accurate personalization based on the individual preferences and emotions of users.

[0943] Feature Overview

[0944] The system includes the following main features:

[0945] 1. Data collection

[0946] 2. Data formatting and cleaning

[0947] 3. Training the AI ​​model

[0948] 4. Customization

[0949] 5. Emotion recognition

[0950] 6. Distribution and Sales

[0951] Data collection

[0952] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this data collection process, the server provides authentication information to each data source and makes the necessary API calls. For example, it collects user visit history from web servers and usage history from mobile apps.

[0953] As a specific example, a server collects the user's smartphone app usage history as a data source, and periodically sends the data to the server, where it is stored in an integrated database.

[0954] Data Shaping and Cleaning

[0955] Since the collected data comes in various formats, the server converts it into a unified format. It also cleans missing or outlier values ​​if they are included. For example, if data associated with a specific user ID is missing, it performs the appropriate imputation process.

[0956] As a specific example, if there are missing values ​​in a user's website browsing history collected by the server, a cleaned dataset is created by filling in the missing values ​​with previous and subsequent data.

[0957] Training an AI model

[0958] Using the cleaned data, the server trains an AI model, which improves its ability to predict user behavior and provide optimal content to users, for example by using machine learning algorithms to find patterns in the training data.

[0959] As a specific example, a server trains a model based on a user's past browsing history to predict which website they are likely to visit next.

[0960] emotion recognition

[0961] The server incorporates an emotion engine that recognizes emotions using the user's voice, facial expressions, and text data as additional data sources, enabling data collection based on the user's emotional state. For example, speech recognition technology can be used to analyze the user's speech to determine the user's emotional state.

[0962] As a specific example, a server analyzes voice data obtained from an app installed on a user's smartphone and determines whether the user is happy or sad.

[0963] Customizing AI models

[0964] It is difficult to apply a trained AI model to all users as is, so the server customizes the AI ​​model based on the user's preference data and recognized emotional data, thereby providing personalized services optimized for each user.

[0965] For example, a server could generate an AI model that prioritizes recommending certain content to a particular user based on the type of content the user prefers, or prioritizes providing different content when the user is in a particular emotional state.

[0966] Distribution and Sales

[0967] The final customized AI models are then distributed and sold to the market or to specific clients. The server commercializes these models by uploading them to the cloud or by deploying them directly to client servers.

[0968] As a specific example, a server provides services by introducing a content recommendation system customized for a company and deploying an AI model directly on the company's server.

[0969] summary

[0970] This invention is a system that integrates data collection, formatting, cleaning, AI model training, emotion recognition, customization, and distribution. The server efficiently executes this series of processes, enabling highly personalized services based on user preferences and emotions. This system provides high-performance AI models based on information aggregated from multiple data sources, creating significant value for users.

[0971] The processing flow will be explained below.

[0972] Detailed description of the process

[0973] The process of data collection, formatting, cleaning, training AI models, emotion recognition, customizing AI models, distribution and sales

[0974] Step 1:

[0975] The server sets authentication information for multiple data sources such as web servers, mobile apps, and social media platforms, and prepares for data collection. Specifically, it sets API keys and secret keys to enable secure data collection from external data sources. For example, it calls an API to obtain user visit history from a web server.

[0976] Step 2:

[0977] The server sends data collection requests from each data source. Specifically, it collects user behavior data through APIs and stores it in an integrated database. During this process, it analyzes data in JSON and XML formats to extract the necessary information. For example, it collects data from mobile apps about user usage time and feature usage.

[0978] Step 3:

[0979] The server normalizes the collected data and formats it into a unified format. Specifically, it unifies different field names and data types and stores the data according to a common schema. For example, it can combine browsing history from a web server and usage history from a mobile app based on the same user ID.

[0980] Step 4:

[0981] The server cleans the formatted data. Specifically, it detects missing values ​​and outliers and performs appropriate imputation or deletion. For example, if the number of visits is abnormally high, that data is removed. If there are missing values, they are imputed with the mean or median.

[0982] Step 5:

[0983] The server splits the cleaned and shaped data into training data and test data by randomly shuffling the entire data and dividing it into 80% training data and 20% test data. This split allows for the evaluation of the model's performance.

[0984] Step 6:

[0985] The server uses the training data to train the AI ​​model. Specifically, it applies machine learning algorithms (e.g., random forests, neural networks) to learn features from the data. During the training process, it adjusts hyperparameters such as the learning rate and the number of epochs.

[0986] Step 7:

[0987] The server evaluates the trained model using test data. Specifically, it calculates metrics such as prediction accuracy, recall, and F-measure to verify the model's performance. Based on these results, the server fine-tunes the model to achieve optimal performance.

[0988] Step 8:

[0989] The server runs an emotion engine that recognizes emotions based on the user's voice, facial expression, and text data, and collects additional data. Specifically, it uses voice recognition, facial recognition, and text analysis technologies to identify the user's emotions and stores that information in an integrated database. For example, it can determine the user's emotional state, such as joy, anger, or sadness, from voice data.

[0990] Step 9:

[0991] The server then incorporates the recognized user emotion data into the training and customization process. Specifically, the emotion data is input as features into the AI ​​model, and a model is generated to make predictions and recommendations that take the user's emotions into account.

[0992] Step 10:

[0993] The server uses the training and emotion data to generate a customized AI model. Specifically, it combines the user's preference data and emotion data to generate an AI model optimized for each user.

[0994] Step 11:

[0995] The server deploys the customized AI model to a cloud server or a client server. Specifically, it exports the model and uploads it to the deployment environment using a secure communication protocol. After deployment, it checks and tests its operation in the production environment.

[0996] Step 12:

[0997] The server distributes and sells customized AI models to markets or specific clients, specifically by processing purchases of the models through marketplaces or sales platforms and providing buyers with license keys and installation instructions.

[0998] This provides a system that collects data from multiple sources, cleans, shapes, trains, recognizes emotions, customizes, and distributes them, enabling highly accurate personalization based on user preferences and emotions.

[0999] Example 2

[1000] 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."

[1001] In modern society, user needs are becoming more diverse, and there is a demand for personalized services based on individual preferences and emotions. However, conventional systems have difficulty collecting and formatting data from a variety of data sources, quickly and accurately grasping user preferences and emotions, and customizing AI models. This has led to issues such as a decline in the quality of service provided to users and a lack of improvement in satisfaction.

[1002] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1003] In this invention, the server includes means for collecting data from multiple information sources, means for shaping and cleaning the collected data, means for training an AI model using the shaped and cleaned data, means for customizing the trained AI model based on a user's preferences and emotions, means for distributing and selling the customized AI model, and means for recognizing the user's emotional state using voice, facial expression, and text data, thereby enabling the provision of highly accurate personalized services that meet the diverse needs of users.

[1004] "Sources" are the locations or systems from which data is collected, specifically web servers, mobile device apps, and social networking services.

[1005] "Data transformation" is the process of converting collected data into a consistent format or structure.

[1006] "Data cleaning" is the process of detecting missing or outliers from collected data and correcting or supplementing them in an appropriate manner.

[1007] An "artificial intelligence model" is a model trained using machine learning algorithms and used to predict user behavior and provide optimal content.

[1008] "Training" is the process of using collected data to teach an artificial intelligence model patterns.

[1009] "User preferences" refer to the type of content or service that a particular user prefers.

[1010] "Emotion recognition" is the process of analyzing a user's voice, facial expressions, and text data to determine their emotional state.

[1011] "Customization" is the process of adjusting a trained artificial intelligence model based on a user's preferences and emotions, optimizing it to provide services tailored to individual needs.

[1012] "Distribution" is the act of offering a customized artificial intelligence model to the market or to specific clients.

[1013] "Sales" is the process of commercially trading and obtaining payment for customized artificial intelligence models.

[1014] This invention is a system that collects data from multiple sources, formats and cleans the data to train an artificial intelligence (AI) model, and then combines it with an emotion engine that recognizes the user's preferences and emotions to provide a customized AI model. This system executes each step centrally on a server, achieving highly accurate personalization based on the user's individual preferences and emotions.

[1015] Data collection

[1016] The server collects data from web servers, mobile communication device apps, and social networking services. During this data collection process, the server provides the necessary authentication information to each information source and makes API calls. For example, the server obtains a user's smartphone app usage history and sends that data to the server to store in an integrated database. This system targets data such as user behavior history, posted content, comments, and usage time.

[1017] Data Shaping and Cleaning

[1018] The server formats the collected data into a unified format and cleans up missing and outlier values ​​within the data. For example, it converts website browsing history collected in different formats into a consistent format. If there is missing data, it complements it by referencing the preceding and following data. This results in a formatted and cleaned dataset, which is then used to train the AI ​​model, as described below.

[1019] Training an AI model

[1020] The server uses the cleaned data to train an AI model. This process uses machine learning algorithms to detect patterns in the collected data and generate a model that can predict behavior and recommend optimal content. For example, the server trains a model that predicts which website a user is likely to visit next based on their past website browsing history. This training process is typically performed using machine learning libraries such as Python's Scikit-learn and TensorFlow.

[1021] emotion recognition

[1022] The server analyzes voice, facial expressions, and text data to recognize the user's emotional state. For example, it uses voice recognition technology to analyze the user's emotional state (happiness, sadness, anger, etc.) from their speech. Analysis tools such as OpenCV and Google Cloud Speech-to-Text are used for this. As a specific example, it can analyze voice data obtained from an app installed on the user's smartphone to determine whether the user is happy or angry.

[1023] Customizing AI models

[1024] The server customizes the trained AI model based on the user's preferences and emotional data. This process allows for services optimized for individual users. For example, the server generates a model that preferentially recommends specific content to a particular user based on the type of content the user prefers. It is also possible to provide different content depending on the user's emotional state.

[1025] Distribution and Sales

[1026] The server provides customized AI models to the market or specific clients. Distribution methods include uploading the model to the cloud or deploying it directly to the client's server. For example, a content recommendation system customized for a company can be deployed to the client's server and provided as a service.

[1027] Example prompt statement

[1028] Examples of prompts to input to a generative AI model include:

[1029] "Predict what site a user might visit next based on their past website browsing data."

[1030] This invention is a system that efficiently collects, formats, cleans, trains AI models, recognizes emotions, customizes models, and distributes data, and provides highly personalized services based on user preferences and emotions, thereby creating high value for users.

[1031] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1032] Step 1:

[1033] Data collection

[1034] The server collects data from web servers, mobile communication device apps, and social networking services. In this process, the server provides authentication information to each information source and calls APIs to retrieve data. For example, the server obtains a user's smartphone app usage history and periodically sends that data to the server. The input data is the user's behavioral history data, and the output is the history data stored in an integrated database. Specifically, the server periodically schedules and executes data collection jobs.

[1035] Step 2:

[1036] Data Shaping and Cleaning

[1037] The server converts the collected data into a consistent format and cleans missing and outlier values. For example, it converts website browsing history in different formats into a common format. The input data is the original collected data, and the output is the converted and cleaned data. Specifically, it performs data format conversion processing and applies a missing value imputation algorithm to impute data.

[1038] Step 3:

[1039] Training an AI model

[1040] The server trains an AI model based on the cleaned data. It uses machine learning algorithms to detect patterns and generate a predictive model of user behavior. The input data is the cleaned and formatted dataset, and the output is the trained AI model. Specifically, it uses Python's Scikit-learn and TensorFlow to execute the model training process.

[1041] Step 4:

[1042] emotion recognition

[1043] The server analyzes voice, facial expressions, and text data to recognize the user's emotional state. For example, it uses voice recognition technology to analyze the user's speech and determine their emotional state. The input data is voice, facial expressions, and text data, and the output is the recognized emotional state. Specifically, it uses tools such as OpenCV and Google Cloud Speech-to-Text to analyze the voice and classify emotions.

[1044] Step 5:

[1045] Customizing AI models

[1046] The server customizes the trained AI model based on the user's preference and emotional data, thereby providing services optimized for each individual user. The input data is the user's preference and emotional data, and the output is a customized AI model. Specifically, the server fine-tunes the model's weights and parameters to generate an optimal model for each user.

[1047] Step 6:

[1048] Distribution and Sales

[1049] The server distributes and sells the customized AI model to the market or to specific clients. Distribution methods include uploading the model to the cloud or deploying it to the client's server. The input data is the customized AI model, and the output is the distributed model and a commercial contract. Specifically, the model is uploaded using a cloud storage service and provided to clients via an API.

[1050] (Application example 2)

[1051] 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."

[1052] In today's online shopping environment, users face difficulties in finding the right products for them from the vast amount of information available. Furthermore, online shopping sites lack the ability to appropriately recommend products based on users' individual preferences and emotional state. As a result, users' purchasing motivation declines and site sales do not increase. It is necessary to solve these problems and provide users with a highly personalized experience.

[1053] The identification processing 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 data from multiple data sources, means for formatting and cleaning the collected data, means for training an AI model using the formatted and cleaned data, means for customizing the AI ​​model based on user preferences and emotions, means for converting voice data into text and recognizing user emotions from the text data, and means for distributing and selling the customized AI model. This makes it possible to recommend products based on the user's individual preferences and emotional state, significantly improving the quality of the online shopping experience, increasing user satisfaction, and increasing sales.

[1054] A "data source" is a source of information for collecting information related to user behavior, preferences, emotions, etc.

[1055] "Formatting" is the process of converting collected data into a unified format and making it suitable for analysis.

[1056] "Cleaning" is a process for removing missing values ​​and outliers from collected data to maintain data quality.

[1057] An "artificial intelligence model" is an algorithm or mathematical model that performs a specific task by analyzing data and learning patterns.

[1058] "Training" is the process of using large amounts of data to train an artificial intelligence model and improve its accuracy.

[1059] "User preferences" refer to the tendency of a particular user to prefer certain products, services, or information.

[1060] "Customization" refers to adjusting a system or model to provide services and content optimized for individual users.

[1061] "Voice data" is a recording of the voice spoken by the user, and is data before it is converted into text information.

[1062] "Text data" refers to data in which voice data or other types of information are expressed as textual information.

[1063] "Emotion recognition" is a technology that analyzes data such as voice and text to determine a user's emotional state.

[1064] "Distribution" refers to providing a customized artificial intelligence model to a user or client.

[1065] "Sale" refers to the commercial trading of customized artificial intelligence models.

[1066] The present invention is a system that analyzes user behavior and emotions to provide personalized product recommendations through the steps of data collection, formatting and cleaning, AI model training, emotion recognition, customization, distribution, and sales.

[1067] Data collection

[1068] The server collects data from multiple sources, such as web servers, mobile apps, and social network platforms. This data includes users' browsing history, purchase history, voice data, and text data. For example, it can collect website visit history, mobile app usage history, and social networking platform posts.

[1069] Data Shaping and Cleaning

[1070] Since the collected data has different formats, the server converts them into a unified format. In addition, it detects and cleans missing and outlier values. Specifically, it complements missing data from the surrounding information and removes outliers.

[1071] Training an AI model

[1072] The server uses the cleaned and formatted data to train an AI model. It uses machine learning algorithms to learn user behavior patterns and build a model that predicts what users will want to buy next. For example, it uses a random forest classifier to predict future behavior based on users' browsing and purchase history.

[1073] emotion recognition

[1074] The server converts the voice data into text and recognizes the user's emotions from the text data. Using voice recognition technology, the server converts the user's speech into text and analyzes the emotions from that text. This makes it possible to determine the emotional state of the user. For example, the category of products the user is interested in can be determined from the text "My recent favorite is gadgets."

[1075] Customizing AI models

[1076] The trained AI model is customized based on the user's preferences and emotions, allowing it to provide optimal product recommendations for each user. For example, if a user is interested in gadgets, new gadget products will be recommended to that user first.

[1077] Distribution and Sales

[1078] Customized AI models are distributed and sold to markets or specific clients by the server. They are commercialized by being uploaded to the cloud or deployed directly to the client's server. For example, if a company provides a customized product recommendation system for a company, the AI ​​model is deployed on the company's server.

[1079] Specific prompt examples

[1080] "User text input: "My current favorite thing is gadgets.""

[1081] "Audio data: 'sample_audio.wav'"

[1082] As described above, the present invention provides a concrete means for implementing personalized product recommendations based on user behavior and emotions.

[1083] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1084] Step 1:

[1085] The server collects data from multiple data sources: user visit history from web servers, usage history from mobile apps, and user comments from social media platforms. These input data are sent to the server and stored in a centralized database. The output is an integrated raw data set.

[1086] Step 2:

[1087] The server formats the collected raw data and converts it into a unified format. In this process, data in different formats is unified. Next, a cleaning process is performed to detect missing values ​​and outliers and to impute or remove them. The input is the raw data collected in step 1, and the output is a clean dataset that has been formatted and cleaned.

[1088] Step 3:

[1089] The server uses the cleaned and sanitized data to train an AI model. This process uses a machine learning algorithm (e.g., a random forest classifier) ​​to learn user behavior patterns. The input is the clean dataset created in step 2, and the output is the trained AI model.

[1090] Step 4:

[1091] The server converts the voice data into text. Specifically, it uses voice recognition technology to analyze the user's speech and obtain it as text. The text data is then analyzed to recognize the user's emotions. The input is the voice data, and the output is the text data and its emotional evaluation.

[1092] Step 5:

[1093] The server customizes the trained AI model based on the user's preferences and emotions, thereby building a product recommendation model optimized for the specific user. The input is the AI ​​model from step 3 and the emotion evaluation from step 4, and the output is the customized AI model.

[1094] Step 6:

[1095] The server distributes and sells the customized AI model by uploading it to the cloud or by deploying it directly to the client's server. The input is the customized AI model created in step 5, and the output is an instance of the distributed AI model.

[1096] Specific operation example

[1097] "User text input: "My current favorite thing is gadgets.""

[1098] "Audio data: 'sample_audio.wav'"

[1099] In this specific example, if a user inputs the text "My recent favorite is gadgets" and provides audio data 'sample_audio.wav', the server can recommend personalized gadgets to this user through the processes of steps 1 to 6.

[1100] 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.

[1101] 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.

[1102] 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.

[1103] [Fourth embodiment]

[1104] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1105] 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.

[1106] 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).

[1107] 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.

[1108] 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.

[1109] 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).

[1110] 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. 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.

[1111] 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.

[1112] 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.

[1113] 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.

[1114] 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.

[1115] 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.

[1116] 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."

[1117] This invention is a system that collects data from multiple data sources, formats and cleans the data, trains an artificial intelligence (AI) model, and provides customized AI models to markets or specific clients. This system executes each step centrally on a server, achieving highly accurate personalization based on the individual preferences of users.

[1118] Feature Overview

[1119] The system includes the following main features:

[1120] 1. Data collection

[1121] 2. Data formatting and cleaning

[1122] 3. Training the AI ​​model

[1123] 4. Customization

[1124] 5. Distribution and Sales

[1125] Data collection

[1126] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this data collection process, the server provides authentication information to each data source and makes the necessary API calls. For example, the server collects user browsing history from web servers and usage history from mobile apps.

[1127] As a specific example, a server collects the user's smartphone app usage history as a data source, and periodically sends the data to the server, where it is stored in an integrated database.

[1128] Data Shaping and Cleaning

[1129] Since the collected data comes in various formats, the server converts it into a unified format. It also cleans missing or outlier values ​​if they are included. For example, if data associated with a specific user ID is missing, it performs the appropriate imputation process.

[1130] As a specific example, if there are missing values ​​in a user's website browsing history collected by the server, a cleaned dataset is created by filling in the missing values ​​with previous and subsequent data.

[1131] Training an AI model

[1132] Using the cleaned data, the server trains an AI model, which improves its ability to predict user behavior and provide optimal content to users, for example by using machine learning algorithms to find patterns in the training data.

[1133] As a specific example, a server trains a model based on a user's past browsing history to predict which website they are likely to visit next.

[1134] Customization

[1135] It is difficult to apply a trained AI model to all users as is, so the server customizes the AI ​​model based on the preferences of each individual user, providing a personalized service optimized for each user.

[1136] As a specific example, a server generates an AI model that preferentially recommends specific content to a particular user based on the type of content the user prefers.

[1137] Distribution and Sales

[1138] The final customized AI models are then distributed and sold to the market or to specific clients. The server commercializes these models by uploading them to the cloud or by deploying them directly to client servers.

[1139] As a specific example, a server provides services by introducing a content recommendation system customized for a company and deploying an AI model directly on the company's server.

[1140] summary

[1141] This invention is a system that integrates data collection, formatting, cleaning, training, customization, and sales. The server efficiently executes this series of processes, realizing highly personalized services based on user preferences. This system provides high-performance AI models based on information aggregated from multiple data sources, creating significant value for users.

[1142] The processing flow will be explained below.

[1143] Detailed description of the process

[1144] Data collection and formatting process

[1145] Step 1:

[1146] The server sets authentication information for multiple data sources such as web servers, mobile apps, and social media platforms, and prepares for data collection. Specifically, by setting API keys and secret keys, data can be securely collected from external data sources. For example, the web server calls an API to obtain user visit history.

[1147] Step 2:

[1148] The server sends data collection requests from each data source. Specifically, it collects user behavior data through APIs and stores it in an integrated database. During this process, it analyzes data in JSON and XML formats to extract the necessary information. For example, it collects data from mobile apps about user usage time and feature usage.

[1149] Step 3:

[1150] The server normalizes the collected data and formats it into a unified format. Specifically, it unifies different field names and data types and stores the data according to a common schema. For example, it can combine browsing history from a web server and usage history from a mobile app based on the same user ID.

[1151] Data Cleaning Process

[1152] Step 4:

[1153] The server cleans the formatted data. Specifically, it detects missing values ​​and outliers and performs appropriate imputation or deletion. For example, if the number of visits is abnormally high, that data is removed. If there are missing values, they are imputed with the mean or median.

[1154] AI model training process

[1155] Step 5:

[1156] The server splits the cleaned and shaped data into training data and test data by randomly shuffling the entire data and dividing it into 80% training data and 20% test data. This split allows for the evaluation of the model's performance.

[1157] Step 6:

[1158] The server uses the training data to train the AI ​​model. Specifically, it applies machine learning algorithms (e.g., random forests, neural networks) to learn features from the data. During the training process, it adjusts hyperparameters such as the learning rate and the number of epochs.

[1159] Step 7:

[1160] The server evaluates the trained model using test data. Specifically, it calculates metrics such as prediction accuracy, recall, and F-measure to verify the model's performance. Based on these results, the server fine-tunes the model to achieve optimal performance.

[1161] Customization and Distribution Process

[1162] Step 8:

[1163] The server collects user preference data and customizes the trained model by adjusting the model parameters based on the user's past behavioral data and explicit preferences (e.g., favorite genres, frequently used features).

[1164] Step 9:

[1165] The server deploys the customized AI model to a cloud server or a client server. Specifically, it exports the model and uploads it to the deployment environment using a secure communication protocol. After deployment, it checks and tests its operation in the production environment.

[1166] Step 10:

[1167] The server distributes and sells customized AI models to markets or specific clients, specifically by processing purchases of the models through marketplaces or sales platforms and providing buyers with license keys and installation instructions.

[1168] In this way, a system is realized that collects, cleans, shapes, trains, customizes, and distributes data from multiple sources, enabling highly accurate personalization based on user preferences.

[1169] Example 1

[1170] 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."

[1171] Existing AI systems require a lot of time and effort to effectively collect, shape, and clean data from multiple disparate sources. It is also difficult to create highly accurate models based on the cleaned data and customize them to suit individual user preferences. Furthermore, efficiently distributing customized models and making them commercially available has been a challenge.

[1172] 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.

[1173] In this invention, the server includes means for collecting information from multiple sources, means for unifying and purifying the collected information, means for training an AI model using the unified and purified information, means for periodically transmitting information from terminals and storing it in an integrated database, means for identifying and completing missing values ​​and outliers in the purification process, and means for predicting the site a user is likely to visit next based on the user's past behavior history. This enables efficient data collection from multiple data sources, shaping and cleaning, the creation and customization of highly accurate AI models based on user preferences, and the efficient distribution and commercialization of the models.

[1174] "Source" refers to the systems or platforms that provide data, such as web servers, mobile applications, and social networking services.

[1175] "Means of collection" refers to the technical means of providing API calls and credentials to retrieve data from multiple sources.

[1176] "Harmonization means" refers to procedures and methods for converting data of different formats into a unified format.

[1177] "Cleaning methods" refers to technical means for shaping and cleaning data and filling in missing or outlier values.

[1178] "Training methods" refers to machine learning algorithms and techniques used to create artificial intelligence models using collected and cleaned data.

[1179] "Means for periodically transmitting information" refers to the settings and technical means for periodically transmitting information from the terminal to the server and storing it in a database.

[1180] "Means for identifying and imputing missing and outliers" refers to technology for detecting missing and outliers in data and imputing them in an appropriate manner.

[1181] "Methods for predicting which site a user is likely to visit next based on their behavioral history" refers to AI models and algorithms that analyze a user's past behavioral history and predict which website they are likely to visit next.

[1182] "Artificial intelligence model" refers to a mathematical model that uses machine learning algorithms to analyze data and perform specific tasks.

[1183] The present invention is a system for collecting information from multiple sources, unifying and purifying the information, training an artificial intelligence (AI) model, and providing customized AI models to the market or specific clients. Detailed embodiments of the present invention are described below.

[1184] 1. Data collection

[1185] The server collects data from multiple sources, such as web servers, mobile applications, social networking services, etc. It retrieves data from each source by making an API call and providing the necessary authentication information.

[1186] Examples:

[1187] The server uses the Google Analytics API to obtain the website browsing history of a specific user.

[1188] The server uses the Firebase Analytics API to obtain mobile app usage data.

[1189] The server performs authentication using OAuth to obtain data from the SNS platform.

[1190] 2. Data formatting and cleaning

[1191] The collected data often comes in different formats, so the server unifies them and also detects missing and outlier values ​​and imputes them in an appropriate way.

[1192] Specific behavior:

[1193] The server uses the Python pandas library to unify data in different formats.

[1194] The server uses the mean or mode to impute missing values.

[1195] The server detects outliers and imputes them using a rational imputation method.

[1196] 3. Training the AI ​​model

[1197] Using the harmonized and cleansed data, the server trains AI models, improving their ability to predict user behavior and deliver optimal content.

[1198] Specific behavior:

[1199] The server builds and trains machine learning models using the Scikit-learn library.

[1200] The server generates a predictive model based on past behavioral data and evaluates its performance.

[1201] 4. Customization

[1202] The trained AI model is then customized based on each user's individual preferences, providing each user with the most personalized service possible.

[1203] Examples:

[1204] The server acquires the user's preference information (movies, music, sports, etc.) from a database.

[1205] The server fine-tunes the AI ​​model for each user segment to provide optimal recommendation results.

[1206] 5. Distribution and Sales

[1207] The final customized AI models are then distributed and sold to the market or to specific clients. The server uploads these models to the cloud or deploys them directly to the client's server.

[1208] Specific behavior:

[1209] The server uploads the completed AI model to an S3 bucket on Amazon Web Services (AWS).

[1210] The server connects to a specific client server via SSH and deploys the AI ​​model.

[1211] Prompt Sentence Examples

[1212] Below are some example prompts to input to a generative AI model:

[1213] "Based on my browsing history over the past month, what website would you recommend next?"

[1214] This system allows for efficient data collection from multiple sources, shaping and cleaning, creating highly accurate AI models, providing personalized services based on user preferences, and efficiently distributing the models to markets and clients.

[1215] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1216] Step 1:

[1217] Data collection

[1218] Input: Data from multiple sources (web servers, mobile applications, social networking services).

[1219] Processing: The server retrieves data by making API calls to each source and providing the necessary authentication information: Google Analytics API to retrieve website browsing history, Firebase Analytics API to collect mobile app usage data, and OAuth authentication to access social media platforms.

[1220] Output: Raw data stored on the server.

[1221] Specific behavior:

[1222] The server uses the Google Analytics API to obtain the user's website browsing history.

[1223] The server uses the Firebase Analytics API to obtain mobile app usage data.

[1224] The server performs OAuth authentication to obtain the social networking data.

[1225] Step 2:

[1226] Data Shaping and Cleaning

[1227] Input: Raw data stored on the server.

[1228] Processing: The server converts the collected data into a unified format, and also detects missing and outlier values ​​and imputes them in a rational way.

[1229] Output: Unified and cleaned data.

[1230] Specific behavior:

[1231] The server uses the Python pandas library to unify data in different formats.

[1232] The server uses the mean or mode to impute missing values.

[1233] The server detects outliers and interpolates them by inferring reasonable values ​​from the surrounding data.

[1234] Step 3:

[1235] Training artificial intelligence models

[1236] Input: Unified and cleaned data.

[1237] Processing: The server uses the formatted data to train an AI model, specifically by building a machine learning model using the Scikit-learn library and applying the training data to generate a predictive model.

[1238] Output: The trained AI model.

[1239] Specific behavior:

[1240] The server builds machine learning models using the Scikit-learn library.

[1241] The server applies past behavioral data as training data to generate a predictive model.

[1242] The server evaluates the performance of the generated model and adjusts the hyperparameters as needed.

[1243] Step 4:

[1244] Customization

[1245] Input: Trained AI model, user preference information.

[1246] Processing: The server customizes the AI ​​model based on the user's individual preferences, thereby providing personalized services optimized for each individual user.

[1247] Output: A customized AI model.

[1248] Specific behavior:

[1249] The server acquires the user's preference information (e.g., movies, music, sports) from a database.

[1250] The server customizes the AI ​​model and recommends the best content for a specific user.

[1251] Step 5:

[1252] Distribution and Sales

[1253] Input: A customized AI model.

[1254] Processing: The server distributes and sells the AI ​​model to a marketplace or specific clients. The model is uploaded to the cloud or deployed directly to the client's server.

[1255] Output: AI model deployed to market or client.

[1256] Specific behavior:

[1257] The server uploads the completed AI model to an AWS S3 bucket.

[1258] The server connects to a specific client server via SSH and deploys the AI ​​model.

[1259] (Application example 1)

[1260] 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."

[1261] Current content distribution services lack sophisticated personalization based on each user's preferences and viewing history, making it difficult to recommend content that matches a user's interests. Furthermore, in addition to primary data processing, they lack integrated training of the AI ​​models required for cleaning and formatting collected data, as well as customization. Therefore, there is a need to provide more appropriate content to users.

[1262] 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.

[1263] In this invention, the server includes means for collecting data from multiple data sources, means for formatting and cleaning the collected data, and means for training an artificial intelligence model using the formatted and cleaned data, thereby enabling the provision of personalized services based on user preferences and enabling smart device applications that individually recommend content based on viewing history.

[1264] "Multiple data sources" refers to different types of information sources, such as web servers, mobile apps, and social media platforms.

[1265] "Means of collecting data" refers to the methods and systems that obtain information from each data source using API calls and authentication information.

[1266] "Shaping and cleaning methods" refers to the processes and techniques used to fill in missing values ​​in collected data and convert it into a unified format.

[1267] "Artificial intelligence model" refers to a model that uses machine learning algorithms to learn patterns from large amounts of data and make predictions and recommendations.

[1268] "Training" refers to the process of applying machine learning algorithms to improve the performance of an artificial intelligence model using collected and cleaned data.

[1269] "Preference-based customization" refers to the process of adjusting and individually optimizing artificial intelligence models based on each user's past behavioral data and interests.

[1270] "Distribution and Sale Means" refers to methods and systems for providing trained and customized artificial intelligence models to markets and clients.

[1271] "Viewing history" refers to a record of content that a user has viewed or listened to in the past.

[1272] "Personalized content recommendation" refers to the process of individually recommending specific content to each user based on their preferences and viewing history.

[1273] "Smart device applications" refers to applications that run on smartphones, smart glasses, and other digital devices.

[1274] This invention is a system that collects data from multiple data sources, shapes and cleans it, and then trains a customized artificial intelligence (AI) model to provide personalized content recommendations for a specific user. This system is particularly intended for use in smart device applications.

[1275] Data collection

[1276] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this process, the server uses authentication information to make the necessary API calls. The collected data includes user viewing history, clicks, browsing history, etc. For example, smartphone app usage history and social media activity can be included as data sources.

[1277] Data Shaping and Cleaning

[1278] The collected data is formatted into a unified format by the server, and missing values ​​and outliers are cleaned. For example, if data on the viewing history linked to a specific user ID is missing, it is supplemented from previous and subsequent data. The data format is also unified, formatted into a data frame, and then converted into a format that can be applied to machine learning algorithms.

[1279] Training an AI model

[1280] The server uses the cleaned data to train an AI model. During this process, machine learning algorithms are applied to build a model based on the user's behavioral patterns and viewing history. For example, algorithms such as Random Forest Classifier are used to create a model that predicts the content that a user is likely to watch next based on their viewing history.

[1281] Customization

[1282] Since it is difficult to apply a trained AI model to all users in its initial state, it is customized based on the preference information of each individual user. The server generates recommendations optimized for each user based on the content type preferred by that particular user. For example, based on the "video1, video2" that a user has previously watched, the server will recommend content that the user is likely to watch next.

[1283] Distribution and Sales

[1284] The final customized AI models are distributed and sold to the market or specific clients. The server uploads these models to the cloud or deploys them directly to client servers for commercialization.

[1285] Examples of concrete examples and prompts

[1286] As a specific example, a history of the user's past viewing of "video1, video2" is collected, and based on that, it is determined which content to recommend next.

[1287] Example prompt sentence:

[1288] Based on the past viewing history "video1, video2", please recommend the next video this user might watch.

[1289] This system makes it possible to provide highly accurate personalized services based on users' viewing history and preferences. The server collects and cleans data, and trains and customizes AI models to provide individually optimized content recommendations to smart device applications.

[1290] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1291] Step 1:

[1292] The server collects data from web servers, mobile apps, and social media platforms. This involves making API calls and using authentication credentials to retrieve the required data (view history, clicks, etc.) from each platform. The collected data is temporarily stored in local storage or a cloud database. The input here is the authentication credentials and API endpoints, and the output is the collected raw data.

[1293] Step 2:

[1294] The server formats and cleans the collected raw data. This includes filling in missing values, correcting outliers, and standardizing data formats. Specifically, it converts the data into a data frame using libraries such as Pandas, and fills in missing values ​​from the surrounding data. The input here is the collected raw data, and the output is a cleaned dataset.

[1295] Step 3:

[1296] The server uses the cleaned data to train an AI model. This process involves applying a machine learning algorithm (e.g., RandomForestClassifier) ​​and optimizing the model's parameters. Specifically, the data is divided into training data and test data, and a model is built using the training data. The input here is the cleaned dataset, and the output is a trained AI model.

[1297] Step 4:

[1298] The server customizes the trained AI model based on the user's preferences. This involves adjusting the model's predictions based on the user's past viewing history and click count. The server then uses the embodied user data to generate personalized recommendations. The input here is the user's preference data, and the output is the customized model's recommendations.

[1299] Step 5:

[1300] The server provides the customized AI model and its recommendation results to the smart device application. This process involves uploading it to the cloud or deploying it directly to the client's server. The user also receives personalized content recommendations through their smartphone or smart glasses. The input here is the customized AI model and the recommendation results, and the output is the display of recommended content on the user's smart device.

[1301] In this way, by linking each step, a highly personalized content recommendation system is realized.

[1302] 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.

[1303] This invention is a system that collects data from multiple data sources, formats and cleans the data to train an artificial intelligence (AI) model, and then combines it with an emotion engine that recognizes user preferences and emotions to provide customized AI models to the market or specific clients. This system executes each step centrally on a server, achieving highly accurate personalization based on the individual preferences and emotions of users.

[1304] Feature Overview

[1305] The system includes the following main features:

[1306] 1. Data collection

[1307] 2. Data formatting and cleaning

[1308] 3. Training the AI ​​model

[1309] 4. Customization

[1310] 5. Emotion recognition

[1311] 6. Distribution and Sales

[1312] Data collection

[1313] The server collects data from multiple data sources, such as web servers, mobile apps, and social media platforms. During this data collection process, the server provides authentication information to each data source and makes the necessary API calls. For example, it collects user visit history from web servers and usage history from mobile apps.

[1314] As a specific example, a server collects the user's smartphone app usage history as a data source, and periodically sends the data to the server, where it is stored in an integrated database.

[1315] Data Shaping and Cleaning

[1316] Since the collected data comes in various formats, the server converts it into a unified format. It also cleans missing or outlier values ​​if they are included. For example, if data associated with a specific user ID is missing, it performs the appropriate imputation process.

[1317] As a specific example, if there are missing values ​​in a user's website browsing history collected by the server, a cleaned dataset is created by filling in the missing values ​​with previous and subsequent data.

[1318] Training an AI model

[1319] Using the cleaned data, the server trains an AI model, which improves its ability to predict user behavior and provide optimal content to users, for example by using machine learning algorithms to find patterns in the training data.

[1320] As a specific example, a server trains a model based on a user's past browsing history to predict which website they are likely to visit next.

[1321] emotion recognition

[1322] The server incorporates an emotion engine that recognizes emotions using the user's voice, facial expressions, and text data as additional data sources, enabling data collection based on the user's emotional state. For example, speech recognition technology can be used to analyze the user's speech to determine the user's emotional state.

[1323] As a specific example, a server analyzes voice data obtained from an app installed on a user's smartphone and determines whether the user is happy or sad.

[1324] Customizing AI models

[1325] It is difficult to apply a trained AI model to all users as is, so the server customizes the AI ​​model based on the user's preference data and recognized emotional data, thereby providing personalized services optimized for each user.

[1326] For example, a server could generate an AI model that prioritizes recommending certain content to a particular user based on the type of content the user prefers, or prioritizes providing different content when the user is in a particular emotional state.

[1327] Distribution and Sales

[1328] The final customized AI models are then distributed and sold to the market or to specific clients. The server commercializes these models by uploading them to the cloud or by deploying them directly to client servers.

[1329] As a specific example, a server provides services by introducing a content recommendation system customized for a company and deploying an AI model directly on the company's server.

[1330] summary

[1331] This invention is a system that integrates data collection, formatting, cleaning, AI model training, emotion recognition, customization, and distribution. The server efficiently executes this series of processes, enabling highly personalized services based on user preferences and emotions. This system provides high-performance AI models based on information aggregated from multiple data sources, creating significant value for users.

[1332] The processing flow will be explained below.

[1333] Detailed description of the process

[1334] The process of data collection, formatting, cleaning, training AI models, emotion recognition, customizing AI models, distribution and sales

[1335] Step 1:

[1336] The server sets authentication information for multiple data sources such as web servers, mobile apps, and social media platforms, and prepares for data collection. Specifically, it sets API keys and secret keys to enable secure data collection from external data sources. For example, it calls an API to obtain user visit history from a web server.

[1337] Step 2:

[1338] The server sends data collection requests from each data source. Specifically, it collects user behavior data through APIs and stores it in an integrated database. During this process, it analyzes data in JSON and XML formats to extract the necessary information. For example, it collects data from mobile apps about user usage time and feature usage.

[1339] Step 3:

[1340] The server normalizes the collected data and formats it into a unified format. Specifically, it unifies different field names and data types and stores the data according to a common schema. For example, it can combine browsing history from a web server and usage history from a mobile app based on the same user ID.

[1341] Step 4:

[1342] The server cleans the formatted data. Specifically, it detects missing values ​​and outliers and performs appropriate imputation or deletion. For example, if the number of visits is abnormally high, that data is removed. If there are missing values, they are imputed with the mean or median.

[1343] Step 5:

[1344] The server splits the cleaned and shaped data into training data and test data by randomly shuffling the entire data and dividing it into 80% training data and 20% test data. This split allows for the evaluation of the model's performance.

[1345] Step 6:

[1346] The server uses the training data to train the AI ​​model. Specifically, it applies machine learning algorithms (e.g., random forests, neural networks) to learn features from the data. During the training process, it adjusts hyperparameters such as the learning rate and the number of epochs.

[1347] Step 7:

[1348] The server evaluates the trained model using test data. Specifically, it calculates metrics such as prediction accuracy, recall, and F-measure to verify the model's performance. Based on these results, the server fine-tunes the model to achieve optimal performance.

[1349] Step 8:

[1350] The server runs an emotion engine that recognizes emotions based on the user's voice, facial expression, and text data, and collects additional data. Specifically, it uses voice recognition, facial recognition, and text analysis technologies to identify the user's emotions and stores that information in an integrated database. For example, it can determine the user's emotional state, such as joy, anger, or sadness, from voice data.

[1351] Step 9:

[1352] The server then incorporates the recognized user emotion data into the training and customization process. Specifically, the emotion data is input as features into the AI ​​model, and a model is generated to make predictions and recommendations that take the user's emotions into account.

[1353] Step 10:

[1354] The server uses the training and emotion data to generate a customized AI model. Specifically, it combines the user's preference data and emotion data to generate an AI model optimized for each user.

[1355] Step 11:

[1356] The server deploys the customized AI model to a cloud server or a client server. Specifically, it exports the model and uploads it to the deployment environment using a secure communication protocol. After deployment, it checks and tests its operation in the production environment.

[1357] Step 12:

[1358] The server distributes and sells customized AI models to markets or specific clients, specifically by processing purchases of the models through marketplaces or sales platforms and providing buyers with license keys and installation instructions.

[1359] This provides a system that collects data from multiple sources, cleans, shapes, trains, recognizes emotions, customizes, and distributes them, enabling highly accurate personalization based on user preferences and emotions.

[1360] Example 2

[1361] 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."

[1362] In modern society, user needs are becoming more diverse, and there is a demand for personalized services based on individual preferences and emotions. However, conventional systems have difficulty collecting and formatting data from a variety of data sources, quickly and accurately grasping user preferences and emotions, and customizing AI models. This has led to issues such as a decline in the quality of service provided to users and a lack of improvement in satisfaction.

[1363] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1364] In this invention, the server includes means for collecting data from multiple information sources, means for shaping and cleaning the collected data, means for training an AI model using the shaped and cleaned data, means for customizing the trained AI model based on a user's preferences and emotions, means for distributing and selling the customized AI model, and means for recognizing the user's emotional state using voice, facial expression, and text data, thereby enabling the provision of highly accurate personalized services that meet the diverse needs of users.

[1365] "Sources" are the locations or systems from which data is collected, specifically web servers, mobile device apps, and social networking services.

[1366] "Data transformation" is the process of converting collected data into a consistent format or structure.

[1367] "Data cleaning" is the process of detecting missing or outliers from collected data and correcting or supplementing them in an appropriate manner.

[1368] An "artificial intelligence model" is a model trained using machine learning algorithms and used to predict user behavior and provide optimal content.

[1369] "Training" is the process of using collected data to teach an artificial intelligence model patterns.

[1370] "User preferences" refer to the type of content or service that a particular user prefers.

[1371] "Emotion recognition" is the process of analyzing a user's voice, facial expressions, and text data to determine their emotional state.

[1372] "Customization" is the process of adjusting a trained artificial intelligence model based on a user's preferences and emotions, optimizing it to provide services tailored to individual needs.

[1373] "Distribution" is the act of offering a customized artificial intelligence model to the market or to specific clients.

[1374] "Sales" is the process of commercially trading and obtaining payment for customized artificial intelligence models.

[1375] This invention is a system that collects data from multiple sources, formats and cleans the data to train an artificial intelligence (AI) model, and then combines it with an emotion engine that recognizes the user's preferences and emotions to provide a customized AI model. This system executes each step centrally on a server, achieving highly accurate personalization based on the user's individual preferences and emotions.

[1376] Data collection

[1377] The server collects data from web servers, mobile communication device apps, and social networking services. During this data collection process, the server provides the necessary authentication information to each information source and makes API calls. For example, the server obtains a user's smartphone app usage history and sends that data to the server to store in an integrated database. This system targets data such as user behavior history, posted content, comments, and usage time.

[1378] Data Shaping and Cleaning

[1379] The server formats the collected data into a unified format and cleans up missing and outlier values ​​within the data. For example, it converts website browsing history collected in different formats into a consistent format. If there is missing data, it complements it by referencing the preceding and following data. This results in a formatted and cleaned dataset, which is then used to train the AI ​​model, as described below.

[1380] Training an AI model

[1381] The server uses the cleaned data to train an AI model. This process uses machine learning algorithms to detect patterns in the collected data and generate a model that can predict behavior and recommend optimal content. For example, the server trains a model that predicts which website a user is likely to visit next based on their past website browsing history. This training process is typically performed using machine learning libraries such as Python's Scikit-learn and TensorFlow.

[1382] emotion recognition

[1383] The server analyzes voice, facial expressions, and text data to recognize the user's emotional state. For example, it uses voice recognition technology to analyze the user's emotional state (happiness, sadness, anger, etc.) from their speech. Analysis tools such as OpenCV and Google Cloud Speech-to-Text are used for this. As a specific example, it can analyze voice data obtained from an app installed on the user's smartphone to determine whether the user is happy or angry.

[1384] Customizing AI models

[1385] The server customizes the trained AI model based on the user's preferences and emotional data. This process allows for services optimized for individual users. For example, the server generates a model that preferentially recommends specific content to a particular user based on the type of content the user prefers. It is also possible to provide different content depending on the user's emotional state.

[1386] Distribution and Sales

[1387] The server provides customized AI models to the market or specific clients. Distribution methods include uploading the model to the cloud or deploying it directly to the client's server. For example, a content recommendation system customized for a company can be deployed to the client's server and provided as a service.

[1388] Example prompt statement

[1389] Examples of prompts to input to a generative AI model include:

[1390] "Predict what site a user might visit next based on their past website browsing data."

[1391] This invention is a system that efficiently collects, formats, cleans, trains AI models, recognizes emotions, customizes models, and distributes data, and provides highly personalized services based on user preferences and emotions, thereby creating high value for users.

[1392] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1393] Step 1:

[1394] Data collection

[1395] The server collects data from web servers, mobile communication device apps, and social networking services. In this process, the server provides authentication information to each information source and calls APIs to retrieve data. For example, the server obtains a user's smartphone app usage history and periodically sends that data to the server. The input data is the user's behavioral history data, and the output is the history data stored in an integrated database. Specifically, the server periodically schedules and executes data collection jobs.

[1396] Step 2:

[1397] Data Shaping and Cleaning

[1398] The server converts the collected data into a consistent format and cleans missing and outlier values. For example, it converts website browsing history in different formats into a common format. The input data is the original collected data, and the output is the converted and cleaned data. Specifically, it performs data format conversion processing and applies a missing value imputation algorithm to impute data.

[1399] Step 3:

[1400] Training an AI model

[1401] The server trains an AI model based on the cleaned data. It uses machine learning algorithms to detect patterns and generate a predictive model of user behavior. The input data is the cleaned and formatted dataset, and the output is the trained AI model. Specifically, it uses Python's Scikit-learn and TensorFlow to execute the model training process.

[1402] Step 4:

[1403] emotion recognition

[1404] The server analyzes voice, facial expressions, and text data to recognize the user's emotional state. For example, it uses voice recognition technology to analyze the user's speech and determine their emotional state. The input data is voice, facial expressions, and text data, and the output is the recognized emotional state. Specifically, it uses tools such as OpenCV and Google Cloud Speech-to-Text to analyze the voice and classify emotions.

[1405] Step 5:

[1406] Customizing AI models

[1407] The server customizes the trained AI model based on the user's preference and emotional data, thereby providing services optimized for each individual user. The input data is the user's preference and emotional data, and the output is a customized AI model. Specifically, the server fine-tunes the model's weights and parameters to generate an optimal model for each user.

[1408] Step 6:

[1409] Distribution and Sales

[1410] The server distributes and sells the customized AI model to the market or to specific clients. Distribution methods include uploading the model to the cloud or deploying it to the client's server. The input data is the customized AI model, and the output is the distributed model and a commercial contract. Specifically, the model is uploaded using a cloud storage service and provided to clients via an API.

[1411] (Application example 2)

[1412] 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."

[1413] In today's online shopping environment, users face difficulties in finding the right products for them from the vast amount of information available. Furthermore, online shopping sites lack the ability to appropriately recommend products based on users' individual preferences and emotional state. As a result, users' purchasing motivation declines and site sales do not increase. It is necessary to solve these problems and provide users with a highly personalized experience.

[1414] The identification processing 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 data from multiple data sources, means for formatting and cleaning the collected data, means for training an AI model using the formatted and cleaned data, means for customizing the AI ​​model based on user preferences and emotions, means for converting voice data into text and recognizing user emotions from the text data, and means for distributing and selling the customized AI model. This makes it possible to recommend products based on the user's individual preferences and emotional state, significantly improving the quality of the online shopping experience, increasing user satisfaction, and increasing sales.

[1415] A "data source" is a source of information for collecting information related to user behavior, preferences, emotions, etc.

[1416] "Formatting" is the process of converting collected data into a unified format and making it suitable for analysis.

[1417] "Cleaning" is a process for removing missing values ​​and outliers from collected data to maintain data quality.

[1418] An "artificial intelligence model" is an algorithm or mathematical model that performs a specific task by analyzing data and learning patterns.

[1419] "Training" is the process of using large amounts of data to train an artificial intelligence model and improve its accuracy.

[1420] "User preferences" refer to the tendency of a particular user to prefer certain products, services, or information.

[1421] "Customization" refers to adjusting a system or model to provide services and content optimized for individual users.

[1422] "Voice data" is a recording of the voice spoken by the user, and is data before it is converted into text information.

[1423] "Text data" refers to data in which voice data or other types of information are expressed as textual information.

[1424] "Emotion recognition" is a technology that analyzes data such as voice and text to determine a user's emotional state.

[1425] "Distribution" refers to providing a customized artificial intelligence model to a user or client.

[1426] "Sale" refers to the commercial trading of customized artificial intelligence models.

[1427] The present invention is a system that analyzes user behavior and emotions to provide personalized product recommendations through the steps of data collection, formatting and cleaning, AI model training, emotion recognition, customization, distribution, and sales.

[1428] Data collection

[1429] The server collects data from multiple sources, such as web servers, mobile apps, and social network platforms. This data includes users' browsing history, purchase history, voice data, and text data. For example, it can collect website visit history, mobile app usage history, and social networking platform posts.

[1430] Data Shaping and Cleaning

[1431] Since the collected data has different formats, the server converts them into a unified format. In addition, it detects and cleans missing and outlier values. Specifically, it complements missing data from the surrounding information and removes outliers.

[1432] Training an AI model

[1433] The server uses the cleaned and formatted data to train an AI model. It uses machine learning algorithms to learn user behavior patterns and build a model that predicts what users will want to buy next. For example, it uses a random forest classifier to predict future behavior based on users' browsing and purchase history.

[1434] emotion recognition

[1435] The server converts the voice data into text and recognizes the user's emotions from the text data. Using voice recognition technology, the server converts the user's speech into text and analyzes the emotions from that text. This makes it possible to determine the emotional state of the user. For example, the category of products the user is interested in can be determined from the text "My recent favorite is gadgets."

[1436] Customizing AI models

[1437] The trained AI model is customized based on the user's preferences and emotions, allowing it to provide optimal product recommendations for each user. For example, if a user is interested in gadgets, new gadget products will be recommended to that user first.

[1438] Distribution and Sales

[1439] Customized AI models are distributed and sold to markets or specific clients by the server. They are commercialized by being uploaded to the cloud or deployed directly to the client's server. For example, if a company provides a customized product recommendation system for a company, the AI ​​model is deployed on the company's server.

[1440] Specific prompt examples

[1441] "User text input: "My current favorite thing is gadgets.""

[1442] "Audio data: 'sample_audio.wav'"

[1443] As described above, the present invention provides a concrete means for implementing personalized product recommendations based on user behavior and emotions.

[1444] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1445] Step 1:

[1446] The server collects data from multiple data sources: user visit history from web servers, usage history from mobile apps, and user comments from social media platforms. These input data are sent to the server and stored in a centralized database. The output is an integrated raw data set.

[1447] Step 2:

[1448] The server formats the collected raw data and converts it into a unified format. In this process, data in different formats is unified. Next, a cleaning process is performed to detect missing values ​​and outliers and to impute or remove them. The input is the raw data collected in step 1, and the output is a clean dataset that has been formatted and cleaned.

[1449] Step 3:

[1450] The server uses the cleaned and sanitized data to train an AI model. This process uses a machine learning algorithm (e.g., a random forest classifier) ​​to learn user behavior patterns. The input is the clean dataset created in step 2, and the output is the trained AI model.

[1451] Step 4:

[1452] The server converts the voice data into text. Specifically, it uses voice recognition technology to analyze the user's speech and obtain it as text. The text data is then analyzed to recognize the user's emotions. The input is the voice data, and the output is the text data and its emotional evaluation.

[1453] Step 5:

[1454] The server customizes the trained AI model based on the user's preferences and emotions, thereby building a product recommendation model optimized for the specific user. The input is the AI ​​model from step 3 and the emotion evaluation from step 4, and the output is the customized AI model.

[1455] Step 6:

[1456] The server distributes and sells the customized AI model by uploading it to the cloud or by deploying it directly to the client's server. The input is the customized AI model created in step 5, and the output is an instance of the distributed AI model.

[1457] Specific operation example

[1458] "User text input: "My current favorite thing is gadgets.""

[1459] "Audio data: 'sample_audio.wav'"

[1460] In this specific example, if a user inputs the text "My recent favorite is gadgets" and provides audio data 'sample_audio.wav', the server can recommend personalized gadgets to this user through the processes of steps 1 to 6.

[1461] 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.

[1462] 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.

[1463] 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.

[1464] 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.

[1465] FIG. 9 is a diagram illustrating 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 actions 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.

[1466] 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.

[1467] 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).

[1468] 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.

[1469] 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."

[1470] 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.

[1471] 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).

[1472] 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.

[1473] 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.

[1474] 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.

[1475] 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.

[1476] The hardware resource for executing a specific process can be any of the following processors: An example of a processor 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 example of a 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.

[1477] 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.

[1478] 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.

[1479] 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.

[1480] 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.

[1481] 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.

[1482] The following is further disclosed regarding the above embodiment.

[1483] (Claim 1)

[1484] a means for collecting data from multiple data sources;

[1485] a means of formatting and cleaning the collected data;

[1486] means for training an artificial intelligence model using the reformulated and cleaned data;

[1487] means for customizing the artificial intelligence model based on user preferences;

[1488] A means to distribute and sell customized artificial intelligence models; and

[1489] A system including:

[1490] (Claim 2)

[1491] The system of claim 1 , wherein the data sources include a web server, a mobile app, and a social networking platform.

[1492] (Claim 3)

[1493] 10. The system of claim 1, wherein the artificial intelligence model is trained using a machine learning algorithm.

[1494] "Example 1"

[1495] (Claim 1)

[1496] a means of collecting information from multiple sources;

[1497] a means of unifying and purifying the collected information;

[1498] means for training an artificial intelligence model using the unified and purified information;

[1499] means for adjusting the artificial intelligence model based on user preferences;

[1500] A means for distributing and selling the adjusted artificial intelligence model; and

[1501] means for periodically transmitting information from the terminal and storing it in an integrated database;

[1502] A means of identifying and imputing missing and outlier values ​​during the cleansing process;

[1503] A method for predicting the next site a user is likely to visit based on their past behavioral history,

[1504] A system including:

[1505] (Claim 2)

[1506] 10. The system of claim 1, wherein the information sources include web servers, mobile applications, and social networking services.

[1507] (Claim 3)

[1508] 10. The system of claim 1, wherein the artificial intelligence model is trained using a machine learning algorithm.

[1509] "Application Example 1"

[1510] (Claim 1)

[1511] a means for collecting data from multiple data sources;

[1512] a means of formatting and cleaning the collected data;

[1513] means for training an artificial intelligence model using the reformulated and cleaned data;

[1514] means for customizing the artificial intelligence model based on user preferences;

[1515] A means to distribute and sell customized artificial intelligence models; and

[1516] A means for providing an application for smart devices that recommends content individually based on a user's viewing history;

[1517] A system including:

[1518] (Claim 2)

[1519] The system of claim 1 , wherein the data sources include a web server, a mobile app, and a social networking platform.

[1520] (Claim 3)

[1521] 10. The system of claim 1, wherein the artificial intelligence model is trained using a machine learning algorithm to provide personalized recommendations based on a user's past behavioral data.

[1522] "Example 2: Combining Emotion Engines"

[1523] (Claim 1)

[1524] a means of collecting data from multiple sources;

[1525] a means of formatting and cleaning the collected data;

[1526] means for training an artificial intelligence model using the reformulated and cleaned data;

[1527] means for customizing the trained artificial intelligence model based on user preferences and emotions;

[1528] A means to distribute and sell customized artificial intelligence models; and

[1529] means for recognizing an emotional state of a user using voice, facial expression, and text data;

[1530] A system including:

[1531] (Claim 2)

[1532] 10. The system of claim 1, wherein the information sources include a web server, a mobile communication device app, and a social networking service.

[1533] (Claim 3)

[1534] 10. The system of claim 1, wherein the artificial intelligence model is trained using a machine learning algorithm.

[1535] "Application example 2 when combining emotion engines"

[1536] (Claim 1)

[1537] a means for collecting data from multiple data sources;

[1538] a means of formatting and cleaning the collected data;

[1539] means for training an artificial intelligence model using the reformulated and cleaned data;

[1540] means for customizing the artificial intelligence model based on user preferences and emotions;

[1541] means for converting voice data into text and recognizing user emotions from the text data;

[1542] A means to distribute and sell customized artificial intelligence models; and

[1543] A system including:

[1544] (Claim 2)

[1545] 10. The system of claim 1, wherein the data sources include a web server, a mobile app, and a social networking platform.

[1546] (Claim 3)

[1547] 10. The system of claim 1, wherein the artificial intelligence model is trained using a machine learning algorithm. [Explanation of symbols]

[1548] 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 data from multiple data sources; a means of formatting and cleaning the collected data; means for training an artificial intelligence model using the reformulated and cleaned data; means for customizing the artificial intelligence model based on user preferences; A means to distribute and sell customized artificial intelligence models; and A system including:

2. The system of claim 1 , wherein the data sources include a web server, a mobile app, and a social networking platform.

3. The system of claim 1 , wherein the artificial intelligence model is trained using a machine learning algorithm.

Images