system

Abstract

We provide the system. [Solution] A means of collecting past survey information, A means for cleaning, classifying, and tagging the aforementioned survey information, A means for training a machine learning model using the cleaned and tagged survey information, A means of searching for survey information related to new ideas entered by users, A means for generating simulation results based on a user's new idea using the aforementioned machine learning model, Means for displaying the generated simulation results on a user terminal, A system that includes this.

Description

Technical Field

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

Background Art

[0002] Patent Document 1 discloses a persona chatbot control method performed by at least one processor, including steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Understanding the true needs of employees and customers and proposing new services and events based on them is an important issue for companies. However, it is not easy to appropriately analyze past questionnaire results and provide specific advice and proposals based on them. In particular, since there is no system that can efficiently process a large amount of questionnaire data and generate highly reliable simulation results, effective means for solving this problem are required.

Means for Solving the Problems

[0005] It should be noted that there seems to be an error in the original text where "[[ID=We]]" is present. It should probably be a misplacement or an incorrect tag. This has been left as is in the translation for the sake of maintaining consistency with the original.The present invention relates to a system for collecting past survey information, preprocessing it, and training a machine learning model based on that information. Specifically, the system includes (1) means for collecting past survey information, (2) means for cleaning, classifying, and tagging the survey information, (3) means for training a machine learning model using the cleaned and tagged survey information, (4) means for searching for survey information related to new ideas entered by the user, (5) means for generating simulation results based on the user's new ideas using the machine learning model, and (6) means for displaying the generated simulation results on a user terminal. By providing such a system, it is possible to efficiently propose services and events that are suitable for the needs of employees and customers.

[0006] "Survey information" refers to data on opinions and feedback collected from employees and customers.

[0007] "Means of collection" refers to the methods and processes for obtaining survey information from an organization's database or other sources.

[0008] "Cleaning" is a process that removes invalid responses and missing data from raw data to improve data consistency and reliability.

[0009] "Classification" is the process of organizing data according to themes or categories.

[0010] "Tagging" is a technique that assigns specific labels or keywords to each piece of data to facilitate searching and analysis.

[0011] A "machine learning model" is a type of algorithm that learns from data and recognizes patterns to make predictions about the future.

[0012] "Training" is the process of using a large amount of data to train a machine learning model so that it can make appropriate predictions and decisions.

[0013] A "user" refers to an individual or group that uses the system to input new ideas or obtain simulation results.

[0014] "Simulation results" refer to predictions and feedback on a user's new idea, generated using a machine learning model.

[0015] A "user terminal" is a device that a user uses to access the system and check the results. [Brief explanation of the drawing]

[0016] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11]It is a sequence diagram showing the processing flow of the data processing system in Embodiment 1. [Figure 12] It is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13] It is a sequence diagram showing the processing flow of the data processing system in Embodiment 2 when combined with an emotion engine. [Figure 14] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 when combined with an emotion engine.

Mode for Carrying Out the Invention

[0017] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0018] First, the terms used in the following description will be explained.

[0019] In the following embodiments, a labeled processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), and the like.

[0020] In the following embodiments, a labeled RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0021] In the following embodiments, the signed storage is one or more non-volatile storage devices that store various programs and various parameters. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes.

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

[0023] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0024] [First Embodiment]

[0025] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0026] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0027] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0028] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0029] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0030] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0031] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0032] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0033] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0034] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0035] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0036] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0037] This invention relates to a system that uses past survey data to clarify the true needs of employees and customers, and based on that, proposes new services and events. Specific embodiments of this system are described below.

[0038] 1. Data Collection

[0039] The server first collects past survey information. This includes data collection from the human resources database, general affairs database, and monitor database. The server connects to these databases through an appropriate authentication process and retrieves the necessary survey information.

[0040] 2. Data preprocessing

[0041] The server then preprocesses the collected survey information. Specifically, it performs processes such as cleaning, classification, and tagging. Cleaning involves removing invalid responses and missing data. Classification organizes the data by theme based on the survey content, dividing it into categories such as "employee benefits," "company events," and "fitness apps." Tagging involves assigning keywords related to each data item to make it easier to search.

[0042] 3. Training machine learning models

[0043] Using the pre-processed survey data, the server trains a machine learning model, specifically a natural language processing model (such as ChatGPT®). This model learns patterns from past survey data and is adjusted to generate appropriate responses to future inputs.

[0044] 4. User input of new ideas

[0045] Users use the system to input ideas for new services or events. They access the system from their terminals and provide ideas in text format. This input data is sent to the server.

[0046] 5. Searching for related data

[0047] The server searches its database for past survey information related to the new idea entered by the user. This allows it to collect similar feedback and opinions and use them for analysis.

[0048] 6. Simulation and Result Generation

[0049] The server uses a trained machine learning model to generate simulation results based on the user's new ideas. For example, if an idea for a specific fitness app is entered, the server analyzes past feedback related to this idea and generates specific advice and suggestions.

[0050] 7. Presentation of Results

[0051] The server constructs the generated simulation results and displays them on the user's terminal. This allows users to easily obtain specific suggestions and advice.

[0052] Specific example

[0053] For example, if a user submits a request to "plan a new sports event," the server searches and analyzes past survey data on sports events. Suppose the results show that "many employees want weekend events, and soccer and basketball are popular." Based on this, the system generates suggestions such as "hold an in-house soccer tournament on Saturday mornings, making it an event that families can also participate in," and displays them on the user's terminal.

[0054] Through these specific means, the system of the present invention can effectively utilize past survey data to propose new services and events based on the needs of employees and customers.

[0055] The following describes the processing flow.

[0056] Step 1: Data Collection

[0057] The server connects to the HR database, general affairs database, and monitor database to collect past survey information. It authenticates with the target databases and establishes secure access. It then executes necessary SQL queries and API calls to retrieve the survey data.

[0058] Step 2: Data Preprocessing

[0059] The server cleans the survey data it has collected. Specifically, it filters out invalid responses and duplicate data, and handles missing values.

[0060] The server sorts the cleaned data by theme. Based on the survey responses, it categorizes the data into categories such as "employee benefits," "company events," and "fitness apps."

[0061] The server tags the classified data and assigns relevant keywords to facilitate searching and analysis. This enables quick searching of data related to specific themes.

[0062] Step 3: Training the machine learning model

[0063] The server uses pre-processed survey data to train a machine learning model (e.g., ChatGPT). This process involves inputting a large amount of data to allow the model to learn patterns.

[0064] The server evaluates the model's performance and fine-tunes the parameters as needed. This ensures that the final model has the ability to generate the optimal response to the user's new ideas.

[0065] Step 4: User input of new ideas

[0066] Users input ideas for new services or events into the system. They provide specific concepts and desired conditions in text format from their user terminals. This input data is then sent to the server.

[0067] Step 5: Search for related data

[0068] The server searches the database for past survey information related to the user's new idea. It extracts highly relevant data using pre-tagged keywords and category information. This allows for efficient collection of past feedback and opinions.

[0069] Step 6: Simulation and result generation

[0070] The server uses a trained machine learning model to generate simulation results based on the user's new idea. The model uses patterns learned from past data to generate predictions and suggestions for the input idea.

[0071] For example, when a new fitness app idea is presented, specific feedback such as, "Many users would like a plan that allows them to exercise effectively in a short amount of time," can be provided.

[0072] Step 7: Presentation of Results

[0073] The server compiles the generated simulation results and displays them on the user's terminal. This allows the user to receive specific suggestions and feedback.

[0074] For example, it might display specific advice such as, "A plan that combines a 10-minute high-intensity interval training session with a game element where users earn points to advance their progress level would likely be well-received."

[0075] This series of steps enables the server to effectively utilize past survey data and create a system that generates concrete and effective suggestions for users' new ideas.

[0076] (Example 1)

[0077] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0078] Traditional systems have made it difficult to effectively utilize past survey data to propose new services and events based on employee and customer needs. Furthermore, there is a lack of means to quickly search and analyze past feedback related to specific new ideas, making it difficult to provide concrete advice to users' new ideas. Solving these problems is essential.

[0079] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0080] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, and means for displaying the generated simulation results on the user's terminal. This makes it possible to quickly provide specific suggestions and advice for the user's new ideas by utilizing past survey information.

[0081] "Past survey information" refers to data based on surveys previously conducted within the organization, including feedback and opinions collected from employees and customers.

[0082] "Means of collection" refers to a combination of hardware and software used to retrieve necessary information from databases or APIs.

[0083] "Cleaning" is a process that improves data quality by removing invalid responses and missing data from the data.

[0084] "Classification" is the process of organizing survey data by theme or category and grouping it according to a specific purpose.

[0085] "Tagging" is a process that improves the searchability of data by assigning keywords related to data items.

[0086] A "machine learning model" is an algorithm that learns patterns and rules from large amounts of data and generates appropriate responses or predictions for future inputs.

[0087] "Training methods" refer to processing equipment and software used to train a machine learning model using collected and pre-processed data.

[0088] "New ideas submitted by users" refer to suggestions for new services or events provided in text format by users of the system.

[0089] "Search methods" refer to search engines and algorithms that retrieve relevant historical data from a database based on ideas entered by the user.

[0090] "Simulation results" refer to specific suggestions and advice for a user's new ideas, generated using a machine learning model.

[0091] A "user terminal" is a computer or mobile device used by a user to access a system and input and view information.

[0092] This invention relates to a system for clarifying the true needs of employees and customers using past survey data, and for proposing new services and events based on those needs. Specific embodiments will be described below.

[0093] First, the server collects past survey information. This process begins by retrieving data from databases such as MySQL® or PostgreSQL. The server connects to these databases using appropriate authentication credentials and executes SQL queries to collect the necessary survey information. This collected data is stored in CSV or JSON format.

[0094] Next, the collected survey information is preprocessed. The server uses Python and the Pandas library to clean the data, removing invalid responses and missing data. After that, data classification and tagging are performed. This classifies the survey content by theme, organizing it into categories such as "employee benefits" and "company events." Tagging assigns keywords related to the data items, improving searchability.

[0095] Using the preprocessed data, the server trains a machine learning model. This process involves training a natural language processing model (e.g., ChatGPT) using libraries such as PyTorch and TENSORFLOW®. The trained model is then optimized to generate appropriate responses and suggestions based on future inputs.

[0096] When users submit ideas for new services or events, they access the system using a web browser or mobile app. Users enter their ideas in text format, and this data is sent to the server as an HTTP request. For example, a request might read, "I want to plan a new sports event."

[0097] The server searches past survey data related to the user's new idea. This search process uses search engines such as Elasticsearch® and Solr. The search results extract the feedback and opinions most relevant to the user's idea and provide them for analysis.

[0098] Based on the retrieved data, the server uses a trained machine learning model to generate simulation results. For example, for the idea of ​​"planning a new sports event," relevant feedback might be obtained such as "many employees want weekend events, and soccer and basketball are popular." Based on this information, the system generates specific suggestions such as "hold an in-house soccer tournament on Saturday mornings, making it an event that families can also participate in."

[0099] The generated simulation results are sent from the server to the user's terminal, allowing the user to view suggestions and advice.

[0100] As a concrete example, the following prompt sentences could be input to the generation AI model:

[0101] "We'd like to plan a new sports event. We're looking for an event that employees can enjoy with their families on the weekend. Please provide suggestions based on past survey data."

[0102] In this way, a system is built that can effectively utilize past survey data to propose new services and events based on the needs of employees and customers.

[0103] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0104] Step 1:

[0105] The server collects past survey information. Specifically, the server connects to databases such as MySQL or PostgreSQL using appropriate authentication credentials and executes SQL queries. The collected data is stored in CSV or JSON format.

[0106] Input: Database connection information and SQL query

[0107] Output: Survey data in CSV or JSON format.

[0108] Specific operation: The server executes an SQL query such as SELECT FROM survey_table and saves the retrieved data as a file.

[0109] Step 2:

[0110] The server preprocesses the collected survey information. This step uses Python and the Pandas library to clean, classify, and tag the data.

[0111] Input: Survey data in CSV or JSON format.

[0112] Output: Preprocessed data frame

[0113] Specific operation: The server reads the data using pd.read_csv("Survey Data.csv"), removes invalid responses and missing data using df.dropna(), and categorizes the data by theme. For example, categorization is performed using df['Category'] = df['Content'].apply(...). Furthermore, keywords are added using df['Tags'] = df['Content'].apply(...).

[0114] Step 3:

[0115] The server trains machine learning models using pre-processed data. It uses PyTorch or TensorFlow to train natural language processing models.

[0116] Input: Preprocessed data frame

[0117] Output: Trained machine learning model

[0118] Specific operation: The server uses the training data to execute model.train(). For example, when using the ChatGPT natural language processing model, the server loads the model and tokenizer using `from transformers import GPT2LMHeadModel, GPT2Tokenizer` and then runs training. After training, the model is saved using model.save_pretrained("model path").

[0119] Step 4:

[0120] Users enter ideas for new services or events. Users access the system via a web browser or mobile app and enter their ideas in text format.

[0121] Input: User's new idea (text format)

[0122] Output: Submitted idea data

[0123] Specific operation: The user enters "I want to plan a new sports event" into a web form and clicks the submit button. This data is then sent to the server as an HTTP request.

[0124] Step 5:

[0125] The server searches past survey information related to the new idea entered by the user. It uses search engines such as Elasticsearch and Solr to extract relevant data.

[0126] Input: New user ideas, search queries such as Elasticsearch.

[0127] Output: Related survey data

[0128] Specific operation: The server executes es.search(index="Survey", body={"query": {"match": {"Content": "User Ideas"}}}) and temporarily saves the results.

[0129] Step 6:

[0130] The server generates simulation results using a trained machine learning model. This generates suggestions and advice based on the user's new ideas.

[0131] Input: New user ideas, relevant survey data, trained machine learning model

[0132] Output: Simulation results (text format)

[0133] Specific operation: The server loads the trained model, inputs the user's ideas and related survey data, and executes model.generate(input_ids). The generated results are constructed in text format.

[0134] Step 7:

[0135] The server displays the generated simulation results on the user's terminal. The user can then view specific suggestions and advice on their terminal.

[0136] Input: Simulation results (text format)

[0137] Output: Suggestions and advice displayed on the user's terminal.

[0138] Specific operation: The server sends the results to the user's terminal in HTML or JSON format. For example, execute response.json({"Suggestion": Suggestion result}). The user's terminal displays these results on the screen.

[0139] (Application Example 1)

[0140] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0141] The objective of this invention is to effectively utilize past survey information to provide proposals for new services and events based on the true needs of customers and employees. Conventional systems simply collect survey data for feedback, but struggle to apply that data effectively. Furthermore, they lacked a mechanism to provide real-time, effective feedback when users entered new ideas. Therefore, there were limitations to improving customer satisfaction and operational efficiency.

[0142] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0143] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, means for displaying the generated simulation results on a user terminal, means for providing a user interface for entering new service ideas, and means for collecting and analyzing feedback in real time through the user interface. This makes it possible to provide users with effective new service and event suggestions based on past survey information in real time.

[0144] Definitions of important words

[0145] "Past survey information" refers to data that includes feedback, opinions, and evaluations collected from customers and employees in the past.

[0146] "Cleaning" is the process of removing invalid responses and missing data from collected data.

[0147] "Classification" is the process of organizing data according to specific themes or categories.

[0148] "Tagging" is the process of assigning keywords related to each data item to make it easier to search.

[0149] A "machine learning model" is a system that uses algorithms to learn patterns and rules from data and perform predictions and analyses.

[0150] "New ideas submitted by users" refer to new service or event concepts and plans proposed by users.

[0151] "Simulation results" refer to predictions and specific proposals based on new ideas, using machine learning models.

[0152] A "user terminal" is a device used by a user to input information or check results.

[0153] A "user interface" refers to the screens and input methods that a user uses to interact with a system.

[0154] "Means for collecting and analyzing feedback in real time" refers to a function that instantly collects data in response to user input and actions, analyzes it, and generates results.

[0155] Preparation of a detailed statement

[0156] This invention provides a specific embodiment of a system for analyzing survey information and proposing new services. The central elements of the system are a server and a user terminal.

[0157] Server Processing

[0158] 1. Data collection:

[0159] The server collects past survey information from various databases within the organization (e.g., HR database, general affairs database, monitor database, etc.). The data collection process uses APIs and SQL queries to appropriately retrieve the necessary information.

[0160] 2. Data preprocessing:

[0161] The collected survey data is cleaned by the server, removing invalid responses and missing data. The data is then categorized into specific themes and categories, and each data item is tagged to facilitate searching. Specifically, this preprocessing is performed using the Python pandas library and scikit-learn.

[0162] 3. Training machine learning models:

[0163] The preprocessed data is trained using a machine learning model (for example, a model based on natural language processing). This model learns past data patterns and becomes capable of generating appropriate responses to future novel ideas. Libraries used include scikit-learn and the OpenAI® GPT model.

[0164] 4. Analysis of new ideas:

[0165] When a user inputs a new service idea into the system, the server searches and analyzes past survey data related to that idea. This utilizes natural language processing technology to analyze past feedback and opinion trends.

[0166] 5. Simulation and result generation:

[0167] The server uses a trained machine learning model to generate simulation results based on the user's new idea. For example, if a new loyalty program is proposed, it analyzes past feedback related to this idea and generates specific advice and suggestions.

[0168] 6. Presentation of results:

[0169] The generated simulation results are displayed on the user's terminal. This allows the user to easily obtain specific suggestions and advice.

[0170] User-side interface

[0171] 1. Enter your new service idea:

[0172] Users utilize an interface to input ideas for new services into the system. This interface is provided as a smartphone app, allowing users to input ideas in text format.

[0173] 2. Collection and analysis of real-time feedback:

[0174] When a new service idea is submitted, feedback is collected in real time and immediately analyzed by the server. This allows users to receive feedback instantly.

[0175] Specific example

[0176] For example, if a store manager inputs a new service idea, such as "I want to introduce a loyalty program," into the system, the server searches and analyzes past customer survey data. Suppose the system finds feedback indicating that "many customers want a points system." Based on this, the system generates a concrete proposal for "introducing a loyalty program with a points system" and displays it on the user's terminal.

[0177] Examples of prompts to input into a generative AI model:

[0178] A new idea for the store: Introduce a loyalty program. How does it compare to past customer feedback?

[0179] In this way, the system of the present invention can effectively utilize past survey data and support users in proposing new services and events in real time.

[0180] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0181] Description of processing steps

[0182] Step 1:

[0183] Data collection

[0184] The server collects past survey information from the organization's HR database, general affairs database, and monitor database, using appropriate APIs and SQL queries. For example, it uses SQL queries to retrieve the necessary survey data and load it into the server's memory.

[0185] Input: Database connection information, query

[0186] Output: Collected survey data

[0187] The server retrieves the survey information from these databases and passes the collected data to the next processing step.

[0188] Step 2:

[0189] Data preprocessing

[0190] The server preprocesses the collected survey information. It cleans, categorizes, and tags the data, removes invalid responses and missing data, and then organizes the data by category.

[0191] Input: Collected survey data

[0192] Output: Preprocessed survey data

[0193] Specifically, we will use Python's pandas library to clean the data and then use scikit-learn's TfidfVectorizer to convert the data into numerical values.

[0194] Step 3:

[0195] Training machine learning models

[0196] The server uses the pre-processed data to train a machine learning model. Specifically, it uses a natural language processing model to learn patterns from past data.

[0197] Input: Preprocessed survey data

[0198] Output: Trained machine learning model

[0199] This process utilizes scikit-learn's KMeans clustering and the OpenAI GPT model. Training improves the accuracy of the model for predictions and analyses.

[0200] Step 4:

[0201] Entering new ideas

[0202] Users input ideas for new services using a smartphone app. This input is sent to the server in real time.

[0203] Input: New idea entered by the user

[0204] Output: Idea data sent to the server

[0205] The user enters new ideas in text format using the interface provided by the user terminal.

[0206] Step 5:

[0207] Searching and analyzing related data

[0208] The server searches and analyzes past survey data related to the new idea entered by the user. It uses natural language processing techniques to extract similar opinions and feedback.

[0209] Input: New ideas entered by users, past survey data

[0210] Output: Similar past feedback, analysis results

[0211] For example, you can use methods such as TF-IDF or cosine similarity to search for highly relevant historical data.

[0212] Step 6:

[0213] Simulation and result generation

[0214] The server uses a trained machine learning model to generate simulation results for new ideas. For example, if the idea is "introduce a loyalty program," it will generate specific proposals based on relevant past feedback.

[0215] Input: New idea, machine learning model, analysis results

[0216] Output: Simulation results, specific proposals

[0217] The AI ​​model is used to analyze the data and concretize the proposed content. An example of a prompt is: "A new idea for the store: Introduce a loyalty program. How does it compare to past customer feedback?"

[0218] Step 7:

[0219] Presentation of results

[0220] The generated simulation results are displayed on the user's device. Users can view specific suggestions and advice through a smartphone app.

[0221] Input: Simulation results

[0222] Output: Suggestions and advice displayed on the user's terminal.

[0223] The user's terminal displays the generated specific suggestions, allowing them to receive immediate feedback. As a result, users can obtain concrete information to quickly implement new, feasible services.

[0224] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0225] This invention relates to a system that utilizes past survey data to clarify the true needs of employees and customers, and then proposes new services and events based on those needs. Furthermore, by combining this with an emotion engine that recognizes user emotions, it is possible to provide even more appropriate suggestions.

[0226] 1. Data Collection

[0227] The server first collects past survey information from the HR database, general affairs database, and monitor database. This involves authenticating to each database, establishing a secure connection, and then executing the necessary SQL queries and API calls to retrieve the data.

[0228] 2. Data preprocessing

[0229] The server cleans the collected survey information. This involves removing invalid responses and missing data, and then classifying each survey by theme. Specifically, it divides them into categories such as "employee benefits," "company events," and "fitness apps." Next, this data is tagged and given relevant keywords to facilitate searching and analysis.

[0230] 3. Training machine learning models

[0231] The server trains a natural language processing model (e.g., ChatGPT) using pre-processed survey data. This model learns patterns from the data and uses them for future predictions and response generation. The model's performance is evaluated, and parameters are fine-tuned as needed to generate optimal responses.

[0232] 4. Introduction of an emotional engine

[0233] The server is equipped with an emotion engine to recognize user emotions. This emotion engine analyzes the text of new ideas entered by the user and determines the user's emotional state from the context. For example, it can identify emotions such as joy, excitement, and confusion. This allows the feedback and suggestions provided to be more accurate and aligned with the user's feelings.

[0234] 5. User input of new ideas

[0235] Users input ideas for new services and events through the system. They submit specific concepts and desired conditions in text format from their user terminals. This input data is sent to the server and simultaneously analyzed by an emotion engine.

[0236] 6. Searching for related data

[0237] The server searches the database for past survey information related to the new idea entered by the user. It efficiently extracts highly relevant data using pre-classified and tagged keywords and category information.

[0238] 7. Simulation and Result Generation

[0239] The server uses trained machine learning models and an emotion engine to generate simulation results based on the user's new idea. For example, if a new idea for a fitness app is entered, it will generate feedback such as "Many users want a plan that allows them to exercise effectively in a short amount of time," based on past feedback. This generated result is adjusted according to the user's emotional state.

[0240] 8. Presentation of Results

[0241] The server constructs the generated simulation results and displays them on the user's terminal. This allows the user to easily obtain specific suggestions and feedback. For example, specific advice such as, "A 10-minute high-intensity interval training plan combined with a game element where the user earns points to advance their level would likely be preferred," might be displayed. Based on sentiment recognition, the tone and details of the suggestions may also be adjusted.

[0242] Through this series of processes, the server can combine past survey data with the results of the emotion engine's analysis to propose new services and events that are better suited to user needs.

[0243] The following describes the processing flow.

[0244] Step 1: Data Collection

[0245] The server connects to the HR database, general affairs database, and monitor database to collect past survey information. The server authenticates to each database and uses authentication protocols to establish a secure connection. Then, it executes the necessary SQL queries and API calls to retrieve the target data.

[0246] Step 2: Data Preprocessing

[0247] The server cleans the survey information it has collected. Specifically, it filters out invalid responses and duplicate data, and imputes or removes missing values. Normalization is also performed to maintain data consistency, for example, by converting response items into a unified format.

[0248] The server then categorizes the cleaned data by theme. This categorization involves analyzing the survey content and assigning it to categories such as "employee benefits," "company events," and "fitness apps."

[0249] The server tags the classified data. This assigns relevant keywords, enabling efficient data retrieval and analysis.

[0250] Step 3: Training the machine learning model

[0251] The server uses pre-processed survey data to train a natural language processing model (e.g., ChatGPT). Training is performed by inputting a large amount of data into the model and having it analyze the results.

[0252] The server evaluates the model's performance and adjusts parameters as needed. For example, it measures the accuracy and relevance of the responses generated by the model and performs an iterative process to find the optimal parameter settings.

[0253] Step 4: Introducing the Emotional Engine

[0254] The server will implement an emotion engine to recognize the user's emotions. The emotion engine will analyze the content of the text entered by the user and determine the emotion from the context. The emotion engine will analyze based on multiple emotion categories (e.g., joy, excitement, confusion, dissatisfaction, etc.).

[0255] Specifically, the server analyzes user input and processes it to recognize the emotion of "joy" from text such as "I think this proposal is great."

[0256] Step 5: User input of new ideas

[0257] Users input ideas for new services or events via their devices. Through the system interface, users submit specific concepts and requirements in text format. This input data is transmitted to the server in real time.

[0258] Step 6: Search for related data

[0259] The server searches the database for past survey information related to the new idea entered by the user. It uses pre-classified and tagged keywords and category information to narrow down the data to the most relevant results.

[0260] For example, if a user enters "ideas for a new fitness app," the server will extract past survey information related to "fitness" and "apps."

[0261] Step 7: Simulation and result generation

[0262] The server uses a trained machine learning model and an emotion engine to generate simulation results based on the user's novel ideas. The server considers patterns obtained from survey data and the user's emotional state to generate responses.

[0263] For example, in response to a user's suggestion, it might generate feedback such as, "Many users want a plan that allows them to exercise effectively in a short amount of time," and if the emotion engine detects the emotion of "joy," it will present the result in a positive tone that takes that into account.

[0264] Step 8: Presentation of Results

[0265] The server constructs the generated simulation results and displays them on the user's terminal. This allows the user to receive specific and emotionally sensitive suggestions and feedback.

[0266] For example, a suggestion might appear stating, "A plan that combines a 10-minute high-intensity interval training session with a game element where users earn points to advance their progress level would likely be well-received." This suggestion is tailored to resonate with the user's emotions.

[0267] This entire process allows the system, combined with the emotion engine, to consider past survey data and the user's emotional state to suggest more appropriate and personalized new services and events.

[0268] (Example 2)

[0269] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0270] Modern companies and organizations are required to accurately understand the needs of their employees and customers and to provide new services and events based on those needs. However, conventional methods have made it difficult to efficiently utilize past survey data and grasp true needs. Furthermore, proposals often did not match the emotions of users, limiting their ability to increase satisfaction. To solve these problems, the present invention aims to provide a system that effectively utilizes past survey data and makes proposals that take user emotions into consideration.

[0271] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0272] In this invention, the server includes means for collecting past survey data, means for cleaning, classifying, and tagging the survey data, means for training a generative model using the cleaned and tagged survey data, means for searching for survey data related to a new idea entered by the user, means for generating simulation results based on the user's new idea using the generative model, means for analyzing the user's emotions and reflecting the analysis results in the feedback, and means for displaying the generated simulation results on the user terminal. This makes it possible to provide highly accurate suggestions based on past survey information and to provide customized feedback that takes into account the user's emotions.

[0273] "Survey data" refers to data that includes questionnaires and feedback information collected from users and customers.

[0274] "Cleaning" is a process that removes invalid responses and missing data from survey data to improve data quality.

[0275] "Classification" is the process of grouping cleaned survey data into specific themes or categories (e.g., employee benefits, company events).

[0276] "Tagging" is the process of assigning keywords related to classified survey data to facilitate searching and analysis.

[0277] A "generative model" is an algorithm or system that uses machine learning or natural language processing techniques to learn patterns from data and generate predictions or responses.

[0278] "New ideas" refer to concepts or suggestions for new services or events that users input into the system.

[0279] "Simulation results" refer to the feedback and suggestions generated as a response to analysis and predictions based on novel ideas using a generative model.

[0280] "Sentiment analysis" is a technology that analyzes the context of text data entered by a user to identify the user's emotional state (e.g., joy, excitement, confusion).

[0281] "Display means" refers to functions or devices for visually presenting generated simulation results and feedback to the user's terminal.

[0282] This invention relates to a system that enables companies and organizations to effectively utilize past survey data to propose new services and events that take user sentiment into consideration. This system functions through the cooperation of a server, terminals, and users.

[0283] Data collection

[0284] The server first collects past survey data. This survey data is obtained from the personnel database, general affairs database, and other monitoring databases within the organization. The server authenticates to each database, establishes a secure connection, and then executes the necessary SQL queries and API calls to obtain the data. The software used at this time is an SQL database management system (e.g., MySQL or PostgreSQL) and a library for API calls (e.g., Axios or Fetch API).

[0285] Data preprocessing

[0286] The received survey data is cleaned by the server. Invalid responses and missing data are removed, and then the data is classified by theme. For example, it is categorized into "welfare", "company events", "fitness apps", etc., and relevant keywords are tagged. The software used for this process is a data mining tool (e.g., the pandas library in Python).

[0287] Training of the machine learning model

[0288] The server trains a generation model using the preprocessed survey data. Here, a natural language processing model (e.g., ChatGPT) is used. The model learns patterns from the data and is used for future predictions and response generation. A high-performance GPU (e.g., NVIDIA Tesla) is used to evaluate the performance of the model and perform fine-tuning of the parameters.

[0289] Introduction of the sentiment engine

[0290] The server introduces a sentiment engine that recognizes the user's sentiment. This sentiment engine analyzes the text of the new idea input by the user and determines the user's sentiment state (e.g., joy, excitement, confusion) from the context. A sentiment analysis library (e.g., Aylien or TextBlob) is used for sentiment analysis.

[0291] User input of new ideas

[0292] Users input ideas for new services or events through the system. They enter specific concepts and desired conditions in text format from their device (e.g., a PC or smartphone) and send them to the server. This transmission uses the HTTPS protocol and encrypts the data. The server analyzes the received data using an emotion engine and evaluates the emotional state.

[0293] Searching for related data

[0294] The server searches past research data related to the new idea entered by the user. It extracts highly relevant data using pre-classified and tagged keywords and category information. A database management system (e.g., Elasticsearch) is used for this search.

[0295] Simulation and result generation

[0296] The server uses generative models and an emotion engine to generate simulation results based on the user's new idea. For example, if a new idea for a fitness app is entered, it will generate feedback such as "Many users want a plan that allows them to exercise effectively in a short amount of time," based on past feedback. This feedback is then refined based on the emotion analysis results.

[0297] Presentation of results

[0298] The server displays the generated simulation results on the user's terminal. The user can visually see specific suggestions and feedback. For example, specific advice might be given such as, "A plan incorporating a 10-minute high-intensity interval training session and a game element where the user earns points to advance their progress level would likely be well-received."

[0299] The following are specific examples of prompt statements:

[0300] Please generate feedback on the idea of a new fitness app using the following questionnaire data.

[0301] Questionnaire Data 1: ...

[0302] Questionnaire Data 2: ...

[0303] ... (Enter multiple questionnaire data) ...

[0304] Idea for the new fitness app:

[0305] 10-minute high-intensity interval training

[0306] Function for users to earn points and increase their progress level

[0307] Please provide feedback taking into account the results of sentiment analysis.

[0308] The flow of the specific process in Example 2 will be described using FIG. 13.

[0309] Step 1: Data collection

[0310] The server collects past survey data. This starts by authenticating to each database and establishing a secure connection. Specifically, authentication is performed using an API key, user ID, and password, and a secure connection is established using the SSL / TLS protocol. Next, the necessary SQL queries or API calls are executed to obtain the data (e.g., SELECT FROM survey_data WHERE date > '2020-01-01'). This data is stored in a secure storage within the company. The input is the database authentication information and connection information, and the output is the collected survey data. The specific operations are the execution of API calls or SQL queries and data acquisition and storage.

[0311] Step 2: Data preprocessing

[0312] The server cleans, classifies, and tags the collected survey data. First, it detects and removes invalid responses and missing data (e.g., empty fields or inconsistent responses). Next, it classifies the data into specific themes (e.g., employee benefits, company events) and tags relevant keywords using TF-IDF (Term Frequency-Inverse Document Frequency). The input is the collected survey data, and the output is the cleaned and tagged survey data. The specific operations are data cleaning, thematic classification, and tagging.

[0313] Step 3: Training the machine learning model

[0314] The server trains a generative AI model (e.g., ChatGPT) using preprocessed survey data. First, the preprocessed data is split into a training dataset and a test dataset. Typically, the training data:test data ratio is 8:2. Next, the model is trained using a high-performance GPU. The model's performance is evaluated, and evaluation metrics such as accuracy, recall, and F-score are checked. Parameters are fine-tuned as needed, and retraining is performed. The input is the cleaned and tagged survey data, and the output is the trained generative AI model.

[0315] Step 4: Introducing the Emotional Engine

[0316] The server implements an emotion engine to recognize user emotions. It installs an emotion engine (e.g., Emotion API), analyzes user-entered text data in real time, and determines the emotional state. The analysis results are saved to data storage and used in subsequent processes. Input is the user's text input, and output is the analyzed emotion information. The specific operations are text analysis and emotion identification.

[0317] Step 5: User input of new ideas

[0318] Users input new ideas using the system. Users enter specific concepts and desired conditions in text format from their devices (e.g., PCs, smartphones) and send them to the server. This data is transmitted encrypted (using the HTTPS protocol). The server passes the received data to an emotion engine, which analyzes the emotional state. The input is the user's new idea, and the output is the submitted idea and the analyzed emotional information.

[0319] Step 6: Search for related data

[0320] The server searches past survey data related to the user's new idea. It searches the database using pre-classified and tagged keywords and category information. Specifically, it executes SQL queries or Elasticsearch search queries to extract highly relevant data (e.g., SELECT FROM survey_data WHERE keywords LIKE '%fitness%'). The search results are temporarily stored in memory storage. The input is the user's new idea and keywords, and the output is the relevant survey data. The specific operations are executing search queries and extracting data.

[0321] Step 7: Simulation and result generation

[0322] The server generates simulation results based on the user's new ideas using a generative model and an emotion engine. The user's new ideas and related data are integrated and input into the generative AI model. Feedback output by the generative model is retrieved and fine-tuned using the emotion engine's analysis results. For example, it might output feedback such as, "Many users would like a plan that allows them to exercise effectively in a short amount of time." The input is the new ideas and related data, and the output is the simulation results.

[0323] Step 8: Presentation of Results

[0324] The server displays the generated simulation results on the user's terminal. First, the generated feedback is converted into HTML or JSON format and sent to the user's terminal. The user's terminal displays the results on an interface, allowing the user to visually confirm specific suggestions and feedback. For example, feedback such as "A 10-minute high-intensity interval training plan combined with a game element where the user earns points to advance their level would likely be preferred" might be displayed. The inputs are the simulation results and analysis results, and the output is the feedback presented to the user.

[0325] (Application Example 2)

[0326] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0327] Currently, systems exist that utilize past survey data to understand users' true needs and propose new services and events based on those needs. However, because these types of systems do not take into account the emotional state of the user, the proposed content does not always meet the user's expectations. In particular, in physical stores, there is a demand for service proposals that reflect the customer's real-time emotional state, but conventional systems have had difficulty achieving this. The present invention aims to solve these problems and provide a system that makes more appropriate service proposals while taking into account the emotional state of the customer.

[0328] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0329] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, means for adjusting the generated simulation results using an emotion engine that analyzes the user's emotions, and means for displaying the generated simulation results on the user terminal. This makes it possible to propose services that take into account the user's emotional state.

[0330] "Survey information" refers to response data obtained from past surveys and feedback.

[0331] "Means of collection" refers to mechanisms for collecting survey information through databases or APIs.

[0332] "Cleaning" is the process of improving data quality by removing invalid responses and missing data.

[0333] "Classification and tagging" is the process of organizing cleaned data by theme and assigning relevant keywords.

[0334] A "machine learning model" is an algorithm that learns patterns from large amounts of data and uses them to make predictions and generate responses.

[0335] "Training methods" refer to methods for training machine learning models using pre-processed data.

[0336] "New ideas" refer to suggestions for new services or events that users input into the system.

[0337] "Search methods" refer to efficient methods for extracting past survey information related to the entered idea from the database.

[0338] "Simulation results" refer to the predictions and feedback generated using a machine learning model based on the proposal.

[0339] An "emotion engine" is a technology that analyzes the user's emotional state from their text input and adjusts its response accordingly.

[0340] A "user terminal" is a device (such as a smartphone or tablet) used to access the system and review the proposed content.

[0341] The system implementing this invention collects past survey information and, based on that, makes service suggestions related to the user's new ideas. Furthermore, it analyzes the user's emotional state using an emotion engine and adjusts the suggestions accordingly to provide more appropriate feedback.

[0342] First, the server collects past survey information from databases. This data is obtained from databases such as the human resources database, general affairs database, and monitor database. The collected data is cleaned, invalid responses and missing data are removed, and the data is categorized and tagged by theme. Programming languages ​​such as Python and SQL are used for this process.

[0343] Next, the server uses the cleaned and tagged survey data to train a machine learning model. Specifically, it uses a generative AI model with natural language processing techniques (e.g., GPT-3(registered trademark).5-turbo). This model learns from a large amount of text data and generates new text and makes predictions about new user ideas.

[0344] When a user inputs a new idea into the system, it searches the database for past survey information related to that new idea. The retrieved data is then fed into a machine learning model, and simulation results are generated.

[0345] Furthermore, the server uses an emotion engine to analyze the user's emotional state. This emotion engine analyzes the text data entered by the user and determines the emotional state (joy, excitement, confusion, etc.) from the context. Based on the analyzed emotional state, the simulation results are adjusted. The emotion engine may use an emotion analysis library, such as Microsoft's Azure Text Analytics API.

[0346] Finally, the server displays the adjusted simulation results on the user's terminal. Users can review the suggestions via devices such as smartphones and tablets. A dedicated application is installed on the terminal, allowing them to proceed with planning new services and events based on the displayed feedback and suggestions.

[0347] As a concrete example, when a user enters the following prompt, the optimal suggestion is generated:

[0348] Example of a prompt:

[0349] I would like to propose a new fitness program. Specifically, I want to create a plan that allows for effective exercise in a short amount of time, and incorporate a game element where users can earn points and advance through the program.

[0350] This allows the server to accurately reflect user needs and provide appropriate service suggestions based on their emotional state through the processes described above.

[0351] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0352] Step 1:

[0353] The server collects past survey information. Specifically, it authenticates to the HR database, general affairs database, and monitor database, establishes a secure connection, and then executes the necessary SQL queries and API calls. This collects the survey information. The input is "survey information from the database," and the output is "collected survey information."

[0354] Step 2:

[0355] The server cleans the collected survey information, removing invalid responses and missing data to improve data quality. Next, it categorizes the data by theme and tags it with relevant keywords. The input is the "collected survey information," and the output is the "cleaned, categorized, and tagged survey information."

[0356] Step 3:

[0357] The server trains a machine learning model using cleaned and tagged survey information. Specifically, it uses a generative AI model based on natural language processing (GPT-3.5-turbo) to learn patterns from a large amount of text data. The input is "cleaned, classified, and tagged survey information," and the output is the "trained machine learning model."

[0358] Step 4:

[0359] Users input new ideas into the system. They submit specific concepts and desired conditions to the system in text format from their terminal. The input is "the text of the user's new idea," and the output is "the user's new idea sent to the server."

[0360] Step 5:

[0361] The server searches the database for past survey information related to the new idea entered by the user. It efficiently extracts highly relevant data using pre-classified and tagged keywords and category information. The input is the "user's new idea" and the "database," and the output is a "list of relevant survey information."

[0362] Step 6:

[0363] The server generates simulation results using a trained machine learning model. Based on the user's new idea and related survey information, it generates future predictions and responses. The inputs are "the user's new idea," "a list of related survey information," and "the trained machine learning model," and the output is "the generated simulation results."

[0364] Step 7:

[0365] The server uses an emotion engine to analyze the user's emotional state. It analyzes the text of the user's new idea and determines the emotional state (joy, excitement, confusion, etc.) from the context. The input is "the text of the user's new idea," and the output is "the analysis result of the user's emotional state."

[0366] Step 8:

[0367] The server adjusts the generated simulation results based on the analysis results of the emotion engine. This optimizes the suggestions according to the user's emotional state. The inputs are the "generated simulation results" and the "analysis results of the user's emotional state," and the output is the "adjusted simulation results."

[0368] Step 9:

[0369] The server displays the adjusted simulation results on the user's terminal. The user can review the suggestions through the terminal and provide further feedback. The input is the "adjusted simulation results," and the output is the "feedback and suggestions displayed to the user."

[0370] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0371] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0372] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0373] [Second Embodiment]

[0374] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0375] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0376] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0377] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0378] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0379] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0380] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0381] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0382] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0383] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0384] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0385] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0386] This invention relates to a system that uses past survey data to clarify the true needs of employees and customers, and based on that, proposes new services and events. Specific embodiments of this system are described below.

[0387] 1. Data Collection

[0388] The server first collects past survey information. This includes data collection from the human resources database, general affairs database, and monitor database. The server connects to these databases through an appropriate authentication process and retrieves the necessary survey information.

[0389] 2. Data preprocessing

[0390] The server then preprocesses the collected survey information. Specifically, it performs processes such as cleaning, classification, and tagging. Cleaning involves removing invalid responses and missing data. Classification organizes the data by theme based on the survey content, dividing it into categories such as "employee benefits," "company events," and "fitness apps." Tagging involves assigning keywords related to each data item to make it easier to search.

[0391] 3. Training machine learning models

[0392] Using the pre-processed survey data, the server trains a machine learning model, specifically a natural language processing model (such as ChatGPT). This model learns patterns from past survey data and is adjusted to generate appropriate responses to future inputs.

[0393] 4. User input of new ideas

[0394] Users use the system to input ideas for new services or events. They access the system from their terminals and provide ideas in text format. This input data is sent to the server.

[0395] 5. Searching for related data

[0396] The server searches its database for past survey information related to the new idea entered by the user. This allows it to collect similar feedback and opinions and use them for analysis.

[0397] 6. Simulation and Result Generation

[0398] The server uses a trained machine learning model to generate simulation results based on the user's new ideas. For example, if an idea for a specific fitness app is entered, the server analyzes past feedback related to this idea and generates specific advice and suggestions.

[0399] 7. Presentation of Results

[0400] The server constructs the generated simulation results and displays them on the user's terminal. This allows users to easily obtain specific suggestions and advice.

[0401] Specific example

[0402] For example, if a user submits a request to "plan a new sports event," the server searches and analyzes past survey data on sports events. Suppose the results show that "many employees want weekend events, and soccer and basketball are popular." Based on this, the system generates suggestions such as "hold an in-house soccer tournament on Saturday mornings, making it an event that families can also participate in," and displays them on the user's terminal.

[0403] Through these specific means, the system of the present invention can effectively utilize past survey data to propose new services and events based on the needs of employees and customers.

[0404] The following describes the processing flow.

[0405] Step 1: Data Collection

[0406] The server connects to the HR database, general affairs database, and monitor database to collect past survey information. It authenticates with the target databases and establishes secure access. It then executes necessary SQL queries and API calls to retrieve the survey data.

[0407] Step 2: Data Preprocessing

[0408] The server cleans the survey data it has collected. Specifically, it filters out invalid responses and duplicate data, and handles missing values.

[0409] The server sorts the cleaned data by theme. Based on the survey responses, it categorizes the data into categories such as "employee benefits," "company events," and "fitness apps."

[0410] The server tags the classified data and assigns relevant keywords to facilitate searching and analysis. This enables quick searching of data related to specific themes.

[0411] Step 3: Training the machine learning model

[0412] The server uses pre-processed survey data to train a machine learning model (e.g., ChatGPT). This process involves inputting a large amount of data to allow the model to learn patterns.

[0413] The server evaluates the model's performance and fine-tunes the parameters as needed. This ensures that the final model has the ability to generate the optimal response to the user's new ideas.

[0414] Step 4: User input of new ideas

[0415] Users input ideas for new services or events into the system. They provide specific concepts and desired conditions in text format from their user terminals. This input data is then sent to the server.

[0416] Step 5: Search for related data

[0417] The server searches the database for past survey information related to the user's new idea. It extracts highly relevant data using pre-tagged keywords and category information. This allows for efficient collection of past feedback and opinions.

[0418] Step 6: Simulation and result generation

[0419] The server uses a trained machine learning model to generate simulation results based on the user's new idea. The model uses patterns learned from past data to generate predictions and suggestions for the input idea.

[0420] For example, when a new fitness app idea is presented, specific feedback such as, "Many users would like a plan that allows them to exercise effectively in a short amount of time," can be provided.

[0421] Step 7: Presentation of Results

[0422] The server compiles the generated simulation results and displays them on the user's terminal. This allows the user to receive specific suggestions and feedback.

[0423] For example, it might display specific advice such as, "A plan that combines a 10-minute high-intensity interval training session with a game element where users earn points to advance their progress level would likely be well-received."

[0424] This series of steps enables the server to effectively utilize past survey data and create a system that generates concrete and effective suggestions for users' new ideas.

[0425] (Example 1)

[0426] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0427] Traditional systems have made it difficult to effectively utilize past survey data to propose new services and events based on employee and customer needs. Furthermore, there is a lack of means to quickly search and analyze past feedback related to specific new ideas, making it difficult to provide concrete advice to users' new ideas. Solving these problems is essential.

[0428] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0429] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, and means for displaying the generated simulation results on the user's terminal. This makes it possible to quickly provide specific suggestions and advice for the user's new ideas by utilizing past survey information.

[0430] "Past survey information" refers to data based on surveys previously conducted within the organization, including feedback and opinions collected from employees and customers.

[0431] "Means of collection" refers to a combination of hardware and software used to retrieve necessary information from databases or APIs.

[0432] "Cleaning" is a process that improves data quality by removing invalid responses and missing data from the data.

[0433] "Classification" is the process of organizing survey data by theme or category and grouping it according to a specific purpose.

[0434] "Tagging" is a process that improves the searchability of data by assigning keywords related to data items.

[0435] A "machine learning model" is an algorithm that learns patterns and rules from large amounts of data and generates appropriate responses or predictions for future inputs.

[0436] "Training methods" refer to processing equipment and software used to train a machine learning model using collected and pre-processed data.

[0437] "New ideas submitted by users" refer to suggestions for new services or events provided in text format by users of the system.

[0438] "Search methods" refer to search engines and algorithms that retrieve relevant historical data from a database based on ideas entered by the user.

[0439] "Simulation results" refer to specific suggestions and advice for a user's new ideas, generated using a machine learning model.

[0440] A "user terminal" is a computer or mobile device used by a user to access a system and input and view information.

[0441] This invention relates to a system for clarifying the true needs of employees and customers using past survey data, and for proposing new services and events based on those needs. Specific embodiments will be described below.

[0442] First, the server collects past survey information. This process begins by retrieving data from databases such as MySQL or PostgreSQL. The server connects to these databases using appropriate authentication credentials and executes SQL queries to collect the necessary survey information. This collected data is stored in CSV or JSON format.

[0443] Next, the collected survey information is preprocessed. The server uses Python and the Pandas library to clean the data, removing invalid responses and missing data. After that, data classification and tagging are performed. This classifies the survey content by theme, organizing it into categories such as "employee benefits" and "company events." Tagging assigns keywords related to the data items, improving searchability.

[0444] Using the preprocessed data, the server trains a machine learning model. This process involves training a natural language processing model (e.g., ChatGPT) using libraries such as PyTorch or TensorFlow. The trained model is then optimized to generate appropriate responses and suggestions based on future inputs.

[0445] When users submit ideas for new services or events, they access the system using a web browser or mobile app. Users enter their ideas in text format, and this data is sent to the server as an HTTP request. For example, a request might read, "I want to plan a new sports event."

[0446] The server searches past survey data related to the user's new idea. This search process uses search engines such as Elasticsearch and Solr. The search results extract the feedback and opinions most relevant to the user's idea and provide them for analysis.

[0447] Based on the retrieved data, the server uses a trained machine learning model to generate simulation results. For example, for the idea of ​​"planning a new sports event," relevant feedback might be obtained such as "many employees want weekend events, and soccer and basketball are popular." Based on this information, the system generates specific suggestions such as "hold an in-house soccer tournament on Saturday mornings, making it an event that families can also participate in."

[0448] The generated simulation results are sent from the server to the user's terminal, allowing the user to view suggestions and advice.

[0449] As a concrete example, the following prompt sentences could be input to the generation AI model:

[0450] "We'd like to plan a new sports event. We're looking for an event that employees can enjoy with their families on the weekend. Please provide suggestions based on past survey data."

[0451] In this way, a system is built that can effectively utilize past survey data to propose new services and events based on the needs of employees and customers.

[0452] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0453] Step 1:

[0454] The server collects past survey information. Specifically, the server connects to databases such as MySQL or PostgreSQL using appropriate authentication credentials and executes SQL queries. The collected data is stored in CSV or JSON format.

[0455] Input: Database connection information and SQL query

[0456] Output: Survey data in CSV or JSON format.

[0457] Specific operation: The server executes an SQL query such as SELECT FROM survey_table and saves the retrieved data as a file.

[0458] Step 2:

[0459] The server preprocesses the collected survey information. This step uses Python and the Pandas library to clean, classify, and tag the data.

[0460] Input: Survey data in CSV or JSON format.

[0461] Output: Preprocessed data frame

[0462] Specific operation: The server reads the data using pd.read_csv("Survey Data.csv"), removes invalid responses and missing data using df.dropna(), and categorizes the data by theme. For example, categorization is performed using df['Category'] = df['Content'].apply(...). Furthermore, keywords are added using df['Tags'] = df['Content'].apply(...).

[0463] Step 3:

[0464] The server trains machine learning models using pre-processed data. It uses PyTorch or TensorFlow to train natural language processing models.

[0465] Input: Preprocessed data frame

[0466] Output: Trained machine learning model

[0467] Specific operation: The server uses the training data to execute model.train(). For example, when using the ChatGPT natural language processing model, the server loads the model and tokenizer using `from transformers import GPT2LMHeadModel, GPT2Tokenizer` and then runs training. After training, the model is saved using model.save_pretrained("model path").

[0468] Step 4:

[0469] Users enter ideas for new services or events. Users access the system via a web browser or mobile app and enter their ideas in text format.

[0470] Input: User's new idea (text format)

[0471] Output: Submitted idea data

[0472] Specific operation: The user enters "I want to plan a new sports event" into a web form and clicks the submit button. This data is then sent to the server as an HTTP request.

[0473] Step 5:

[0474] The server searches past survey information related to the new idea entered by the user. It uses search engines such as Elasticsearch and Solr to extract relevant data.

[0475] Input: New user ideas, search queries such as Elasticsearch.

[0476] Output: Related survey data

[0477] Specific operation: The server executes es.search(index="Survey", body={"query": {"match": {"Content": "User Ideas"}}}) and temporarily saves the results.

[0478] Step 6:

[0479] The server generates simulation results using a trained machine learning model. This generates suggestions and advice based on the user's new ideas.

[0480] Input: New user ideas, relevant survey data, trained machine learning model

[0481] Output: Simulation results (text format)

[0482] Specific operation: The server loads the trained model, inputs the user's ideas and related survey data, and executes model.generate(input_ids). The generated results are constructed in text format.

[0483] Step 7:

[0484] The server displays the generated simulation results on the user's terminal. The user can then view specific suggestions and advice on their terminal.

[0485] Input: Simulation results (text format)

[0486] Output: Suggestions and advice displayed on the user's terminal.

[0487] Specific operation: The server sends the results to the user's terminal in HTML or JSON format. For example, execute response.json({"Suggestion": Suggestion result}). The user's terminal displays these results on the screen.

[0488] (Application Example 1)

[0489] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0490] The objective of this invention is to effectively utilize past survey information to provide proposals for new services and events based on the true needs of customers and employees. Conventional systems simply collect survey data for feedback, but struggle to apply that data effectively. Furthermore, they lacked a mechanism to provide real-time, effective feedback when users entered new ideas. Therefore, there were limitations to improving customer satisfaction and operational efficiency.

[0491] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0492] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, means for displaying the generated simulation results on a user terminal, means for providing a user interface for entering new service ideas, and means for collecting and analyzing feedback in real time through the user interface. This makes it possible to provide users with effective new service and event suggestions based on past survey information in real time.

[0493] Definitions of important words

[0494] "Past survey information" refers to data that includes feedback, opinions, and evaluations collected from customers and employees in the past.

[0495] "Cleaning" is the process of removing invalid responses and missing data from collected data.

[0496] "Classification" is the process of organizing data according to specific themes or categories.

[0497] "Tagging" is the process of assigning keywords related to each data item to make it easier to search.

[0498] A "machine learning model" is a system that uses algorithms to learn patterns and rules from data and perform predictions and analyses.

[0499] "New ideas submitted by users" refer to new service or event concepts and plans proposed by users.

[0500] "Simulation results" refer to predictions and specific proposals based on new ideas, using machine learning models.

[0501] A "user terminal" is a device used by a user to input information or check results.

[0502] A "user interface" refers to the screens and input methods that a user uses to interact with a system.

[0503] "Means for collecting and analyzing feedback in real time" refers to a function that instantly collects data in response to user input and actions, analyzes it, and generates results.

[0504] Preparation of a detailed statement

[0505] This invention provides a specific embodiment of a system for analyzing survey information and proposing new services. The central elements of the system are a server and a user terminal.

[0506] Server Processing

[0507] 1. Data collection:

[0508] The server collects past survey information from various databases within the organization (e.g., HR database, general affairs database, monitor database, etc.). The data collection process uses APIs and SQL queries to appropriately retrieve the necessary information.

[0509] 2. Data preprocessing:

[0510] The collected survey data is cleaned by the server, removing invalid responses and missing data. The data is then categorized into specific themes and categories, and each data item is tagged to facilitate searching. Specifically, this preprocessing is performed using the Python pandas library and scikit-learn.

[0511] 3. Training machine learning models:

[0512] The preprocessed data is trained using a machine learning model (for example, a model based on natural language processing). This model learns past data patterns and becomes capable of generating appropriate responses to future novel ideas. The libraries used include scikit-learn and the OpenAI GPT model.

[0513] 4. Analysis of new ideas:

[0514] When a user inputs a new service idea into the system, the server searches and analyzes past survey data related to that idea. This utilizes natural language processing technology to analyze past feedback and opinion trends.

[0515] 5. Simulation and result generation:

[0516] The server uses a trained machine learning model to generate simulation results based on the user's new idea. For example, if a new loyalty program is proposed, it analyzes past feedback related to this idea and generates specific advice and suggestions.

[0517] 6. Presentation of results:

[0518] The generated simulation results are displayed on the user's terminal. This allows the user to easily obtain specific suggestions and advice.

[0519] User-side interface

[0520] 1. Enter your new service idea:

[0521] Users utilize an interface to input ideas for new services into the system. This interface is provided as a smartphone app, allowing users to input ideas in text format.

[0522] 2. Collection and analysis of real-time feedback:

[0523] When a new service idea is submitted, feedback is collected in real time and immediately analyzed by the server. This allows users to receive feedback instantly.

[0524] Specific example

[0525] For example, if a store manager inputs a new service idea, such as "I want to introduce a loyalty program," into the system, the server searches and analyzes past customer survey data. Suppose the system finds feedback indicating that "many customers want a points system." Based on this, the system generates a concrete proposal for "introducing a loyalty program with a points system" and displays it on the user's terminal.

[0526] Examples of prompts to input into a generative AI model:

[0527] A new idea for the store: Introduce a loyalty program. How does it compare to past customer feedback?

[0528] In this way, the system of the present invention can effectively utilize past survey data and support users in proposing new services and events in real time.

[0529] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0530] Description of processing steps

[0531] Step 1:

[0532] Data collection

[0533] The server collects past survey information from the organization's HR database, general affairs database, and monitor database, using appropriate APIs and SQL queries. For example, it uses SQL queries to retrieve the necessary survey data and load it into the server's memory.

[0534] Input: Database connection information, query

[0535] Output: Collected survey data

[0536] The server retrieves the survey information from these databases and passes the collected data to the next processing step.

[0537] Step 2:

[0538] Data preprocessing

[0539] The server preprocesses the collected survey information. It cleans, categorizes, and tags the data, removes invalid responses and missing data, and then organizes the data by category.

[0540] Input: Collected survey data

[0541] Output: Preprocessed survey data

[0542] Specifically, we will use Python's pandas library to clean the data and then use scikit-learn's TfidfVectorizer to convert the data into numerical values.

[0543] Step 3:

[0544] Training machine learning models

[0545] The server uses the pre-processed data to train a machine learning model. Specifically, it uses a natural language processing model to learn patterns from past data.

[0546] Input: Preprocessed survey data

[0547] Output: Trained machine learning model

[0548] This process utilizes scikit-learn's KMeans clustering and the OpenAI GPT model. Training improves the accuracy of the model for predictions and analyses.

[0549] Step 4:

[0550] Entering new ideas

[0551] Users input ideas for new services using a smartphone app. This input is sent to the server in real time.

[0552] Input: New idea entered by the user

[0553] Output: Idea data sent to the server

[0554] The user enters new ideas in text format using the interface provided by the user terminal.

[0555] Step 5:

[0556] Searching and analyzing related data

[0557] The server searches and analyzes past survey data related to the new idea entered by the user. It uses natural language processing techniques to extract similar opinions and feedback.

[0558] Input: New ideas entered by users, past survey data

[0559] Output: Similar past feedback, analysis results

[0560] For example, you can use methods such as TF-IDF or cosine similarity to search for highly relevant historical data.

[0561] Step 6:

[0562] Simulation and result generation

[0563] The server uses a trained machine learning model to generate simulation results for new ideas. For example, if the idea is "introduce a loyalty program," it will generate specific proposals based on relevant past feedback.

[0564] Input: New idea, machine learning model, analysis results

[0565] Output: Simulation results, specific proposals

[0566] The AI ​​model is used to analyze the data and concretize the proposed content. An example of a prompt is: "A new idea for the store: Introduce a loyalty program. How does it compare to past customer feedback?"

[0567] Step 7:

[0568] Presentation of results

[0569] The generated simulation results are displayed on the user's device. Users can view specific suggestions and advice through a smartphone app.

[0570] Input: Simulation results

[0571] Output: Suggestions and advice displayed on the user's terminal.

[0572] The user's terminal displays the generated specific suggestions, allowing them to receive immediate feedback. As a result, users can obtain concrete information to quickly implement new, feasible services.

[0573] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0574] This invention relates to a system that utilizes past survey data to clarify the true needs of employees and customers, and then proposes new services and events based on those needs. Furthermore, by combining this with an emotion engine that recognizes user emotions, it is possible to provide even more appropriate suggestions.

[0575] 1. Data Collection

[0576] The server first collects past survey information from the HR database, general affairs database, and monitor database. This involves authenticating to each database, establishing a secure connection, and then executing the necessary SQL queries and API calls to retrieve the data.

[0577] 2. Data preprocessing

[0578] The server cleans the collected survey information. This involves removing invalid responses and missing data, and then classifying each survey by theme. Specifically, it divides them into categories such as "employee benefits," "company events," and "fitness apps." Next, this data is tagged and given relevant keywords to facilitate searching and analysis.

[0579] 3. Training machine learning models

[0580] The server trains a natural language processing model (e.g., ChatGPT) using pre-processed survey data. This model learns patterns from the data and uses them for future predictions and response generation. The model's performance is evaluated, and parameters are fine-tuned as needed to generate optimal responses.

[0581] 4. Introduction of an emotional engine

[0582] The server is equipped with an emotion engine to recognize user emotions. This emotion engine analyzes the text of new ideas entered by the user and determines the user's emotional state from the context. For example, it can identify emotions such as joy, excitement, and confusion. This allows the feedback and suggestions provided to be more accurate and aligned with the user's feelings.

[0583] 5. User input of new ideas

[0584] Users input ideas for new services and events through the system. They submit specific concepts and desired conditions in text format from their user terminals. This input data is sent to the server and simultaneously analyzed by an emotion engine.

[0585] 6. Searching for related data

[0586] The server searches the database for past survey information related to the new idea entered by the user. It efficiently extracts highly relevant data using pre-classified and tagged keywords and category information.

[0587] 7. Simulation and Result Generation

[0588] The server uses trained machine learning models and an emotion engine to generate simulation results based on the user's new idea. For example, if a new idea for a fitness app is entered, it will generate feedback such as "Many users want a plan that allows them to exercise effectively in a short amount of time," based on past feedback. This generated result is adjusted according to the user's emotional state.

[0589] 8. Presentation of Results

[0590] The server constructs the generated simulation results and displays them on the user's terminal. This allows the user to easily obtain specific suggestions and feedback. For example, specific advice such as, "A 10-minute high-intensity interval training plan combined with a game element where the user earns points to advance their level would likely be preferred," might be displayed. Based on sentiment recognition, the tone and details of the suggestions may also be adjusted.

[0591] Through this series of processes, the server can combine past survey data with the results of the emotion engine's analysis to propose new services and events that are better suited to user needs.

[0592] The following describes the processing flow.

[0593] Step 1: Data Collection

[0594] The server connects to the HR database, general affairs database, and monitor database to collect past survey information. The server authenticates to each database and uses authentication protocols to establish a secure connection. Then, it executes the necessary SQL queries and API calls to retrieve the target data.

[0595] Step 2: Data Preprocessing

[0596] The server cleans the survey information it has collected. Specifically, it filters out invalid responses and duplicate data, and imputes or removes missing values. Normalization is also performed to maintain data consistency, for example, by converting response items into a unified format.

[0597] The server then categorizes the cleaned data by theme. This categorization involves analyzing the survey content and assigning it to categories such as "employee benefits," "company events," and "fitness apps."

[0598] The server tags the classified data. This assigns relevant keywords, enabling efficient data retrieval and analysis.

[0599] Step 3: Training the machine learning model

[0600] The server uses pre-processed survey data to train a natural language processing model (e.g., ChatGPT). Training is performed by inputting a large amount of data into the model and having it analyze the results.

[0601] The server evaluates the model's performance and adjusts parameters as needed. For example, it measures the accuracy and relevance of the responses generated by the model and performs an iterative process to find the optimal parameter settings.

[0602] Step 4: Introducing the Emotional Engine

[0603] The server will implement an emotion engine to recognize the user's emotions. The emotion engine will analyze the content of the text entered by the user and determine the emotion from the context. The emotion engine will analyze based on multiple emotion categories (e.g., joy, excitement, confusion, dissatisfaction, etc.).

[0604] Specifically, the server analyzes user input and processes it to recognize the emotion of "joy" from text such as "I think this proposal is great."

[0605] Step 5: User input of new ideas

[0606] Users input ideas for new services or events via their devices. Through the system interface, users submit specific concepts and requirements in text format. This input data is transmitted to the server in real time.

[0607] Step 6: Search for related data

[0608] The server searches the database for past survey information related to the new idea entered by the user. It uses pre-classified and tagged keywords and category information to narrow down the data to the most relevant results.

[0609] For example, if a user enters "ideas for a new fitness app," the server will extract past survey information related to "fitness" and "apps."

[0610] Step 7: Simulation and result generation

[0611] The server uses a trained machine learning model and an emotion engine to generate simulation results based on the user's novel ideas. The server considers patterns obtained from survey data and the user's emotional state to generate responses.

[0612] For example, in response to a user's suggestion, it might generate feedback such as, "Many users want a plan that allows them to exercise effectively in a short amount of time," and if the emotion engine detects the emotion of "joy," it will present the result in a positive tone that takes that into account.

[0613] Step 8: Presentation of Results

[0614] The server constructs the generated simulation results and displays them on the user's terminal. This allows the user to receive specific and emotionally sensitive suggestions and feedback.

[0615] For example, a suggestion might appear stating, "A plan that combines a 10-minute high-intensity interval training session with a game element where users earn points to advance their progress level would likely be well-received." This suggestion is tailored to resonate with the user's emotions.

[0616] This entire process allows the system, combined with the emotion engine, to consider past survey data and the user's emotional state to suggest more appropriate and personalized new services and events.

[0617] (Example 2)

[0618] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0619] Modern companies and organizations are required to accurately understand the needs of their employees and customers and to provide new services and events based on those needs. However, conventional methods have made it difficult to efficiently utilize past survey data and grasp true needs. Furthermore, proposals often did not match the emotions of users, limiting their ability to increase satisfaction. To solve these problems, the present invention aims to provide a system that effectively utilizes past survey data and makes proposals that take user emotions into consideration.

[0620] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0621] In this invention, the server includes means for collecting past survey data, means for cleaning, classifying, and tagging the survey data, means for training a generative model using the cleaned and tagged survey data, means for searching for survey data related to a new idea entered by the user, means for generating simulation results based on the user's new idea using the generative model, means for analyzing the user's emotions and reflecting the analysis results in the feedback, and means for displaying the generated simulation results on the user terminal. This makes it possible to provide highly accurate suggestions based on past survey information and to provide customized feedback that takes into account the user's emotions.

[0622] "Survey data" refers to data that includes questionnaires and feedback information collected from users and customers.

[0623] "Cleaning" is a process that removes invalid responses and missing data from survey data to improve data quality.

[0624] "Classification" is the process of grouping cleaned survey data into specific themes or categories (e.g., employee benefits, company events).

[0625] "Tagging" is the process of assigning keywords related to classified survey data to facilitate searching and analysis.

[0626] A "generative model" is an algorithm or system that uses machine learning or natural language processing techniques to learn patterns from data and generate predictions or responses.

[0627] "New ideas" refer to concepts or suggestions for new services or events that users input into the system.

[0628] "Simulation results" refer to the feedback and suggestions generated as a response to analysis and predictions based on novel ideas using a generative model.

[0629] "Sentiment analysis" is a technology that analyzes the context of text data entered by a user to identify the user's emotional state (e.g., joy, excitement, confusion).

[0630] "Display means" refers to functions or devices for visually presenting generated simulation results and feedback to the user's terminal.

[0631] This invention relates to a system that enables companies and organizations to effectively utilize past survey data to propose new services and events that take user sentiment into consideration. This system functions through the cooperation of a server, terminals, and users.

[0632] Data collection

[0633] The server first collects historical survey data. This data is obtained from the organization's HR database, general affairs database, and other monitoring databases. The server authenticates to each database, establishes a secure connection, and then executes the necessary SQL queries and API calls to retrieve the data. The software used for this includes SQL database management systems (e.g., MySQL and PostgreSQL) and libraries for API calls (e.g., Axios and Fetch API).

[0634] Data preprocessing

[0635] The received survey data is cleaned by the server. Invalid responses and missing data are removed, and then the data is categorized by theme. For example, it may be categorized into "employee benefits," "company events," "fitness apps," etc., and relevant keywords are tagged. The software used for this process is a data mining tool (e.g., the pandas library in Python).

[0636] Training machine learning models

[0637] The server trains a generative model using pre-processed survey data. A natural language processing model (e.g., ChatGPT) is used here. The model learns patterns from the data and uses them for future predictions and response generation. A high-performance GPU (e.g., NVIDIA Tesla) is used to evaluate the model's performance and fine-tune its parameters.

[0638] Introducing an emotional engine

[0639] The server implements an emotion engine that recognizes the user's emotions. This emotion engine analyzes the text of new ideas entered by the user and determines the user's emotional state (e.g., joy, excitement, confusion) from the context. Emotion analysis libraries (e.g., Aylien or TextBlob) are used for emotion analysis.

[0640] User input of new ideas

[0641] Users input ideas for new services or events through the system. They enter specific concepts and desired conditions in text format from their device (e.g., a PC or smartphone) and send them to the server. This transmission uses the HTTPS protocol and encrypts the data. The server analyzes the received data using an emotion engine and evaluates the emotional state.

[0642] Searching for related data

[0643] The server searches past research data related to the new idea entered by the user. It extracts highly relevant data using pre-classified and tagged keywords and category information. A database management system (e.g., Elasticsearch) is used for this search.

[0644] Simulation and result generation

[0645] The server uses generative models and an emotion engine to generate simulation results based on the user's new idea. For example, if a new idea for a fitness app is entered, it will generate feedback such as "Many users want a plan that allows them to exercise effectively in a short amount of time," based on past feedback. This feedback is then refined based on the emotion analysis results.

[0646] Presentation of results

[0647] The server displays the generated simulation results on the user's terminal. The user can visually see specific suggestions and feedback. For example, specific advice might be given such as, "A plan incorporating a 10-minute high-intensity interval training session and a game element where the user earns points to advance their level would likely be well-received."

[0648] The following are specific examples of prompt statements:

[0649] Please use the following survey data to generate feedback on your ideas for a new fitness app.

[0650] Survey data 1: ...

[0651] Survey data 2: ...

[0652] ...(Enter multiple survey data)...

[0653] Ideas for a new fitness app:

[0654] 10-minute high-intensity interval training

[0655] A feature that allows users to earn points and advance their progress level.

[0656] Please provide feedback while taking into account the results of the sentiment analysis.

[0657] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0658] Step 1: Data Collection

[0659] The server collects historical survey data. This begins with authenticating to each database and establishing a secure connection. Specifically, authentication is performed using an API key, user ID, and password, and a secure connection is established using the SSL / TLS protocol. Next, the server retrieves data by executing the necessary SQL queries or API calls (e.g., SELECT FROM survey_data WHERE date > '2020-01-01'). This data is stored in secure internal storage. The input is the database authentication and connection information, and the output is the collected survey data. The specific operations involve executing API calls or SQL queries, retrieving data, and saving it.

[0660] Step 2: Data Preprocessing

[0661] The server cleans, classifies, and tags the collected survey data. First, it detects and removes invalid responses and missing data (e.g., empty fields or inconsistent responses). Next, it classifies the data into specific themes (e.g., employee benefits, company events) and tags relevant keywords using TF-IDF (Term Frequency-Inverse Document Frequency). The input is the collected survey data, and the output is the cleaned and tagged survey data. The specific operations are data cleaning, thematic classification, and tagging.

[0662] Step 3: Training the machine learning model

[0663] The server trains a generative AI model (e.g., ChatGPT) using preprocessed survey data. First, the preprocessed data is split into a training dataset and a test dataset. Typically, the training data:test data ratio is 8:2. Next, the model is trained using a high-performance GPU. The model's performance is evaluated, and evaluation metrics such as accuracy, recall, and F-score are checked. Parameters are fine-tuned as needed, and retraining is performed. The input is the cleaned and tagged survey data, and the output is the trained generative AI model.

[0664] Step 4: Introducing the Emotional Engine

[0665] The server implements an emotion engine to recognize user emotions. It installs an emotion engine (e.g., Emotion API), analyzes user-entered text data in real time, and determines the emotional state. The analysis results are saved to data storage and used in subsequent processes. Input is the user's text input, and output is the analyzed emotion information. The specific operations are text analysis and emotion identification.

[0666] Step 5: User input of new ideas

[0667] Users input new ideas using the system. Users enter specific concepts and desired conditions in text format from their devices (e.g., PCs, smartphones) and send them to the server. This data is transmitted encrypted (using the HTTPS protocol). The server passes the received data to an emotion engine, which analyzes the emotional state. The input is the user's new idea, and the output is the submitted idea and the analyzed emotional information.

[0668] Step 6: Search for related data

[0669] The server searches past survey data related to the user's new idea. It searches the database using pre-classified and tagged keywords and category information. Specifically, it executes SQL queries or Elasticsearch search queries to extract highly relevant data (e.g., SELECT FROM survey_data WHERE keywords LIKE '%fitness%'). The search results are temporarily stored in memory storage. The input is the user's new idea and keywords, and the output is the relevant survey data. The specific operations are executing search queries and extracting data.

[0670] Step 7: Simulation and result generation

[0671] The server generates simulation results based on the user's new ideas using a generative model and an emotion engine. The user's new ideas and related data are integrated and input into the generative AI model. Feedback output by the generative model is retrieved and fine-tuned using the emotion engine's analysis results. For example, it might output feedback such as, "Many users would like a plan that allows them to exercise effectively in a short amount of time." The input is the new ideas and related data, and the output is the simulation results.

[0672] Step 8: Presentation of Results

[0673] The server displays the generated simulation results on the user's terminal. First, the generated feedback is converted into HTML or JSON format and sent to the user's terminal. The user's terminal displays the results on an interface, allowing the user to visually confirm specific suggestions and feedback. For example, feedback such as "A 10-minute high-intensity interval training plan combined with a game element where the user earns points to advance their level would likely be preferred" might be displayed. The inputs are the simulation results and analysis results, and the output is the feedback presented to the user.

[0674] (Application Example 2)

[0675] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0676] Currently, systems exist that utilize past survey data to understand users' true needs and propose new services and events based on those needs. However, because these types of systems do not take into account the emotional state of the user, the proposed content does not always meet the user's expectations. In particular, in physical stores, there is a demand for service proposals that reflect the customer's real-time emotional state, but conventional systems have had difficulty achieving this. The present invention aims to solve these problems and provide a system that makes more appropriate service proposals while taking into account the emotional state of the customer.

[0677] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0678] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, means for adjusting the generated simulation results using an emotion engine that analyzes the user's emotions, and means for displaying the generated simulation results on the user terminal. This makes it possible to propose services that take into account the user's emotional state.

[0679] "Survey information" refers to response data obtained from past surveys and feedback.

[0680] "Means of collection" refers to mechanisms for collecting survey information through databases or APIs.

[0681] "Cleaning" is the process of improving data quality by removing invalid responses and missing data.

[0682] "Classification and tagging" is the process of organizing cleaned data by theme and assigning relevant keywords.

[0683] A "machine learning model" is an algorithm that learns patterns from large amounts of data and uses them to make predictions and generate responses.

[0684] "Training methods" refer to methods for training machine learning models using pre-processed data.

[0685] "New ideas" refer to suggestions for new services or events that users input into the system.

[0686] "Search methods" refer to efficient methods for extracting past survey information related to the entered idea from the database.

[0687] "Simulation results" refer to the predictions and feedback generated using a machine learning model based on the proposal.

[0688] An "emotion engine" is a technology that analyzes the user's emotional state from their text input and adjusts its response accordingly.

[0689] A "user terminal" is a device (such as a smartphone or tablet) used to access the system and review the proposed content.

[0690] The system implementing this invention collects past survey information and, based on that, makes service suggestions related to the user's new ideas. Furthermore, it analyzes the user's emotional state using an emotion engine and adjusts the suggestions accordingly to provide more appropriate feedback.

[0691] First, the server collects past survey information from databases. This data is obtained from databases such as the human resources database, general affairs database, and monitor database. The collected data is cleaned, invalid responses and missing data are removed, and the data is categorized and tagged by theme. Programming languages ​​such as Python and SQL are used for this process.

[0692] Next, the server uses the cleaned and tagged survey data to train a machine learning model. Specifically, it uses a generative AI model with natural language processing techniques (e.g., GPT-3.5-turbo). This model learns from a large amount of text data and performs new text generation and predictions for new user ideas.

[0693] When a user inputs a new idea into the system, it searches the database for past survey information related to that new idea. The retrieved data is then fed into a machine learning model, and simulation results are generated.

[0694] Furthermore, the server uses an emotion engine to analyze the user's emotional state. This emotion engine analyzes the text data entered by the user and determines the emotional state (joy, excitement, confusion, etc.) from the context. Based on the analyzed emotional state, the simulation results are adjusted. The emotion engine uses, for example, an emotion analysis library (such as Microsoft's Azure Text Analytics API).

[0695] Finally, the server displays the adjusted simulation results on the user's terminal. Users can review the suggestions via devices such as smartphones and tablets. A dedicated application is installed on the terminal, allowing them to proceed with planning new services and events based on the displayed feedback and suggestions.

[0696] As a concrete example, when a user enters the following prompt, the optimal suggestion is generated:

[0697] Example of a prompt:

[0698] I would like to propose a new fitness program. Specifically, I want to create a plan that allows for effective exercise in a short amount of time, and incorporate a game element where users can earn points and advance through the program.

[0699] This allows the server to accurately reflect user needs and provide appropriate service suggestions based on their emotional state through the processes described above.

[0700] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0701] Step 1:

[0702] The server collects past survey information. Specifically, it authenticates to the HR database, general affairs database, and monitor database, establishes a secure connection, and then executes the necessary SQL queries and API calls. This collects the survey information. The input is "survey information from the database," and the output is "collected survey information."

[0703] Step 2:

[0704] The server cleans the collected survey information, removing invalid responses and missing data to improve data quality. Next, it categorizes the data by theme and tags it with relevant keywords. The input is the "collected survey information," and the output is the "cleaned, categorized, and tagged survey information."

[0705] Step 3:

[0706] The server trains a machine learning model using cleaned and tagged survey information. Specifically, it uses a generative AI model based on natural language processing (GPT-3.5-turbo) to learn patterns from a large amount of text data. The input is "cleaned, classified, and tagged survey information," and the output is the "trained machine learning model."

[0707] Step 4:

[0708] Users input new ideas into the system. They submit specific concepts and desired conditions to the system in text format from their terminal. The input is "the text of the user's new idea," and the output is "the user's new idea sent to the server."

[0709] Step 5:

[0710] The server searches the database for past survey information related to the new idea entered by the user. It efficiently extracts highly relevant data using pre-classified and tagged keywords and category information. The input is the "user's new idea" and the "database," and the output is a "list of relevant survey information."

[0711] Step 6:

[0712] The server generates simulation results using a trained machine learning model. Based on the user's new idea and related survey information, it generates future predictions and responses. The inputs are "the user's new idea," "a list of related survey information," and "the trained machine learning model," and the output is "the generated simulation results."

[0713] Step 7:

[0714] The server uses an emotion engine to analyze the user's emotional state. It analyzes the text of the user's new idea and determines the emotional state (joy, excitement, confusion, etc.) from the context. The input is "the text of the user's new idea," and the output is "the analysis result of the user's emotional state."

[0715] Step 8:

[0716] The server adjusts the generated simulation results based on the analysis results of the emotion engine. This optimizes the suggestions according to the user's emotional state. The inputs are the "generated simulation results" and the "analysis results of the user's emotional state," and the output is the "adjusted simulation results."

[0717] Step 9:

[0718] The server displays the adjusted simulation results on the user's terminal. The user can review the suggestions through the terminal and provide further feedback. The input is the "adjusted simulation results," and the output is the "feedback and suggestions displayed to the user."

[0719] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0720] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0721] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0722] [Third Embodiment]

[0723] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0724] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0725] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0726] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0727] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0728] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0729] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0730] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0731] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0732] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0733] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0734] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0735] This invention relates to a system that uses past survey data to clarify the true needs of employees and customers, and based on that, proposes new services and events. Specific embodiments of this system are described below.

[0736] 1. Data Collection

[0737] The server first collects past survey information. This includes data collection from the human resources database, general affairs database, and monitor database. The server connects to these databases through an appropriate authentication process and retrieves the necessary survey information.

[0738] 2. Data preprocessing

[0739] The server then preprocesses the collected survey information. Specifically, it performs processes such as cleaning, classification, and tagging. Cleaning involves removing invalid responses and missing data. Classification organizes the data by theme based on the survey content, dividing it into categories such as "employee benefits," "company events," and "fitness apps." Tagging involves assigning keywords related to each data item to make it easier to search.

[0740] 3. Training machine learning models

[0741] Using the pre-processed survey data, the server trains a machine learning model, specifically a natural language processing model (such as ChatGPT). This model learns patterns from past survey data and is adjusted to generate appropriate responses to future inputs.

[0742] 4. User input of new ideas

[0743] Users use the system to input ideas for new services or events. They access the system from their terminals and provide ideas in text format. This input data is sent to the server.

[0744] 5. Searching for related data

[0745] The server searches its database for past survey information related to the new idea entered by the user. This allows it to collect similar feedback and opinions and use them for analysis.

[0746] 6. Simulation and Result Generation

[0747] The server uses a trained machine learning model to generate simulation results based on the user's new ideas. For example, if an idea for a specific fitness app is entered, the server analyzes past feedback related to this idea and generates specific advice and suggestions.

[0748] 7. Presentation of Results

[0749] The server constructs the generated simulation results and displays them on the user's terminal. This allows users to easily obtain specific suggestions and advice.

[0750] Specific example

[0751] For example, if a user submits a request to "plan a new sports event," the server searches and analyzes past survey data on sports events. Suppose the results show that "many employees want weekend events, and soccer and basketball are popular." Based on this, the system generates suggestions such as "hold an in-house soccer tournament on Saturday mornings, making it an event that families can also participate in," and displays them on the user's terminal.

[0752] Through these specific means, the system of the present invention can effectively utilize past survey data to propose new services and events based on the needs of employees and customers.

[0753] The following describes the processing flow.

[0754] Step 1: Data Collection

[0755] The server connects to the HR database, general affairs database, and monitor database to collect past survey information. It authenticates with the target databases and establishes secure access. It then executes necessary SQL queries and API calls to retrieve the survey data.

[0756] Step 2: Data Preprocessing

[0757] The server cleans the survey data it has collected. Specifically, it filters out invalid responses and duplicate data, and handles missing values.

[0758] The server sorts the cleaned data by theme. Based on the survey responses, it categorizes the data into categories such as "employee benefits," "company events," and "fitness apps."

[0759] The server tags the classified data and assigns relevant keywords to facilitate searching and analysis. This enables quick searching of data related to specific themes.

[0760] Step 3: Training the machine learning model

[0761] The server uses pre-processed survey data to train a machine learning model (e.g., ChatGPT). This process involves inputting a large amount of data to allow the model to learn patterns.

[0762] The server evaluates the model's performance and fine-tunes the parameters as needed. This ensures that the final model has the ability to generate the optimal response to the user's new ideas.

[0763] Step 4: User input of new ideas

[0764] Users input ideas for new services or events into the system. They provide specific concepts and desired conditions in text format from their user terminals. This input data is then sent to the server.

[0765] Step 5: Search for related data

[0766] The server searches the database for past survey information related to the user's new idea. It extracts highly relevant data using pre-tagged keywords and category information. This allows for efficient collection of past feedback and opinions.

[0767] Step 6: Simulation and result generation

[0768] The server uses a trained machine learning model to generate simulation results based on the user's new idea. The model uses patterns learned from past data to generate predictions and suggestions for the input idea.

[0769] For example, when a new fitness app idea is presented, specific feedback such as, "Many users would like a plan that allows them to exercise effectively in a short amount of time," can be provided.

[0770] Step 7: Presentation of Results

[0771] The server compiles the generated simulation results and displays them on the user's terminal. This allows the user to receive specific suggestions and feedback.

[0772] For example, it might display specific advice such as, "A plan that combines a 10-minute high-intensity interval training session with a game element where users earn points to advance their progress level would likely be well-received."

[0773] This series of steps enables the server to effectively utilize past survey data and create a system that generates concrete and effective suggestions for users' new ideas.

[0774] (Example 1)

[0775] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0776] Traditional systems have made it difficult to effectively utilize past survey data to propose new services and events based on employee and customer needs. Furthermore, there is a lack of means to quickly search and analyze past feedback related to specific new ideas, making it difficult to provide concrete advice to users' new ideas. Solving these problems is essential.

[0777] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0778] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, and means for displaying the generated simulation results on the user's terminal. This makes it possible to quickly provide specific suggestions and advice for the user's new ideas by utilizing past survey information.

[0779] "Past survey information" refers to data based on surveys previously conducted within the organization, including feedback and opinions collected from employees and customers.

[0780] "Means of collection" refers to a combination of hardware and software used to retrieve necessary information from databases or APIs.

[0781] "Cleaning" is a process that improves data quality by removing invalid responses and missing data from the data.

[0782] "Classification" is the process of organizing survey data by theme or category and grouping it according to a specific purpose.

[0783] "Tagging" is a process that improves the searchability of data by assigning keywords related to data items.

[0784] A "machine learning model" is an algorithm that learns patterns and rules from large amounts of data and generates appropriate responses or predictions for future inputs.

[0785] "Training methods" refer to processing equipment and software used to train a machine learning model using collected and pre-processed data.

[0786] "New ideas submitted by users" refer to suggestions for new services or events provided in text format by users of the system.

[0787] "Search methods" refer to search engines and algorithms that retrieve relevant historical data from a database based on ideas entered by the user.

[0788] "Simulation results" refer to specific suggestions and advice for a user's new ideas, generated using a machine learning model.

[0789] A "user terminal" is a computer or mobile device used by a user to access a system and input and view information.

[0790] This invention relates to a system for clarifying the true needs of employees and customers using past survey data, and for proposing new services and events based on those needs. Specific embodiments will be described below.

[0791] First, the server collects past survey information. This process begins by retrieving data from databases such as MySQL or PostgreSQL. The server connects to these databases using appropriate authentication credentials and executes SQL queries to collect the necessary survey information. This collected data is stored in CSV or JSON format.

[0792] Next, the collected survey information is preprocessed. The server uses Python and the Pandas library to clean the data, removing invalid responses and missing data. After that, data classification and tagging are performed. This classifies the survey content by theme, organizing it into categories such as "employee benefits" and "company events." Tagging assigns keywords related to the data items, improving searchability.

[0793] Using the preprocessed data, the server trains a machine learning model. This process involves training a natural language processing model (e.g., ChatGPT) using libraries such as PyTorch or TensorFlow. The trained model is then optimized to generate appropriate responses and suggestions based on future inputs.

[0794] When users submit ideas for new services or events, they access the system using a web browser or mobile app. Users enter their ideas in text format, and this data is sent to the server as an HTTP request. For example, a request might read, "I want to plan a new sports event."

[0795] The server searches past survey data related to the user's new idea. This search process uses search engines such as Elasticsearch and Solr. The search results extract the feedback and opinions most relevant to the user's idea and provide them for analysis.

[0796] Based on the retrieved data, the server uses a trained machine learning model to generate simulation results. For example, for the idea of ​​"planning a new sports event," relevant feedback might be obtained such as "many employees want weekend events, and soccer and basketball are popular." Based on this information, the system generates specific suggestions such as "hold an in-house soccer tournament on Saturday mornings, making it an event that families can also participate in."

[0797] The generated simulation results are sent from the server to the user's terminal, allowing the user to view suggestions and advice.

[0798] As a concrete example, the following prompt sentences could be input to the generation AI model:

[0799] "We'd like to plan a new sports event. We're looking for an event that employees can enjoy with their families on the weekend. Please provide suggestions based on past survey data."

[0800] In this way, a system is built that can effectively utilize past survey data to propose new services and events based on the needs of employees and customers.

[0801] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0802] Step 1:

[0803] The server collects past survey information. Specifically, the server connects to databases such as MySQL or PostgreSQL using appropriate authentication credentials and executes SQL queries. The collected data is stored in CSV or JSON format.

[0804] Input: Database connection information and SQL query

[0805] Output: Survey data in CSV or JSON format.

[0806] Specific operation: The server executes an SQL query such as SELECT FROM survey_table and saves the retrieved data as a file.

[0807] Step 2:

[0808] The server preprocesses the collected survey information. This step uses Python and the Pandas library to clean, classify, and tag the data.

[0809] Input: Survey data in CSV or JSON format.

[0810] Output: Preprocessed data frame

[0811] Specific operation: The server reads the data using pd.read_csv("Survey Data.csv"), removes invalid responses and missing data using df.dropna(), and categorizes the data by theme. For example, categorization is performed using df['Category'] = df['Content'].apply(...). Furthermore, keywords are added using df['Tags'] = df['Content'].apply(...).

[0812] Step 3:

[0813] The server trains machine learning models using pre-processed data. It uses PyTorch or TensorFlow to train natural language processing models.

[0814] Input: Preprocessed data frame

[0815] Output: Trained machine learning model

[0816] Specific operation: The server uses the training data to execute model.train(). For example, when using the ChatGPT natural language processing model, the server loads the model and tokenizer using `from transformers import GPT2LMHeadModel, GPT2Tokenizer` and then runs training. After training, the model is saved using model.save_pretrained("model path").

[0817] Step 4:

[0818] Users enter ideas for new services or events. Users access the system via a web browser or mobile app and enter their ideas in text format.

[0819] Input: User's new idea (text format)

[0820] Output: Submitted idea data

[0821] Specific operation: The user enters "I want to plan a new sports event" into a web form and clicks the submit button. This data is then sent to the server as an HTTP request.

[0822] Step 5:

[0823] The server searches past survey information related to the new idea entered by the user. It uses search engines such as Elasticsearch and Solr to extract relevant data.

[0824] Input: New user ideas, search queries such as Elasticsearch.

[0825] Output: Related survey data

[0826] Specific operation: The server executes es.search(index="Survey", body={"query": {"match": {"Content": "User Ideas"}}}) and temporarily saves the results.

[0827] Step 6:

[0828] The server generates simulation results using a trained machine learning model. This generates suggestions and advice based on the user's new ideas.

[0829] Input: New user ideas, relevant survey data, trained machine learning model

[0830] Output: Simulation results (text format)

[0831] Specific operation: The server loads the trained model, inputs the user's ideas and related survey data, and executes model.generate(input_ids). The generated results are constructed in text format.

[0832] Step 7:

[0833] The server displays the generated simulation results on the user's terminal. The user can then view specific suggestions and advice on their terminal.

[0834] Input: Simulation results (text format)

[0835] Output: Suggestions and advice displayed on the user's terminal.

[0836] Specific operation: The server sends the results to the user's terminal in HTML or JSON format. For example, execute response.json({"Suggestion": Suggestion result}). The user's terminal displays these results on the screen.

[0837] (Application Example 1)

[0838] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0839] The objective of this invention is to effectively utilize past survey information to provide proposals for new services and events based on the true needs of customers and employees. Conventional systems simply collect survey data for feedback, but struggle to apply that data effectively. Furthermore, they lacked a mechanism to provide real-time, effective feedback when users entered new ideas. Therefore, there were limitations to improving customer satisfaction and operational efficiency.

[0840] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0841] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, means for displaying the generated simulation results on a user terminal, means for providing a user interface for entering new service ideas, and means for collecting and analyzing feedback in real time through the user interface. This makes it possible to provide users with effective new service and event suggestions based on past survey information in real time.

[0842] Definitions of important words

[0843] "Past survey information" refers to data that includes feedback, opinions, and evaluations collected from customers and employees in the past.

[0844] "Cleaning" is the process of removing invalid responses and missing data from collected data.

[0845] "Classification" is the process of organizing data according to specific themes or categories.

[0846] "Tagging" is the process of assigning keywords related to each data item to make it easier to search.

[0847] A "machine learning model" is a system that uses algorithms to learn patterns and rules from data and perform predictions and analyses.

[0848] "New ideas submitted by users" refer to new service or event concepts and plans proposed by users.

[0849] "Simulation results" refer to predictions and specific proposals based on new ideas, using machine learning models.

[0850] A "user terminal" is a device used by a user to input information or check results.

[0851] A "user interface" refers to the screens and input methods that a user uses to interact with a system.

[0852] "Means for collecting and analyzing feedback in real time" refers to a function that instantly collects data in response to user input and actions, analyzes it, and generates results.

[0853] Preparation of a detailed statement

[0854] This invention provides a specific embodiment of a system for analyzing survey information and proposing new services. The central elements of the system are a server and a user terminal.

[0855] Server Processing

[0856] 1. Data collection:

[0857] The server collects past survey information from various databases within the organization (e.g., HR database, general affairs database, monitor database, etc.). The data collection process uses APIs and SQL queries to appropriately retrieve the necessary information.

[0858] 2. Data preprocessing:

[0859] The collected survey data is cleaned by the server, removing invalid responses and missing data. The data is then categorized into specific themes and categories, and each data item is tagged to facilitate searching. Specifically, this preprocessing is performed using the Python pandas library and scikit-learn.

[0860] 3. Training machine learning models:

[0861] The preprocessed data is trained using a machine learning model (for example, a model based on natural language processing). This model learns past data patterns and becomes capable of generating appropriate responses to future novel ideas. The libraries used include scikit-learn and the OpenAI GPT model.

[0862] 4. Analysis of new ideas:

[0863] When a user inputs a new service idea into the system, the server searches and analyzes past survey data related to that idea. This utilizes natural language processing technology to analyze past feedback and opinion trends.

[0864] 5. Simulation and result generation:

[0865] The server uses a trained machine learning model to generate simulation results based on the user's new idea. For example, if a new loyalty program is proposed, it analyzes past feedback related to this idea and generates specific advice and suggestions.

[0866] 6. Presentation of results:

[0867] The generated simulation results are displayed on the user's terminal. This allows the user to easily obtain specific suggestions and advice.

[0868] User-side interface

[0869] 1. Enter your new service idea:

[0870] Users utilize an interface to input ideas for new services into the system. This interface is provided as a smartphone app, allowing users to input ideas in text format.

[0871] 2. Collection and analysis of real-time feedback:

[0872] When a new service idea is submitted, feedback is collected in real time and immediately analyzed by the server. This allows users to receive feedback instantly.

[0873] Specific example

[0874] For example, if a store manager inputs a new service idea, such as "I want to introduce a loyalty program," into the system, the server searches and analyzes past customer survey data. Suppose the system finds feedback indicating that "many customers want a points system." Based on this, the system generates a concrete proposal for "introducing a loyalty program with a points system" and displays it on the user's terminal.

[0875] Examples of prompts to input into a generative AI model:

[0876] A new idea for the store: Introduce a loyalty program. How does it compare to past customer feedback?

[0877] In this way, the system of the present invention can effectively utilize past survey data and support users in proposing new services and events in real time.

[0878] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0879] Description of processing steps

[0880] Step 1:

[0881] Data collection

[0882] The server collects past survey information from the organization's HR database, general affairs database, and monitor database, using appropriate APIs and SQL queries. For example, it uses SQL queries to retrieve the necessary survey data and load it into the server's memory.

[0883] Input: Database connection information, query

[0884] Output: Collected survey data

[0885] The server retrieves the survey information from these databases and passes the collected data to the next processing step.

[0886] Step 2:

[0887] Data preprocessing

[0888] The server preprocesses the collected survey information. It cleans, categorizes, and tags the data, removes invalid responses and missing data, and then organizes the data by category.

[0889] Input: Collected survey data

[0890] Output: Preprocessed survey data

[0891] Specifically, we will use Python's pandas library to clean the data and then use scikit-learn's TfidfVectorizer to convert the data into numerical values.

[0892] Step 3:

[0893] Training machine learning models

[0894] The server uses the pre-processed data to train a machine learning model. Specifically, it uses a natural language processing model to learn patterns from past data.

[0895] Input: Preprocessed survey data

[0896] Output: Trained machine learning model

[0897] This process utilizes scikit-learn's KMeans clustering and the OpenAI GPT model. Training improves the accuracy of the model for predictions and analyses.

[0898] Step 4:

[0899] Entering new ideas

[0900] Users input ideas for new services using a smartphone app. This input is sent to the server in real time.

[0901] Input: New idea entered by the user

[0902] Output: Idea data sent to the server

[0903] The user enters new ideas in text format using the interface provided by the user terminal.

[0904] Step 5:

[0905] Searching and analyzing related data

[0906] The server searches and analyzes past survey data related to the new idea entered by the user. It uses natural language processing techniques to extract similar opinions and feedback.

[0907] Input: New ideas entered by users, past survey data

[0908] Output: Similar past feedback, analysis results

[0909] For example, you can use methods such as TF-IDF or cosine similarity to search for highly relevant historical data.

[0910] Step 6:

[0911] Simulation and result generation

[0912] The server uses a trained machine learning model to generate simulation results for new ideas. For example, if the idea is "introduce a loyalty program," it will generate specific proposals based on relevant past feedback.

[0913] Input: New idea, machine learning model, analysis results

[0914] Output: Simulation results, specific proposals

[0915] The AI ​​model is used to analyze the data and concretize the proposed content. An example of a prompt is: "A new idea for the store: Introduce a loyalty program. How does it compare to past customer feedback?"

[0916] Step 7:

[0917] Presentation of results

[0918] The generated simulation results are displayed on the user's device. Users can view specific suggestions and advice through a smartphone app.

[0919] Input: Simulation results

[0920] Output: Suggestions and advice displayed on the user's terminal.

[0921] The user's terminal displays the generated specific suggestions, allowing them to receive immediate feedback. As a result, users can obtain concrete information to quickly implement new, feasible services.

[0922] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0923] This invention relates to a system that utilizes past survey data to clarify the true needs of employees and customers, and then proposes new services and events based on those needs. Furthermore, by combining this with an emotion engine that recognizes user emotions, it is possible to provide even more appropriate suggestions.

[0924] 1. Data Collection

[0925] The server first collects past survey information from the HR database, general affairs database, and monitor database. This involves authenticating to each database, establishing a secure connection, and then executing the necessary SQL queries and API calls to retrieve the data.

[0926] 2. Data preprocessing

[0927] The server cleans the collected survey information. This involves removing invalid responses and missing data, and then classifying each survey by theme. Specifically, it divides them into categories such as "employee benefits," "company events," and "fitness apps." Next, this data is tagged and given relevant keywords to facilitate searching and analysis.

[0928] 3. Training machine learning models

[0929] The server trains a natural language processing model (e.g., ChatGPT) using pre-processed survey data. This model learns patterns from the data and uses them for future predictions and response generation. The model's performance is evaluated, and parameters are fine-tuned as needed to generate optimal responses.

[0930] 4. Introduction of an emotional engine

[0931] The server is equipped with an emotion engine to recognize user emotions. This emotion engine analyzes the text of new ideas entered by the user and determines the user's emotional state from the context. For example, it can identify emotions such as joy, excitement, and confusion. This allows the feedback and suggestions provided to be more accurate and aligned with the user's feelings.

[0932] 5. User input of new ideas

[0933] Users input ideas for new services and events through the system. They submit specific concepts and desired conditions in text format from their user terminals. This input data is sent to the server and simultaneously analyzed by an emotion engine.

[0934] 6. Searching for related data

[0935] The server searches the database for past survey information related to the new idea entered by the user. It efficiently extracts highly relevant data using pre-classified and tagged keywords and category information.

[0936] 7. Simulation and Result Generation

[0937] The server uses trained machine learning models and an emotion engine to generate simulation results based on the user's new idea. For example, if a new idea for a fitness app is entered, it will generate feedback such as "Many users want a plan that allows them to exercise effectively in a short amount of time," based on past feedback. This generated result is adjusted according to the user's emotional state.

[0938] 8. Presentation of Results

[0939] The server constructs the generated simulation results and displays them on the user's terminal. This allows the user to easily obtain specific suggestions and feedback. For example, specific advice such as, "A 10-minute high-intensity interval training plan combined with a game element where the user earns points to advance their level would likely be preferred," might be displayed. Based on sentiment recognition, the tone and details of the suggestions may also be adjusted.

[0940] Through this series of processes, the server can combine past survey data with the results of the emotion engine's analysis to propose new services and events that are better suited to user needs.

[0941] The following describes the processing flow.

[0942] Step 1: Data Collection

[0943] The server connects to the HR database, general affairs database, and monitor database to collect past survey information. The server authenticates to each database and uses authentication protocols to establish a secure connection. Then, it executes the necessary SQL queries and API calls to retrieve the target data.

[0944] Step 2: Data Preprocessing

[0945] The server cleans the survey information it has collected. Specifically, it filters out invalid responses and duplicate data, and imputes or removes missing values. Normalization is also performed to maintain data consistency, for example, by converting response items into a unified format.

[0946] The server then categorizes the cleaned data by theme. This categorization involves analyzing the survey content and assigning it to categories such as "employee benefits," "company events," and "fitness apps."

[0947] The server tags the classified data. This assigns relevant keywords, enabling efficient data retrieval and analysis.

[0948] Step 3: Training the machine learning model

[0949] The server uses pre-processed survey data to train a natural language processing model (e.g., ChatGPT). Training is performed by inputting a large amount of data into the model and having it analyze the results.

[0950] The server evaluates the model's performance and adjusts parameters as needed. For example, it measures the accuracy and relevance of the responses generated by the model and performs an iterative process to find the optimal parameter settings.

[0951] Step 4: Introducing the Emotional Engine

[0952] The server will implement an emotion engine to recognize the user's emotions. The emotion engine will analyze the content of the text entered by the user and determine the emotion from the context. The emotion engine will analyze based on multiple emotion categories (e.g., joy, excitement, confusion, dissatisfaction, etc.).

[0953] Specifically, the server analyzes user input and processes it to recognize the emotion of "joy" from text such as "I think this proposal is great."

[0954] Step 5: User input of new ideas

[0955] Users input ideas for new services or events via their devices. Through the system interface, users submit specific concepts and requirements in text format. This input data is transmitted to the server in real time.

[0956] Step 6: Search for related data

[0957] The server searches the database for past survey information related to the new idea entered by the user. It uses pre-classified and tagged keywords and category information to narrow down the data to the most relevant results.

[0958] For example, if a user enters "ideas for a new fitness app," the server will extract past survey information related to "fitness" and "apps."

[0959] Step 7: Simulation and result generation

[0960] The server uses a trained machine learning model and an emotion engine to generate simulation results based on the user's novel ideas. The server considers patterns obtained from survey data and the user's emotional state to generate responses.

[0961] For example, in response to a user's suggestion, it might generate feedback such as, "Many users want a plan that allows them to exercise effectively in a short amount of time," and if the emotion engine detects the emotion of "joy," it will present the result in a positive tone that takes that into account.

[0962] Step 8: Presentation of Results

[0963] The server constructs the generated simulation results and displays them on the user's terminal. This allows the user to receive specific and emotionally sensitive suggestions and feedback.

[0964] For example, a suggestion might appear stating, "A plan that combines a 10-minute high-intensity interval training session with a game element where users earn points to advance their progress level would likely be well-received." This suggestion is tailored to resonate with the user's emotions.

[0965] This entire process allows the system, combined with the emotion engine, to consider past survey data and the user's emotional state to suggest more appropriate and personalized new services and events.

[0966] (Example 2)

[0967] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0968] Modern companies and organizations are required to accurately understand the needs of their employees and customers and to provide new services and events based on those needs. However, conventional methods have made it difficult to efficiently utilize past survey data and grasp true needs. Furthermore, proposals often did not match the emotions of users, limiting their ability to increase satisfaction. To solve these problems, the present invention aims to provide a system that effectively utilizes past survey data and makes proposals that take user emotions into consideration.

[0969] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0970] In this invention, the server includes means for collecting past survey data, means for cleaning, classifying, and tagging the survey data, means for training a generative model using the cleaned and tagged survey data, means for searching for survey data related to a new idea entered by the user, means for generating simulation results based on the user's new idea using the generative model, means for analyzing the user's emotions and reflecting the analysis results in the feedback, and means for displaying the generated simulation results on the user terminal. This makes it possible to provide highly accurate suggestions based on past survey information and to provide customized feedback that takes into account the user's emotions.

[0971] "Survey data" refers to data that includes questionnaires and feedback information collected from users and customers.

[0972] "Cleaning" is a process that removes invalid responses and missing data from survey data to improve data quality.

[0973] "Classification" is the process of grouping cleaned survey data into specific themes or categories (e.g., employee benefits, company events).

[0974] "Tagging" is the process of assigning keywords related to classified survey data to facilitate searching and analysis.

[0975] A "generative model" is an algorithm or system that uses machine learning or natural language processing techniques to learn patterns from data and generate predictions or responses.

[0976] "New ideas" refer to concepts or suggestions for new services or events that users input into the system.

[0977] "Simulation results" refer to the feedback and suggestions generated as a response to analysis and predictions based on novel ideas using a generative model.

[0978] "Sentiment analysis" is a technology that analyzes the context of text data entered by a user to identify the user's emotional state (e.g., joy, excitement, confusion).

[0979] "Display means" refers to functions or devices for visually presenting generated simulation results and feedback to the user's terminal.

[0980] This invention relates to a system that enables companies and organizations to effectively utilize past survey data to propose new services and events that take user sentiment into consideration. This system functions through the cooperation of a server, terminals, and users.

[0981] Data collection

[0982] The server first collects historical survey data. This data is obtained from the organization's HR database, general affairs database, and other monitoring databases. The server authenticates to each database, establishes a secure connection, and then executes the necessary SQL queries and API calls to retrieve the data. The software used for this includes SQL database management systems (e.g., MySQL and PostgreSQL) and libraries for API calls (e.g., Axios and Fetch API).

[0983] Data preprocessing

[0984] The received survey data is cleaned by the server. Invalid responses and missing data are removed, and then the data is categorized by theme. For example, it may be categorized into "employee benefits," "company events," "fitness apps," etc., and relevant keywords are tagged. The software used for this process is a data mining tool (e.g., the pandas library in Python).

[0985] Training machine learning models

[0986] The server trains a generative model using pre-processed survey data. A natural language processing model (e.g., ChatGPT) is used here. The model learns patterns from the data and uses them for future predictions and response generation. A high-performance GPU (e.g., NVIDIA Tesla) is used to evaluate the model's performance and fine-tune its parameters.

[0987] Introducing an emotional engine

[0988] The server implements an emotion engine that recognizes the user's emotions. This emotion engine analyzes the text of new ideas entered by the user and determines the user's emotional state (e.g., joy, excitement, confusion) from the context. Emotion analysis libraries (e.g., Aylien or TextBlob) are used for emotion analysis.

[0989] User input of new ideas

[0990] Users input ideas for new services or events through the system. They enter specific concepts and desired conditions in text format from their device (e.g., a PC or smartphone) and send them to the server. This transmission uses the HTTPS protocol and encrypts the data. The server analyzes the received data using an emotion engine and evaluates the emotional state.

[0991] Searching for related data

[0992] The server searches past research data related to the new idea entered by the user. It extracts highly relevant data using pre-classified and tagged keywords and category information. A database management system (e.g., Elasticsearch) is used for this search.

[0993] Simulation and result generation

[0994] The server uses generative models and an emotion engine to generate simulation results based on the user's new idea. For example, if a new idea for a fitness app is entered, it will generate feedback such as "Many users want a plan that allows them to exercise effectively in a short amount of time," based on past feedback. This feedback is then refined based on the emotion analysis results.

[0995] Presentation of results

[0996] The server displays the generated simulation results on the user's terminal. The user can visually see specific suggestions and feedback. For example, specific advice might be given such as, "A plan incorporating a 10-minute high-intensity interval training session and a game element where the user earns points to advance their level would likely be well-received."

[0997] The following are specific examples of prompt statements:

[0998] Please use the following survey data to generate feedback on your ideas for a new fitness app.

[0999] Survey data 1: ...

[1000] Survey data 2: ...

[1001] ...(Enter multiple survey data)...

[1002] Ideas for a new fitness app:

[1003] 10-minute high-intensity interval training

[1004] A feature that allows users to earn points and advance their progress level.

[1005] Please provide feedback while taking into account the results of the sentiment analysis.

[1006] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1007] Step 1: Data Collection

[1008] The server collects historical survey data. This begins with authenticating to each database and establishing a secure connection. Specifically, authentication is performed using an API key, user ID, and password, and a secure connection is established using the SSL / TLS protocol. Next, the server retrieves data by executing the necessary SQL queries or API calls (e.g., SELECT FROM survey_data WHERE date > '2020-01-01'). This data is stored in secure internal storage. The input is the database authentication and connection information, and the output is the collected survey data. The specific operations involve executing API calls or SQL queries, retrieving data, and saving it.

[1009] Step 2: Data Preprocessing

[1010] The server cleans, classifies, and tags the collected survey data. First, it detects and removes invalid responses and missing data (e.g., empty fields or inconsistent responses). Next, it classifies the data into specific themes (e.g., employee benefits, company events) and tags relevant keywords using TF-IDF (Term Frequency-Inverse Document Frequency). The input is the collected survey data, and the output is the cleaned and tagged survey data. The specific operations are data cleaning, thematic classification, and tagging.

[1011] Step 3: Training the machine learning model

[1012] The server trains a generative AI model (e.g., ChatGPT) using preprocessed survey data. First, the preprocessed data is split into a training dataset and a test dataset. Typically, the training data:test data ratio is 8:2. Next, the model is trained using a high-performance GPU. The model's performance is evaluated, and evaluation metrics such as accuracy, recall, and F-score are checked. Parameters are fine-tuned as needed, and retraining is performed. The input is the cleaned and tagged survey data, and the output is the trained generative AI model.

[1013] Step 4: Introducing the Emotional Engine

[1014] The server implements an emotion engine to recognize user emotions. It installs an emotion engine (e.g., Emotion API), analyzes user-entered text data in real time, and determines the emotional state. The analysis results are saved to data storage and used in subsequent processes. Input is the user's text input, and output is the analyzed emotion information. The specific operations are text analysis and emotion identification.

[1015] Step 5: User input of new ideas

[1016] Users input new ideas using the system. Users enter specific concepts and desired conditions in text format from their devices (e.g., PCs, smartphones) and send them to the server. This data is transmitted encrypted (using the HTTPS protocol). The server passes the received data to an emotion engine, which analyzes the emotional state. The input is the user's new idea, and the output is the submitted idea and the analyzed emotional information.

[1017] Step 6: Search for related data

[1018] The server searches past survey data related to the user's new idea. It searches the database using pre-classified and tagged keywords and category information. Specifically, it executes SQL queries or Elasticsearch search queries to extract highly relevant data (e.g., SELECT FROM survey_data WHERE keywords LIKE '%fitness%'). The search results are temporarily stored in memory storage. The input is the user's new idea and keywords, and the output is the relevant survey data. The specific operations are executing search queries and extracting data.

[1019] Step 7: Simulation and result generation

[1020] The server generates simulation results based on the user's new ideas using a generative model and an emotion engine. The user's new ideas and related data are integrated and input into the generative AI model. Feedback output by the generative model is retrieved and fine-tuned using the emotion engine's analysis results. For example, it might output feedback such as, "Many users would like a plan that allows them to exercise effectively in a short amount of time." The input is the new ideas and related data, and the output is the simulation results.

[1021] Step 8: Presentation of Results

[1022] The server displays the generated simulation results on the user's terminal. First, the generated feedback is converted into HTML or JSON format and sent to the user's terminal. The user's terminal displays the results on an interface, allowing the user to visually confirm specific suggestions and feedback. For example, feedback such as "A 10-minute high-intensity interval training plan combined with a game element where the user earns points to advance their level would likely be preferred" might be displayed. The inputs are the simulation results and analysis results, and the output is the feedback presented to the user.

[1023] (Application Example 2)

[1024] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1025] Currently, systems exist that utilize past survey data to understand users' true needs and propose new services and events based on those needs. However, because these types of systems do not take into account the emotional state of the user, the proposed content does not always meet the user's expectations. In particular, in physical stores, there is a demand for service proposals that reflect the customer's real-time emotional state, but conventional systems have had difficulty achieving this. The present invention aims to solve these problems and provide a system that makes more appropriate service proposals while taking into account the emotional state of the customer.

[1026] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1027] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, means for adjusting the generated simulation results using an emotion engine that analyzes the user's emotions, and means for displaying the generated simulation results on the user terminal. This makes it possible to propose services that take into account the user's emotional state.

[1028] "Survey information" refers to response data obtained from past surveys and feedback.

[1029] "Means of collection" refers to mechanisms for collecting survey information through databases or APIs.

[1030] "Cleaning" is the process of improving data quality by removing invalid responses and missing data.

[1031] "Classification and tagging" is the process of organizing cleaned data by theme and assigning relevant keywords.

[1032] A "machine learning model" is an algorithm that learns patterns from large amounts of data and uses them to make predictions and generate responses.

[1033] "Training methods" refer to methods for training machine learning models using pre-processed data.

[1034] "New ideas" refer to suggestions for new services or events that users input into the system.

[1035] "Search methods" refer to efficient methods for extracting past survey information related to the entered idea from the database.

[1036] "Simulation results" refer to the predictions and feedback generated using a machine learning model based on the proposal.

[1037] An "emotion engine" is a technology that analyzes the user's emotional state from their text input and adjusts its response accordingly.

[1038] A "user terminal" is a device (such as a smartphone or tablet) used to access the system and review the proposed content.

[1039] The system implementing this invention collects past survey information and, based on that, makes service suggestions related to the user's new ideas. Furthermore, it analyzes the user's emotional state using an emotion engine and adjusts the suggestions accordingly to provide more appropriate feedback.

[1040] First, the server collects past survey information from databases. This data is obtained from databases such as the human resources database, general affairs database, and monitor database. The collected data is cleaned, invalid responses and missing data are removed, and the data is categorized and tagged by theme. Programming languages ​​such as Python and SQL are used for this process.

[1041] Next, the server uses the cleaned and tagged survey data to train a machine learning model. Specifically, it uses a generative AI model with natural language processing techniques (e.g., GPT-3.5-turbo). This model learns from a large amount of text data and performs new text generation and predictions for new user ideas.

[1042] When a user inputs a new idea into the system, it searches the database for past survey information related to that new idea. The retrieved data is then fed into a machine learning model, and simulation results are generated.

[1043] Furthermore, the server uses an emotion engine to analyze the user's emotional state. This emotion engine analyzes the text data entered by the user and determines the emotional state (joy, excitement, confusion, etc.) from the context. Based on the analyzed emotional state, the simulation results are adjusted. The emotion engine uses, for example, an emotion analysis library (such as Microsoft's Azure Text Analytics API).

[1044] Finally, the server displays the adjusted simulation results on the user's terminal. Users can review the suggestions via devices such as smartphones and tablets. A dedicated application is installed on the terminal, allowing them to proceed with planning new services and events based on the displayed feedback and suggestions.

[1045] As a concrete example, when a user enters the following prompt, the optimal suggestion is generated:

[1046] Example of a prompt:

[1047] I would like to propose a new fitness program. Specifically, I want to create a plan that allows for effective exercise in a short amount of time, and incorporate a game element where users can earn points and advance through the program.

[1048] This allows the server to accurately reflect user needs and provide appropriate service suggestions based on their emotional state through the processes described above.

[1049] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1050] Step 1:

[1051] The server collects past survey information. Specifically, it authenticates to the HR database, general affairs database, and monitor database, establishes a secure connection, and then executes the necessary SQL queries and API calls. This collects the survey information. The input is "survey information from the database," and the output is "collected survey information."

[1052] Step 2:

[1053] The server cleans the collected survey information, removing invalid responses and missing data to improve data quality. Next, it categorizes the data by theme and tags it with relevant keywords. The input is the "collected survey information," and the output is the "cleaned, categorized, and tagged survey information."

[1054] Step 3:

[1055] The server trains a machine learning model using cleaned and tagged survey information. Specifically, it uses a generative AI model based on natural language processing (GPT-3.5-turbo) to learn patterns from a large amount of text data. The input is "cleaned, classified, and tagged survey information," and the output is the "trained machine learning model."

[1056] Step 4:

[1057] Users input new ideas into the system. They submit specific concepts and desired conditions to the system in text format from their terminal. The input is "the text of the user's new idea," and the output is "the user's new idea sent to the server."

[1058] Step 5:

[1059] The server searches the database for past survey information related to the new idea entered by the user. It efficiently extracts highly relevant data using pre-classified and tagged keywords and category information. The input is the "user's new idea" and the "database," and the output is a "list of relevant survey information."

[1060] Step 6:

[1061] The server generates simulation results using a trained machine learning model. Based on the user's new idea and related survey information, it generates future predictions and responses. The inputs are "the user's new idea," "a list of related survey information," and "the trained machine learning model," and the output is "the generated simulation results."

[1062] Step 7:

[1063] The server uses an emotion engine to analyze the user's emotional state. It analyzes the text of the user's new idea and determines the emotional state (joy, excitement, confusion, etc.) from the context. The input is "the text of the user's new idea," and the output is "the analysis result of the user's emotional state."

[1064] Step 8:

[1065] The server adjusts the generated simulation results based on the analysis results of the emotion engine. This optimizes the suggestions according to the user's emotional state. The inputs are the "generated simulation results" and the "analysis results of the user's emotional state," and the output is the "adjusted simulation results."

[1066] Step 9:

[1067] The server displays the adjusted simulation results on the user's terminal. The user can review the suggestions through the terminal and provide further feedback. The input is the "adjusted simulation results," and the output is the "feedback and suggestions displayed to the user."

[1068] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset 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 audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1069] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1070] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[1071] [Fourth Embodiment]

[1072] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1073] As shown in Figure 7, 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.

[1074] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1075] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[1076] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1077] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1078] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1079] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[1080] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1081] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1082] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1083] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1084] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1085] This invention relates to a system that uses past survey data to clarify the true needs of employees and customers, and based on that, proposes new services and events. Specific embodiments of this system are described below.

[1086] 1. Data Collection

[1087] The server first collects past survey information. This includes data collection from the human resources database, general affairs database, and monitor database. The server connects to these databases through an appropriate authentication process and retrieves the necessary survey information.

[1088] 2. Data preprocessing

[1089] The server then preprocesses the collected survey information. Specifically, it performs processes such as cleaning, classification, and tagging. Cleaning involves removing invalid responses and missing data. Classification organizes the data by theme based on the survey content, dividing it into categories such as "employee benefits," "company events," and "fitness apps." Tagging involves assigning keywords related to each data item to make it easier to search.

[1090] 3. Training machine learning models

[1091] Using the pre-processed survey data, the server trains a machine learning model, specifically a natural language processing model (such as ChatGPT). This model learns patterns from past survey data and is adjusted to generate appropriate responses to future inputs.

[1092] 4. User input of new ideas

[1093] Users use the system to input ideas for new services or events. They access the system from their terminals and provide ideas in text format. This input data is sent to the server.

[1094] 5. Searching for related data

[1095] The server searches its database for past survey information related to the new idea entered by the user. This allows it to collect similar feedback and opinions and use them for analysis.

[1096] 6. Simulation and Result Generation

[1097] The server uses a trained machine learning model to generate simulation results based on the user's new ideas. For example, if an idea for a specific fitness app is entered, the server analyzes past feedback related to this idea and generates specific advice and suggestions.

[1098] 7. Presentation of Results

[1099] The server constructs the generated simulation results and displays them on the user's terminal. This allows users to easily obtain specific suggestions and advice.

[1100] Specific example

[1101] For example, if a user submits a request to "plan a new sports event," the server searches and analyzes past survey data on sports events. Suppose the results show that "many employees want weekend events, and soccer and basketball are popular." Based on this, the system generates suggestions such as "hold an in-house soccer tournament on Saturday mornings, making it an event that families can also participate in," and displays them on the user's terminal.

[1102] Through these specific means, the system of the present invention can effectively utilize past survey data to propose new services and events based on the needs of employees and customers.

[1103] The following describes the processing flow.

[1104] Step 1: Data Collection

[1105] The server connects to the HR database, general affairs database, and monitor database to collect past survey information. It authenticates with the target databases and establishes secure access. It then executes necessary SQL queries and API calls to retrieve the survey data.

[1106] Step 2: Data Preprocessing

[1107] The server cleans the survey data it has collected. Specifically, it filters out invalid responses and duplicate data, and handles missing values.

[1108] The server sorts the cleaned data by theme. Based on the survey responses, it categorizes the data into categories such as "employee benefits," "company events," and "fitness apps."

[1109] The server tags the classified data and assigns relevant keywords to facilitate searching and analysis. This enables quick searching of data related to specific themes.

[1110] Step 3: Training the machine learning model

[1111] The server uses pre-processed survey data to train a machine learning model (e.g., ChatGPT). This process involves inputting a large amount of data to allow the model to learn patterns.

[1112] The server evaluates the model's performance and fine-tunes the parameters as needed. This ensures that the final model has the ability to generate the optimal response to the user's new ideas.

[1113] Step 4: User input of new ideas

[1114] Users input ideas for new services or events into the system. They provide specific concepts and desired conditions in text format from their user terminals. This input data is then sent to the server.

[1115] Step 5: Search for related data

[1116] The server searches the database for past survey information related to the user's new idea. It extracts highly relevant data using pre-tagged keywords and category information. This allows for efficient collection of past feedback and opinions.

[1117] Step 6: Simulation and result generation

[1118] The server uses a trained machine learning model to generate simulation results based on the user's new idea. The model uses patterns learned from past data to generate predictions and suggestions for the input idea.

[1119] For example, when a new fitness app idea is presented, specific feedback such as, "Many users would like a plan that allows them to exercise effectively in a short amount of time," can be provided.

[1120] Step 7: Presentation of Results

[1121] The server compiles the generated simulation results and displays them on the user's terminal. This allows the user to receive specific suggestions and feedback.

[1122] For example, it might display specific advice such as, "A plan that combines a 10-minute high-intensity interval training session with a game element where users earn points to advance their progress level would likely be well-received."

[1123] This series of steps enables the server to effectively utilize past survey data and create a system that generates concrete and effective suggestions for users' new ideas.

[1124] (Example 1)

[1125] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1126] Traditional systems have made it difficult to effectively utilize past survey data to propose new services and events based on employee and customer needs. Furthermore, there is a lack of means to quickly search and analyze past feedback related to specific new ideas, making it difficult to provide concrete advice to users' new ideas. Solving these problems is essential.

[1127] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1128] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, and means for displaying the generated simulation results on the user's terminal. This makes it possible to quickly provide specific suggestions and advice for the user's new ideas by utilizing past survey information.

[1129] "Past survey information" refers to data based on surveys previously conducted within the organization, including feedback and opinions collected from employees and customers.

[1130] "Means of collection" refers to a combination of hardware and software used to retrieve necessary information from databases or APIs.

[1131] "Cleaning" is a process that improves data quality by removing invalid responses and missing data from the data.

[1132] "Classification" is the process of organizing survey data by theme or category and grouping it according to a specific purpose.

[1133] "Tagging" is a process that improves the searchability of data by assigning keywords related to data items.

[1134] A "machine learning model" is an algorithm that learns patterns and rules from large amounts of data and generates appropriate responses or predictions for future inputs.

[1135] "Training methods" refer to processing equipment and software used to train a machine learning model using collected and pre-processed data.

[1136] "New ideas submitted by users" refer to suggestions for new services or events provided in text format by users of the system.

[1137] "Search methods" refer to search engines and algorithms that retrieve relevant historical data from a database based on ideas entered by the user.

[1138] "Simulation results" refer to specific suggestions and advice for a user's new ideas, generated using a machine learning model.

[1139] A "user terminal" is a computer or mobile device used by a user to access a system and input and view information.

[1140] This invention relates to a system for clarifying the true needs of employees and customers using past survey data, and for proposing new services and events based on those needs. Specific embodiments will be described below.

[1141] First, the server collects past survey information. This process begins by retrieving data from databases such as MySQL or PostgreSQL. The server connects to these databases using appropriate authentication credentials and executes SQL queries to collect the necessary survey information. This collected data is stored in CSV or JSON format.

[1142] Next, the collected survey information is preprocessed. The server uses Python and the Pandas library to clean the data, removing invalid responses and missing data. After that, data classification and tagging are performed. This classifies the survey content by theme, organizing it into categories such as "employee benefits" and "company events." Tagging assigns keywords related to the data items, improving searchability.

[1143] Using the preprocessed data, the server trains a machine learning model. This process involves training a natural language processing model (e.g., ChatGPT) using libraries such as PyTorch or TensorFlow. The trained model is then optimized to generate appropriate responses and suggestions based on future inputs.

[1144] When users submit ideas for new services or events, they access the system using a web browser or mobile app. Users enter their ideas in text format, and this data is sent to the server as an HTTP request. For example, a request might read, "I want to plan a new sports event."

[1145] The server searches past survey data related to the user's new idea. This search process uses search engines such as Elasticsearch and Solr. The search results extract the feedback and opinions most relevant to the user's idea and provide them for analysis.

[1146] Based on the retrieved data, the server uses a trained machine learning model to generate simulation results. For example, for the idea of ​​"planning a new sports event," relevant feedback might be obtained such as "many employees want weekend events, and soccer and basketball are popular." Based on this information, the system generates specific suggestions such as "hold an in-house soccer tournament on Saturday mornings, making it an event that families can also participate in."

[1147] The generated simulation results are sent from the server to the user's terminal, allowing the user to view suggestions and advice.

[1148] As a concrete example, the following prompt sentences could be input to the generation AI model:

[1149] "We'd like to plan a new sports event. We're looking for an event that employees can enjoy with their families on the weekend. Please provide suggestions based on past survey data."

[1150] In this way, a system is built that can effectively utilize past survey data to propose new services and events based on the needs of employees and customers.

[1151] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1152] Step 1:

[1153] The server collects past survey information. Specifically, the server connects to databases such as MySQL or PostgreSQL using appropriate authentication credentials and executes SQL queries. The collected data is stored in CSV or JSON format.

[1154] Input: Database connection information and SQL query

[1155] Output: Survey data in CSV or JSON format.

[1156] Specific operation: The server executes an SQL query such as SELECT FROM survey_table and saves the retrieved data as a file.

[1157] Step 2:

[1158] The server preprocesses the collected survey information. This step uses Python and the Pandas library to clean, classify, and tag the data.

[1159] Input: Survey data in CSV or JSON format.

[1160] Output: Preprocessed data frame

[1161] Specific operation: The server reads the data using pd.read_csv("Survey Data.csv"), removes invalid responses and missing data using df.dropna(), and categorizes the data by theme. For example, categorization is performed using df['Category'] = df['Content'].apply(...). Furthermore, keywords are added using df['Tags'] = df['Content'].apply(...).

[1162] Step 3:

[1163] The server trains machine learning models using pre-processed data. It uses PyTorch or TensorFlow to train natural language processing models.

[1164] Input: Preprocessed data frame

[1165] Output: Trained machine learning model

[1166] Specific operation: The server uses the training data to execute model.train(). For example, when using the ChatGPT natural language processing model, the server loads the model and tokenizer using `from transformers import GPT2LMHeadModel, GPT2Tokenizer` and then runs training. After training, the model is saved using model.save_pretrained("model path").

[1167] Step 4:

[1168] Users enter ideas for new services or events. Users access the system via a web browser or mobile app and enter their ideas in text format.

[1169] Input: User's new idea (text format)

[1170] Output: Submitted idea data

[1171] Specific operation: The user enters "I want to plan a new sports event" into a web form and clicks the submit button. This data is then sent to the server as an HTTP request.

[1172] Step 5:

[1173] The server searches past survey information related to the new idea entered by the user. It uses search engines such as Elasticsearch and Solr to extract relevant data.

[1174] Input: New user ideas, search queries such as Elasticsearch.

[1175] Output: Related survey data

[1176] Specific operation: The server executes es.search(index="Survey", body={"query": {"match": {"Content": "User Ideas"}}}) and temporarily saves the results.

[1177] Step 6:

[1178] The server generates simulation results using a trained machine learning model. This generates suggestions and advice based on the user's new ideas.

[1179] Input: New user ideas, relevant survey data, trained machine learning model

[1180] Output: Simulation results (text format)

[1181] Specific operation: The server loads the trained model, inputs the user's ideas and related survey data, and executes model.generate(input_ids). The generated results are constructed in text format.

[1182] Step 7:

[1183] The server displays the generated simulation results on the user's terminal. The user can then view specific suggestions and advice on their terminal.

[1184] Input: Simulation results (text format)

[1185] Output: Suggestions and advice displayed on the user's terminal.

[1186] Specific operation: The server sends the results to the user's terminal in HTML or JSON format. For example, execute response.json({"Suggestion": Suggestion result}). The user's terminal displays these results on the screen.

[1187] (Application Example 1)

[1188] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1189] The objective of this invention is to effectively utilize past survey information to provide proposals for new services and events based on the true needs of customers and employees. Conventional systems simply collect survey data for feedback, but struggle to apply that data effectively. Furthermore, they lacked a mechanism to provide real-time, effective feedback when users entered new ideas. Therefore, there were limitations to improving customer satisfaction and operational efficiency.

[1190] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1191] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, means for displaying the generated simulation results on a user terminal, means for providing a user interface for entering new service ideas, and means for collecting and analyzing feedback in real time through the user interface. This makes it possible to provide users with effective new service and event suggestions based on past survey information in real time.

[1192] Definitions of important words

[1193] "Past survey information" refers to data that includes feedback, opinions, and evaluations collected from customers and employees in the past.

[1194] "Cleaning" is the process of removing invalid responses and missing data from collected data.

[1195] "Classification" is the process of organizing data according to specific themes or categories.

[1196] "Tagging" is the process of assigning keywords related to each data item to make it easier to search.

[1197] A "machine learning model" is a system that uses algorithms to learn patterns and rules from data and perform predictions and analyses.

[1198] "New ideas submitted by users" refer to new service or event concepts and plans proposed by users.

[1199] "Simulation results" refer to predictions and specific proposals based on new ideas, using machine learning models.

[1200] A "user terminal" is a device used by a user to input information or check results.

[1201] A "user interface" refers to the screens and input methods that a user uses to interact with a system.

[1202] "Means for collecting and analyzing feedback in real time" refers to a function that instantly collects data in response to user input and actions, analyzes it, and generates results.

[1203] Preparation of a detailed statement

[1204] This invention provides a specific embodiment of a system for analyzing survey information and proposing new services. The central elements of the system are a server and a user terminal.

[1205] Server Processing

[1206] 1. Data collection:

[1207] The server collects past survey information from various databases within the organization (e.g., HR database, general affairs database, monitor database, etc.). The data collection process uses APIs and SQL queries to appropriately retrieve the necessary information.

[1208] 2. Data preprocessing:

[1209] The collected survey data is cleaned by the server, removing invalid responses and missing data. The data is then categorized into specific themes and categories, and each data item is tagged to facilitate searching. Specifically, this preprocessing is performed using the Python pandas library and scikit-learn.

[1210] 3. Training machine learning models:

[1211] The preprocessed data is trained using a machine learning model (for example, a model based on natural language processing). This model learns past data patterns and becomes capable of generating appropriate responses to future novel ideas. The libraries used include scikit-learn and the OpenAI GPT model.

[1212] 4. Analysis of new ideas:

[1213] When a user inputs a new service idea into the system, the server searches and analyzes past survey data related to that idea. This utilizes natural language processing technology to analyze past feedback and opinion trends.

[1214] 5. Simulation and result generation:

[1215] The server uses a trained machine learning model to generate simulation results based on the user's new idea. For example, if a new loyalty program is proposed, it analyzes past feedback related to this idea and generates specific advice and suggestions.

[1216] 6. Presentation of results:

[1217] The generated simulation results are displayed on the user's terminal. This allows the user to easily obtain specific suggestions and advice.

[1218] User-side interface

[1219] 1. Enter your new service idea:

[1220] Users utilize an interface to input ideas for new services into the system. This interface is provided as a smartphone app, allowing users to input ideas in text format.

[1221] 2. Collection and analysis of real-time feedback:

[1222] When a new service idea is submitted, feedback is collected in real time and immediately analyzed by the server. This allows users to receive feedback instantly.

[1223] Specific example

[1224] For example, if a store manager inputs a new service idea, such as "I want to introduce a loyalty program," into the system, the server searches and analyzes past customer survey data. Suppose the system finds feedback indicating that "many customers want a points system." Based on this, the system generates a concrete proposal for "introducing a loyalty program with a points system" and displays it on the user's terminal.

[1225] Examples of prompts to input into a generative AI model:

[1226] A new idea for the store: Introduce a loyalty program. How does it compare to past customer feedback?

[1227] In this way, the system of the present invention can effectively utilize past survey data and support users in proposing new services and events in real time.

[1228] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1229] Description of processing steps

[1230] Step 1:

[1231] Data collection

[1232] The server collects past survey information from the organization's HR database, general affairs database, and monitor database, using appropriate APIs and SQL queries. For example, it uses SQL queries to retrieve the necessary survey data and load it into the server's memory.

[1233] Input: Database connection information, query

[1234] Output: Collected survey data

[1235] The server retrieves the survey information from these databases and passes the collected data to the next processing step.

[1236] Step 2:

[1237] Data preprocessing

[1238] The server preprocesses the collected survey information. It cleans, categorizes, and tags the data, removes invalid responses and missing data, and then organizes the data by category.

[1239] Input: Collected survey data

[1240] Output: Preprocessed survey data

[1241] Specifically, we will use Python's pandas library to clean the data and then use scikit-learn's TfidfVectorizer to convert the data into numerical values.

[1242] Step 3:

[1243] Training machine learning models

[1244] The server uses the pre-processed data to train a machine learning model. Specifically, it uses a natural language processing model to learn patterns from past data.

[1245] Input: Preprocessed survey data

[1246] Output: Trained machine learning model

[1247] This process utilizes scikit-learn's KMeans clustering and the OpenAI GPT model. Training improves the accuracy of the model for predictions and analyses.

[1248] Step 4:

[1249] Entering new ideas

[1250] Users input ideas for new services using a smartphone app. This input is sent to the server in real time.

[1251] Input: New idea entered by the user

[1252] Output: Idea data sent to the server

[1253] The user enters new ideas in text format using the interface provided by the user terminal.

[1254] Step 5:

[1255] Searching and analyzing related data

[1256] The server searches and analyzes past survey data related to the new idea entered by the user. It uses natural language processing techniques to extract similar opinions and feedback.

[1257] Input: New ideas entered by users, past survey data

[1258] Output: Similar past feedback, analysis results

[1259] For example, you can use methods such as TF-IDF or cosine similarity to search for highly relevant historical data.

[1260] Step 6:

[1261] Simulation and result generation

[1262] The server uses a trained machine learning model to generate simulation results for new ideas. For example, if the idea is "introduce a loyalty program," it will generate specific proposals based on relevant past feedback.

[1263] Input: New idea, machine learning model, analysis results

[1264] Output: Simulation results, specific proposals

[1265] The AI ​​model is used to analyze the data and concretize the proposed content. An example of a prompt is: "A new idea for the store: Introduce a loyalty program. How does it compare to past customer feedback?"

[1266] Step 7:

[1267] Presentation of results

[1268] The generated simulation results are displayed on the user's device. Users can view specific suggestions and advice through a smartphone app.

[1269] Input: Simulation results

[1270] Output: Suggestions and advice displayed on the user's terminal.

[1271] The user's terminal displays the generated specific suggestions, allowing them to receive immediate feedback. As a result, users can obtain concrete information to quickly implement new, feasible services.

[1272] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1273] This invention relates to a system that utilizes past survey data to clarify the true needs of employees and customers, and then proposes new services and events based on those needs. Furthermore, by combining this with an emotion engine that recognizes user emotions, it is possible to provide even more appropriate suggestions.

[1274] 1. Data Collection

[1275] The server first collects past survey information from the HR database, general affairs database, and monitor database. This involves authenticating to each database, establishing a secure connection, and then executing the necessary SQL queries and API calls to retrieve the data.

[1276] 2. Data preprocessing

[1277] The server cleans the collected survey information. This involves removing invalid responses and missing data, and then classifying each survey by theme. Specifically, it divides them into categories such as "employee benefits," "company events," and "fitness apps." Next, this data is tagged and given relevant keywords to facilitate searching and analysis.

[1278] 3. Training machine learning models

[1279] The server trains a natural language processing model (e.g., ChatGPT) using pre-processed survey data. This model learns patterns from the data and uses them for future predictions and response generation. The model's performance is evaluated, and parameters are fine-tuned as needed to generate optimal responses.

[1280] 4. Introduction of an emotional engine

[1281] The server is equipped with an emotion engine to recognize user emotions. This emotion engine analyzes the text of new ideas entered by the user and determines the user's emotional state from the context. For example, it can identify emotions such as joy, excitement, and confusion. This allows the feedback and suggestions provided to be more accurate and aligned with the user's feelings.

[1282] 5. User input of new ideas

[1283] Users input ideas for new services and events through the system. They submit specific concepts and desired conditions in text format from their user terminals. This input data is sent to the server and simultaneously analyzed by an emotion engine.

[1284] 6. Searching for related data

[1285] The server searches the database for past survey information related to the new idea entered by the user. It efficiently extracts highly relevant data using pre-classified and tagged keywords and category information.

[1286] 7. Simulation and Result Generation

[1287] The server uses trained machine learning models and an emotion engine to generate simulation results based on the user's new idea. For example, if a new idea for a fitness app is entered, it will generate feedback such as "Many users want a plan that allows them to exercise effectively in a short amount of time," based on past feedback. This generated result is adjusted according to the user's emotional state.

[1288] 8. Presentation of Results

[1289] The server constructs the generated simulation results and displays them on the user's terminal. This allows the user to easily obtain specific suggestions and feedback. For example, specific advice such as, "A 10-minute high-intensity interval training plan combined with a game element where the user earns points to advance their level would likely be preferred," might be displayed. Based on sentiment recognition, the tone and details of the suggestions may also be adjusted.

[1290] Through this series of processes, the server can combine past survey data with the results of the emotion engine's analysis to propose new services and events that are better suited to user needs.

[1291] The following describes the processing flow.

[1292] Step 1: Data Collection

[1293] The server connects to the HR database, general affairs database, and monitor database to collect past survey information. The server authenticates to each database and uses authentication protocols to establish a secure connection. Then, it executes the necessary SQL queries and API calls to retrieve the target data.

[1294] Step 2: Data Preprocessing

[1295] The server cleans the survey information it has collected. Specifically, it filters out invalid responses and duplicate data, and imputes or removes missing values. Normalization is also performed to maintain data consistency, for example, by converting response items into a unified format.

[1296] The server then categorizes the cleaned data by theme. This categorization involves analyzing the survey content and assigning it to categories such as "employee benefits," "company events," and "fitness apps."

[1297] The server tags the classified data. This assigns relevant keywords, enabling efficient data retrieval and analysis.

[1298] Step 3: Training the machine learning model

[1299] The server uses pre-processed survey data to train a natural language processing model (e.g., ChatGPT). Training is performed by inputting a large amount of data into the model and having it analyze the results.

[1300] The server evaluates the model's performance and adjusts parameters as needed. For example, it measures the accuracy and relevance of the responses generated by the model and performs an iterative process to find the optimal parameter settings.

[1301] Step 4: Introducing the Emotional Engine

[1302] The server will implement an emotion engine to recognize the user's emotions. The emotion engine will analyze the content of the text entered by the user and determine the emotion from the context. The emotion engine will analyze based on multiple emotion categories (e.g., joy, excitement, confusion, dissatisfaction, etc.).

[1303] Specifically, the server analyzes user input and processes it to recognize the emotion of "joy" from text such as "I think this proposal is great."

[1304] Step 5: User input of new ideas

[1305] Users input ideas for new services or events via their devices. Through the system interface, users submit specific concepts and requirements in text format. This input data is transmitted to the server in real time.

[1306] Step 6: Search for related data

[1307] The server searches the database for past survey information related to the new idea entered by the user. It uses pre-classified and tagged keywords and category information to narrow down the data to the most relevant results.

[1308] For example, if a user enters "ideas for a new fitness app," the server will extract past survey information related to "fitness" and "apps."

[1309] Step 7: Simulation and result generation

[1310] The server uses a trained machine learning model and an emotion engine to generate simulation results based on the user's novel ideas. The server considers patterns obtained from survey data and the user's emotional state to generate responses.

[1311] For example, in response to a user's suggestion, it might generate feedback such as, "Many users want a plan that allows them to exercise effectively in a short amount of time," and if the emotion engine detects the emotion of "joy," it will present the result in a positive tone that takes that into account.

[1312] Step 8: Presentation of Results

[1313] The server constructs the generated simulation results and displays them on the user's terminal. This allows the user to receive specific and emotionally sensitive suggestions and feedback.

[1314] For example, a suggestion might appear stating, "A plan that combines a 10-minute high-intensity interval training session with a game element where users earn points to advance their progress level would likely be well-received." This suggestion is tailored to resonate with the user's emotions.

[1315] This entire process allows the system, combined with the emotion engine, to consider past survey data and the user's emotional state to suggest more appropriate and personalized new services and events.

[1316] (Example 2)

[1317] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1318] Modern companies and organizations are required to accurately understand the needs of their employees and customers and to provide new services and events based on those needs. However, conventional methods have made it difficult to efficiently utilize past survey data and grasp true needs. Furthermore, proposals often did not match the emotions of users, limiting their ability to increase satisfaction. To solve these problems, the present invention aims to provide a system that effectively utilizes past survey data and makes proposals that take user emotions into consideration.

[1319] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1320] In this invention, the server includes means for collecting past survey data, means for cleaning, classifying, and tagging the survey data, means for training a generative model using the cleaned and tagged survey data, means for searching for survey data related to a new idea entered by the user, means for generating simulation results based on the user's new idea using the generative model, means for analyzing the user's emotions and reflecting the analysis results in the feedback, and means for displaying the generated simulation results on the user terminal. This makes it possible to provide highly accurate suggestions based on past survey information and to provide customized feedback that takes into account the user's emotions.

[1321] "Survey data" refers to data that includes questionnaires and feedback information collected from users and customers.

[1322] "Cleaning" is a process that removes invalid responses and missing data from survey data to improve data quality.

[1323] "Classification" is the process of grouping cleaned survey data into specific themes or categories (e.g., employee benefits, company events).

[1324] "Tagging" is the process of assigning keywords related to classified survey data to facilitate searching and analysis.

[1325] A "generative model" is an algorithm or system that uses machine learning or natural language processing techniques to learn patterns from data and generate predictions or responses.

[1326] "New ideas" refer to concepts or suggestions for new services or events that users input into the system.

[1327] "Simulation results" refer to the feedback and suggestions generated as a response to analysis and predictions based on novel ideas using a generative model.

[1328] "Sentiment analysis" is a technology that analyzes the context of text data entered by a user to identify the user's emotional state (e.g., joy, excitement, confusion).

[1329] "Display means" refers to functions or devices for visually presenting generated simulation results and feedback to the user's terminal.

[1330] This invention relates to a system that enables companies and organizations to effectively utilize past survey data to propose new services and events that take user sentiment into consideration. This system functions through the cooperation of a server, terminals, and users.

[1331] Data collection

[1332] The server first collects historical survey data. This data is obtained from the organization's HR database, general affairs database, and other monitoring databases. The server authenticates to each database, establishes a secure connection, and then executes the necessary SQL queries and API calls to retrieve the data. The software used for this includes SQL database management systems (e.g., MySQL and PostgreSQL) and libraries for API calls (e.g., Axios and Fetch API).

[1333] Data preprocessing

[1334] The received survey data is cleaned by the server. Invalid responses and missing data are removed, and then the data is categorized by theme. For example, it may be categorized into "employee benefits," "company events," "fitness apps," etc., and relevant keywords are tagged. The software used for this process is a data mining tool (e.g., the pandas library in Python).

[1335] Training machine learning models

[1336] The server trains a generative model using pre-processed survey data. A natural language processing model (e.g., ChatGPT) is used here. The model learns patterns from the data and uses them for future predictions and response generation. A high-performance GPU (e.g., NVIDIA Tesla) is used to evaluate the model's performance and fine-tune its parameters.

[1337] Introducing an emotional engine

[1338] The server implements an emotion engine that recognizes the user's emotions. This emotion engine analyzes the text of new ideas entered by the user and determines the user's emotional state (e.g., joy, excitement, confusion) from the context. Emotion analysis libraries (e.g., Aylien or TextBlob) are used for emotion analysis.

[1339] User input of new ideas

[1340] Users input ideas for new services or events through the system. They enter specific concepts and desired conditions in text format from their device (e.g., a PC or smartphone) and send them to the server. This transmission uses the HTTPS protocol and encrypts the data. The server analyzes the received data using an emotion engine and evaluates the emotional state.

[1341] Searching for related data

[1342] The server searches past research data related to the new idea entered by the user. It extracts highly relevant data using pre-classified and tagged keywords and category information. A database management system (e.g., Elasticsearch) is used for this search.

[1343] Simulation and result generation

[1344] The server uses generative models and an emotion engine to generate simulation results based on the user's new idea. For example, if a new idea for a fitness app is entered, it will generate feedback such as "Many users want a plan that allows them to exercise effectively in a short amount of time," based on past feedback. This feedback is then refined based on the emotion analysis results.

[1345] Presentation of results

[1346] The server displays the generated simulation results on the user's terminal. The user can visually see specific suggestions and feedback. For example, specific advice might be given such as, "A plan incorporating a 10-minute high-intensity interval training session and a game element where the user earns points to advance their level would likely be well-received."

[1347] The following are specific examples of prompt statements:

[1348] Please use the following survey data to generate feedback on your ideas for a new fitness app.

[1349] Survey data 1: ...

[1350] Survey data 2: ...

[1351] ...(Enter multiple survey data)...

[1352] Ideas for a new fitness app:

[1353] 10-minute high-intensity interval training

[1354] A feature that allows users to earn points and advance their progress level.

[1355] Please provide feedback while taking into account the results of the sentiment analysis.

[1356] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1357] Step 1: Data Collection

[1358] The server collects historical survey data. This begins with authenticating to each database and establishing a secure connection. Specifically, authentication is performed using an API key, user ID, and password, and a secure connection is established using the SSL / TLS protocol. Next, the server retrieves data by executing the necessary SQL queries or API calls (e.g., SELECT FROM survey_data WHERE date > '2020-01-01'). This data is stored in secure internal storage. The input is the database authentication and connection information, and the output is the collected survey data. The specific operations involve executing API calls or SQL queries, retrieving data, and saving it.

[1359] Step 2: Data Preprocessing

[1360] The server cleans, classifies, and tags the collected survey data. First, it detects and removes invalid responses and missing data (e.g., empty fields or inconsistent responses). Next, it classifies the data into specific themes (e.g., employee benefits, company events) and tags relevant keywords using TF-IDF (Term Frequency-Inverse Document Frequency). The input is the collected survey data, and the output is the cleaned and tagged survey data. The specific operations are data cleaning, thematic classification, and tagging.

[1361] Step 3: Training the machine learning model

[1362] The server trains a generative AI model (e.g., ChatGPT) using preprocessed survey data. First, the preprocessed data is split into a training dataset and a test dataset. Typically, the training data:test data ratio is 8:2. Next, the model is trained using a high-performance GPU. The model's performance is evaluated, and evaluation metrics such as accuracy, recall, and F-score are checked. Parameters are fine-tuned as needed, and retraining is performed. The input is the cleaned and tagged survey data, and the output is the trained generative AI model.

[1363] Step 4: Introducing the Emotional Engine

[1364] The server implements an emotion engine to recognize user emotions. It installs an emotion engine (e.g., Emotion API), analyzes user-entered text data in real time, and determines the emotional state. The analysis results are saved to data storage and used in subsequent processes. Input is the user's text input, and output is the analyzed emotion information. The specific operations are text analysis and emotion identification.

[1365] Step 5: User input of new ideas

[1366] Users input new ideas using the system. Users enter specific concepts and desired conditions in text format from their devices (e.g., PCs, smartphones) and send them to the server. This data is transmitted encrypted (using the HTTPS protocol). The server passes the received data to an emotion engine, which analyzes the emotional state. The input is the user's new idea, and the output is the submitted idea and the analyzed emotional information.

[1367] Step 6: Search for related data

[1368] The server searches past survey data related to the user's new idea. It searches the database using pre-classified and tagged keywords and category information. Specifically, it executes SQL queries or Elasticsearch search queries to extract highly relevant data (e.g., SELECT FROM survey_data WHERE keywords LIKE '%fitness%'). The search results are temporarily stored in memory storage. The input is the user's new idea and keywords, and the output is the relevant survey data. The specific operations are executing search queries and extracting data.

[1369] Step 7: Simulation and result generation

[1370] The server generates simulation results based on the user's new ideas using a generative model and an emotion engine. The user's new ideas and related data are integrated and input into the generative AI model. Feedback output by the generative model is retrieved and fine-tuned using the emotion engine's analysis results. For example, it might output feedback such as, "Many users would like a plan that allows them to exercise effectively in a short amount of time." The input is the new ideas and related data, and the output is the simulation results.

[1371] Step 8: Presentation of Results

[1372] The server displays the generated simulation results on the user's terminal. First, the generated feedback is converted into HTML or JSON format and sent to the user's terminal. The user's terminal displays the results on an interface, allowing the user to visually confirm specific suggestions and feedback. For example, feedback such as "A 10-minute high-intensity interval training plan combined with a game element where the user earns points to advance their level would likely be preferred" might be displayed. The inputs are the simulation results and analysis results, and the output is the feedback presented to the user.

[1373] (Application Example 2)

[1374] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1375] Currently, systems exist that utilize past survey data to understand users' true needs and propose new services and events based on those needs. However, because these types of systems do not take into account the emotional state of the user, the proposed content does not always meet the user's expectations. In particular, in physical stores, there is a demand for service proposals that reflect the customer's real-time emotional state, but conventional systems have had difficulty achieving this. The present invention aims to solve these problems and provide a system that makes more appropriate service proposals while taking into account the emotional state of the customer.

[1376] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1377] In this invention, the server includes means for collecting past survey information, means for cleaning, classifying, and tagging the survey information, means for training a machine learning model using the cleaned and tagged survey information, means for searching for survey information related to new ideas entered by the user, means for generating simulation results based on the user's new ideas using the machine learning model, means for adjusting the generated simulation results using an emotion engine that analyzes the user's emotions, and means for displaying the generated simulation results on the user terminal. This makes it possible to propose services that take into account the user's emotional state.

[1378] "Survey information" refers to response data obtained from past surveys and feedback.

[1379] "Means of collection" refers to mechanisms for collecting survey information through databases or APIs.

[1380] "Cleaning" is the process of improving data quality by removing invalid responses and missing data.

[1381] "Classification and tagging" is the process of organizing cleaned data by theme and assigning relevant keywords.

[1382] A "machine learning model" is an algorithm that learns patterns from large amounts of data and uses them to make predictions and generate responses.

[1383] "Training methods" refer to methods for training machine learning models using pre-processed data.

[1384] "New ideas" refer to suggestions for new services or events that users input into the system.

[1385] "Search methods" refer to efficient methods for extracting past survey information related to the entered idea from the database.

[1386] "Simulation results" refer to the predictions and feedback generated using a machine learning model based on the proposal.

[1387] An "emotion engine" is a technology that analyzes the user's emotional state from their text input and adjusts its response accordingly.

[1388] A "user terminal" is a device (such as a smartphone or tablet) used to access the system and review the proposed content.

[1389] The system implementing this invention collects past survey information and, based on that, makes service suggestions related to the user's new ideas. Furthermore, it analyzes the user's emotional state using an emotion engine and adjusts the suggestions accordingly to provide more appropriate feedback.

[1390] First, the server collects past survey information from databases. This data is obtained from databases such as the human resources database, general affairs database, and monitor database. The collected data is cleaned, invalid responses and missing data are removed, and the data is categorized and tagged by theme. Programming languages ​​such as Python and SQL are used for this process.

[1391] Next, the server uses the cleaned and tagged survey data to train a machine learning model. Specifically, it uses a generative AI model with natural language processing techniques (e.g., GPT-3.5-turbo). This model learns from a large amount of text data and performs new text generation and predictions for new user ideas.

[1392] When a user inputs a new idea into the system, it searches the database for past survey information related to that new idea. The retrieved data is then fed into a machine learning model, and simulation results are generated.

[1393] Furthermore, the server uses an emotion engine to analyze the user's emotional state. This emotion engine analyzes the text data entered by the user and determines the emotional state (joy, excitement, confusion, etc.) from the context. Based on the analyzed emotional state, the simulation results are adjusted. The emotion engine uses, for example, an emotion analysis library (such as Microsoft's Azure Text Analytics API).

[1394] Finally, the server displays the adjusted simulation results on the user's terminal. Users can review the suggestions via devices such as smartphones and tablets. A dedicated application is installed on the terminal, allowing them to proceed with planning new services and events based on the displayed feedback and suggestions.

[1395] As a concrete example, when a user enters the following prompt, the optimal suggestion is generated:

[1396] Example of a prompt:

[1397] I would like to propose a new fitness program. Specifically, I want to create a plan that allows for effective exercise in a short amount of time, and incorporate a game element where users can earn points and advance through the program.

[1398] This allows the server to accurately reflect user needs and provide appropriate service suggestions based on their emotional state through the processes described above.

[1399] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1400] Step 1:

[1401] The server collects past survey information. Specifically, it authenticates to the HR database, general affairs database, and monitor database, establishes a secure connection, and then executes the necessary SQL queries and API calls. This collects the survey information. The input is "survey information from the database," and the output is "collected survey information."

[1402] Step 2:

[1403] The server cleans the collected survey information, removing invalid responses and missing data to improve data quality. Next, it categorizes the data by theme and tags it with relevant keywords. The input is the "collected survey information," and the output is the "cleaned, categorized, and tagged survey information."

[1404] Step 3:

[1405] The server trains a machine learning model using cleaned and tagged survey information. Specifically, it uses a generative AI model based on natural language processing (GPT-3.5-turbo) to learn patterns from a large amount of text data. The input is "cleaned, classified, and tagged survey information," and the output is the "trained machine learning model."

[1406] Step 4:

[1407] Users input new ideas into the system. They submit specific concepts and desired conditions to the system in text format from their terminal. The input is "the text of the user's new idea," and the output is "the user's new idea sent to the server."

[1408] Step 5:

[1409] The server searches the database for past survey information related to the new idea entered by the user. It efficiently extracts highly relevant data using pre-classified and tagged keywords and category information. The input is the "user's new idea" and the "database," and the output is a "list of relevant survey information."

[1410] Step 6:

[1411] The server generates simulation results using a trained machine learning model. Based on the user's new idea and related survey information, it generates future predictions and responses. The inputs are "the user's new idea," "a list of related survey information," and "the trained machine learning model," and the output is "the generated simulation results."

[1412] Step 7:

[1413] The server uses an emotion engine to analyze the user's emotional state. It analyzes the text of the user's new idea and determines the emotional state (joy, excitement, confusion, etc.) from the context. The input is "the text of the user's new idea," and the output is "the analysis result of the user's emotional state."

[1414] Step 8:

[1415] The server adjusts the generated simulation results based on the analysis results of the emotion engine. This optimizes the suggestions according to the user's emotional state. The inputs are the "generated simulation results" and the "analysis results of the user's emotional state," and the output is the "adjusted simulation results."

[1416] Step 9:

[1417] The server displays the adjusted simulation results on the user's terminal. The user can review the suggestions through the terminal and provide further feedback. The input is the "adjusted simulation results," and the output is the "feedback and suggestions displayed to the user."

[1418] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1419] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1420] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[1421] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1422] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. In the upper and lower directions of the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. Also, the upper side of the concentric circles is where "pleasant" emotions are located, and the lower side is where "unpleasant" emotions are located. In this way, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[1423] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[1424] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[1425] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[1426] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[1427] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[1428] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[1429] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[1430] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[1431] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1432] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[1433] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[1434] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[1435] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[1436] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[1437] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[1438] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted as being incorporated by reference.

[1439] The following is further disclosed regarding the embodiments described above.

[1440] (Claim 1)

[1441] Means of collecting past survey information,

[1442] A means for cleaning, classifying, and tagging the aforementioned survey information,

[1443] A means for training a machine learning model using the cleaned and tagged survey information,

[1444] A means of searching for survey information related to new ideas entered by users,

[1445] A means for generating simulation results based on a user's new idea using the aforementioned machine learning model,

[1446] Means for displaying the generated simulation results on a user terminal,

[1447] A system that includes this.

[1448] (Claim 2)

[1449] The system according to claim 1, wherein the aforementioned survey information is collected from an internal personnel database, a general affairs database, and a monitor database within the organization.

[1450] (Claim 3)

[1451] The system according to claim 1, wherein the machine learning model performs simulations based on natural language processing.

[1452] "Example 1"

[1453] (Claim 1)

[1454] Means of collecting past survey information,

[1455] A means for cleaning, classifying, and tagging the aforementioned survey information,

[1456] A means for training a machine learning model using the cleaned and tagged survey information,

[1457] A means of searching for survey information related to new ideas entered by users,

[1458] A means for generating simulation results based on a user's new idea using the aforementioned machine learning model,

[1459] Means for displaying the generated simulation results on a user terminal,

[1460] A system that includes this.

[1461] (Claim 2)

[1462] The system according to claim 1, wherein the aforementioned survey information is collected from a database within the organization.

[1463] (Claim 3)

[1464] The system according to claim 1, wherein the machine learning model performs simulations based on natural language processing.

[1465] "Application Example 1"

[1466] Rewrite

[1467] (Claim 1)

[1468] Means of collecting past survey information,

[1469] A means for cleaning, classifying, and tagging the aforementioned survey information,

[1470] A means for training a machine learning model using the cleaned and tagged survey information,

[1471] A means of searching for survey information related to new ideas entered by users,

[1472] A means for generating simulation results based on a user's new idea using the aforementioned machine learning model,

[1473] Means for displaying the generated simulation results on a user terminal,

[1474] A means of providing a user interface for inputting ideas for new services,

[1475] A means for collecting and analyzing feedback in real time through the aforementioned user interface,

[1476] A system that includes this.

[1477] (Claim 2)

[1478] The system according to claim 1, wherein the aforementioned survey information is collected from a database within the organization.

[1479] (Claim 3)

[1480] The system according to claim 1, wherein the machine learning model performs simulations based on natural language processing.

[1481] "Example 2 of combining an emotion engine"

[1482] (Claim 1)

[1483] Means of collecting past survey data,

[1484] A means for cleaning, classifying, and tagging the aforementioned survey data,

[1485] A means for training a generative model using the cleaned and tagged survey data,

[1486] A means of searching for survey data related to new ideas entered by the user,

[1487] A means for generating simulation results based on a user's new idea using the aforementioned generation model,

[1488] A means of analyzing user emotions and reflecting the analysis results in feedback,

[1489] Means for displaying the generated simulation results on a user terminal,

[1490] A system that includes this.

[1491] (Claim 2)

[1492] The system according to claim 1, wherein the survey data is collected from a database within the organization, and the database includes a human resources database, a general affairs database, and other databases.

[1493] (Claim 3)

[1494] The system according to claim 1, wherein the generative model performs simulations based on natural language processing.

[1495] "Application example 2 of combining emotional engines"

[1496] (Claim 1)

[1497] Means of collecting past survey information,

[1498] A means for cleaning, classifying, and tagging the aforementioned survey information,

[1499] A means for training a machine learning model using the cleaned and tagged survey information,

[1500] A means of searching for survey information related to new ideas entered by users,

[1501] A means for generating simulation results based on a user's new idea using the aforementioned machine learning model,

[1502] A means for adjusting the generated simulation results using an emotion engine that analyzes the user's emotions,

[1503] Means for displaying the generated simulation results on a user terminal,

[1504] A system that includes this.

[1505] (Claim 2)

[1506] The system according to claim 1, wherein the aforementioned survey information is collected from an internal personnel database, a general affairs database, and a monitor database within the organization.

[1507] (Claim 3)

[1508] The system according to claim 1, wherein the machine learning model performs simulations based on natural language processing, and the emotion engine adjusts the simulation results based on the user's emotional state. [Explanation of Symbols]

[1509] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. Means of collecting past survey information, A means for cleaning, classifying, and tagging the aforementioned survey information, A means for training a machine learning model using the cleaned and tagged survey information, A means of searching for survey information related to new ideas entered by users, A means for generating simulation results based on a user's novel idea using the aforementioned machine learning model, Means for displaying the generated simulation results on a user terminal, A system that includes this.

2. The system according to claim 1, wherein the aforementioned survey information is collected from an internal personnel database, a general affairs database, and a monitor database within the organization.

3. The system according to claim 1, wherein the machine learning model performs simulations based on natural language processing.

Images