system

Abstract

The system according to the present embodiment includes a learning unit, a generating unit, a testing unit, and an analyzing unit. The learning unit learns data related to a market. The generating unit generates a pseudo personality (pseudo user profile) based on features learned by the learning unit. The testing unit performs a reaction test using the pseudo personality generated by the generating unit. The analyzing unit analyzes a test result obtained by the testing unit.

Description

Technical Field

[0001] The technology disclosed herein relates to a system. Background Technology

[0002] Patent Document 1 discloses a personalized chatbot control method executed by at least one processor, the method comprising: receiving user speech; adding the user speech to a prompt containing instructions related to a chatbot role; encoding the prompt; and inputting the encoded prompt into a language model to generate chatbot speech in response to the user speech.

[0003] Patent document 1: Japanese Patent Application Publication No. 2022-180282.

[0004] The problem with existing technologies is that it is difficult to accurately predict the market's reaction to a company's products when launching new products or entering new markets. Summary of the Invention

[0005] The system described in this embodiment includes a learning unit, a generation unit, a testing unit, and an analysis unit. The learning unit learns market-related data. The generation unit generates pseudo-personalities (pseudo-user profiles) based on the features learned by the learning unit. The testing unit uses the pseudo-personalities generated by the generation unit to conduct reaction tests. The analysis unit analyzes the test results obtained by the testing unit. Attached Figure Description

[0006] Figure 1 This is a conceptual diagram illustrating an example of the configuration of a data processing system according to the first embodiment.

[0007] Figure 2 This is a conceptual diagram illustrating an example of the main functions of the data processing apparatus and smart device according to the first embodiment.

[0008] Figure 3 This is a conceptual diagram illustrating an example of the data processing system configuration in the second embodiment.

[0009] Figure 4 This is a conceptual diagram illustrating an example of the main functions of the data processing device and smart glasses according to the second embodiment.

[0010] Figure 5 This is a conceptual diagram illustrating an example of the data processing system configuration in the third embodiment.

[0011] Figure 6 This is a conceptual diagram illustrating an example of the main functions of the data processing device and head-mounted terminal according to the third embodiment.

[0012] Figure 7 This is a conceptual diagram illustrating an example of the data processing system configuration in the fourth embodiment.

[0013] Figure 8 This is a conceptual diagram illustrating an example of the functions of the main parts of the data processing device and robot according to the fourth embodiment.

[0014] Figure 9 It represents an emotion graph that maps multiple emotions.

[0015] Figure 10 It represents an emotion graph that maps multiple emotions.

[0016] Explanation of reference numerals in the attached figures Data processing systems 10, 210, 310, and 410 12 Data processing device 14 Smart devices 214 Smart Glasses 314 Head-mounted terminal 414 Robot. Detailed Implementation

[0017] Hereinafter, an example of an implementation of the system involved in this disclosure will be described with reference to the accompanying drawings.

[0018] First, let's explain the terms used in the following description.

[0019] In the following embodiments, the processor (hereinafter referred to as "processor") can be a single computing device or a combination of multiple computing devices. Furthermore, a processor can be a single computing device or a combination of multiple computing devices. Examples of computing devices include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), APU (Accelerated Processing Unit), or TPU (Tensor Processing Unit), etc.

[0020] In the following implementation, the labeled RAM (Random Access Memory) is a memory that temporarily stores information and is used by the processor as working memory.

[0021] In the following embodiments, the labeled memory is one or more non-volatile storage devices used to store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), disk (e.g., hard disk) or magnetic tape, etc.

[0022] In the following implementation, the labeled Communication I / F (Interface) is an interface that includes a communication processor and an antenna, etc. The Communication I / F is responsible for communication between multiple computers. Examples of communication standards applicable to the Communication I / F include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0023] In the following implementation, "A and / or B" is synonymous with "at least one of A and B". That is, "A and / or B" means that it can be only A, only B, or a combination of A and B. Furthermore, in this specification, when "and / or" connects more than three items, the same approach as "A and / or B" applies.

[0024] [First Implementation] Figure 1 An example of the configuration of the data processing system 10 according to the first embodiment is shown.

[0025] like Figure 1 As shown, the data processing system 10 includes a data processing device 12 and an intelligent 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 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, 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. An example of the network 54 includes a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0027] The smart device 14 includes a computer 36, a receiver 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. In addition, the receiver 38, output device 40, and camera 42 are also connected to the bus 52.

[0028] The receiving device 38 includes a touchscreen 38A and a microphone 38B, etc., for receiving user input. The touchscreen 38A receives user input generated by contact with an indicator (e.g., a pen or finger). The microphone 38B receives user input generated by sound by detecting the user's voice. The control unit 46A sends data representing user input received via the touchscreen 38A and microphone 38B to the data processing unit 12. In the data processing unit 12, a specific processing unit 290 (see...) Figure 2 Get the data that represents the user input.

[0029] The output device 40 includes a display 40A and a speaker 40B, etc., and presents data to the user by outputting data in a user-perceptible form (e.g., sound and / or text). The display 40A displays visual information such as text and images according to instructions from the processor 46. The speaker 40B outputs sound according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0030] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for sending and receiving various information between processor 46 and processor 28 via network 54.

[0031] Figure 2 An example of the main functions of the data processing device 12 and the smart device 14 is shown.

[0032] like Figure 2 As shown, in the data processing apparatus 12, specific processing is performed by the processor 28. A specific processing program 56 is stored in the memory 32. The specific processing program 56 is an example of a "program" as understood in this disclosure. The processor 28 reads the specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. Specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0033] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0034] In the smart device 14, specific processing is performed by the processor 46. A specific processing program 60 is stored in the memory 50. The specific processing program 60 is used in conjunction with the data processing system 10. The processor 46 reads the specific processing program 60 from the memory 50 and executes the read specific processing program 60 on the RAM 48. Specific processing is implemented by the processor 46 operating as a control unit 46A based on the specific processing program 60 executed on the RAM 48. Furthermore, the smart device 14 may also have the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and use these models to perform the same processing as the specific processing unit 290.

[0035] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device (e.g., a generation server) may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.). Next, an example of the processing performed by the data processing system 10 of the first embodiment will be described.

[0036] (Example) The market reaction prediction system described in this invention is used to understand the market reaction of a product when it enters a new market or new market. This system utilizes generative AI to generate pseudo-personalities (pseudo-user profiles) that closely resemble actual people in the market, and then uses these pseudo-personalities to conduct reaction tests. First, market-related data is input into the generative AI, which learns the characteristics of people existing in the market based on this data. Next, a pseudo-personality is generated based on the characteristics learned by the generative AI. This pseudo-personality possesses characteristics very similar to those of people in the actual market, and by conducting reaction tests, the market reaction to the product can be predicted. For example, introducing a new product to the pseudo-personality and observing its reaction can predict the actual market reaction. This service can reduce the risk of entering a new product or new market and enable the development of more effective marketing strategies. For example, a learning unit is set up to input market-related data into the generative AI, allowing the AI ​​to learn the characteristics of people existing in the market; a generation unit is set up to generate pseudo-personalities based on the characteristics learned by the generative AI; a testing unit is set up to conduct reaction tests using the pseudo-personalities; and an analysis unit is set up to analyze the test results and develop marketing strategies. Therefore, the market reaction prediction system can reduce the risks of entering new products or new markets and formulate effective marketing strategies. Specifically, unlike traditional human resource market surveys or questionnaire statistics, this market reaction prediction system utilizes computer-based high-dimensional data analysis and generative modeling to vectorize and tensorize large amounts of market data (such as millions of purchase records, thousands of competitor information, and time-series economic indicators) as input data. The system's learning unit can employ a multilayer perceptron or Transformer-type neural network, accepting inputs such as consumer attribute vectors (e.g., age, gender, purchase records, region, preference scores, etc., 100-dimensional vectors), competitor trend tensors (e.g., time-series sales changes, price fluctuations, advertising volume, etc., 50×12 tensors), and economic indicator arrays (e.g., GDP growth rate, unemployment rate, consumer confidence index, etc., 20-dimensional arrays). A specific example is an input data vector of "30 years old, female, purchased cosmetics 5 times in the past year, Kanto region, preference score 0.8," and a time-series tensor of "competitor A company's sales changes from January to December 2023." The learning department normalizes and encodes this data, and extracts latent features at the feature extraction layer. Based on these extracted features, the generative AI uses generative models such as VAEs (Variational Autoencoders) or diffusion models to output pseudo-personality attribute vectors that approximate real consumers (e.g., age 35, male, purchasing tendency: new product-oriented, price sensitivity: high, SNS usage frequency: 5 times per week, etc.). An example output might be: "Pseudo-personality A: 40-year-old male, purchased home appliances 10 times in the past year, price sensitivity 0.7, high SNS posting frequency."The testing department inputs new product information (such as product description text, images, and pricing information) into the generated pseudo-personality, and the generative AI outputs personality reactions (such as a purchase intention score of 0.85, positive reviews, and negative feedback). Outputs might include, for example, "Purchase intention score 0.92" or "Review: Good design but high price." The analysis department aggregates this reaction data, performs statistical analysis (such as cluster analysis, principal component analysis, and regression analysis) and data mining (such as association analysis and outlier detection), and outputs decision support information for developing marketing strategies (such as target group reaction distribution, price optimization suggestions, and advertising channel recommendations). This series of processes is executed at high speed on a GPU-based parallel computing cluster, significantly improving processing speed and accuracy compared to traditional manual surveys, statistics, and analysis. Furthermore, the application of cutting-edge technologies such as weight optimization, loss function minimization, data augmentation, and transfer learning in each stage of AI model learning, generation, testing, and analysis enhances the model's generalization performance and adaptability. The technical effect is that this system can accurately and quickly predict the reactions of diverse consumer profiles before a product is launched on the market, thereby optimizing product development and marketing strategy decisions based on scientific evidence. It is applicable to a wide range of industries and business models, including new product development for consumer goods manufacturers, new market entry for service industries, event planning for advertising agencies, and risk assessment when financial institutions introduce new services.

[0037] The market reaction prediction system described in this embodiment includes a learning unit, a generation unit, a testing unit, and an analysis unit. The learning unit learns market-related data. Market-related data includes, but is not limited to, consumer purchase records, competitor activities, and economic indicators. The learning unit can use algorithms such as machine learning and deep learning to learn market data. Generative AI learns the characteristics of individuals existing in the market based on a large amount of market data. The generation unit generates pseudo-personalities based on the characteristics learned by the generative AI. Pseudo-personalities can be generated, for example, based on target customer profiles or behavioral patterns, but are not limited to these. The generation unit uses generative AI to generate pseudo-personalities with characteristics highly similar to those of individuals in the actual market. The testing unit uses the pseudo-personalities to conduct reaction tests. Reaction tests include, for example, questionnaires and A / B tests, but are not limited to these. The testing unit uses generative AI to observe the reactions of the pseudo-personalities and collect data. The analysis unit analyzes the test results obtained by the testing unit. Analysis includes, for example, statistical analysis and data mining, but is not limited to these. The analysis unit uses generative AI to provide information for developing marketing strategies based on the test results. Therefore, the market reaction prediction system of this embodiment can reduce the risk of entering new products or new markets and formulate effective marketing strategies. Specifically, the learning unit of this market reaction prediction system can accept diverse market data such as consumer attribute vectors (e.g., 100-dimensional vectors such as age, gender, purchase history, region, preference score, etc.), competitor trend tensors (e.g., 50×12 tensors such as time-series sales changes, price fluctuations, advertising volume, etc.), and economic indicator arrays (e.g., 20-dimensional arrays such as GDP growth rate, unemployment rate, consumer confidence index, etc.). The learning unit normalizes and encodes this data and uses a multilayer perceptron or Transformer-type neural network for feature extraction. Inputs include, for example, a vector of "30 years old, female, purchased cosmetics 5 times in the past year, Kanto region, preference score 0.8" and a time-series tensor of "sales changes of competitor company A from January to December 2023". Generative AI, based on extracted latent features, utilizes generative models such as VAEs (Variational Autoencoders) or diffusion models to learn pseudo-personality attribute vectors that approximate real consumers (e.g., age 35, male, purchasing tendency: new product-oriented, price sensitivity: high, SNS usage frequency: 5 times per week, etc.). Output, for example, "Pseudo-personality A: 40-year-old male, purchased home appliances 10 times in the past year, price sensitivity 0.7, high SNS posting frequency." The testing department inputs new product information (e.g., product description text, images, price information, etc.) into the generated pseudo-personality, and the generative AI outputs personality reactions (e.g., purchase intention score 0.85, generating positive reviews, generating negative feedback, etc.). Output, for example, "Purchase intention score 0.92," "Review: Good design but high price." The analysis department summarizes these reaction data, performs statistical analyses such as cluster analysis, principal component analysis, and regression analysis, as well as data mining such as association analysis and outlier detection, and outputs decision support information such as target group reaction distribution, price optimization suggestions, and advertising channel recommendations.This series of processes executes at high speed on a GPU-based parallel computing cluster, significantly improving processing speed and accuracy compared to traditional manual surveys, statistics, and analysis. Furthermore, the application of cutting-edge technologies such as weight optimization, loss function minimization, data augmentation, and transfer learning in each stage of AI model learning, generation, testing, and analysis enhances the model's generalization performance and adaptability. The technical effect is that this system can accurately and quickly predict the reactions of diverse consumer profiles before a product is launched to the market, thereby optimizing product development and marketing strategy decisions based on scientific evidence. Applicable fields include new product development for consumer goods manufacturers, new market entry for service industries, event planning for advertising agencies, and risk assessment for the introduction of new services by financial institutions, applicable to a wide range of industries and business models.

[0038] The learning unit can input market-related data and enable generative AI to learn the characteristics of market participants based on that data. For example, the learning unit can input market data such as consumer purchase records, competitor activities, and economic indicators, and the generative AI can learn the characteristics of market participants based on this data. The generative AI can use algorithms such as machine learning and deep learning to learn market data. The generative AI extracts and learns the characteristics of market participants from a large amount of market data. Thus, the learning unit can learn the characteristics of market participants based on market-related data. Some or all of the above processing in the learning unit can be implemented using generative AI, or it can be done without it. For example, the learning unit can input market data into the generative AI, which will then learn the characteristics of market participants. Specifically, this learning unit can accept diverse market data as input, such as consumer attribute vectors (e.g., 100-dimensional vectors such as age, gender, purchase records, region, preference scores, etc.), competitor activity tensors (e.g., 50×12 tensors such as time-series sales changes, price fluctuations, advertising volume, etc.), and economic indicator arrays (e.g., 20-dimensional arrays such as GDP growth rate, unemployment rate, consumer confidence index, etc.). The input consists of a vector such as "30 years old, female, purchased cosmetics 5 times in the past year, Kanto region, preference score 0.8" and a time series tensor of "sales changes of competitor A company from January to December 2023". The learning department normalizes and encodes this data and extracts features using a multilayer perceptron or Transformer-type neural network. Generative AI, based on the extracted latent features, uses generative models such as VAEs (Variational Autoencoders) or diffusion models to learn feature vectors that approximate real consumers. The learning department uses cross-entropy or mean squared error as the loss function and updates the weights using algorithms such as gradient descent or Adam optimization. Thus, the learning department can automate feature extraction and pattern recognition of high-dimensional, multivariate data, significantly improving the model's generalization performance and adaptability—something difficult to achieve with traditional simple statistical or manual analysis. The technical benefits include strong adaptability to the diversity and variability of market data, enabling rapid response to real-time market changes, thereby significantly improving the accuracy and speed of product development and marketing strategies. Applicable fields include market analysis and demand forecasting in various industries such as consumer goods, services, advertising, and finance.

[0039] The generation department can generate pseudo-personalities based on features learned by generative AI. For example, the generation department generates pseudo-personalities based on features learned by generative AI. Pseudo-personalities can be generated based on, for example, the profiles or behavioral patterns of target customers, but are not limited to these. The generation department uses generative AI to generate pseudo-personalities with characteristics highly similar to people in the actual market. For example, generative AI learns features such as the target customer's age, gender, and purchase history, and generates pseudo-personalities based on these. Thus, the generation department can generate pseudo-personalities based on features learned by generative AI. Some or all of the above processing in the generation department can be implemented using generative AI, or it can be done without generative AI. For example, the generation department generates pseudo-personalities based on features learned by generative AI. Specifically, this generation department can accept latent feature vectors extracted by the learning department (such as age 35, gender male, purchase tendency: new product orientation, price sensitivity: high, SNS usage frequency: 5 times per week, etc.) as input. Input example: "Age 40, male, purchased home appliances 10 times in the past year, price sensitivity 0.7, high SNS posting frequency." The generation department utilizes generative models such as VAEs (Variational Autoencoders) or diffusion models to generate pseudo-personality attribute vectors that approximate real consumers from these feature quantities. Outputs might include, for example, "Pseudo-personality A: 40-year-old male, high frequency of appliance purchases, price sensitivity 0.7, high frequency of SNS posting." During the generation process, the department can apply latent space sampling or conditional generation to generate diverse personalities tailored to specific target groups or behavioral patterns. Furthermore, the generation department can leverage generative AI's weight optimization and data augmentation techniques to improve the model's generalization performance and diversity. The technical effect is that the generation department can automatically generate high-dimensional, multivariate consumer profiles, which is difficult to achieve with traditional simple attribute combinations or manual design, thus enabling more realistic and diverse market response simulations. Applicable fields include new product development, advertising targeting, service design, and financial product design.

[0040] The testing department can utilize pseudo-personalities for reaction testing. For example, the testing department uses pseudo-personalities for reaction testing. Reaction testing includes, but is not limited to, questionnaires, A / B testing, etc. The testing department uses generative AI to observe the pseudo-personalities' reactions and collect data. For example, the generative AI introduces a new product to the pseudo-personalities and observes their reactions. The generative AI predicts the actual market reaction based on the pseudo-personalities' reactions. Thus, by using pseudo-personalities for reaction testing, the testing department can predict the market's reaction to its own products. Some or all of the above processing in the testing department can be implemented using generative AI, or it can be done without it. For example, the testing department inputs the pseudo-personalities' reactions into the generative AI, which then performs the reaction test. Specifically, this testing department can accept pseudo-personality attribute vectors generated by the generative department (such as age 40, male, high frequency of home appliance purchases, price sensitivity 0.7, high frequency of SNS posting) and new product information (such as product description text, images, price information, etc.) as input. For example, the input can be a combination of "description text + images + price information of new product A" and "attribute vector of pseudo-personality A". The testing department utilizes generative AI (such as large-scale language models or multimodal generative models) to output personality responses (e.g., a purchase intention score of 0.85, positive reviews, negative feedback, etc.). Outputs might include phrases like "Purchase intention score 0.92" or "Review: Well-designed but expensive." The testing department then passes the output response data to subsequent processing methods such as thresholding or clustering to detect trends or outliers in the target group's responses. The technological advantage is that the testing department can automatically and rapidly predict the responses of large-scale, diverse consumer profiles—something difficult to achieve with traditional manual questionnaires or A / B testing—thus optimizing product development and marketing strategy decisions based on scientific evidence. Applicable fields include new product development, advertising effectiveness measurement, service design, and financial product evaluation.

[0041] The analysis department can analyze test results and provide information for developing marketing strategies. For example, the analysis department analyzes test results obtained by the testing department. Analysis includes, but is not limited to, statistical analysis and data mining. The analysis department utilizes generative AI to provide information for developing marketing strategies based on the test results. For example, generative AI analyzes consumer reactions based on the test results and provides information for developing marketing strategies. Thus, by analyzing test results and providing information for developing marketing strategies, the analysis department can develop more effective marketing strategies. Some or all of the above processing in the analysis department can be implemented using generative AI, or it can be done without generative AI. For example, the analysis department inputs test results into generative AI, which then analyzes the test results. Specifically, this analysis department can accept response data output by the testing department (such as structured data like purchase intention scores, positive reviews, and negative feedback) as input. Inputs include, for example, "Purchase intention score 0.92" and "Review: Well-designed but expensive." The analysis department performs statistical analyses such as cluster analysis, principal component analysis, and regression analysis, as well as data mining such as association analysis and outlier detection, outputting decision support information such as target group response distribution, price optimization suggestions, and advertising channel recommendations. Outputs might include phrases like "Target group A's average purchase intention score is 0.85" and "Price optimization suggestion: It is recommended to reduce the current price by 5%." The analytics department then transmits the output decision support information to dashboard displays or report generation modules to assist management or marketing personnel in decision-making. The technological benefits include the ability for the analytics department to automate and rapidly extract patterns and propose strategies from high-dimensional, multivariate data—capabilities difficult to achieve with traditional simple statistical or manual analysis—significantly improving the accuracy and speed of marketing strategies. Applicable areas include new product development, advertising campaign design, service improvement, and financial product strategy formulation, among others.

[0042] The learning department can infer users' emotions and adjust the timing of market data learning based on these inferred emotions. For example, the learning department can infer users' emotions and adjust the timing of market data learning accordingly. User emotions can be inferred, for example, through techniques such as sentiment analysis and facial expression recognition. Generative AI can adjust to learn market data quickly when users are excited, slowly when users are relaxed, and temporarily stop learning when users are stressed, resuming only after the users have calmed down. Thus, the learning department can adjust the timing of market data learning based on user emotions, thereby learning market data at a more appropriate time. Emotion inference can be achieved through emotion inference functions such as emotion engines or generative AI. Generative AI can be text generation AI (such as LLM) or multimodal generation AI, but is not limited to these. Some or all of the above processing in the learning department can be implemented using generative AI, or it can be implemented without generative AI. For example, the learning department can input user emotion data into generative AI, which can then adjust the timing of market data learning. Specifically, this learning unit can accept inputs such as facial image tensors (e.g., 224×224×3 RGB images) for user sentiment inference, speech waveform data (e.g., a 16000-dimensional array sampled at 16kHz per second), and text sentence data (e.g., a natural language sentence with 100 tokens). Inputs include, for example, "smiling face image," "high-pitched speech waveform," and "text 'excited about the new product.'" The learning unit extracts features from these multimodal data using CNN or Transformer-type neural networks, outputting sentiment labels or scores such as "excitement," "relaxation," and "stress" (e.g., excitement 0.85, relaxation 0.10, stress 0.05) at the sentiment classification layer. Outputs include, for example, "sentiment label: excitement, score 0.92" and "sentiment label: relaxation, score 0.75." Based on these sentiment inference results, the learning unit dynamically adjusts hyperparameters such as learning start timing, batch size, number of rounds, and learning rate in the learning scheduling module. For example, when excitement is high, the batch size is increased to improve the learning frequency; when stress is high, learning is paused and resumed later. These controls differ from traditional fixed-schedule learning, achieving adaptive learning control based on user state. The technical effect is that this learning unit can learn market data at the optimal time while reducing the user's psychological burden, thereby improving the model's learning efficiency and accuracy. Furthermore, the linkage between real-time sentiment inference and learning control enhances the quality of user experience. Applicable fields include consumer survey systems, educational AI assistants, patient monitoring in the medical field, and interactive advertising in the entertainment field, among others.

[0043] The learning department can adjust its learning algorithm by referencing past market trends when learning market data. For example, the learning department adjusts its learning algorithm based on past market trends. Past market trends include, but are not limited to, past sales data and economic indicators. Generative AI can optimize its learning algorithm by referencing market trend data from the past 5 years; it can also adjust the learning algorithm by referring to market trend data related to specific seasons or events; and it can extract trends from past market trend data to improve the learning algorithm. Thus, the learning department can achieve more accurate market data learning by optimizing its learning algorithm by referring to past market trends. Some or all of the above processing in the learning department can be implemented using generative AI, or it can be done without generative AI. For example, the learning department can input past market trend data into the generative AI, which will then adjust the learning algorithm. Specifically, this learning department can accept past market trend data as input, such as time-series sales tensors (e.g., 60 months of sales data, 60×1 tensor), seasonal event flag arrays (e.g., 12-month event flags, 12-dimensional binary array), and time-series economic indicators (e.g., GDP growth rate, unemployment rate, consumer confidence index, etc., 60×3 tensor). Inputs include, for example, "monthly sales data from 2019 to 2023," "event markers such as Christmas, New Year's Day, and summer vacation," and "monthly GDP growth rates." The learning unit extracts features from this time-series data using LSTM or Transformer-type neural networks, calculating features such as moving averages, seasonal variations, and event impact at the trend extraction layer. Based on these features, the learning unit automatically optimizes the hyperparameters of the learning algorithm (such as learning rate, batch size, weight initialization method, and regularization coefficient). For example, it adaptively controls the learning rate by increasing it during months with sales surges and decreasing it during stable periods. The learning unit can also detect past trend changes and dynamically adjust the model structure (such as the number of layers and hidden units). Outputs include, for example, "learning rate adjusted from 0.01 to 0.05" and "batch size adjusted from 32 to 64." These processes differ from traditional static learning algorithm settings, achieving dynamic optimization based on historical data. The technical effect is that this learning unit can reflect market seasonal variations and event impacts, achieving high-precision model learning and significantly improving prediction accuracy and adaptability. It is applicable to various fields including retail demand forecasting, advertising optimization, financial product risk assessment, and supply chain management.

[0044] The learning department can focus on specific market segments when learning market data. For example, the learning department can focus on specific market segments, such as youth, senior, and regional market segments, but is not limited to these. Generative AI can focus on learning data from the youth market segment; it can also focus on the senior market segment; and it can also focus on learning data from regional market segments. Thus, the learning department can achieve more targeted market data learning by focusing on specific market segments. Some or all of the above processing in the learning department can be implemented using generative AI, or it can be done without it. For example, the learning department inputs specific market segment data into the generative AI, which then learns from that data. Specifically, this learning department can accept inputs such as segmentation attribute vectors (e.g., 50-dimensional vectors such as age group, gender, region, income level, and interest preferences), detailed purchase record tensors (e.g., 10×12 tensors such as monthly purchase frequency for the youth group and sales changes by region), and detailed response score arrays (e.g., 10-dimensional arrays such as new product interest scores). Input is a vector such as "18-25 years old, female, city, middle income, fashion preference," representing "purchase records of the elderly in the Kanto region." The learning unit extracts features from this segmented data using a Transformer-type neural network or clustering algorithm, learning the potential patterns and purchasing tendencies of each segment. The learning unit sets different loss function weights and data sampling ratios for each segment to achieve optimal model learning for the target group. Outputs include, for example, "Purchasing tendency of the youth segment: new product orientation" and "Price sensitivity of the elderly segment: high." This processing differs from traditional overall averaging learning, achieving individual optimization for the target group. The technical effect is that this learning unit can achieve high-precision market forecasting and demand analysis for specific segments, maximizing the effectiveness of marketing measures. Applicable fields include targeted advertising, regional product development, age-based service design, and personalized financial products.

[0045] The learning department can infer users' emotions and determine the priority of learning data based on these inferred emotions. For example, the learning department can infer users' emotions and determine the priority of learning data based on these inferred emotions. User emotions can be inferred, for example, through techniques such as emotion analysis and facial expression recognition. Generative AI, for example, prioritizes learning important market data when users are excited; prioritizes learning detailed market data when users are relaxed; and prioritizes learning simple market data when users are stressed. Thus, the learning department can determine the priority of learning data based on user emotions, prioritizing the learning of more important market data. Emotion inference can be achieved through emotion inference functions such as emotion engines or generative AI. Generative AI can be text generation AI (such as LLM) or multimodal generation AI, but is not limited to these. Some or all of the above processing in the learning department can be implemented using generative AI, or it can be implemented without generative AI. For example, the learning department inputs user emotion data into generative AI, which then determines the priority of the learning data. Specifically, this learning unit can accept inputs such as facial image tensors (e.g., 224×224×3), speech waveform data (e.g., 16000-dimensional), and text sentence data (e.g., 100 tokens) for user sentiment inference. Inputs include, for example, "smiling face image" or "text 'Looking forward to the new product'". The learning unit processes this data using a CNN or Transformer-type neural network, outputting sentiment labels or scores (e.g., excitement 0.9, relaxation 0.7, stress 0.2). Based on the sentiment inference results, the learning unit assigns priority scores to each data point in the learning dataset, performing batch learning in descending order of priority. For example, high excitement prioritizes learning data related to new products or important market changes; high relaxation prioritizes learning detailed attribute data or auxiliary data; and high stress prioritizes learning simple data or basic data. Outputs include, for example, "Priority 1: New Product Data" and "Priority 2: Detailed Attribute Data". This processing differs from traditional random sampling or fixed-order learning, achieving dynamic data priority control based on user state. The technical benefits include the ability of this learning unit to select optimal learning data based on user attention and psychological state, thereby improving learning efficiency and model accuracy. Applicable fields include educational AI, personalized marketing, medical diagnostic support, and interactive advertising.

[0046] The learning department can learn market data while considering geographic market characteristics. For example, the learning department considers geographic market characteristics, which may include, but are not limited to, urban market characteristics, rural market characteristics, and international market characteristics. Generative AI can learn the data considering urban market characteristics, rural market characteristics, and international market characteristics. Thus, the learning department can achieve more regionally specific market data learning by considering geographic market characteristics. Some or all of the above processing in the learning department can be implemented using generative AI, or it can be done without it. For example, the learning department can input geographic market characteristic data into the generative AI, which will then learn from the data. Specifically, this learning department can accept inputs such as geographic attribute vectors (e.g., 20-dimensional vectors such as city / rural / overseas markers, region codes, population density, average income, etc.), region-specific purchase record tensors (e.g., monthly sales data for cities, rural areas, and overseas, 3×12 tensors), and region-specific response score arrays (e.g., 3-dimensional arrays such as new product regional interest scores, etc.). Inputs include, for example, "Cities: Tokyo, population 14 million, average income 6 million yen" and "Rural areas: Nagano Prefecture, population 200,000, average income 3.5 million yen." The learning department extracts features from this geographical data using Transformer-type neural networks or geospatial clustering algorithms to learn the purchasing tendencies and response patterns of each region. The learning department sets different loss function weights and data sampling ratios for each region to achieve region-specific model learning. Outputs include, for example, "Urban new product orientation: High," "Rural price sensitivity: High," and "Overseas market brand orientation: Strong." This processing differs from traditional nationwide uniform learning, achieving individual optimization that reflects regional characteristics. The technical effect is that this learning department can accurately reflect the consumption behavior and market characteristics of each region, contributing to the optimization of regional marketing and globalization strategies. Applicable areas include local revitalization support, urban service development, overseas market entry strategies, and tourism demand forecasting, among others.

[0047] The learning department can analyze social media trends and reflect them in its learning process when studying market data. For example, it analyzes and reflects social media trends, including but not limited to trending hashtags, influencer posts, and user comments. Generative AI can analyze trending hashtags on social media and learn from them; it can also analyze influencer posts and learn from them; and it can analyze user comments and learn from them. Thus, the learning department can achieve more timely market data learning by analyzing and reflecting social media trends. Some or all of the above processing in the learning department can be implemented using generative AI, or it can be done without it. For example, the learning department can input social media trend data into the generative AI, which will then learn from that data. Specifically, the learning department can accept inputs such as hashtag frequency vectors (e.g., the number of occurrences of 1000 hashtags, a 1000-dimensional vector), influencer posting tensors (e.g., the number of posts by 100 people over 12 months, a 100×12 tensor), and user comment embedding vectors (e.g., BERT-encoded comment vectors, 768-dimensional). Inputs include the frequency of tags such as "#newproduct" and "#popularword," the monthly posting count of well-known influencer A, and user comments like "This product is innovative." The learning department extracts features from this social media data using natural language processing models or graph neural networks, calculating hot words and topicality scores at the trend detection layer. Data with high trend scores is prioritized for inclusion in the learning dataset, adapting the model weights to the latest trends. Outputs include, for example, "Trend tag: #newproduct, score 0.95" and "Topic influencer: A, influence 0.88." This processing differs from traditional static data learning, achieving real-time trend-responsive learning. The technical effect is that this learning department can instantly reflect the latest market dynamics and changes in consumer attention into the model, significantly improving prediction accuracy and the responsiveness of marketing measures. Applicable fields include SNS marketing, brand monitoring, advertising effectiveness measurement, and consumer behavior analysis.

[0048] The generation unit can infer the user's emotions and adjust the characteristics of the pseudo-personality based on the inferred emotions. For example, the generation unit can infer the user's emotions and adjust the characteristics of the pseudo-personality accordingly. User emotions can be inferred, for example, through techniques such as emotion analysis and facial expression recognition. Generative AI can generate pseudo-personalities with positive characteristics when the user is excited; with mild characteristics when the user is relaxed; and with calm characteristics when the user is stressed. Thus, the generation unit can adjust the characteristics of the pseudo-personality based on the user's emotions to generate a more suitable pseudo-personality. Emotion inference can be achieved through emotion inference functions such as emotion engines or generative AI. Generative AI can be text generation AI (such as LLM) or multimodal generation AI, but is not limited to these. Some or all of the above processing in the generation unit can be implemented using generative AI, or it can be implemented without generative AI. For example, the generation unit can input user emotion data into the generative AI, which then adjusts the characteristics of the pseudo-personality. Specifically, this generation unit can accept multimodal data as input, including facial image tensors (such as 224×224×3 RGB images), speech waveform data (such as 16000-dimensional arrays sampled at 16kHz per second), and text sentence data (such as natural language sentences with 100 tokens). Inputs include, for example, "smiling face images," "text 'excited about the new product,'" and "high-pitched speech waveforms." The generation unit processes this data using CNN or Transformer-type neural networks, outputting emotion labels or scores such as "excitement," "relaxation," and "stress" at the emotion classification layer (e.g., excitement 0.85, relaxation 0.10, stress 0.05). Outputs include, for example, "emotion label: excitement, score 0.92" and "emotion label: relaxation, score 0.75." Based on these emotion inference results, the generation unit assigns emotion parameters as generation conditions for generative AI to latent feature vectors, and adjusts pseudo-personality attribute vectors (such as positivity scores, cooperativeness scores, calmness, etc.) using generative models such as VAEs or diffusion models. For example, when excitement is high, the positivity score is set to 0.9; when relaxation is high, the cooperation score is set to 0.8; and when stress is high, the calmness score is set to 0.95. Outputs might include "Pseudo-personality A: positivity 0.9, cooperation 0.6, calmness 0.2" and "Pseudo-personality B: positivity 0.3, cooperation 0.8, calmness 0.9". The generation department utilizes generative AI's weight optimization and conditional generation techniques to generate diverse personalities based on user emotions. These processes differ from traditional fixed attribute settings or manual adjustments, achieving dynamic and automatic feature adjustments based on user states. The technical effect is that this generation department can generate pseudo-personalities in real time, reflecting the user's psychological state, significantly improving the personalization and accuracy of response tests and marketing initiatives. Applicable fields include consumer surveys, educational AI, patient simulation in the medical field, and interactive character generation in the entertainment field, among others.

[0049] The generation department can adjust features by referring to past market data when generating pseudo-personalities. For example, the generation department adjusts features by referring to past market data. Past market data includes, but is not limited to, past sales data and economic indicators. Generative AI can generate pseudo-personalities with the most common features based on past market data; it can also generate pseudo-personalities with specific trend features based on past market data; and it can also generate pseudo-personalities with specific consumer behavior features based on past market data. Thus, the generation department can generate more accurate pseudo-personalities by optimizing features by referring to past market data. Some or all of the above processing in the generation department can be implemented using generative AI, or it can be done without generative AI. For example, the generation department inputs past market data into the generative AI, which then adjusts the features of the pseudo-personality. Specifically, this generation department can accept past market data as input, such as time-series sales tensors (e.g., 60 months of sales data, 60×1 tensor), time-series economic indicators (e.g., GDP growth rate, unemployment rate, consumer confidence index, etc., 60×3 tensor), and consumer behavior record tensors (e.g., purchase frequency, product category purchase tendency, etc., 100×12 tensor), etc. Inputs include, for example, "monthly sales data from 2019 to 2023," "monthly GDP growth rate," and "Consumer A's purchase history over the past year." The generation department extracts features from this time-series data using LSTM or Transformer-type neural networks, calculating features such as moving averages, seasonal variations, and consumer behavior patterns at the trend extraction layer. These features are then used as conditions for generative AI, assigning them to latent feature vectors. Generative models such as VAEs or diffusion models generate pseudo-personality attribute vectors (e.g., age, gender, purchasing tendency, price sensitivity, trend adaptation). For example, it can generate pseudo-personality emphasizing purchasing tendencies in popular product categories over the past five years, or pseudo-personality adjusting price sensitivity based on economic indicator changes. Outputs include, for example, "Pseudo-personality A: Appliance-oriented, Price Sensitivity 0.7, Trend Adaptability 0.9" and "Pseudo-personality B: Food-oriented, Price Sensitivity 0.4, Trend Adaptability 0.6." The generation department utilizes weight optimization and conditional generation techniques in generative AI to generate diverse personalities based on historical data. This processing differs from traditional simple mean setting or manual attribute adjustment, achieving dynamic and automatic feature optimization based on historical data. The technological benefits include the ability of this generation department to reflect market trends and changes in consumer behavior, achieving high-precision personality generation and significantly improving the accuracy and adaptability of response testing and marketing initiatives. Applicable fields include retail demand forecasting, advertising targeting, financial product risk assessment, and service design, among others.

[0050] The generation department can focus on specific market segment characteristics when generating pseudo-personalities. For example, the generation department can focus on specific market segment characteristics. Specific market segments include, but are not limited to, youth market segments, senior market segments, and regional market segments. The generative AI can, for example, focus on the youth market segment to generate pseudo-personalities with those characteristics; it can also focus on the senior market segment to generate pseudo-personalities with those characteristics; and it can also focus on regional market segments to generate pseudo-personalities with those characteristics. Thus, the generation department can generate more targeted pseudo-personalities by focusing on specific market segment characteristics. Some or all of the above processing in the generation department can be implemented using generative AI, or it can be done without generative AI. For example, the generation department inputs specific market segment data into the generative AI, which then sets the pseudo-personality characteristics based on that data. Specifically, this generation department can accept inputs such as segmented attribute vectors (e.g., 50-dimensional vectors of age group, gender, region, income level, and interest preferences), segmented purchase record tensors (e.g., 10×12 tensors of monthly purchase frequency for the youth group and sales changes by region), and segmented response score arrays (e.g., 10-dimensional arrays of new product interest scores). Inputs include, for example, vectors for "18-25 years old, female, city, middle income, fashion preferences" and "purchase records of the elderly in the Kanto region." The generation department extracts features from these segmented data using Transformer-type neural networks or clustering algorithms to learn the potential patterns and purchasing tendencies of each segment. The generation department sets different loss function weights and data sampling ratios for each segment to achieve optimal generative model learning for the target group. Based on the extracted segmented features, the generative AI generates segmented specialized pseudo-personality attribute vectors using generative models such as VAEs or diffusion models (e.g., youth group: new product orientation 0.9, SNS usage frequency 0.8; elderly group: price sensitivity 0.7, health orientation 0.85, etc.). Outputs such as "Pseudo-personality A: Youth, Fashion-oriented, New Product-oriented 0.9" and "Pseudo-personality B: Elderly, Health-oriented 0.85, Price Sensitivity 0.7" demonstrate how this approach differs from traditional overall average generation or artificial attribute setting, achieving individual optimization for specific target groups. The technical effect is that this generation department can achieve high-precision personality generation for specific sub-groups, maximizing the effectiveness of marketing initiatives. Applicable fields include targeted advertising, geographic product development, age-based service design, and personalized financial products, among others.

[0051] The generation unit can infer the user's emotions and determine the priority of pseudo-personalities based on these inferred emotions. For example, the generation unit can infer the user's emotions and determine the priority of pseudo-personalities based on these inferred emotions. User emotions can be inferred, for example, through techniques such as emotion analysis and facial expression recognition. Generative AI, for example, prioritizes generating pseudo-personalities with positive characteristics when the user is excited; prioritizes generating pseudo-personalities with mild characteristics when the user is relaxed; and prioritizes generating pseudo-personalities with calm characteristics when the user is stressed. Thus, the generation unit can determine the priority of pseudo-personalities based on the user's emotions, prioritizing the generation of more important pseudo-personalities. Emotion inference can be achieved through emotion inference functions such as emotion engines or generative AI. Generative AI can be text generation AI (such as LLM) or multimodal generation AI, but is not limited to these. Some or all of the above processing in the generation unit can be implemented using generative AI, or it can be implemented without generative AI. For example, the generation unit inputs user emotion data into the generative AI, which then determines the priority of pseudo-personalities. Specifically, this generation unit can accept multimodal data as input, such as facial image tensors (e.g., 224×224×3), speech waveform data (e.g., 16000-dimensional), and text sentence data (e.g., 100 tokens). Inputs include, for example, "smiling face image" or "text 'Looking forward to the new product'". The generation unit processes this data using a CNN or Transformer-type neural network, outputting sentiment labels or scores (e.g., excitement 0.9, relaxation 0.7, stress 0.2). Based on the sentiment inference results, the generation unit assigns priority scores to the generative AI's candidate personalities, generating and testing personalities in descending order of priority. For example, high excitement prioritizes generating positive personality traits, high relaxation prioritizes generating mild personality traits, and high stress prioritizes generating calm personality traits. Outputs include, for example, "Priority 1: Positive Personality" and "Priority 2: Mild Personality". This processing differs from traditional random or fixed-order generation, achieving dynamic personality priority control based on the user's state. The technological benefits include the ability of this generator to achieve optimal personality selection based on user attention and psychological state, thereby improving the efficiency and accuracy of reaction tests and simulations. Applicable fields include educational AI, personalized marketing, medical diagnostic support, and interactive advertising.

[0052] The generation department can consider geographic market characteristics when generating pseudo-personalities. For example, the generation department considers geographic market characteristics when setting features. Geographic market characteristics include, but are not limited to, urban market characteristics, rural market characteristics, and international market characteristics. The generative AI can generate pseudo-personalities with these characteristics, for example, considering urban market characteristics; it can also consider rural market characteristics; and it can also consider international market characteristics. Thus, by considering geographic market characteristics when setting features, the generation department can generate pseudo-personalities with more regional characteristics. Some or all of the above processing in the generation department can be implemented using generative AI, or it can be done without generative AI. For example, the generation department inputs geographic market characteristic data into the generative AI, which then sets the pseudo-personality features based on this data. Specifically, this generation unit accepts inputs such as regional attribute vectors (e.g., 20-dimensional vectors including city / rural / overseas markers, region codes, population density, average income, etc.), regional purchase record tensors (e.g., monthly sales data for cities, rural areas, and overseas, 3×12 tensors), and regional response score arrays (e.g., 3-dimensional arrays including new product regional interest scores, etc.). Inputs could be, for example, "City: Tokyo, population 14 million, average income 6 million yen" or "Rural: Nagano Prefecture, population 200,000, average income 3.5 million yen." The generation unit extracts features from this geographic data using Transformer-type neural networks or geospatial clustering algorithms to learn the purchasing tendencies and response patterns of each region. The generation unit sets different loss function weights and data sampling ratios for each region to achieve the learning of a regionally specialized generation model. Generative AI, based on extracted regional features, generates regionally specific pseudo-personality attribute vectors (e.g., urban groups: new product orientation 0.8, rural groups: price sensitivity 0.9, overseas groups: brand orientation 0.85, etc.) using generative models such as VAE or diffusion models. Outputs include, for example, "Pseudo-personality A: urban group, trend orientation 0.8," "Pseudo-personality B: rural group, price sensitivity 0.9," and "Pseudo-personality C: overseas group, brand orientation 0.85." This process differs from traditional nationwide uniform generation or manual attribute setting, achieving individual optimization that reflects regional characteristics. The technical effect is that this generation department can accurately reflect the consumption behavior and market characteristics of various regions, contributing to the optimization of regional marketing and globalization strategies. Applicable areas include local revitalization support, urban service development, overseas market entry strategies, and tourism demand forecasting, among others.

[0053] The generation department can analyze social media trends and reflect them in its features when generating pseudo-personalities. For example, the generation department analyzes social media trends and reflects them in its features. Social media trends include, but are not limited to, trending hashtags, influencer posts, and user comments. Generative AI, for example, analyzes trending hashtags on social media to generate pseudo-personalities with those features; it can also analyze influencer posts on social media to generate pseudo-personalities with those features; and it can also analyze user comments on social media to generate pseudo-personalities with those features. Thus, the generation department can generate pseudo-personalities based on the latest market data by analyzing social media trends and reflecting them in its features. Some or all of the above processing in the generation department can be implemented using generative AI, or it can be done without generative AI. For example, the generation department inputs social media trend data into the generative AI, which then reflects pseudo-personality features based on that data. Specifically, this generation department accepts inputs such as tag frequency vectors (e.g., 1000-dimensional vectors representing the occurrence counts of 1000 tags), influencer posting tensors (e.g., 100×12 tensors representing the number of posts by 100 people over 12 months), and user comment embedding vectors (e.g., 768-dimensional comment vectors encoded with BERT, etc.). Inputs include, for example, the frequency of tags such as "#newproduct" and "#popularwords," the monthly posting count of well-known influencer A, and user comments like "This product is innovative." The generation department extracts features from this social media data using natural language processing models or graph neural networks, calculating hot words and topicality scores at the trend detection layer. Features with high trend scores are used as generation conditions for generative AI and assigned to latent feature vectors. Generative models such as VAEs or diffusion models generate trend-reflective pseudo-personality attribute vectors (e.g., topicality score 0.95, influencer influence 0.88, etc.). Outputs include, for example, "Pseudo-personality A: Topicality 0.95, Influencer Influence 0.88" and "Pseudo-personality B: Trend-oriented 0.9." These processes differ from traditional static data generation or manual attribute setting, achieving real-time trend-reflective generation. The technical effect is that this generation department can instantly reflect the latest market dynamics and changes in consumer attention into the personality model, significantly improving prediction accuracy and the responsiveness of marketing measures. Applicable fields include SNS marketing, brand monitoring, advertising effectiveness measurement, and consumer behavior analysis.

[0054] The testing department can infer users' emotions and adjust the reaction testing methods based on these inferred emotions. For example, the testing department can infer users' emotions and adjust the reaction testing methods accordingly. User emotions can be inferred, for example, through techniques such as emotion analysis and facial expression recognition. Generative AI can perform rapid reaction tests when users are excited; detailed reaction tests when users are relaxed; and simple reaction tests when users are stressed. Thus, the testing department can adjust the reaction testing methods based on user emotions to conduct more appropriate reaction tests. Emotion inference can be achieved through emotion inference functions such as emotion engines or generative AI. Generative AI can be text generation AI (such as LLM) or multimodal generation AI, but is not limited to these. Some or all of the above processing in the testing department can be implemented using generative AI, or it can be done without generative AI. For example, the testing department can input user emotion data into generative AI, which can then adjust the reaction testing methods. Specifically, this testing department accepts multimodal data as input, including facial image tensors (e.g., 224×224×3 RGB images) for user emotion inference, speech waveform data (e.g., a 16000-dimensional array sampled at 16kHz per second), and text sentence data (e.g., a natural language sentence with 100 tokens). Inputs include, for example, "smiling face images," "text 'excited about the new product,'" and "high-pitched speech waveforms." The testing department processes this data using CNN or Transformer-type neural networks, outputting emotion labels or scores such as "excitement," "relaxation," and "stress" at the emotion classification layer (e.g., excitement 0.85, relaxation 0.10, stress 0.05). Outputs include, for example, "Emotion label: Excitement, score 0.92" and "Emotion label: Relaxation, score 0.75." Based on these emotion inference results, the testing department dynamically adjusts the implementation method of the response test (e.g., the number of test items, test detail, test time, and test presentation order). For example, when excitement levels are high, the number of questions is reduced, prioritizing rapid A / B tests or simple questionnaires; when relaxation levels are high, detailed questions or scenario-based tests are prioritized; and when stress levels are high, only the fewest questions or multiple-choice questions are tested. Based on the output of generative AI (such as a list of test questions, test scenarios, and estimated test time), the testing department transmits the test results to the subsequent response data collection module or analysis department. This processing differs from traditional fixed test designs or manual test adjustments, achieving dynamic and automated optimization of testing methods based on user states. The technical benefits include that this testing department can automatically design optimal tests based on user psychological states and attention, significantly improving the reliability and collection efficiency of response data, and helping to improve user experience quality and reduce churn. Applicable fields include consumer surveys, adaptive testing of educational AI, patient response assessment in the medical field, and interactive experience design in the entertainment field, among others.

[0055] The testing department can adjust testing algorithms by referring to past test data when conducting reaction tests. This past test data includes, but is not limited to, past test results and conditions. For example, generative AI can select the optimal testing algorithm based on past test data. Furthermore, generative AI can adjust testing algorithms that conform to specific trends based on past test data. Moreover, generative AI can improve testing algorithms based on specific consumer behaviors based on past test data. Thus, by optimizing testing algorithms by referring to past test data, the testing department can conduct more accurate reaction tests. Some or all of the above-mentioned processing in the testing department can be performed using generative AI, or it can be performed without it. For example, the testing department can input past test data into generative AI, which then adjusts the testing algorithm. Specifically, the testing department can receive test result tensors (e.g., 1000 test results × 10 items, a 1000 × 10 tensor), test condition arrays (e.g., a 1000 × 5 array containing test implementation dates, target personality attributes, number of test questions, etc.), and consumer behavior history tensors (e.g., a 1000 × 12 tensor containing purchase frequency, response scores, etc.) as input. Input examples include "test result history from January to December 2022," "past test responses of Personality A," and "test conditions with 10 questions." The testing department utilizes this time-series and multivariate data to extract features using LSTM or Transformer-type neural networks. In the trend extraction layer, it calculates past test patterns, response tendencies, and algorithm performance metrics (e.g., accuracy, response speed, churn rate, etc.). Based on these features, the testing department automatically optimizes the hyperparameters of the testing algorithm (e.g., number of questions, question difficulty, test branch conditions, presentation order, etc.). For example, it avoids question patterns with high past churn rates and prioritizes question structures with high response accuracy. Furthermore, the testing department can detect past trend changes and consumer behavior patterns, dynamically modifying the logic of the testing algorithm (e.g., branching conditions, scoring methods). Output examples include "the number of questions changed from 8 to 6" and "the branching condition changed from A to B." These processes differ from previous static test designs or manual algorithm adjustments, achieving dynamic optimization based on historical data. Technically, this testing department can reflect past testing history and changes in consumer behavior, achieving high-precision test design and significantly improving the reliability and collection efficiency of response data. Applicable areas include consumer survey optimization, adaptive testing of educational AI, patient assessment protocol design in the medical field, and user experience optimization in the entertainment field, among others.

[0056] The testing department can focus on specific market segments when conducting reaction tests. For example, the testing department can focus on specific market segments, such as youth market segments, senior market segments, and regional market segments, but is not limited to these. For instance, generative AI can focus on the youth market segment and conduct reaction tests based on that data. Furthermore, generative AI can also focus on the senior market segment and conduct reaction tests based on that data. Further, generative AI can also focus on regional market segments and conduct reaction tests based on that data. Thus, by focusing on specific market segments, the testing department can conduct more targeted reaction tests. Some or all of the processing in the testing department described above can be performed using generative AI, or it can be performed without it. For example, the testing department can input data from a specific market segment into the generative AI, which then conducts reaction tests based on that data. Specifically, this testing department can accept detailed attribute vectors (e.g., a 50-dimensional vector representing age group, gender, region, income level, interests, etc.), detailed purchase history tensors (e.g., a 10×12 tensor representing monthly purchase frequency of young people, regional sales changes, etc.), and detailed response score arrays (e.g., a 10-dimensional array representing interest scores for new products, etc.) as input. Input examples include "18-25 years old". female urban areas Middle-income Vectors such as "fashion preferences" and "purchasing history of the elderly in the Kanto region" are used. The testing department utilizes this segmented data, employing Transformer-type neural networks or clustering algorithms for feature extraction, to learn the potential patterns and purchasing tendencies of each segment. The testing department automatically generates different test questions, scenarios, and evaluation criteria for each segment, implementing response tests optimized for the target group. For example, questions about new product preferences or SNS usage are posed to young people, while questions about price sensitivity or health preferences are posed to the elderly. Output examples include "a list of test questions for young people" and "test scenarios for the elderly." Based on the output of generative AI (such as the list of test questions, test scenarios, and evaluation criteria), the testing department passes the test results to the subsequent response data collection or analysis department. This processing differs from previous overall average testing or manual question design, achieving individual optimization for the target group. Technically, this testing department can achieve high-precision response testing for specific segments, maximizing the effectiveness of marketing initiatives and product development. Applicable areas include targeted advertising, geographically intensive product development, age-based service design, and personalized financial products.

[0057] The testing department can infer users' emotions and determine the priority of reaction tests based on the inferred user emotions. For example, the testing department infers users' emotions and determines the priority of reaction tests based on the inferred user emotions. User emotions are inferred, for example, through techniques such as emotion analysis or facial expression recognition. For instance, generative AI prioritizes important reaction tests when the user is excited. Furthermore, generative AI can also prioritize detailed reaction tests when the user is relaxed. Further, generative AI can prioritize simple reaction tests when the user is under stress. Thus, by prioritizing reaction tests based on users' emotions, the testing department can prioritize more important reaction tests. Emotion inference is achieved, for example, through the emotion inference function of an emotion engine or generative AI. Generative AI can be text-generating AI (such as LLM) or multimodal generative AI, but is not limited to these. Some or all of the processing in the testing department described above can be performed using generative AI, or it can be performed without generative AI. For example, the testing department inputs users' emotional data into generative AI, which then determines the priority of reaction tests. Specifically, this testing department can input multimodal data for user emotion inference, including facial image tensors (e.g., 224×224×3), speech waveform data (e.g., 16000-dimensional), and text sentence data (e.g., 100 tokens). Input examples include "smiling facial image" and "text 'Looking forward to the new product'". The testing department uses this data to process through CNN or Transformer-type neural networks, outputting emotion labels or scores (e.g., excitement 0.9, relaxation 0.7, stress 0.2). Based on the emotion inference results, the testing department assigns priority scores to the candidate list of reaction tests and implements tests in order of priority. For example, when excitement is high, tests related to new products or important market changes are prioritized; when relaxation is high, detailed attribute tests or auxiliary tests are prioritized; and when stress is high, simple tests or basic tests are prioritized. Output examples include "Priority 1: New product test" and "Priority 2: Detailed attribute test". Based on the output results of generative AI (e.g., test priority list, test implementation order, etc.), the testing department passes the test results to the subsequent reaction data collection or analysis department. These processes differ from traditional random sampling or fixed-order testing, enabling dynamic test prioritization based on user status. Technically, this testing department can optimize test selection based on user attention and psychological state, improving the reliability and collection efficiency of response data. Applicable fields include educational AI, personalized marketing, medical diagnostic support, and interactive advertising, among others.

[0058] The testing department can consider geographic market characteristics when conducting reaction tests. For example, the testing department can consider geographic market characteristics, such as urban market characteristics, rural market characteristics, and international market characteristics, but is not limited to these. For instance, generative AI can consider urban market characteristics and conduct reaction tests based on that data. Furthermore, generative AI can also consider rural market characteristics and conduct reaction tests based on that data. Further, generative AI can also consider international market characteristics and conduct reaction tests based on that data. Thus, by considering geographic market characteristics, the testing department can conduct more regionally specific reaction tests. Some or all of the processing in the testing department described above can be performed using generative AI, or it can be performed without generative AI. For example, the testing department can input geographic market characteristic data into generative AI, which then conducts reaction tests based on that data. Specifically, this testing department can accept regional attribute vectors (e.g., a 20-dimensional vector of city / rural / overseas identifiers, region codes, population density, average income, etc.), regional purchase history tensors (e.g., monthly sales data for cities, rural areas, and overseas, a 3×12 tensor), and regional response score arrays (e.g., a 3-dimensional array of regional interest scores for new products, etc.) as input. An example input is "City: Tokyo". Population 14 million Average income of 6 million yen; Rural areas: Nagano Prefecture Population 200,000 The testing department utilizes geographical data such as "average income of 3.5 million yen" to extract features through Transformer-type neural networks or geospatial clustering algorithms, learning the purchasing tendencies and response patterns of different regions. The testing department automatically generates different test questions, scenarios, and evaluation criteria for each region, implementing region-specific response tests. For example, questions about trend tendencies or brand awareness are asked for urban areas, questions about price sensitivity or local preferences for rural areas, and questions about cultural adaptation or local needs for overseas markets. Output examples include "test question list for cities," "test scenarios for rural areas," and "evaluation criteria for overseas markets." Based on the output of generative AI (such as test question list, test scenarios, and evaluation criteria), the testing department transmits the test results to the subsequent response data collection or analysis department. This process differs from previous nationwide standardized tests or manually designed questions, achieving individual optimization that reflects regional characteristics. Technically, this testing department can accurately reflect the test designs of consumer behavior and market characteristics in different regions, contributing to the optimization of region-intensive marketing and globalization strategies. Applicable areas include local revitalization support, urban service development, overseas market entry strategies, and tourism demand forecasting.

[0059] The testing department can analyze social media trends and incorporate them into the testing process during reaction testing. For example, the testing department analyzes social media trends and incorporates them into the testing. Social media trends include, but are not limited to, trending hashtags, influencer posts, and user comments. For instance, generative AI can analyze trending hashtags on social media and conduct reaction testing based on that data. Furthermore, generative AI can analyze influencer posts on social media and conduct reaction testing based on that data. Further, generative AI can analyze user comments on social media and conduct reaction testing based on that data. Thus, by analyzing social media trends and incorporating them into the testing, the testing department can conduct reaction testing based on the latest market data. Some or all of the above-mentioned processing in the testing department can be performed using generative AI, or it can be performed without it. For example, the testing department can input social media trend data into generative AI, which then conducts reaction testing based on that data. Specifically, the testing department can receive inputs such as tag frequency vectors (e.g., the number of occurrences of 1000 tags, a 1000-dimensional vector), influencer posting tensors (e.g., the number of posts by 100 people over 12 months, a 100×12 tensor), and user comment embedding vectors (e.g., comment vectors encoded using BERT, etc., 768-dimensional). Input examples include the frequency of tags such as "#newproduct" and "#popularwords," the monthly posting count of well-known influencer A, and user comments such as "This product is innovative." The testing department uses this social media data to extract features through natural language processing models or graph neural networks, and calculates surging words and topicality scores in the trend detection layer. The testing department uses topics or influencer posts with high trend scores as the generation conditions for generative AI, and uses generative models such as VAEs or diffusion models to generate test questions or scenarios that reflect trends. Output examples include "Trend Tag Question: #newproduct," "Influencer Posting Reflection Scenario," and "Topicality Score 0.95." Based on the output of generative AI (such as a list of test questions, test scenarios, and topicality scores), the testing department transmits the test results to the subsequent response data collection or analysis department. This process differs from previous static data testing or manual question design, enabling real-time trend-reflective testing. Technically, this testing department can instantly reflect the latest market dynamics and changes in consumer attention in the test design, significantly improving the freshness of response data and the responsiveness of marketing initiatives. Applicable areas include SNS marketing, brand monitoring, advertising effectiveness measurement, and consumer behavior analysis.

[0060] The analysis department can infer users' emotions and adjust the analysis methods for test results based on the inferred user emotions. For example, the analysis department infers users' emotions and adjusts the analysis methods for test results based on the inferred user emotions. Users' emotions are inferred, for example, through techniques such as emotion analysis or facial expression recognition. For instance, generative AI performs rapid analysis when a user is excited. Furthermore, generative AI can perform detailed analysis when a user is relaxed. Further, generative AI can perform simplified analysis when a user is stressed. Thus, by adjusting the analysis methods for test results based on users' emotions, the analysis department can perform more appropriate analysis. Emotion inference is achieved, for example, through the emotion inference function of an emotion engine or generative AI. Generative AI can be text-generating AI (such as LLM) or multimodal generative AI, but is not limited to these. Some or all of the processing in the analysis department described above can be performed using generative AI, or it can be performed without generative AI. For example, the analysis department inputs users' emotion data into generative AI, which then adjusts the analysis methods for test results. Specifically, this analysis unit can input multimodal data for user sentiment inference, including facial image tensors (e.g., 224×224×3 RGB images), speech waveform data (e.g., a 16000-dimensional array sampled at 16kHz per second), and text sentence data (e.g., a natural language sentence with 100 tokens). Input examples include "smiling facial images," "text 'excited about the new product,'" and "voice waveforms with a high-pitched tone." The analysis unit uses this data to process through CNN or Transformer-type neural networks, outputting sentiment labels or sentiment scores such as "excitement," "relaxation," and "stress" at the sentiment classification layer (e.g., excitement 0.85, relaxation 0.10, stress 0.05). Output examples include "sentiment label: excitement, score 0.92" and "sentiment label: relaxation, score 0.75." Based on these sentiment inferences, the analysis department dynamically adjusts the analysis methods (e.g., simple summarization, detailed multivariate analysis, summary report generation, detailed report generation, etc.) and granularity (e.g., summarization unit, number of variables, visualization methods, etc.) within the analysis method selection module. For example, when excitement levels are high, priority is given to quickly summarizing key indicators or displaying them on a dashboard; when relaxation levels are high, detailed cluster analysis, regression analysis, principal component analysis, and other multivariate analyses are implemented; and when stress levels are high, a simple summary report is generated or only key indicators are analyzed. The analysis department then transmits the analysis results (e.g., target group response distribution, price optimization suggestions, advertising channel recommendations, etc.) to subsequent decision support or report generation modules. This processing differs from previous fixed analysis methods or manual analysis designs, achieving dynamic and automated optimization of analysis methods based on user states. Technically, this analysis department can automatically design optimal analysis based on user psychological states and attention, significantly improving the reliability and utilization efficiency of analysis results, while also contributing to improved user experience quality and analysis business efficiency.It is applicable to multiple fields, including consumer survey analysis, adaptive performance analysis in education AI, patient evaluation analysis in the medical field, and user experience analysis in the entertainment field.

[0061] The analysis department can adjust its analysis algorithms by referring to past analysis data when analyzing test results. For example, the analysis department adjusts the analysis algorithms by referring to past analysis data. Past analysis data includes, but is not limited to, past analysis results and analysis conditions. For example, generative AI can select the optimal analysis algorithm based on past analysis data. Furthermore, generative AI can also adjust analysis algorithms that conform to specific trends based on past analysis data. Further, generative AI can also improve analysis algorithms based on specific consumer behaviors based on past analysis data. Thus, by optimizing analysis algorithms by referring to past analysis data, the analysis department can perform more accurate analysis. Some or all of the above-mentioned processing in the analysis department can be performed using generative AI, or it can be performed without generative AI. For example, the analysis department inputs past analysis data into the generative AI, which then adjusts the analysis algorithm. Specifically, this analysis department can receive analysis result tensors (e.g., 1000 × 20 tensors representing analysis results of 1000 items × 20 items), analysis condition arrays (e.g., 1000 × 5 arrays representing analysis implementation dates, target personality attributes, and types of analysis methods), and consumer behavior history tensors (e.g., 1000 × 12 tensors representing purchase frequency and response scores). Input examples include "historical analysis results from January to December 2022," "past analysis results for Personality A," and "analysis conditions for cluster analysis methods." The analysis department utilizes this time-series and multivariate data to extract features using LSTM or Transformer-type neural networks. In the trend extraction layer, it calculates past analysis patterns and algorithm performance indicators (e.g., analysis accuracy, processing speed, error rate). Based on these features, the analysis department automatically optimizes the hyperparameters of the analysis algorithm (e.g., number of clusters, number of principal components, regularization coefficients of the regression model) and the analysis methods (e.g., cluster analysis, regression analysis, principal component analysis, association analysis). For example, it prioritizes methods with high historical accuracy and avoids methods with high error rates. Furthermore, the analysis department can detect past trend changes and consumer behavior patterns, dynamically adjusting the logic of the analysis algorithm (e.g., feature selection, weighting methods). Output examples include "cluster number changed from 5 to 8" and "regularization coefficient of regression model changed from 0.1 to 0.05." These processes differ from previous static analysis designs or manual algorithm adjustments, achieving dynamic optimization based on historical data. Technically, this analysis department can reflect past analysis history and changes in consumer behavior, achieving high-precision analysis design and significantly improving the reliability and efficiency of analysis results. Applicable fields include consumer survey optimization, performance analysis in educational AI, patient assessment protocol design in the medical field, and user experience optimization in the entertainment field, among others.

[0062] The analytics department can focus on specific market segments when analyzing test results. For example, it can focus on specific market segments, such as youth, seniors, and regional segments, but is not limited to these. For instance, generative AI can focus on the youth market segment and analyze the data based on that. Furthermore, it can focus on the senior market segment and analyze the data based on that. Further, it can focus on regional market segments and analyze the data based on that. Thus, by focusing on specific market segments, the analytics department can conduct more targeted analysis. Some or all of the processing described above in the analytics department can be performed using generative AI, or it can be performed without it. For example, the analytics department can input data from a specific market segment into the generative AI, which then performs analysis based on that data. Specifically, this analysis department can accept detailed attribute vectors (e.g., a 50-dimensional vector representing age group, gender, region, income level, interests, etc.), detailed purchase history tensors (e.g., a 10×12 tensor representing monthly purchase frequency of young people, regional sales changes, etc.), and detailed response score arrays (e.g., a 10-dimensional array representing interest scores for new products, etc.) as input. An example input includes "18-25 years old". female urban areas Middle-income Vectors such as "fashion preferences" and "purchasing history of the elderly in the Kanto region" are used. The analytics department utilizes these segmented data, employing Transformer-type neural networks or clustering algorithms for feature extraction, to analyze the potential patterns and purchasing tendencies of each segment. The analytics department sets different analytical methods, weighting, and visualization techniques for each segment to achieve optimized analysis for the target group. Output examples include "purchasing tendencies of the young segment: new product preference" and "price sensitivity of the elderly segment: high." This processing differs from previous overall average analysis or manual analysis design, achieving individual optimization for the target group. Technically, this analytics department can achieve high-precision market and demand analysis for specific segments, maximizing the effectiveness of marketing initiatives. Applicable areas include targeted advertising, geographic-intensive product development, age-based service design, and personalized financial products.

[0063] The analysis unit can infer the user's emotions and determine the priority of test results based on the inferred emotions. For example, the analysis unit infers the user's emotions and determines the priority of test results based on the inferred emotions. The user's emotions are inferred, for example, through techniques such as emotion analysis or facial expression recognition. For instance, generative AI prioritizes analyzing important test results when the user is excited. Furthermore, generative AI can also prioritize analyzing detailed test results when the user is relaxed. Further, generative AI can prioritize analyzing simple test results when the user is stressed. Thus, by determining the priority of test results based on the user's emotions, the analysis unit can prioritize analyzing more important test results. Emotion inference is achieved, for example, through the emotion inference function of an emotion engine or generative AI. Generative AI can be text generation AI (such as LLM) or multimodal generation AI, but is not limited to these. Some or all of the processing in the analysis unit described above can be performed using generative AI, or it can be performed without generative AI. For example, the analysis unit inputs the user's emotion data into the generative AI, which then determines the priority of the test results. Specifically, this analysis unit can input multimodal data for user sentiment inference, including facial image tensors (e.g., 224×224×3), speech waveform data (e.g., 16000-dimensional), and text sentence data (e.g., 100 tokens). Input examples include "smiling facial image" and "text 'Looking forward to the new product'". The analysis unit uses this data to process through CNN or Transformer-type neural networks, outputting sentiment labels or scores (e.g., excitement 0.9, relaxation 0.7, stress 0.2). Based on the sentiment inference results, the analysis unit assigns priority scores to the test result list, analyzing the test results in order of priority. For example, when excitement is high, test results related to new products or important market changes are analyzed first; when relaxation is high, detailed attribute tests or auxiliary test results are analyzed first; and when stress is high, simple tests or basic test results are analyzed first. Output examples include "Priority 1: New product test result" and "Priority 2: Detailed attribute test result". This processing differs from previous random sampling or fixed-order analysis, achieving dynamic priority control of analysis based on user state. In terms of technical effectiveness, this analytics department can achieve optimized analysis based on user attention and psychological state, improving analysis efficiency and the reliability of results. Applicable fields include educational AI, personalized marketing, medical diagnostic support, interactive advertising, and more.

[0064] The analysis department can consider geographic market characteristics when analyzing test results. For example, the analysis department considers geographic market characteristics, which may include, but are not limited to, urban market characteristics, rural market characteristics, and international market characteristics. For instance, generative AI can consider urban market characteristics and analyze based on that data. Furthermore, generative AI can also consider rural market characteristics and analyze based on that data. Further, generative AI can also consider international market characteristics and analyze based on that data. Thus, by considering geographic market characteristics, the analysis department can conduct more regionally specific analyses. Some or all of the processing in the analysis department described above can be performed using generative AI, or it can be performed without using generative AI. For example, the analysis department inputs geographic market characteristic data into generative AI, which then performs analysis based on that data. Specifically, this analysis department can accept regional attribute vectors (e.g., a 20-dimensional vector of city / rural / overseas indicators, region codes, population density, average income, etc.), regional purchase history tensors (e.g., monthly sales data for cities, rural areas, and overseas, a 3×12 tensor), and regional response score arrays (e.g., a 3-dimensional array of regional interest scores for new products, etc.) as input. An example input is "City: Tokyo". Population 14 million Average income of 6 million yen; Rural areas: Nagano Prefecture Population 200,000 The analysis department utilizes geographical data such as "average income of 3.5 million yen" to extract features through Transformer-type neural networks or geospatial clustering algorithms, analyzing purchasing tendencies and response patterns in various regions. The analysis department employs different analytical methods, weighting, and visualization techniques for each region to achieve region-specific analysis. Output examples include "New product preference in cities: High," "Price sensitivity in rural areas: High," and "Brand preference in overseas markets: Strong." This processing differs from previous nationwide uniform analysis or manual analysis designs, achieving individual optimization that reflects regional characteristics. Technically, this analysis department can accurately reflect the consumption behavior and market characteristics of each region, contributing to the optimization of region-intensive marketing and globalization strategies. Applicable areas include local revitalization support, urban service development, overseas market entry strategies, and tourism demand forecasting.

[0065] The analytics department can analyze social media trends and reflect them in the results when analyzing test results. For example, the analytics department analyzes social media trends and reflects them in the results. Social media trends include, but are not limited to, trending hashtags, influencer posts, and user comments. For instance, generative AI can analyze trending hashtags on social media and perform analysis based on that data. Furthermore, generative AI can analyze influencer posts on social media and perform analysis based on that data. Further, generative AI can analyze user comments on social media and perform analysis based on that data. Thus, by analyzing social media trends and reflecting them in the results, the analytics department can perform analysis based on the latest market data. Some or all of the above processing in the analytics department can be performed using generative AI, or it can be performed without generative AI. For example, the analytics department inputs social media trend data into generative AI, which then performs analysis based on that data. Specifically, this analysis department can receive inputs such as tag frequency vectors (e.g., the number of occurrences of 1000 tags, a 1000-dimensional vector), influencer posting tensors (e.g., the number of posts by 100 people over 12 months, a 100×12 tensor), and user comment embedding vectors (e.g., comment vectors encoded using BERT, etc., 768-dimensional). Input examples include tag frequencies such as "#newproduct" and "#popularwords," monthly posting counts of well-known influencer A, and user comments like "This product is innovative." The analysis department utilizes this social media data to extract features through natural language processing models or graph neural networks, calculating surging words and topicality scores in the trend detection layer. The analysis department reflects high-trend-score topics or influencer posts in the analysis results, outputting decision support information considering the latest trends (e.g., trend tag response distribution, influencer influence analysis, reports with topicality scores, etc.). Output examples include "Trend tag: #newproduct, score 0.95" and "Popular influencer: A, influence 0.88." These processes differ from previous static data analysis or manual analysis designs, enabling real-time trend-reflective analysis. Technically, this analysis department can instantly reflect the latest market dynamics and changes in consumer attention in the analysis results, significantly improving the timeliness of the results and the responsiveness of marketing initiatives. Applicable fields include SNS marketing, brand monitoring, advertising effectiveness measurement, and consumer behavior analysis.

[0066] The system described in this embodiment is not limited to the examples above. For instance, various modifications can be made. Specifically, the system can be diversified in both hardware and software aspects, such as modularization of constituent elements, API linkage, introduction of distributed processing architecture, and adoption of cloud-based data storage or high-speed computing environments using GPU clusters. The system can combine various neural network architectures (such as CNN, RNN, Transformer, VAE, diffusion models, etc.). Furthermore, the system can flexibly accept various data types as input and output formats, such as image tensors (e.g., 224×224×3), time-series arrays (e.g., 60×1), natural language text (e.g., 100 tokens), and category vectors (e.g., 50-dimensional). Further, the system can automate the learning and inference processes of AI models by adding hyperparameter optimization modules or automatic feature extraction engines. For example, the learning unit can introduce advanced learning methods such as online learning, transfer learning, and meta-learning to achieve real-time adaptation to market changes. The generation unit can enhance the conditional generation and diversity control functions of the generative model to achieve more diversified personality generation. The testing department can add features such as automatic scenario generation or user-state adaptive test design to improve the flexibility and accuracy of reaction testing. The analysis department can improve the reliability and interpretability of analysis results by integrating anomaly detection, causal inference, and interpretable AI (XAI) modules. Through these extensions, this system differs from previous single-model, static processing systems, achieving a high level of computer technology integration, including multi-AI model linkage, dynamic process control, and real-time analysis of high-dimensional data. Technically, this system significantly improves responsiveness to market changes, model accuracy, data processing efficiency, and user experience quality. It is applicable to a wide range of fields, including consumer research, advertising optimization, financial risk assessment, medical diagnostic support, educational AI, and interactive experience design in the entertainment industry.

[0067] The market reaction prediction system can also include a feedback collection department. This department collects feedback from actual consumers and inputs it into the generative AI to improve the accuracy of the pseudo-personality model. For example, it collects consumer opinions on new products from surveys or review websites and inputs this data into the generative AI. Additionally, it can collect consumer posts or comments on social media and input them into the generative AI. Furthermore, it can collect consumers' actual experiences and satisfaction with using the new product and input this data into the generative AI. Thus, the feedback collection department can improve the accuracy of the pseudo-personality model based on actual consumer feedback. Specifically, this feedback collection department can automatically collect and standardize various types of data, such as consumer survey data (e.g., 5-level ratings for each question, free comments), rating arrays from review websites (e.g., 1000 ratings per product, 1000×1 tensor), social media post text (e.g., natural language text of up to 280 characters per post), image / video data (e.g., 224×224×3 images of product usage scenarios, video frame sequences), etc., and performs noise removal, language unification, anonymization, and other processing in the preprocessing module, structuring it into input tensors for the generative AI. Input examples include comments such as "'This product is easy to use'", "rating 4.5", and "#NewProductExperience". The feedback collection department uses natural language processing models (such as BERT and Transformer), image recognition models (such as CNN), and time series analysis models (such as LSTM) to extract features, inputting multi-dimensional feature vectors such as consumer sentiment scores, satisfaction indices, and topicality scores into the AI. The generated AI reflects these features from real-world data sources in the personality generation conditions or model weight optimization, generating high-precision pseudo-personalities that more closely resemble real consumer behavior and preferences through generative models such as VAEs or diffusion models. Output examples include "Pseudo-personality A: Satisfaction 0.92, Topicality 0.85" and "Pseudo-personality B: Dissatisfaction 0.3, Improvement Need: Price". The feedback collection department can also perform time-series changes and outlier detection on the collected data for continuous model updates and abnormal behavior pattern detection. These processes differ from previous simple manual summarization or static attribute settings; by directly reflecting real-time and diverse consumer feedback into the AI ​​model, the model's real-world adaptability, accuracy, and adaptability are significantly improved. In terms of technical effectiveness, this feedback collection department can instantly reflect the latest market trends and changes in consumer psychology in personality generation, significantly improving the accuracy and responsiveness of marketing measures and product development. Applicable areas include consumer research, product development, advertising effectiveness measurement, brand monitoring, customer support analysis, and many others.

[0068] The Learning Department also possesses real-time data collection capabilities. This capability allows for the real-time gathering of market dynamics and input into generative AI, enabling faster responses to market changes. For example, it can collect sales and inventory data in real-time and input them into generative AI. Furthermore, it can collect consumer purchasing behavior data in real-time and input it into generative AI. Moreover, it can collect competitor dynamic data in real-time and input it into generative AI. Thus, the Learning Department can rapidly respond to market changes through real-time data collection. Specifically, this Learning Department can automatically collect sales time-series tensors (e.g., 1440×1 sales data per minute), inventory quantity arrays (e.g., 100×1 inventory quantity per SKU), purchase behavior event logs (e.g., 10000×4 arrays containing user ID, product ID, time, action type, etc.), and competitor price and promotion information (e.g., price change time series of competitor products, 50×24 tensors), automatically collected by second or minute. Input examples include "10 units sold at 10:00 AM on June 1, 2024", "50 units of product A in stock", and "price change event of competitor B". The learning department uses stream processing platforms (such as Kafka and Spark Streaming) to batch process and standardize this real-time data, performing outlier detection and missing value completion, and then extracting temporal features using LSTM or Transformer neural networks. Based on the extracted features (such as sales surges, inventory shortage signals, and competitor price change trends), the learning department dynamically adjusts model weights and hyperparameters (such as learning rate and batch size) in real-time during the AI ​​learning process. Output examples include "Learning rate changed from 0.01 to 0.05" and "New trend detection: competitor price reduction". Furthermore, the learning department can automatically execute model relearning or ensemble model switching based on real-time anomaly detection or trend change point detection. These processes differ from previous batch-processing, static data learning, enabling real-time AI learning control that responds to market changes. In terms of technical effectiveness, this learning unit can respond instantly to market upheavals and unforeseen events, significantly improving the model's predictive accuracy, adaptability, and operational efficiency. Applicable areas include demand forecasting, inventory optimization, dynamic pricing, real-time advertising, and financial transaction monitoring, among others.

[0069] The generation unit may also have a multilingual support function. The multilingual support function can learn market data in different languages and generate pseudo-personas in multiple languages. For example, it learns market data such as English, Spanish, Chinese, etc., and generates pseudo-personas in the corresponding languages. In addition, the multilingual support function can also generate pseudo-personas based on different cultures and habits. Further, the multilingual support function can also generate pseudo-personas considering the market characteristics of different regions. Thus, through the multilingual support function, the generation unit can generate pseudo-personas that adapt to different languages and cultures. Specifically, this generation unit can receive multilingual text data (such as: 100 tokens of review texts in English, Spanish, and Chinese respectively), multilingual attribute vectors (such as: 10-dimensional vectors of language ID, region code, cultural markers, etc.), multilingual purchase history tensors (such as: monthly purchase times in each language region, 5×12 tensor), etc. as inputs. Input examples include "English: 'Easy to use'", "Chinese: '非常好用'", "Spanish: 'Muy práctico'", etc. The generation unit uses these multilingual data to extract features through multilingual pre-trained models (such as mBERT, XLM-R), multilingual Transformers, cross-lingual embedding models, etc., and learns the consumer behavior patterns and preference tendencies of each language and culture. The generation unit assigns the extracted multilingual and multicultural feature quantities as generation conditions of the generation AI to the latent feature vector, and generates multilingual and multicultural adaptable pseudo-persona attribute vectors (such as: language fitness, cultural fitness, regional characteristic scores, etc.) through generation models such as VAE or diffusion models. Output examples include "Pseudo-persona A: English region, trend tendency 0.8", "Pseudo-persona B: Chinese region, price sensitivity 0.7", "Pseudo-persona C: Spanish region, health tendency 0.9", etc. In addition, the generation unit can also set different loss function weights and data sampling ratios for each region to achieve learning of language and culture specialized generation models. These processes are different from the previous single-language and single-culture generation or artificial attribute setting, and achieve the generation of pseudo-personas that accurately reflect the market characteristics of multiple languages, multiple cultures, and multiple regions. In terms of technical effects, this generation unit can greatly improve the accuracy and adaptability of global market expansion and multicultural marketing. The applicable fields include multiple fields such as global product development, multilingual advertising placement, region-specialized service design, and international brand strategy.

[0070] The testing department can also be equipped with virtual reality (VR) testing capabilities. This VR testing capability allows for the testing of pseudo-personalities within a virtual reality environment. For example, a new product can be introduced to a pseudo-personality in a VR environment, and their reaction observed. Furthermore, it can simulate consumer purchasing behavior in a VR environment and collect relevant data. Further, it can observe consumer experience and satisfaction in the VR environment and collect related data. Thus, the testing department can conduct reaction tests in a more realistic environment through the VR testing capability. Specifically, this testing department can receive VR environment scene data (such as 3D spatial coordinate sequences, user gaze tracking data, and interaction event logs), VR in-app purchase behavior tensors (such as 100×10 tensors of product selection frequency, dwell time, and gaze movement patterns), and VR in-app emotional response data (such as facial expression change frame sequences, voice tone changes, and biometric data) as input. Input examples include "the action of picking up product A in VR," "3 minutes of gaze concentration," and "saying 'It's easy to use,'" etc. The testing department utilizes this VR data to extract features through 3D spatial analysis models (such as PointNet and 3D-CNN), temporal analysis models (such as LSTM), and multimodal emotion inference models (such as integrated models of voice, facial expressions, and biosignals), calculating consumer purchase intention scores, satisfaction indices, and interaction patterns. The testing department inputs the reaction data from the VR environment into AI to model the behavioral patterns and emotional changes of each personality type, applying this model to VR scene optimization or product design feedback. Output examples include "VR purchase intention score 0.88," "satisfaction score 0.92," and "gaze focus mode: product A→B→C." Furthermore, the testing department can automatically adjust VR environment parameters (such as product layout, lighting, and sound effects) to efficiently conduct A / B testing or multivariate testing. These processes differ from traditional questionnaires or 2D interface testing, enabling high-precision reproduction and analysis of consumer behavior in an immersive, highly immersive virtual environment, driving the advancement of computer technology. Technically, this testing department can significantly improve the accuracy and reliability of UX evaluation, purchase behavior analysis, and emotional response measurement for new products or services. It is applicable to multiple fields, including product development, store design, advertising experience design, VR simulation in education and medical fields, etc.

[0071] The analysis department can also possess predictive analytics capabilities. These capabilities can predict future market dynamics based on test results. For example, it can forecast sales of new products based on test results. Furthermore, it can predict changes in consumer purchase intentions based on test results. Moreover, it can predict competitor dynamics based on test results. Thus, the analysis department can use predictive analytics to forecast future market dynamics and develop more effective marketing strategies. Specifically, this analysis department can receive test result tensors (e.g., response data from 1000 items × 10 items), time-series purchase intention score arrays (e.g., intention scores for each user over 12 months, 1000 × 12 tensors), and competitor dynamic time-series data (e.g., sales and price changes of competing products, 50 × 24 tensors) as input. Input examples include "new product test response in June 2024," "time-series purchase intention score for user A," and "price changes for competitor B." The analysis department uses this data to process it through time-series predictive models (e.g., LSTM, Transformer), regression analysis models, Bayesian inference models, etc., to calculate future sales forecasts, purchase intention trends, and probability distributions of competitive share changes. Output examples include "New Product Sales Forecast: 5000 units in the first month," "Upward Trend in Purchase Intent: 0.85," and "Probability of Declining Market Share: 0.7." Based on the forecast results, the analysis department transmits information to the marketing strategy suggestion module or dashboard display module for optimizing advertising budget allocation, product launch timing, and promotional measures. Furthermore, the analysis department can automatically perform accuracy assessments of the forecasting model, outlier detection, and scenario simulations (such as sales forecasts during price changes). These processes differ from previous simple aggregations or manual forecasts, enabling AI-driven dynamic predictive analysis of high-dimensional, multivariate data. Technically, this analysis department significantly improves responsiveness to market changes, forecast accuracy, and strategy development efficiency. Applicable areas include demand forecasting, advertising effectiveness forecasting, competitive analysis, risk assessment, and supply chain optimization.

[0072] The learning department can infer users' emotions and adjust the learning content of market data based on the inferred user emotions. For example, it prioritizes learning positive market data when users are excited. Furthermore, it can learn detailed market data when users are relaxed. Moreover, it can learn simplified market data when users are stressed. Thus, by adjusting the learning content of market data according to users' emotions, the learning department can conduct more appropriate learning. Specifically, this learning department can input multimodal data for user emotion inference, such as facial image tensors (e.g., 224×224×3), speech waveform data (e.g., 16000 dimensions), and text sentence data (e.g., 100 tokens). Input examples include "smiling facial image" and "text 'excited about the new product'". The learning department uses this data to process through CNN or Transformer-type neural networks, outputting emotion labels or emotion scores such as "excitement", "relaxation", and "stress" (e.g., excitement 0.85, relaxation 0.10, stress 0.05) at the emotion classification layer. Output examples include "Emotional label: Excitement, score 0.92" and "Emotional label: Relaxation, score 0.75," etc. Based on the emotion inference results, the learning department assigns priority scores to each data point in the learning dataset. High excitement levels prioritize learning new products or positive market data; high relaxation levels prioritize learning detailed attribute data or auxiliary data; and high stress levels prioritize learning simple or basic data. The learning department dynamically adjusts the batch learning and data sampling ratios according to the priority scores, constructing a learning process best suited to the user's state. These processes differ from previous random sampling or fixed-order learning, achieving dynamic data priority control based on the user's state. Technically, this learning department can achieve optimized learning data selection based on user attention and psychological state, improving learning efficiency and model accuracy. Applicable fields include educational AI, personalized marketing, medical diagnostic support, interactive advertising, and more.

[0073] The generation unit can infer the user's emotions and adjust the behavior patterns of the pseudo-personality based on the inferred emotions. For example, it generates a pseudo-personality with a positive behavior pattern when the user is excited. Furthermore, it generates a pseudo-personality with a mild behavior pattern when the user is relaxed. Moreover, it generates a pseudo-personality with a calm behavior pattern when the user is stressed. Thus, by adjusting the behavior patterns of the pseudo-personality according to the user's emotions, the generation unit can generate more appropriate pseudo-personalities. Specifically, the generation unit can input multimodal data such as facial image tensors (e.g., 224×224×3), speech waveform data (e.g., 16000-dimensional), and text sentence data (e.g., 100 tokens) for user emotion inference. Input examples include "smiling facial images" and "text 'excited about the new product'". The generation unit uses this data to process through a CNN or Transformer-type neural network, outputting emotion labels or scores such as "excitement," "relaxation," and "stress" (e.g., excitement 0.85, relaxation 0.10, stress 0.05) at the emotion classification layer. Output examples include "Emotional label: Excitement, score 0.92" and "Emotional label: Relaxation, score 0.75". Based on the emotion inference results, the generation department assigns behavioral pattern parameters (such as positivity score, cooperativeness score, calmness score, etc.) as generation conditions to the AI, assigning them to latent feature vectors. It then generates pseudo-personalities with behavioral patterns that match the emotional state using generative models such as VAEs or diffusion models. For example, when excitement is high, the positivity score is set to 0.9; when relaxation is high, the cooperativeness score is set to 0.8; and when stress is high, the calmness score is set to 0.95. Output examples include "Pseudo-personality A: positivity 0.9, cooperativeness 0.6, calmness 0.2" and "Pseudo-personality B: positivity 0.3, cooperativeness 0.8, calmness 0.9". This processing differs from previous fixed attribute settings or manual adjustments, achieving dynamic and automatic behavioral pattern adjustments based on the user's state. Technically, this generation department can achieve real-time personality generation that reflects the user's psychological state, significantly improving the personalization and accuracy of response tests and marketing measures. It is applicable to multiple fields, including consumer research, educational AI, patient simulation in the medical field, and interactive character generation in the entertainment field.

[0074] The testing department can infer users' emotions and adjust the reaction test scenarios based on the inferred emotions. For example, a positive scenario can be used for reaction testing when users are excited. Furthermore, a detailed scenario can be used when users are relaxed. Additionally, a simplified scenario can be used when users are stressed. Thus, by adjusting the reaction test scenarios according to users' emotions, the testing department can conduct more appropriate reaction tests. Specifically, the testing department can input multimodal data for user emotion inference, such as facial image tensors (e.g., 224×224×3), speech waveform data (e.g., 16000 dimensions), and text sentence data (e.g., 100 tokens). Input examples include "smiling facial image" and "text 'excited about the new product'". The testing department uses this data to process through a CNN or Transformer-type neural network, outputting emotion labels or scores such as "excitement," "relaxation," and "stress" (e.g., excitement 0.85, relaxation 0.10, stress 0.05) at the emotion classification layer. Output examples include "Emotional label: Excitement, score 0.92" and "Emotional label: Relaxation, score 0.75," etc. Based on the emotion inference results, the testing department dynamically adjusts the test scenario content (e.g., number of questions, question content, scenario branches, required time, etc.) in the scenario generation module. When excitement is high, positive scenarios or simple questions are prioritized; when relaxation is high, detailed scenarios or multivariate questions are provided; and when stress is high, only simple scenarios or multiple-choice questions are provided. Based on the AI's output (e.g., test scenario list, question content, branch conditions, etc.), the testing department passes the test results to the subsequent reaction data collection or analysis department. This processing differs from previous fixed test designs or manual scenario adjustments, achieving dynamic and automatic optimization of test scenarios based on user states. Technically, this testing department can automatically design optimal test scenarios based on user psychological states and attention, significantly improving the reliability and collection efficiency of reaction data. Applicable fields include consumer surveys, adaptive testing of educational AI, patient response assessment in the medical field, and interactive experience design in the entertainment field, among others.

[0075] The analysis unit can infer users' emotions and adjust the reporting method of test results based on the inferred user emotions. For example, when users are excited, a positive report is given; when users are relaxed, a detailed report can be given; and when users feel stressed, a brief report can be given. Thus, the analysis unit can adjust the reporting method of test results according to the user's emotions, thereby achieving more appropriate reporting. Specifically, this analysis unit can receive multimodal data for user emotion inference as input, such as face image tensors (e.g., 224×224×3), speech waveform data (e.g., 16000 dimensions), and text sentence data (e.g., 100 tokens). Input examples include "smiling face image" and "text 'excited about new products'". The analysis unit processes this data using CNN or Transformer-type neural networks, outputting emotion labels such as "excitement", "relaxation", and "stress" and emotion scores (e.g., excitement 0.85, relaxation 0.10, stress 0.05) at the emotion classification layer. Output examples include "Emotional Tag: Excitement, Score 0.92" and "Emotional Tag: Relaxation, Score 0.75". Based on the emotion inference results, the analysis department dynamically adjusts the report content (e.g., summary report, detailed report, positive emphasis report) and report granularity (e.g., statistical units, number of variables, visualization method) in the report generation module. When excitement is high, a report emphasizing positive elements is generated; when relaxation is high, a report containing detailed analysis results is generated; and when stress is high, a brief summary report is generated. The analysis department uses the output results of generative AI (e.g., report PDF, dashboard data) for subsequent decision support and user feedback. This processing differs from previous fixed report generation or manual content adjustment, achieving dynamic and automatic optimization of report methods based on user status. As a technical effect, this analysis department can automatically design optimal reports based on the user's psychological state and focus, thereby significantly improving the acceptability and utilization efficiency of report content. Applicable fields include consumer survey analysis, educational AI performance reports, patient evaluation reports in the medical field, and user experience reports in the entertainment field, among others.

[0076] The analysis department can infer users' emotions and adjust the content of marketing strategy proposals based on the inferred user emotions. For example, when users are excited, a positive marketing strategy can be proposed; when users are relaxed, a detailed marketing strategy can be proposed; and when users are stressed, a simple marketing strategy can be proposed. Thus, the analysis department can adjust the content of marketing strategy proposals according to users' emotions, thereby achieving more appropriate proposals. Specifically, this analysis department can receive multimodal data for user emotion inference as input, such as face image tensors (e.g., 224×224×3), speech waveform data (e.g., 16000 dimensions), and text sentence data (e.g., 100 tokens). Input examples include "smiling face image" and "text 'excited about the new product'". The analysis department processes this data using CNN or Transformer-type neural networks, outputting emotion labels such as "excitement", "relaxation", and "stress" and emotion scores (e.g., excitement 0.85, relaxation 0.10, stress 0.05) at the emotion classification layer. Output examples include "Emotional Tag: Excitement, Score 0.92" and "Emotional Tag: Relaxation, Score 0.75". Based on the emotion inference results, the analysis department dynamically adjusts the proposal content (e.g., proactive strategy, detailed strategy, simplified strategy) and proposal granularity (e.g., number of measures, measure content, execution priority) in the strategy proposal generation module. When excitement is high, proactive strategies such as new product launches or large-scale promotions are proposed; when relaxation is high, detailed target positioning or multi-stage measures are proposed; when stress is high, simplified measures or phased implementation are proposed. Based on the output results of generative AI (e.g., strategy proposal list, measure content, priority, etc.), the analysis department transmits the information to subsequent decision support or execution departments. This processing differs from previous fixed strategy proposals or manual content adjustments, achieving dynamic and automatic optimization of strategy proposals based on user status. As a technical effect, this analysis department can automatically propose optimal strategy proposals based on the user's psychological state and focus, thereby significantly improving the acceptability of proposal content and execution efficiency. It is applicable to a variety of fields, including consumer research and analysis, advertising strategy formulation, product development measures, learning strategy proposals for educational AI, and treatment policy proposals in the medical field.

[0077] The following is a brief description of the implementation process. Specifically, this system implements a pipelined processing mechanism where multiple AI modules (learning department, generation department, testing department, and analysis department) work collaboratively. Each department clearly defines the types, formats, and dimensions of input data, achieving automatic control of the entire data flow while dynamically optimizing the AI ​​model architecture and parameters. For example, the learning department uses LSTM or Transformer-type neural networks to extract features from market data (such as purchase history tensors, competitive dynamic time series, and economic indicator arrays) to generate user feature vectors. The generation department inputs the feature vectors output by the learning department into generative models such as VAEs or diffusion models to generate diverse pseudo-personalities reflecting the target customer profile and behavioral patterns. The testing department uses the pseudo-personalities generated by the generation department to conduct diverse response tests such as A / B testing, situational questionnaires, and VR testing, collecting response data (such as purchase intention scores, reviews, and behavioral logs). The analysis department uses multivariate analysis models (such as clustering, regression analysis, and trend extraction) to process the response data obtained from the testing department and output decision support information (such as target group response distribution, price optimization suggestions, and advertising channel recommendations). Each component can dynamically adjust the hyperparameters of the AI ​​model (such as learning rate, batch size, loss function weights, etc.) and data sampling ratio, rapidly responding to user sentiment inferences and market trend changes. Furthermore, by combining extended modules such as feedback collection, real-time data collection, multi-language support, VR testing, and predictive analytics, the overall accuracy, adaptability, and operational efficiency of the system can be significantly improved. These processes differ from previous manual static design or single-model applications; through the collaboration of multiple AI models, dynamic control, and high-dimensional data analysis, the system achieves a high level of sophistication in computer technology itself. As a technical result, this system significantly improves responsiveness to market changes, model accuracy, data processing efficiency, and user experience quality. Applicable fields include a wide range of areas such as consumer research, advertising optimization, financial risk assessment, medical diagnostic support, educational AI, and interactive experience design in the entertainment industry.

[0078] Step 1: The Learning Department studies market-related data. This data includes consumer purchase history, competitor activities, and economic indicators. The Learning Department uses machine learning or deep learning algorithms to learn market data and, based on a large amount of market data, learns the characteristics of individuals existing in the market. Step 2: The Generation Department generates pseudo-personalities based on the characteristics learned by the Learning Department. These pseudo-personalities are generated based on the target customer profile and behavioral patterns, using generative AI to create pseudo-personalities with characteristics highly similar to those existing in the actual market. Step 3: The Testing Department conducts reaction tests using the pseudo-personalities generated by the Generation Department. Reaction tests include questionnaires and A / B testing, using generative AI to observe the pseudo-personalities' reactions and collect data. Step 4: The Analysis Department analyzes the test results obtained by the Testing Department. Analysis includes statistical analysis and data mining, using generative AI to provide information for developing marketing strategies based on the test results. Specifically, in step 1, the learning unit uses LSTM or Transformer-type neural networks to extract time-series features from consumer purchase history tensors (e.g., purchase frequency of 10,000 people × 12 months), competitive dynamic time-series data (e.g., sales volume and price changes of 50 products × 24 months), and economic indicator arrays (e.g., 12-month GDP growth rate, unemployment rate, etc.), generating user feature vectors (e.g., a 100-dimensional vector of age, gender, purchasing tendency, price sensitivity, etc.). In step 2, the generation unit inputs the feature vectors output by the learning unit into generative models such as VAE or diffusion models to generate diverse pseudo-personalities reflecting the target customer profile and behavioral patterns (e.g., new product tendency 0.8, price sensitivity 0.7, health tendency 0.6, etc.). In step 3, the testing unit uses the pseudo-personals generated by the generation unit to conduct diverse response tests such as A / B testing, situational questionnaires, and VR testing, collecting response data (e.g., purchase intention scores, reviews, behavioral logs, etc.). The testing unit dynamically adjusts test questions and scenarios based on user sentiment inferences and market segmentation characteristics to improve the reliability and diversity of response data. In step 4, the analysis department uses multivariate analysis models (such as clustering, regression analysis, and trend extraction) to process the response data obtained from the testing department, outputting decision support information (such as target group response distribution, price optimization suggestions, and advertising channel recommendations). The analysis department dynamically optimizes analysis methods and report content based on user sentiment inferences and market trend changes, improving the accuracy and responsiveness of marketing strategy formulation. Furthermore, by combining extended modules such as the feedback collection department, real-time data collection function, multi-language support function, VR testing function, and predictive analysis function, the overall accuracy, adaptability, and operational efficiency of the system can be significantly improved. These processes differ from previous manual static design or single-model application; through the collaboration of multiple AI models, dynamic control, and high-dimensional data analysis, the system achieves a high level of computer technology itself. As a technical effect, this system can significantly improve responsiveness to market changes, model accuracy, data processing efficiency, and user experience quality.It is applicable to a wide range of fields, including consumer research, advertising optimization, financial risk assessment, medical diagnostic support, educational AI, and interactive experience design in the entertainment industry.

[0079] The specific processing unit 290 sends 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 voice representing the user's input to the result of the specific processing. The control unit 46A sends the voice data representing the user's 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 voice data.

[0080] Data generation model 58 is what is known as generative AI (Artificial Intelligence). An example of data generation model 58 includes ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> Generative AI, such as data generation model 58, is obtained by deep learning through a neural network. The data generation model 58 is input with a prompt containing instructions, and with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the above-described specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts without instructions; in this case, the data generation model 58 can output inference results from prompts without instructions. The data processing device 12, etc., includes multiple data generation models 58, including AI other than generative AI. AI other than generative AI includes, but is not limited to, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes. Furthermore, AI can also act as an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to this example. Moreover, processing performed by AI, including generative AI, can be replaced by rule-based processing, and vice versa.

[0081] Furthermore, the processing performed by the aforementioned data processing system 10 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart device 14, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart device 14. Additionally, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart device 14 or external devices, and the smart device 14 acquires or collects information required for processing from the data processing device 12 or external devices.

[0082] Each of the aforementioned elements, such as the learning unit, generation unit, testing unit, and analysis unit, can be implemented, for example, in at least one of the smart device 14 and the data processing device 12. For instance, the learning unit, implemented by a specific processing unit 290 of the data processing device 12, is used to learn market-related data. The generation unit, for example, implemented by the specific processing unit 290 of the data processing device 12, generates pseudo-personalities (pseudo-user profiles) based on features learned by generative AI. The testing unit, for example, implemented by the control unit 46A of the smart device 14, uses the pseudo-personalities to conduct reaction tests. The analysis unit, for example, implemented by the specific processing unit 290 of the data processing device 12, analyzes the test results and provides information for developing marketing strategies. The correspondence between the various units and the device or control unit is not limited to the above examples and can be varied.

[0083] [Second Implementation] Figure 3 An example of the configuration of the data processing system 210 according to the second embodiment is shown.

[0084] like Figure 3 As shown, the data processing system 210 includes a data processing device 12 and smart glasses 214. One example of the data processing device 12 is a server.

[0085] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, 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. An example of the network 54 includes a WAN and / or LAN, etc.

[0086] 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, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0087] Microphone 238 receives user commands by receiving the user's voice. Microphone 238 captures the user's voice, converts the captured sound into speech data, and outputs it to processor 46. Speaker 240 outputs sound according to commands from processor 46.

[0088] Camera 42 is a small digital camera equipped with an optical system such as 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, used to capture the user's surroundings (e.g., the shooting range defined by an angle of view equivalent to the field of vision of an average healthy person).

[0089] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F 44 and 26 is performed in a secure state.

[0090] Figure 4 An example of the main functions of the data processing device 12 and the smart glasses 214 is shown. Figure 4 As shown, in the data processing device 12, specific processing is performed by the processor 28. The memory 32 stores a specific processing program 56.

[0091] The processor 28 reads a specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0092] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0093] In the smart glasses 214, specific processing is performed by the processor 46. A specific processing program 60 is stored in the memory 50. The processor 46 reads the specific processing program 60 from the memory 50 and executes the read specific processing program 60 on the RAM 48. Specific processing is implemented by the processor 46 operating as a control unit 46A based on the specific processing program 60 executed on the RAM 48. Furthermore, the smart glasses 214 may also have the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and use these models to perform the same processing as the specific processing unit 290.

[0094] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.).

[0095] The specific processing unit 290 sends 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 voice input representing the user's input to the specific processing result. The control unit 46A sends the voice data representing the user's 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.

[0096] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 includes generative AIs such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with a prompt containing instructions, and also with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the aforementioned specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts that do not contain instructions; in this case, the data generation model 58 is able to output inference results from prompts that do not contain instructions. The data processing device 12, etc., includes various data generation models 58, which include AI other than generative AI. These AIs include, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and are capable of various processing methods, but are not limited to these examples. Furthermore, AI can also be an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to these examples. Moreover, processing performed by AI including generative AI can be replaced by rule-based processing, and vice versa.

[0097] The data processing system 210 of the second embodiment performs the same processing as the data processing system 10 of the first embodiment. The processing performed by the data processing system 210 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart glasses 214, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart glasses 214. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart glasses 214 or external devices, and the smart glasses 214 acquires or collects information required for processing from the data processing device 12 or external devices.

[0098] Each of the aforementioned elements, such as the learning unit, generation unit, testing unit, and analysis unit, can be implemented, for example, in at least one of the smart glasses 214 and the data processing device 12. For instance, the learning unit, implemented by a specific processing unit 290 of the data processing device 12, is used to learn market-related data. The generation unit, for example, implemented by the specific processing unit 290 of the data processing device 12, generates pseudo-personalities (pseudo-user profiles) based on features learned by generative AI. The testing unit, for example, implemented by the control unit 46A of the smart glasses 214, uses the pseudo-personalities to conduct reaction tests. The analysis unit, for example, implemented by the specific processing unit 290 of the data processing device 12, analyzes the test results and provides information for developing marketing strategies. The correspondence between the various units and the device or control unit is not limited to the above examples and can be varied.

[0099] [Third Implementation] Figure 5 An example of the configuration of the data processing system 310 according to the third embodiment is shown.

[0100] like Figure 5 As shown, the data processing system 310 includes a data processing device 12 and a head-mounted terminal 314. One example of the data processing device 12 is a server.

[0101] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, 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. An example of the network 54 includes a WAN and / or LAN, etc.

[0102] The head-mounted 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, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0103] Microphone 238 receives user commands by receiving the user's voice. Microphone 238 captures the user's voice, converts the captured sound into speech data, and outputs it to processor 46. Speaker 240 outputs sound according to commands from processor 46.

[0104] Camera 42 is a small digital camera equipped with an optical system such as 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, used to capture the user's surroundings (e.g., the shooting range defined by an angle of view equivalent to the field of vision of an average healthy person).

[0105] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F 44 and 26 is performed in a secure state.

[0106] Figure 6 An example of the main functions of the data processing device 12 and the head-mounted terminal 314 is shown. Figure 6 As shown, in the data processing device 12, specific processing is performed by the processor 28. The memory 32 stores a specific processing program 56.

[0107] The processor 28 reads a specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0108] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0109] In the head-mounted terminal 314, specific processing is performed by the processor 46. A specific program 60 is stored in the memory 50. The processor 46 reads the specific program 60 from the memory 50 and executes the read specific program 60 on the RAM 48. Specific processing is implemented by the processor 46 operating as a control unit 46A based on the specific program 60 executed on the RAM 48. Furthermore, the head-mounted terminal 314 may also have the same data generation model and emotion-specific model as the data generation model 58 and the emotion-specific model 59, and use these models to perform the same processing as the specific processing unit 290.

[0110] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.).

[0111] The specific processing unit 290 sends the result of the specific processing to the head-mounted terminal 314. In the head-mounted terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires voice representing the user's input to the specific processing result. The control unit 46A sends the voice data representing the user's 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.

[0112] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 includes generative AIs such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with a prompt containing instructions, and also with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the aforementioned specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts that do not contain instructions; in this case, the data generation model 58 is able to output inference results from prompts that do not contain instructions. The data processing device 12, etc., includes various data generation models 58, which include AI other than generative AI. These AIs include, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and are capable of various processing methods, but are not limited to these examples. Furthermore, AI can also be an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to these examples. Moreover, processing performed by AI including generative AI can be replaced by rule-based processing, and vice versa.

[0113] The data processing system 310 of the third embodiment performs the same processing as the data processing system 10 of the first embodiment. The processing performed by the data processing system 310 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the head-mounted terminal 314, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the head-mounted terminal 314. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the head-mounted terminal 314 or external devices, and the head-mounted terminal 314 acquires or collects information required for processing from the data processing device 12 or external devices.

[0114] Each of the aforementioned elements, such as the learning unit, generation unit, testing unit, and analysis unit, can be implemented, for example, in at least one of the head-mounted terminal 314 and the data processing device 12. For instance, the learning unit, implemented by a specific processing unit 290 of the data processing device 12, is used to learn market-related data. The generation unit, for example, implemented by the specific processing unit 290 of the data processing device 12, generates pseudo-personalities (pseudo-user profiles) based on features learned by generative AI. The testing unit, for example, implemented by the control unit 46A of the head-mounted terminal 314, uses the pseudo-personalities to conduct reaction tests. The analysis unit, for example, implemented by the specific processing unit 290 of the data processing device 12, analyzes the test results and provides information for developing marketing strategies. The correspondence between the various units and the device or control unit is not limited to the above examples and can be varied.

[0115] [Fourth Implementation] Figure 7 An example of the configuration of the data processing system 410 according to the fourth embodiment is shown.

[0116] like Figure 7 As shown, the 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.

[0117] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, RAM 30, and memory 32. The processor 28, RAM 30, and memory 32 are connected to a bus 34. Furthermore, 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. An example of the network 54 includes a WAN and / or LAN, etc.

[0118] Robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control object 443. The computer 36 includes a processor 46, RAM 48, and memory 50. The processor 46, RAM 48, and memory 50 are connected to a bus 52. Furthermore, the microphone 238, speaker 240, camera 42, and control object 443 are also connected to the bus 52.

[0119] Microphone 238 receives user commands by receiving the user's voice. Microphone 238 captures the user's voice, converts the captured sound into speech data, and outputs it to processor 46. Speaker 240 outputs sound according to commands from processor 46.

[0120] The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture and shutter, and an imaging element such as a CMOS image sensor or a CCD image sensor, used to photograph the user's surroundings (e.g., the shooting range defined by an angle of view equivalent to the field of vision of an average healthy person).

[0121] Communication I / F 44 is connected to network 54. Communication I / F 44 and 26 are responsible for the transmission and reception of various information between processor 46 and processor 28 via network 54. The transmission and reception of various information between processor 46 and processor 28 using communication I / F 44 and 26 is performed in a secure state.

[0122] The controlled object 443 includes a display device, LEDs for the eyes, and motors for driving the arms, hands, and feet. The posture and movements of the robot 414 are controlled by controlling the motors for the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. In addition, facial expressions of the robot 414 can also be expressed by controlling the illumination state of the LEDs for the robot 414's eyes.

[0123] Figure 8 An example of the main functions of the data processing device 12 and the robot 414 is shown. For example... Figure 8 As shown, in the data processing device 12, specific processing is performed by the processor 28. The memory 32 stores a specific processing program 56.

[0124] The processor 28 reads a specific processing program 56 from the memory 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is implemented by the processor 28 operating as a specific processing unit 290 based on the specific processing program 56 executed on the RAM 30.

[0125] The memory 32 stores a data generation model 58 and an emotion-specific model 59. The data generation model 58 and the emotion-specific model 59 are used by the specific processing unit 290. The specific processing unit 290 can use the emotion-specific model 59 to infer the user's emotions and perform specific processing based on the user's emotions. The emotion inference function using the emotion-specific model 59 (emotion-specific function) includes various inferences and predictions related to the user's emotions, but is not limited to this example. Furthermore, emotion inference and prediction also include, for example, emotion analysis (analysis).

[0126] In robot 414, specific processing is performed by processor 46. A specific program 60 is stored in memory 50. Processor 46 reads the specific program 60 from memory 50 and executes the read specific program 60 on RAM 48. Specific processing is achieved by processor 46 acting as control unit 46A based on the specific program 60 executed on RAM 48. Furthermore, robot 414 may also have the same data generation model and emotion-specific model as data generation model 58 and emotion-specific model 59, and use these models to perform the same processing as specific processing unit 290.

[0127] Furthermore, other devices besides the data processing device 12 may also have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 obtains the processing results (prediction results, etc.) using the data generation model 58 by communicating with the server device that has the data generation model 58. In addition, the data processing device 12 may be a server device or a user-owned terminal device (e.g., a mobile phone, robot, home appliance, etc.).

[0128] The specific processing unit 290 sends the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires voice representing the user's input regarding the result of the specific processing. The control unit 46A sends the voice data representing 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.

[0129] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 includes generative AIs such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with a prompt containing instructions, and also with inference data such as speech data representing speech, text data representing text, and image data representing images (e.g., data of still images or data of moving images). The data generation model 58 infers the inference data based on the instructions shown in the prompt and outputs the inference result in one or more data forms such as speech data, text data, and image data. The data generation model 58 includes, for example, text generation AI, image generation AI, and multimodal generation AI. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specific processing unit 290 performs the aforementioned specific processing while using the data generation model 58. The data generation model 58 can also be a fine-tuned model to output inference results from prompts that do not contain instructions; in this case, the data generation model 58 is able to output inference results from prompts that do not contain instructions. The data processing device 12, etc., includes various data generation models 58, which include AI other than generative AI. These AIs include, for example, linear regression, logistic regression, decision trees, random forests, support vector machines (SVM), k-means clustering, convolutional neural networks (CNN), recurrent neural networks (RNN), generative adversarial networks (GAN), or Naive Bayes, and are capable of various processing methods, but are not limited to these examples. Furthermore, AI can also be an AI agent. Additionally, when the processing described above is performed by AI, this processing may be partially or entirely performed by AI, but is not limited to these examples. Moreover, processing performed by AI including generative AI can be replaced by rule-based processing, and vice versa.

[0130] The data processing system 410 of the fourth embodiment performs the same processing as the data processing system 10 of the first embodiment. The processing performed by the data processing system 410 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the robot 414, but it can also be executed jointly by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the robot 414. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the robot 414 or external devices, and the robot 414 acquires or collects information required for processing from the data processing device 12 or external devices.

[0131] Each of the aforementioned elements, such as the learning unit, generation unit, testing unit, and analysis unit, can be implemented, for example, in at least one of the robot 414 and the data processing device 12. For instance, the learning unit, implemented by a specific processing unit 290 of the data processing device 12, is used to learn market-related data. The generation unit, for example, implemented by the specific processing unit 290 of the data processing device 12, generates pseudo-personalities (pseudo-user profiles) based on features learned by generative AI. The testing unit, for example, implemented by the control unit 46A of the robot 414, uses the pseudo-personalities to conduct reaction tests. The analysis unit, for example, implemented by the specific processing unit 290 of the data processing device 12, analyzes the test results and provides information for developing marketing strategies. The correspondence between the various units and the device or control unit is not limited to the above examples and can be varied.

[0132] Furthermore, the emotion-specific model 59, serving as an emotion engine, can determine the user's emotion based on a specific mapping. Specifically, the emotion-specific model 59 can determine the user's emotion based on an emotion graph that serves as a specific mapping (see...). Figure 9 The robot's emotions can be determined by the emotion-specific model 59. In addition, the emotion-specific model 59 can also determine the robot's emotions in the same way, and the specific processing unit 290 can also perform specific processing using the robot's emotions.

[0133] Figure 9 This is a diagram representing an emotion map 400 that maps various emotions. In the emotion map 400, emotions are arranged radially from the center in concentric circles. The closer to the center of the concentric circles, the more primitive the emotion is. Further out on the concentric circles, emotions are arranged representing states or actions arising from mood. Emotion is a concept that includes both feelings and mental states. To the left of the concentric circles, emotions generated by reactions occurring in the brain are arranged roughly. To the right of the concentric circles, emotions guided by situational judgments are arranged roughly. Above and below the concentric circles, emotions generated by reactions occurring in the brain and guided by situational judgments are arranged roughly. Furthermore, the emotion of "pleasure" is arranged above the concentric circles, and the emotion of "unpleasantness" is arranged below. Thus, in the emotion map 400, various emotions are mapped according to the structure of emotion generation, while easily generated emotions are mapped nearby.

[0134] These emotions are distributed at the 3 o'clock position on the Emotion Chart 400, and usually fluctuate between peace and anxiety. In the right half of the Emotion Chart 400, because situational awareness is more dominant than internal feelings, it gives a sense of calm.

[0135] The inner side of the emotion diagram 400 represents the mind, and the outer side of the emotion diagram 400 represents actions. Therefore, the further you go to the outer side of the emotion diagram 400, the more the emotion can be seen (manifested in actions).

[0136] Here, human emotions are based on a balance of various factors such as posture and blood sugar levels. When these balances deviate from the ideal, it indicates unhappiness; when they approach the ideal, it indicates pleasure. In robots, cars, and motorcycles, emotions can also be created based on a balance of factors such as posture and remaining battery power. When these balances deviate from the ideal, it indicates unhappiness; when they approach the ideal, it indicates pleasure. Emotion maps can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Speech Emotion Recognition and Brain Physiological Signal Analysis Systems for Emotion, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). In the left half of the emotion map, emotions belonging to the "Reaction" domain, where sensation is dominant, are arranged. Furthermore, in the right half of the emotion map, emotions belonging to the "Situation" domain, where situational cognition is dominant, are arranged.

[0137] The emotion map defines two types of emotions that promote learning. One is a negative emotion located near the middle of "repentance" or "reflection" on the situation side. That is, when the robot experiences negative emotions such as "I never want to feel this way again" or "I never want to be scolded again." The other is a positive emotion located near "desire" on the response side. That is, when the robot experiences positive feelings such as "wanting more" or "wanting to know more."

[0138] The emotion-specific model 59 feeds user input into a pre-learned neural network to obtain emotion values ​​representing each emotion shown in the emotion graph 400, and determines the user's emotion. This neural network is pre-learned based on multiple learning data sets that combine user input with emotion values ​​representing each emotion shown in the emotion graph 400. Furthermore, this neural network is learned to... Figure 10 As shown in sentiment graph 900, sentiment values ​​in nearby configurations are similar to each other. Figure 10 Examples show that multiple emotions such as "peace of mind", "stability", and "reassurance" have similar emotional values.

[0139] In the above embodiments, a specific processing is described by a single computer 22, but the technology disclosed herein is not limited to this, and distributed processing by multiple computers, including computer 22, is also possible.

[0140] In the above embodiments, an example of storing a specific processing program 56 in memory 32 is illustrated, but the technology disclosed herein is not limited thereto. For example, the specific processing program 56 may also 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 performs specific processing according to the specific processing program 56.

[0141] Alternatively, the specific processing program 56 can be stored in a storage device such as a server connected to the data processing device 12 via a network 54, and the specific processing program 56 can be downloaded and installed into the computer 22 upon request from the data processing device 12.

[0142] Furthermore, it is not necessary to store the entire specific process 56 in a storage device such as a server connected to the data processing device 12 via the network 54, nor is it necessary to store the entire specific process 56 in the memory 32; a portion of the specific process 56 may also be stored.

[0143] As a hardware resource for performing specific processing, various processors can be used. For example, a CPU is a general-purpose processor that functions as a hardware resource for performing specific processing by executing software, i.e., programs. Additionally, a dedicated circuit can be listed as a processor; it is a processor with a circuit structure specifically designed for performing specific processing, such as a FPGA (Field-Programmable Gate Array), PLD (Programmable Logic Device), or ASIC (Application Specific Integrated Circuit). Every processor has built-in or connected memory, and every processor executes specific processing by using that memory.

[0144] The hardware resources for performing a specific process can consist of one of these various processors, or a combination of two or more processors of the same or different types (e.g., a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Furthermore, the hardware resources for performing a specific process can also be a single processor.

[0145] As an example of a single processor, the first type consists of a combination of one or more CPUs and software, which functions as a hardware resource to perform specific processing. The second type uses a processor, such as a System-on-a-chip (SoC), which implements the entire system functionality, including multiple hardware resources performing specific processing, using a single IC chip. In this case, the specific processing is implemented using one or more of the aforementioned processors that serve as hardware resources.

[0146] Furthermore, as the hardware architecture of these various processors, more specifically, circuits combining semiconductor elements and other circuit components can be used. Moreover, the specific process described above is merely an example. Therefore, it goes without saying that, without departing from the main point, unnecessary steps can be removed, new steps can be added, or the processing order can be changed.

[0147] Furthermore, although the above examples have been described in terms of first to fourth embodiments, some or all of these embodiments can also be combined. Additionally, the smart device 14, smart glasses 214, head-mounted terminal 314, and robot 414 are just examples; they can be combined separately or are other devices.

[0148] The foregoing descriptions and illustrations are detailed explanations of the parts covered by this disclosure and are merely one example of this disclosure. For instance, the descriptions of the above-described structure, function, role, and effect are just one example of the structure, function, role, and effect of the parts covered by this disclosure. Therefore, it goes without saying that, without departing from the spirit of this disclosure, unnecessary parts can be deleted, new elements can be added, or replacements can be made to the foregoing descriptions and illustrations. Furthermore, to avoid confusion and facilitate understanding of the parts covered by this disclosure, explanations of technical common sense that does not require special explanation for implementing this disclosure have been omitted from the foregoing descriptions and illustrations.

[0149] All documents, patent applications and technical standards described in this specification are incorporated herein by reference as if they were specifically and individually described as incorporated by reference.

[0150] (Note 1) A system comprising: The Learning Department is responsible for learning market-related data. The generation department generates pseudo-personalities (pseudo-user profiles) based on the features learned by the learning department. The testing department uses the pseudo-personality generated by the generation department to conduct reaction tests; The analysis unit analyzes the test results obtained by the testing unit.

[0151] (Note 2) The system as described in Appendix 1 is characterized in that, The learning unit takes in market-related data, and the generative AI learns the characteristics of people existing in the market based on this data.

[0152] (Note 3) The system as described in Appendix 1 is characterized in that, The generation department generates pseudo-personalities based on features learned by generative AI.

[0153] (Note 4) The system as described in Appendix 1 is characterized in that, The testing department uses pseudo-personalities to conduct reaction tests.

[0154] (Note 5) The system as described in Appendix 1 is characterized in that, The analysis unit analyzes the test results and provides information for developing marketing strategies.

[0155] (Note 6) The learning unit is a system used to infer user sentiment and adjust the timing of market data learning based on the inferred user sentiment.

[0156] (Note 7) The learning unit is a system that adjusts its learning algorithm by referring to past market trends when learning market data.

[0157] (Postscript 8) The system as described in Appendix 1 is characterized in that, When studying market data, the learning department focuses on specific market segments.

[0158] (Note 9) The learning unit is a system used to infer the user's emotions and determine the priority of learning data based on the inferred user emotions.

[0159] (Postscript 10) The system as described in Appendix 1 is characterized in that, When learning market data, the learning department takes into account the characteristics of geographical markets.

[0160] (Postscript 11) The system as described in Appendix 1 is characterized in that, The learning department analyzes social media trends and incorporates them into its learning process when studying market data.

[0161] (Postscript 12) The generation unit is a system used to infer the user's emotions and adjust the characteristics of the pseudo-personality based on the inferred user emotions.

[0162] (Postscript 13) The generation department adjusts the characteristics of the pseudo-personality by referring to past market data when generating it.

[0163] (Postscript 14) The system as described in Appendix 1 is characterized in that, When generating pseudo-personalities, the generation department focuses on specific market segmentation features.

[0164] (Postscript 15) The generation unit is a system used to infer the user's emotions and determine the priority of pseudo-personalities based on the inferred user emotions.

[0165] (Postscript 16) The system as described in Appendix 1 is characterized in that, When generating pseudo-personalities, the generation department considers the characteristics of the geographic market when setting features.

[0166] (Postscript 17) The system as described in Appendix 1 is characterized in that, When generating pseudo-personalities, the generation department analyzes social media trends and reflects them in the characteristics.

[0167] (Postscript 18) The testing unit is a system used to infer the user's emotions and adjust the reaction testing method according to the inferred user emotions.

[0168] (Postscript 19) The testing unit is a system that adjusts the testing algorithm by referring to past test data when conducting reaction tests.

[0169] (Postscript 20) The system as described in Appendix 1 is characterized in that, When conducting reaction tests, the testing department focuses on specific market segments.

[0170] (Postscript 21) The testing unit is a system used to infer the user's emotions and determine the priority of response tests based on the inferred user emotions.

[0171] (Postscript 22) The system as described in Appendix 1 is characterized in that, The testing department takes into account the characteristics of the geographic market when conducting reaction tests.

[0172] (Postscript 23) The system as described in Appendix 1 is characterized in that, The testing unit analyzes social media trends and reflects them in the testing process when conducting reaction tests.

[0173] (Postscript 24) The analysis unit is a system for inferring user emotions and adjusting the analysis method of test results based on the inferred user emotions.

[0174] (Postscript 25) The analysis unit is a system that adjusts the analysis algorithm by referring to past analysis data when analyzing test results.

[0175] (Postscript 26) The system as described in Appendix 1 is characterized in that, When analyzing test results, the analysis department focuses on specific market segments.

[0176] (Postscript 27) The analysis unit is used to infer the user's emotions and determine the priority of test results based on the inferred user emotions.

[0177] (Postscript 28) The system as described in Appendix 1 is characterized in that, The analysis department takes into account the characteristics of the geographic market when analyzing the test results.

[0178] (Postscript 29) The system as described in Appendix 1 is characterized in that, When analyzing test results, the analysis unit analyzes social media trends and reflects them in the results.

Claims

1. A system comprising: The Learning Department is responsible for learning market-related data. The generation department generates pseudo-personalities (pseudo-user profiles) based on the features learned by the learning department. The testing department uses the pseudo-personality generated by the generation department to conduct reaction tests; The analysis unit analyzes the test results obtained by the testing unit.

2. The system as described in claim 1, characterized in that, The learning unit takes in market-related data, and the generative AI learns the characteristics of people existing in the market based on this data.

3. The system as described in claim 1, characterized in that, The generation department generates pseudo-personalities based on features learned by generative AI.

4. The system as described in claim 1, characterized in that, The testing department uses pseudo-personalities to conduct reaction tests.

5. The system as described in claim 1, characterized in that, The analysis unit analyzes the test results and provides information for developing marketing strategies.

6. The system as described in claim 1, characterized in that, When studying market data, the learning department focuses on specific market segments.

Images