system

Abstract

Provide a system. A data collection means; a means for cleaning the collected data; A means of analyzing the cleaned data and extracting important features; a means for training a generative AI model based on the extracted features; and means for generating information in response to a user request using the trained model; means for obtaining user feedback on the generated information; A means to retune the generative AI model based on the feedback obtained; and A system including:

Description

[Technical Field]

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

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

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

[0004] In recent years, technology that uses generative AI to generate creative works about the past has been attracting attention. However, the accuracy and quality of the generated content is not always high, and it often fails to accurately reflect historical facts and context. Furthermore, it is difficult to accurately generate the specific information and depictions that users need, and there is no established method for utilizing feedback on the generated content. Solving this problem would make it possible to improve the quality of creative works based on past events. [Means for solving the problem]

[0005] The present invention provides a system including a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, and a feedback-based generative AI model readjustment means. First, the data collection means automatically collects relevant literature, records, papers, etc. from the Internet and databases. Next, the collected data is converted into a consistent format by the cleaning means, and duplicate data is removed and missing information is filled in. Then, important features and topics are extracted from the cleaned data using the data analysis means. Based on this feature information, a generative AI model is optimized and trained for a specific topic. The generation means uses the trained model to generate specific information in response to a user request. Furthermore, user feedback on the generated information is obtained, and the model is readjusted based on the feedback to improve the accuracy and quality of the generated content. This series of processes effectively improves the quality of past creative works.

[0006] "Data collection means" refers to a system that automatically retrieves relevant data such as literature, records, and papers from the Internet or databases.

[0007] "Data cleaning methods" are mechanisms for converting collected data into a consistent format, removing unnecessary data, and filling in missing information.

[0008] "Data analysis tools" are mechanisms that use natural language processing techniques to extract important features, topics, and patterns from cleaned data.

[0009] The "training method for generative AI models" is a mechanism for training generative AI models based on extracted feature information and optimizing them for specific topics.

[0010] The "generative means" is a mechanism that uses a trained generative AI model to generate specific content based on user requests.

[0011] The "feedback acquisition means" is a mechanism for acquiring user evaluations of the generated content and points for improvement.

[0012] The "generative AI model readjustment means" is a mechanism for readjusting the generative AI model based on the obtained feedback to improve the accuracy and quality of the generated results. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0014] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0015] First, the terms used in the following description will be explained.

[0016] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0017] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0018] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0019] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0020] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0021] [First embodiment]

[0022] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0023] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0024] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0025] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0026] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0027] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0028] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0029] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0030] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0031] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0032] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0033] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0034] The present invention relates to a system for improving the accuracy and quality of past creative works using generative AI. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, and a generative AI model readjustment means. An embodiment of the system is described below.

[0035] 1. Data Collection

[0036] Users upload relevant documents and records to create their creative works, for example, providing relevant historical documents to write a novel about World War I.

[0037] The server automatically collects relevant materials from the Internet and databases, for example, retrieving relevant information from online academic journal databases and encyclopedias.

[0038] 2. Data Cleaning

[0039] The server converts the collected data into a consistent format, removes duplicates, and completes missing information, for example, by unifying descriptions of the same event from multiple sources and completing missing dates.

[0040] 3. Data Analysis

[0041] The server uses natural language processing techniques to extract key features from the cleaned data, such as identifying the date, time, location, and people involved in a particular battle.

[0042] The server categorizes the data by relevant topic for later generation, for example by stage of a war or important event.

[0043] 4. Model training

[0044] The server trains a generative AI model based on the extracted data features, for example, training the model to generate specific World War I battle scenes and soldier dialogue.

[0045] The server optimizes the model's parameters to generate creations tailored to the user's needs, such as enhancing detailed tactical descriptions or emotional expression.

[0046] 5. Creation of creative works

[0047] A user inputs a specific request into the system, for example, "Please depict the Gallipoli landings on April 25, 1915."

[0048] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0049] 6. User Feedback and Refinement

[0050] The user provides feedback on the generated creation, for example, "I would like more detailed descriptions of tactics added" or "I would like the expression of emotions improved."

[0051] The server readjusts the generative AI model based on user feedback to improve the accuracy and quality of the generated results, for example by adding training data to enhance emotional expression and readjusting the model parameters.

[0052] By using these methods, the system of the present invention can improve the accuracy and quality of creative works about the past and generate historically accurate and compelling content. For example, when creating a novel or movie scenario about World War I, this system can be used to generate detailed and accurate battle scenes and character dialogue, increasing the appeal to readers and viewers.

[0053] The processing flow will be explained below.

[0054] Step 1:

[0055] Users upload relevant documents and records to create their creative works. For example, if a user is writing a novel about World War I, they provide the system with digital files of historical documents they have on hand.

[0056] Step 2:

[0057] The server automatically collects relevant materials from the Internet and databases, for example, retrieves relevant data from online academic paper databases and encyclopedias, and incorporates them into the system.

[0058] Step 3:

[0059] The server cleans the collected data, which includes converting it into a consistent format, removing duplicates, and filling in missing information (for example, unifying descriptions of the same event collected from multiple sources and filling in missing information such as dates and locations).

[0060] Step 4:

[0061] The server then analyzes the cleaned data using natural language processing techniques to extract key features, such as the date, time, location, and people involved in a particular battle.

[0062] Step 5:

[0063] The server trains the generative AI model based on the analyzed data. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[0064] Step 6:

[0065] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[0066] Step 7:

[0067] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0068] Step 8:

[0069] Users can provide feedback to the system about the generated creations, such as "I'd like more detailed descriptions of tactics" or "I'd like the expression of emotions to be improved."

[0070] Step 9:

[0071] The server readjusts the generative AI model based on user feedback. Based on the feedback, it readjusts the model parameters to improve the accuracy and quality of the generated results. For example, it adds training data to enhance emotional expression and retrains the model.

[0072] Step 10:

[0073] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[0074] Through this series of steps, the system of the present invention can effectively generate highly accurate and high-quality creative works about the past that meet the needs of the user.

[0075] Example 1

[0076] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0077] With conventional technologies, it is difficult to efficiently collect detailed historical information, analyze it in a unified manner, and automatically generate highly satisfying creative works using a highly accurate generative AI model. Furthermore, there is a lack of a process for incorporating user feedback on the generated content and continuously improving the generative AI model, which limits the improvement in the quality of the deliverables. Therefore, there is an urgent need to resolve these issues.

[0078] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0079] In this invention, the server includes a means for users to upload documents using their terminals, a means for the server to extract features from the cleaned data using natural language processing technology, a means for the server to classify the data by specific topic, and a means for the server to generate creative works using a Transformer model. This allows users to efficiently create high-quality creative works based on detailed historical information. Furthermore, by incorporating user feedback, the generative AI model can be continuously improved, thereby increasing the accuracy and quality of the generated content.

[0080] "Data collection means" means a means that has the function of uploading documents and records provided by users and automatically collecting related information from the Internet and databases.

[0081] "Means for cleaning collected data" means means for converting collected data into a consistent format, removing duplicate content, and completing missing information.

[0082] The "means for analyzing cleaned data and extracting important features" refers to a means having a function for extracting important features from cleaned data using natural language processing technology.

[0083] "Means for training a generative AI model" refers to means that trains a generative AI model based on extracted feature data and has the function of improving the accuracy of information generation.

[0084] "Means for generating information in response to a user's request using a trained model" means means that have the function of generating information using a trained generative AI model based on a specific request from a user.

[0085] The "means for obtaining user feedback" refers to a means having a function for collecting user evaluations and opinions about the generated information.

[0086] "Means for readjusting the generative AI model" refers to means that have the function of retraining or adjusting the generative AI model based on obtained user feedback to improve the accuracy and quality of the generated content.

[0087] "Means for users to upload documents using their terminals" refers to means that allow users to send documents and records to the system from the devices they use.

[0088] "Natural language processing technology" is a technology that allows computers to process and understand natural language used by humans.

[0089] A "Transformer model" is a neural network architecture that has particularly powerful capabilities in natural language processing, and is a model for context-aware text generation and translation.

[0090] A "means for categorizing data by specific topics" is a means that has the ability to organize and group cleaned data based on specific themes or categories.

[0091] MODE FOR CARRYING OUT THE INVENTION

[0092] The present invention is a system for improving the accuracy and quality of past creative works using generative AI. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, and a generative AI model readjustment means. An embodiment of the present invention will be described in detail below.

[0093] Data collection

[0094] A user uploads relevant documents and records to the system to create a creative work using their terminal. This process is performed using the user's device (PC, smartphone, etc.). For example, a user may upload historical documents in PDF format to the system in order to write a novel about World War II.

[0095] The server then automatically collects relevant material from the internet and existing databases based on the uploaded documents, using Python libraries (e.g., BeautifulSoup) to scrape relevant information from websites and retrieve information from online academic paper databases and encyclopedias.

[0096] Data Cleaning

[0097] The server converts the collected data into a consistent format, removes duplicates, and completes missing information. It converts the data into a data frame using the Pandas library and standardizes it to a consistent date and time format. It also uses a natural language processing library (e.g., spaCy) to automatically complete missing information.

[0098] Data analysis

[0099] The server uses natural language processing techniques to extract important features from the cleaned data, such as topic modeling (e.g., LDA) to extract features related to specific events or people.

[0100] The server then classifies the data into relevant topics based on the extracted features, using a clustering algorithm (e.g., K-means) to categorize the data into categories such as battle scenes, diplomatic events, and daily life.

[0101] Model training

[0102] The server trains a generative AI model based on the cleaned feature data. Using Tensorflow, we train an RNN-based generative AI model to generate battle scenes and dialogue.

[0103] The server then adjusts and optimizes the model parameters, performing hyperparameter tuning to determine the optimal learning rate and number of epochs to improve generation accuracy.

[0104] Creation

[0105] A user enters a specific request into the system, such as "Please portray a scene from the Normandy landings in 1944," into a web form.

[0106] The server uses trained generative AI models to generate depictions based on user requests, using Transformer models to detail soldier movements and scenery from the Normandy landings.

[0107] User feedback and refinements

[0108] Users can provide feedback on the generated creations, such as "more detailed tactical descriptions are needed" or "improvement in emotional expression."

[0109] The server retunes the generative AI model based on user feedback, retraining the model with additional training data to improve the accuracy and quality of the generated content.

[0110] Prompt Sentence Examples

[0111] "Please describe the Gallipoli landings on 25 April 1915. Specifically, we are looking for a description of the landing boats approaching the shore, the tension among the soldiers, and the fighting that followed."

[0112] Through the system of the present invention, users can efficiently create high-quality creative works based on detailed historical information, and by incorporating user feedback, the generative AI model can be continuously improved, thereby increasing the accuracy and quality of the generated content.

[0113] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0114] Step 1:

[0115] Users upload documents and records to the system.

[0116] Input: Literature data such as PDFs or text files selected by the user from their device.

[0117] Specific operation: The user uses the system's upload function from their PC or smartphone to select historical documents and records and send them to the server.

[0118] Output: Uploaded bibliographic data stored on the server.

[0119] Step 2:

[0120] The server collects additional data from the internet and databases.

[0121] Input: A search query based on the uploaded literature data.

[0122] What it does: The server uses Python's BeautifulSoup library to scrape information from relevant websites and retrieve relevant data from sources like online academic paper databases.

[0123] Output: Additional data collected (e.g., text of web pages, abstracts of matching academic papers).

[0124] Step 3:

[0125] The server cleans the collected data.

[0126] Input: The raw text of the collected data.

[0127] What it does: The server uses the Pandas library to convert the data into a data frame, removes duplicates, converts to a consistent date and time format, and imputes missing data. It also uses a natural language processing library (e.g., spaCy) to auto-complete missing information.

[0128] Output: A cleaned and consistent dataset.

[0129] Step 4:

[0130] The server parses the cleaned data.

[0131] Input: The cleaned dataset.

[0132] Specific operations: Conduct topic modeling (e.g., LDA) to extract important features (e.g., involvement of specific events or people), and use machine learning algorithms to classify topics.

[0133] Output: Extracted features and classified topics.

[0134] Step 5:

[0135] The server trains the generative AI model.

[0136] Input: Extracted feature data.

[0137] What it does: Use Tensorflow to train an RNN-based generative AI model to generate specific scenarios and dialogues, and use hyperparameter tuning to select optimal model settings.

[0138] Output: A trained generative AI model.

[0139] Step 6:

[0140] A user inputs a specific request into the system.

[0141] Input: A prompt from the user (e.g., "Please describe a scene from the Normandy landings in 1944").

[0142] Specific action: The user enters a prompt sentence about a specific scene or situation into a web form and submits it to the system.

[0143] Output: The prompt received by the server.

[0144] Step 7:

[0145] The server generates information based on the user's request.

[0146] Input: A trained generative AI model and a user prompt.

[0147] Specific behavior: Use trained generative AI models to generate specific scenes and dialogues based on prompts, and use Transformer models to generate context-aware text.

[0148] Output: The generated text in response to the user's request.

[0149] Step 8:

[0150] The user provides feedback on the generated information.

[0151] Input: Generated text and user rating.

[0152] Specific actions: The user evaluates the generated text and inputs suggestions for improvement, such as detailed tactical descriptions and emotional expressions.

[0153] Output: Feedback provided by the user.

[0154] Step 9:

[0155] The server readjusts the generative AI model based on the feedback.

[0156] Input: User feedback and retraining data.

[0157] What it does: Based on the feedback, it retrains the generative AI model with additional training data and optimizes the model's parameters.

[0158] Output: An improved generative AI model.

[0159] Through these steps, the system can generate high-quality creative works tailored to the user's needs and continuously improve its accuracy and quality.

[0160] (Application example 1)

[0161] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0162] Previous systems using generative AI technology struggled to sufficiently improve the accuracy and quality of historical creations. They also lacked a process for effectively utilizing user feedback to optimize generative AI models. Furthermore, they lacked a means to visually display the generated information and improve the user experience.

[0163] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0164] In this invention, the server includes a data collection means, a means for cleaning the collected data, a means for analyzing the cleaned data and extracting important features, a means for training a generative AI model based on the extracted features, a means for generating information according to a user request using the trained model, a means for obtaining user feedback on the generated information, a means for readjusting the generative AI model based on the obtained feedback, and a means for visually displaying the generated information. This enables improvement in the accuracy and quality of historical creations using generative AI, optimization of the model based on effective feedback, and an improvement in user experience.

[0165] "Data collection methods" are methods for automatically collecting relevant literature, records, and papers from the Internet and databases.

[0166] "Data cleaning procedures" are procedures that convert collected data into a consistent format, remove duplicate content, and complete missing information.

[0167] "Data analysis tools" are tools for extracting important features from cleaned data and classifying the data by related topics.

[0168] The "generative AI model training means" is a means for training a generative AI model based on extracted features and optimizing the model parameters.

[0169] "Information generation means" refers to a means for generating information in response to a user request using a trained generative AI model.

[0170] The "feedback acquisition means" is a means for acquiring user feedback on the generated information.

[0171] The "readjustment means" is a means for readjusting the generative AI model based on the obtained feedback.

[0172] The "visual display means" is a means for visually displaying the generated information.

[0173] The present invention relates to a system for improving the accuracy and quality of historical fiction using generative AI, which is implemented using several means:

[0174] First, the server uses a data collection tool to automatically collect relevant literature, records, and papers from the Internet and databases. The software used for this is the Python requests library, which retrieves data through an API.

[0175] The collected data is then converted into a consistent format through data cleaning procedures, using tools such as Pandas to remove duplicate content and fill in missing information, for example by unifying descriptions of the same event from multiple documents and filling in missing date information.

[0176] The cleaned data is then analyzed using natural language processing techniques to extract key features, such as the date, time, location, and people involved in a particular historical event, using NLP libraries such as spaCy.

[0177] Based on the extracted features, the server trains a generative AI model using a machine learning framework such as TENSORFLOW® or PyTorch. For example, the server trains the model to generate specific combat scenes or dialogues.

[0178] The server then uses the trained model to generate information based on the user's request. For example, if a user requests, "Generate a detailed description of the Gallipoli Landings on April 25, 1915," the generative AI model will generate a detailed description of the scene.

[0179] The feedback acquisition means acquires user feedback on the generated information, such as "I would like more detailed descriptions of tactics to be added."

[0180] Finally, the recalibration means recalibrates the generative AI model based on the obtained feedback. In this process, the training dataset is reconstructed based on the collected feedback and the model is retrained.

[0181] Using visual display means, the generated information is visually displayed on a user device. For example, it can be visually confirmed through a smartphone or head-mounted display. As a result, the accuracy and quality of historical creations using generative AI can be improved, models can be optimized based on effective feedback, and the user experience can be improved.

[0182] Example prompt sentence:

[0183] Generate a depiction of the Allied landing forces approaching the beaches of Gallipoli at sunrise on April 25, 1915. Detail specific soldier movements, the scenery, and emotions.

[0184] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0185] Step 1:

[0186] Data collection

[0187] The server uses data collection tools to automatically collect relevant literature, records, and papers from the Internet and databases, for example, using the Python requests library to retrieve data through APIs. The input is the relevant URL or API endpoint, and the output is the retrieved raw data.

[0188] Step 2:

[0189] Data Cleaning

[0190] The server uses data cleaning techniques to convert the collected data into a consistent format, remove duplicates, and fill in missing information. Specifically, it converts the data into a data frame using Pandas and removes duplicates using the .drop_duplicates() method. The input is raw data, and the output is cleaned data.

[0191] Step 3:

[0192] Data analysis

[0193] The server then uses data analysis tools and natural language processing techniques to extract important features from the cleaned data. For example, it uses an NLP library such as spaCy to extract proper nouns such as specific events or people. The input is the cleaned data, and the output is the data with important features extracted.

[0194] Step 4:

[0195] Training generative AI models

[0196] The server trains a generative AI model based on the extracted features. Using a machine learning framework such as TensorFlow or PyTorch, the model is trained to generate specific dialogues or scenes. The input is the extracted feature data, and the output is a trained generative AI model.

[0197] Step 5:

[0198] information generation

[0199] The server uses the trained model to generate information based on the user's request. The user inputs a specific prompt, and the generated information is a detailed description. The input is the user's prompt, and the output is the generated text content. For example, the prompt "Please generate a detailed description of the Gallipoli landings on April 25, 1915" is input, and a specific description is generated.

[0200] Step 6:

[0201] Get feedback

[0202] The server obtains user feedback on the generated information. It uses a feedback form or questionnaire to collect user opinions and correction requests. The input is the user's feedback content, and the output is the feedback data.

[0203] Step 7:

[0204] Model Rebalancing

[0205] The server retunes the generative AI model based on the feedback it receives. It reconstructs the training dataset based on the collected feedback and retrains the model. The input is the feedback data, and the output is the retuned generative AI model.

[0206] Step 8:

[0207] Visual Indication

[0208] The device visually displays the generated information using a visual display means. The user can view the generated detailed depiction through a smartphone or head-mounted display. The input is the generated text content, and the output is the visually displayed information.

[0209] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0210] The present invention further combines a system for improving the accuracy and quality of past creative works using generative AI with an emotion engine. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, a generative AI model readjustment means, and an emotion engine. An embodiment of this system is shown below.

[0211] 1. Data Collection

[0212] Users upload relevant documents and records to create their creative works. For example, if a user is writing a novel about World War I, they provide the system with digital files of historical documents they have on hand.

[0213] The server automatically collects relevant materials from the Internet and databases, for example, retrieves relevant data from online academic paper databases and encyclopedias, and incorporates them into the system.

[0214] 2. Data Cleaning

[0215] The server converts the collected data into a consistent format, removes duplicates, and fills in missing information, for example by unifying descriptions of the same event collected from multiple sources and filling in missing information such as dates and locations.

[0216] 3. Data Analysis

[0217] The server uses natural language processing techniques to extract key features from the cleaned data, such as the date, time, location, and people involved in a particular battle.

[0218] The server categorizes the data by relevant topic for later generation, for example by stage of a war or important event.

[0219] 4. Model training

[0220] The server trains a generative AI model based on the extracted data features. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[0221] The server optimizes the model's parameters to generate creations tailored to the user's needs, such as enhancing detailed tactical descriptions or emotional expression.

[0222] 5. Creation of creative works

[0223] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[0224] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0225] 6. User Feedback and Emotion Recognition

[0226] The user provides feedback on the generated creation, for example, "I would like more detailed descriptions of tactics added" or "I would like the expression of emotions improved."

[0227] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, it extracts positive, negative, and neutral emotions from the feedback content and further analyzes the user's detailed emotional state (e.g., satisfaction, dissatisfaction, surprise, etc.).

[0228] 7. Readjust based on feedback

[0229] The server readjusts the generative AI model based on user feedback and the results of emotion analysis by the emotion engine. For example, if the emotion engine detects the emotion "dissatisfaction," it identifies the cause of that emotion and optimizes the model parameters accordingly.

[0230] 8. Final generation and verification

[0231] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[0232] Check whether the user is satisfied with the final generated creation and provide further feedback if necessary to iterate and improve the quality of the process.

[0233] In this way, the system of the present invention not only uses generative AI to generate creative works about the past, but also uses an emotion engine to analyze user emotions and readjust the model based on feedback, thereby providing even more accurate and high-quality creative works. For example, when creating a novel or movie scenario about World War I, this system can generate detailed and accurate battle scenes and character dialogue, increasing the appeal to readers and viewers.

[0234] The processing flow will be explained below.

[0235] Step 1:

[0236] Users upload relevant documents and records to the system from their devices in order to create their creative works. For example, a user may submit digital files of historical documents they have on hand to write a novel about World War I.

[0237] Step 2:

[0238] The server automatically collects relevant materials from the Internet and databases, for example, retrieving relevant information from online academic paper databases and encyclopedias, and integrating it into the system.

[0239] Step 3:

[0240] The server cleans the collected data, which includes standardizing the data format, removing duplicate content, and completing missing information (for example, standardizing descriptions of the same event collected from multiple sources and completing missing information such as dates and locations).

[0241] Step 4:

[0242] The server then uses natural language processing techniques to extract key features from the cleaned data, such as the date, time, location, and people involved in a particular battle.

[0243] Step 5:

[0244] The server trains the generative AI model based on the analyzed data. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[0245] Step 6:

[0246] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[0247] Step 7:

[0248] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0249] Step 8:

[0250] The user provides feedback to the system about the generated creation, such as "I would like more detailed descriptions of tactics" or "I would like the expression of emotions to be improved."

[0251] Step 9:

[0252] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, it extracts positive, negative, and neutral emotions from the feedback content and analyzes more detailed emotional states (e.g., satisfaction, dissatisfaction, surprise, etc.).

[0253] Step 10:

[0254] The server readjusts the generative AI model based on user feedback and the results of emotion analysis. For example, if the emotion engine detects the emotion "dissatisfaction," it identifies the cause of that emotion and readjusts the model parameters accordingly.

[0255] Step 11:

[0256] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[0257] Step 12:

[0258] Check whether the user is satisfied with the final generated creation and provide further feedback if necessary to iterate and improve the quality of the process.

[0259] Through these steps, the system of the present invention can provide highly accurate and high-quality creative works by analyzing users' emotions in addition to the process of generating creative works related to the past using generative AI.

[0260] Example 2

[0261] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0262] Conventional creative creation systems using generative AI models lack the ability to analyze user feedback and readjust the model based on that feedback. As a result, the accuracy and quality of the created creations often do not meet user requirements. In particular, it is difficult to generate creations that appropriately reflect the user's emotions, and there is a need to improve user satisfaction.

[0263] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0264] In this invention, the server includes a data collection means, a means for cleaning the collected data into a consistent format, removing duplication, and completing missing information, a means for analyzing the cleaned data and extracting important features using natural language processing technology, a means for training a generative AI model based on the extracted features and optimizing model parameters, a means for generating information according to a user request using the trained model, a means for acquiring user feedback on the generated information, analyzing the feedback using an emotion engine, and recognizing the user's emotions, and a means for readjusting the generative AI model based on the acquired feedback and the emotion analysis results. This makes it possible to generate creative works with high accuracy and quality by appropriately readjusting the generative AI model based on the user's feedback and emotions.

[0265] "Data collection means" refers to the means by which user-provided documents and records are uploaded to the system and further related materials are automatically collected from the internet and databases.

[0266] "Data cleaning procedures" are procedures used to convert collected data into a consistent format, remove duplication, and complete missing information.

[0267] The "data analysis means" is a means for extracting important features from the cleaned data using natural language processing technology and classifying the data based on those features.

[0268] A "generative AI model" is an artificial intelligence model that is trained based on cleaned and analyzed data to generate information in response to user requests.

[0269] The "feedback acquisition means" is a means for collecting user feedback on the generated information or creation.

[0270] An "emotion engine" is a means for analyzing feedback provided by a user and recognizing the user's emotions based on that feedback.

[0271] "Model readjustment means" refers to a means for readjusting the generative AI model based on the obtained feedback and sentiment analysis results to improve its accuracy and quality.

[0272] The present invention provides a creative creation system that uses a generative AI model and an emotion engine to generate accurate and high-quality creative works based on user-provided documents and records. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a creative creation means, a feedback acquisition means, an emotion engine, and a model readjustment means.

[0273] An embodiment of this system is described below.

[0274] 1. Data Collection

[0275] A user uploads relevant documents and records related to their creative work to the system. For example, a user is writing a novel about World War I and provides the system with digital files (PDF, Word files, etc.) of historical documents.

[0276] The server collects relevant materials from Internet resources and existing databases, for example, pulls necessary information from Google Scholar or Wikipedia, and stores it in the system.

[0277] 2. Data Cleaning

[0278] The server converts the collected data into a consistent format, removes duplicates, and completes missing information, for example, unifying documents with different formats, removing duplicate content, and adding missing dates and other details.

[0279] 3. Data Analysis

[0280] The server uses natural language processing (NLP) techniques to extract key features from the cleaned data, specifically through text analysis, to extract information such as the date, time, location, and people involved in a particular battle.

[0281] The server uses the extracted features to categorize the data into related topics, for example, organizing data by phases of a war (start, major battles, intermediate ceasefires, etc.) or important events.

[0282] 4. Training the generative AI model

[0283] The server trains a generative AI model based on features extracted from the data, focusing on the scene the user wants to depict, for example, using data on the landing scene of the Gallipoli War.

[0284] The server optimizes the model parameters to generate a creation that meets the user's needs, for example by setting specific parameters to enhance tactical detail or emotional expression.

[0285] 5. Creation of creative works

[0286] The user inputs a specific request into the system, for example, a prompt such as "Please describe in detail the scenes from the Gallipoli landings on April 25, 1915."

[0287] The server uses a trained generative AI model to generate depictions based on user requests, such as detailed descriptions of soldier movements and scenery during a landing scene.

[0288] 6. User Feedback and Emotion Recognition

[0289] Users can provide feedback on the generated creation, for example by entering specific feedback in text, such as "I would like more detailed tactical descriptions."

[0290] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, the server can analyze the emotion of "dissatisfaction" extracted from the feedback, such as "I want more detailed tactical descriptions."

[0291] 7. Re-adjusting the model based on feedback

[0292] The server readjusts the generative AI model based on the feedback and emotion analysis results it receives. For example, if the emotion "dissatisfaction" is detected, the cause is identified and the model parameters are optimized.

[0293] 8. Final generation and verification

[0294] The server then uses the retuned model to generate the creation again and provides it to the user. Specifically, the adjusted model is used to generate a more detailed and emotionally rich Gallipoli War landing scene.

[0295] The user checks whether they are satisfied with the final generated creation and provides feedback again if necessary to iterate the process and improve the quality.

[0296] In this way, the system of the present invention can efficiently generate highly accurate and high-quality creative works by readjusting the generative AI model based on feedback provided by the user and the results of emotion analysis.

[0297] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0298] Step 1:

[0299] Data collection

[0300] Input: User-provided digital files of documents and records (e.g. PDF, Word files).

[0301] Specific operation: The user submits historical documents and record files they have on hand using the system's upload function.

[0302] Output: Literature and records stored in the system's database.

[0303] The server automatically collects relevant literature, records, and papers from the Internet and databases, specifically by accessing open data sources such as Google Scholar and Wikipedia, and stores the necessary information in the system.

[0304] Step 2:

[0305] Cleaning the data

[0306] Input: Collected literature and archival data.

[0307] What it does: It converts the data collected by the server into a consistent format (e.g., UTF-8 encoded text), detects and removes duplicate data, and fills in missing information (e.g., by filling in missing details like date or location from other resources).

[0308] Output: Consistent, deduplicated, and imputed bibliographic and archival data.

[0309] Step 3:

[0310] Data analysis

[0311] Input: Cleaned data.

[0312] What it does: The server uses natural language processing (NLP) techniques to extract important features from the data (e.g., battle dates, locations, and people involved), and then categorizes the data by related topics based on the extracted features. For example, it organizes data about each stage of a war or key events separately.

[0313] Output: Data with important features extracted and categorized by relevant topics.

[0314] Step 4:

[0315] Training generative AI models

[0316] Input: Extracted feature data.

[0317] How it works: The server trains a generative AI model based on the extracted feature data and optimizes it to generate information based on the user's request. For example, if a user wants a detailed depiction of the landing scene from the Gallipoli War, the server trains the model using data related to that scene.

[0318] Output: A trained generative AI model.

[0319] Step 5:

[0320] Creation

[0321] Input: The specific request (prompt) that the user provides to the system.

[0322] What it does: The user enters a prompt, such as "Please give me a detailed description of the Gallipoli Landings on April 25, 1915." The server uses a trained generative AI model to generate sentences and scene descriptions based on the request.

[0323] Output: The creative work generated based on the user's request.

[0324] Step 6:

[0325] User Feedback and Emotion Recognition

[0326] Input: Feedback provided by the user about the generated creation.

[0327] Specific operation: The user inputs feedback in text format, such as "I would like more detailed tactical descriptions." The server uses the emotion engine to analyze the feedback, extracts the user's emotion (e.g., "dissatisfied"), and analyzes it in detail.

[0328] Output: Parsed feedback and user sentiment data.

[0329] Step 7:

[0330] Retune the model based on feedback

[0331] Input: Feedback and sentiment analysis results.

[0332] Specific operation: The server readjusts the generative AI model based on the feedback and emotion analysis results it receives. For example, if the emotion "dissatisfaction" is detected, it identifies the cause of this and adjusts the model parameters.

[0333] Output: A retuned generative AI model.

[0334] Step 8:

[0335] Final generation and confirmation

[0336] Input: A retuned generative AI model and a new user request.

[0337] What happens: The server uses the refined model to generate new drawings and sentences, and serves them to the user. The server checks whether the user is satisfied with the final generated work, providing further feedback if necessary.

[0338] Output: High-quality creations that satisfy users.

[0339] (Application example 2)

[0340] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0341] Conventional generative AI models have limitations in their ability to generate content based on past data, making it particularly difficult to adjust the content to reflect emotional expressions and user feedback. As a result, the quality of the generated content varies and user feedback cannot be properly utilized. Furthermore, when users request specific scenes or historical events, it is difficult to respond to their requests accurately and in detail.

[0342] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a data collection means, a data cleaning means, and a means for analyzing the cleaned data and extracting important features. This enables the collected data to be organized into a consistent format and necessary features to be extracted. The server also includes a means for training a generative AI model, a means for generating information in response to a user request, a means for obtaining user feedback on the generated information, and a means for analyzing the user's emotional feedback using an emotion analysis means and reflecting the analysis results in readjusting the model. This allows the generated content to reflect user feedback and include emotional elements, making it possible to provide higher quality content.

[0343] "Data collection means" has the function of automatically collecting data provided by users and related materials from the Internet and databases.

[0344] "Data cleaning procedures" are procedures that convert collected data into a consistent format, remove duplicate content, and complete missing information.

[0345] "Data analysis means" extracts important features from the cleaned data so that they can be used in subsequent processes.

[0346] The "means for training a generative AI model" refers to optimizing the generative AI model based on the extracted features so that it can generate specific content.

[0347] An "information generation means" is a means that uses a trained generative AI model to generate information or content in response to a user's request.

[0348] The "user feedback acquisition means" has a function of collecting user opinions and impressions regarding the generated information.

[0349] The "means for readjusting the generative AI model" refers to the re-optimization of the generative AI model based on feedback obtained from users and the results of sentiment analysis, in order to improve its accuracy and quality.

[0350] The "emotion analysis means" analyzes the user's feedback from an emotional perspective and identifies the emotional state, such as positive, negative, or neutral.

[0351] This invention is a system for improving the accuracy and quality of past creative works using generative AI combined with an emotion engine. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, an information generation means, a user feedback acquisition means, a generative AI model readjustment means, and an emotion analysis means.

[0352] System Programs and Processing

[0353] Hardware and software used

[0354] Hardware: Server with high-performance CPU and GPU

[0355] Software: Python, TensorFlow / PyTorch, natural language processing libraries (spaCy, NLTK), sentiment analysis engine API

[0356] Data collection methods

[0357] The server collects documents and records uploaded by users and automatically retrieves related materials from the Internet and databases, thereby comprehensively obtaining the necessary data.

[0358] Data Cleaning Methods

[0359] The collected data is converted into a consistent format, duplicates are removed, and missing information is filled in. This cleaning process improves the quality of the data and makes it easier to analyze.

[0360] Data Analysis Methods

[0361] The server then uses natural language processing technology to analyze the cleaned data and extract important features, such as the date, time, and location of a particular event, as well as the people involved, which are then used in subsequent processes.

[0362] A means of training generative AI models

[0363] The server trains a generative AI model based on the extracted data, and the trained model is optimized to generate specific scenes and interactions requested by the user.

[0364] Information generation means

[0365] The server uses a trained generative AI model to generate information based on requests from users' devices. For example, a user might input a prompt such as, "Please describe the Normandy landings on June 6, 1944."

[0366] User feedback acquisition method

[0367] The user provides feedback on the generated information, which the server receives and uses for further improvement.

[0368] Emotion analysis means

[0369] The server analyzes the user feedback using an emotion engine to extract emotions such as positive, negative, or neutral, and passes the analysis results to the recalibration means.

[0370] Generative AI model readjustment method

[0371] Based on the information obtained from the user feedback acquisition means and the results of the sentiment analysis means, the server readjusts the generative AI model so that the next and subsequent generated content will better meet the user's needs.

[0372] Specific examples

[0373] For example, a user might enter a prompt request such as, "Please describe the Normandy landings on June 6, 1944." Based on this request, the server uses a trained generative AI model to generate a detailed and accurate scene description.

[0374] In this way, the system of the present invention is able to continually evolve and provide high quality creative works through user feedback and sentiment analysis.

[0375] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0376] Step 1: Data collection

[0377] The server uploads documents and records provided by users and automatically collects related materials from the Internet and databases.

[0378] Input: Digital files uploaded by users, data collected from the internet and databases.

[0379] Output: The collected dataset.

[0380] What it does: Users upload literature files about World War I, and the server also collects data from online academic paper databases.

[0381] Step 2: Data cleaning

[0382] The server cleans the collected data, standardizing the format, removing duplicates, and completing missing information.

[0383] Input: The collected dataset.

[0384] Output: A cleaned dataset.

[0385] What it does: The server unifies descriptions of the same event collected from multiple sources and completes missing information such as dates and locations.

[0386] Step 3: Data analysis

[0387] The server then analyzes the cleaned data using natural language processing techniques to extract key features.

[0388] Input: The cleaned dataset.

[0389] Output: Extracted feature dataset.

[0390] What it does: The server extracts data about specific fight scenes, such as the date, time, location, and people involved.

[0391] Step 4: Training the generative AI model

[0392] The server trains a generative AI model based on features extracted through data analysis.

[0393] Input: Extracted feature dataset.

[0394] Output: A trained generative AI model.

[0395] What it does: The server feeds the model with training data so that it can accurately depict scenes from the Gallipoli War.

[0396] Step 5: Information Generation

[0397] The user terminal inputs a prompt sentence, and the server generates information using a trained generative AI model.

[0398] Input: The prompt text entered by the user.

[0399] Output: The generated information (text or scene).

[0400] What it does: The user types in "Please describe the Normandy landings on June 6, 1944," and the server generates the scene.

[0401] Step 6: Get user feedback

[0402] The user provides feedback on the generated information, which is captured by the server.

[0403] Input: User feedback.

[0404] Output: Collected feedback data.

[0405] Specific action: The user enters feedback, such as "I need a detailed tactical description."

[0406] Step 7: Sentiment Analysis

[0407] The server uses an emotion engine to analyze the user feedback and extract positive, negative, or neutral emotions.

[0408] Input: Collected feedback data.

[0409] Output: Sentiment analysis result data.

[0410] Specific behavior: The server detects the emotion "dissatisfied" from the feedback.

[0411] Step 8: Refining the generative AI model

[0412] The server readjusts the generative AI model based on the emotion analysis results and feedback data.

[0413] Input: Sentiment analysis result data, feedback data.

[0414] Output: A retuned generative AI model.

[0415] Specific operation: The server adjusts parameters to "enhance tactical depiction."

[0416] Step 9: Final generation and verification

[0417] The prompt sentence is input again from the user terminal, and information is generated using the re-adjusted model and provided to the user.

[0418] Input: The prompt statement typed again.

[0419] Output: Improved generated information.

[0420] What it does: The user again types in "Describe the Normandy landings on June 6, 1944," generating a more detailed and emotive description of the scene.

[0421] The above are the specific processing steps for carrying out the invention.

[0422] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0423] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0424] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0425] [Second embodiment]

[0426] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0427] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0428] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0429] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0430] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0431] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0432] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0433] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0434] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0435] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0436] In the smart glasses 214, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0437] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0438] The present invention relates to a system for improving the accuracy and quality of past creative works using generative AI. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, and a generative AI model readjustment means. An embodiment of the system is described below.

[0439] 1. Data Collection

[0440] Users upload relevant documents and records to create their creative works, for example, providing relevant historical documents to write a novel about World War I.

[0441] The server automatically collects relevant materials from the Internet and databases, for example, retrieving relevant information from online academic journal databases and encyclopedias.

[0442] 2. Data Cleaning

[0443] The server converts the collected data into a consistent format, removes duplicates, and completes missing information, for example, by unifying descriptions of the same event from multiple sources and completing missing dates.

[0444] 3. Data Analysis

[0445] The server uses natural language processing techniques to extract key features from the cleaned data, such as identifying the date, time, location, and people involved in a particular battle.

[0446] The server categorizes the data by relevant topic for later generation, for example by stage of a war or important event.

[0447] 4. Model training

[0448] The server trains a generative AI model based on the extracted data features, for example, training the model to generate specific World War I battle scenes and soldier dialogue.

[0449] The server optimizes the model's parameters to generate creations tailored to the user's needs, such as enhancing detailed tactical descriptions or emotional expression.

[0450] 5. Creation of creative works

[0451] A user inputs a specific request into the system, for example, "Please depict the Gallipoli landings on April 25, 1915."

[0452] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0453] 6. User Feedback and Refinement

[0454] The user provides feedback on the generated creation, for example, "I would like more detailed descriptions of tactics added" or "I would like the expression of emotions improved."

[0455] The server readjusts the generative AI model based on user feedback to improve the accuracy and quality of the generated results, for example by adding training data to enhance emotional expression and readjusting the model parameters.

[0456] By using these methods, the system of the present invention can improve the accuracy and quality of creative works about the past and generate historically accurate and compelling content. For example, when creating a novel or movie scenario about World War I, this system can be used to generate detailed and accurate battle scenes and character dialogue, increasing the appeal to readers and viewers.

[0457] The processing flow will be explained below.

[0458] Step 1:

[0459] Users upload relevant documents and records to create their creative works. For example, if a user is writing a novel about World War I, they provide the system with digital files of historical documents they have on hand.

[0460] Step 2:

[0461] The server automatically collects relevant materials from the Internet and databases, for example, retrieves relevant data from online academic paper databases and encyclopedias, and incorporates them into the system.

[0462] Step 3:

[0463] The server cleans the collected data, which includes converting it into a consistent format, removing duplicates, and filling in missing information (for example, unifying descriptions of the same event collected from multiple sources and filling in missing information such as dates and locations).

[0464] Step 4:

[0465] The server then analyzes the cleaned data using natural language processing techniques to extract key features, such as the date, time, location, and people involved in a particular battle.

[0466] Step 5:

[0467] The server trains the generative AI model based on the analyzed data. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[0468] Step 6:

[0469] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[0470] Step 7:

[0471] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0472] Step 8:

[0473] Users can provide feedback to the system about the generated creations, such as "I'd like more detailed descriptions of tactics" or "I'd like the expression of emotions to be improved."

[0474] Step 9:

[0475] The server readjusts the generative AI model based on user feedback. Based on the feedback, it readjusts the model parameters to improve the accuracy and quality of the generated results. For example, it adds training data to enhance emotional expression and retrains the model.

[0476] Step 10:

[0477] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[0478] Through this series of steps, the system of the present invention can effectively generate highly accurate and high-quality creative works about the past that meet the needs of the user.

[0479] Example 1

[0480] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0481] With conventional technologies, it is difficult to efficiently collect detailed historical information, analyze it in a unified manner, and automatically generate highly satisfying creative works using a highly accurate generative AI model. Furthermore, there is a lack of a process for incorporating user feedback on the generated content and continuously improving the generative AI model, which limits the improvement in the quality of the deliverables. Therefore, there is an urgent need to resolve these issues.

[0482] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0483] In this invention, the server includes a means for users to upload documents using their terminals, a means for the server to extract features from the cleaned data using natural language processing technology, a means for the server to classify the data by specific topic, and a means for the server to generate creative works using a Transformer model. This allows users to efficiently create high-quality creative works based on detailed historical information. Furthermore, by incorporating user feedback, the generative AI model can be continuously improved, thereby increasing the accuracy and quality of the generated content.

[0484] "Data collection means" means a means that has the function of uploading documents and records provided by users and automatically collecting related information from the Internet and databases.

[0485] "Means for cleaning collected data" means means for converting collected data into a consistent format, removing duplicate content, and completing missing information.

[0486] The "means for analyzing cleaned data and extracting important features" refers to a means having a function for extracting important features from cleaned data using natural language processing technology.

[0487] "Means for training a generative AI model" refers to means that trains a generative AI model based on extracted feature data and has the function of improving the accuracy of information generation.

[0488] "Means for generating information in response to a user's request using a trained model" means means that have the function of generating information using a trained generative AI model based on a specific request from a user.

[0489] The "means for obtaining user feedback" refers to a means having a function for collecting user evaluations and opinions about the generated information.

[0490] "Means for readjusting the generative AI model" refers to means that have the function of retraining or adjusting the generative AI model based on obtained user feedback to improve the accuracy and quality of the generated content.

[0491] "Means for users to upload documents using their terminals" refers to means that allow users to send documents and records to the system from the devices they use.

[0492] "Natural language processing technology" is a technology that allows computers to process and understand natural language used by humans.

[0493] A "Transformer model" is a neural network architecture that has particularly powerful capabilities in natural language processing, and is a model for context-aware text generation and translation.

[0494] A "means for categorizing data by specific topics" is a means that has the ability to organize and group cleaned data based on specific themes or categories.

[0495] MODE FOR CARRYING OUT THE INVENTION

[0496] The present invention is a system for improving the accuracy and quality of past creative works using generative AI. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, and a generative AI model readjustment means. An embodiment of the present invention will be described in detail below.

[0497] Data collection

[0498] A user uploads relevant documents and records to the system to create a creative work using their terminal. This process is performed using the user's device (PC, smartphone, etc.). For example, a user may upload historical documents in PDF format to the system in order to write a novel about World War II.

[0499] The server then automatically collects relevant material from the internet and existing databases based on the uploaded documents, using Python libraries (e.g., BeautifulSoup) to scrape relevant information from websites and retrieve information from online academic paper databases and encyclopedias.

[0500] Data Cleaning

[0501] The server converts the collected data into a consistent format, removes duplicates, and completes missing information. It converts the data into a data frame using the Pandas library and standardizes it to a consistent date and time format. It also uses a natural language processing library (e.g., spaCy) to automatically complete missing information.

[0502] Data analysis

[0503] The server uses natural language processing techniques to extract important features from the cleaned data, such as topic modeling (e.g., LDA) to extract features related to specific events or people.

[0504] The server then classifies the data into relevant topics based on the extracted features, using a clustering algorithm (e.g., K-means) to categorize the data into categories such as battle scenes, diplomatic events, and daily life.

[0505] Model training

[0506] The server trains a generative AI model based on the cleaned feature data. Using Tensorflow, we train an RNN-based generative AI model to generate battle scenes and dialogue.

[0507] The server then adjusts and optimizes the model parameters, performing hyperparameter tuning to determine the optimal learning rate and number of epochs to improve generation accuracy.

[0508] Creation

[0509] A user enters a specific request into the system, such as "Please portray a scene from the Normandy landings in 1944," into a web form.

[0510] The server uses trained generative AI models to generate depictions based on user requests, using Transformer models to detail soldier movements and scenery from the Normandy landings.

[0511] User feedback and refinements

[0512] Users can provide feedback on the generated creations, such as "more detailed tactical descriptions are needed" or "improvement in emotional expression."

[0513] The server retunes the generative AI model based on user feedback, retraining the model with additional training data to improve the accuracy and quality of the generated content.

[0514] Prompt Sentence Examples

[0515] "Please describe the Gallipoli landings on 25 April 1915. Specifically, we are looking for a description of the landing boats approaching the shore, the tension among the soldiers, and the fighting that followed."

[0516] Through the system of the present invention, users can efficiently create high-quality creative works based on detailed historical information, and by incorporating user feedback, the generative AI model can be continuously improved, thereby increasing the accuracy and quality of the generated content.

[0517] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0518] Step 1:

[0519] Users upload documents and records to the system.

[0520] Input: Literature data such as PDFs or text files selected by the user from their device.

[0521] Specific operation: The user uses the system's upload function from their PC or smartphone to select historical documents and records and send them to the server.

[0522] Output: Uploaded bibliographic data stored on the server.

[0523] Step 2:

[0524] The server collects additional data from the internet and databases.

[0525] Input: A search query based on the uploaded literature data.

[0526] What it does: The server uses Python's BeautifulSoup library to scrape information from relevant websites and retrieve relevant data from sources like online academic paper databases.

[0527] Output: Additional data collected (e.g., text of web pages, abstracts of matching academic papers).

[0528] Step 3:

[0529] The server cleans the collected data.

[0530] Input: The raw text of the collected data.

[0531] What it does: The server uses the Pandas library to convert the data into a data frame, removes duplicates, converts to a consistent date and time format, and imputes missing data. It also uses a natural language processing library (e.g., spaCy) to auto-complete missing information.

[0532] Output: A cleaned and consistent dataset.

[0533] Step 4:

[0534] The server parses the cleaned data.

[0535] Input: The cleaned dataset.

[0536] Specific operations: Conduct topic modeling (e.g., LDA) to extract important features (e.g., involvement of specific events or people), and use machine learning algorithms to classify topics.

[0537] Output: Extracted features and classified topics.

[0538] Step 5:

[0539] The server trains the generative AI model.

[0540] Input: Extracted feature data.

[0541] What it does: Use Tensorflow to train an RNN-based generative AI model to generate specific scenarios and dialogues, and use hyperparameter tuning to select optimal model settings.

[0542] Output: A trained generative AI model.

[0543] Step 6:

[0544] A user inputs a specific request into the system.

[0545] Input: A prompt from the user (e.g., "Please describe a scene from the Normandy landings in 1944").

[0546] Specific action: The user enters a prompt sentence about a specific scene or situation into a web form and submits it to the system.

[0547] Output: The prompt received by the server.

[0548] Step 7:

[0549] The server generates information based on the user's request.

[0550] Input: A trained generative AI model and a user prompt.

[0551] Specific behavior: Use trained generative AI models to generate specific scenes and dialogues based on prompts, and use Transformer models to generate context-aware text.

[0552] Output: The generated text in response to the user's request.

[0553] Step 8:

[0554] The user provides feedback on the generated information.

[0555] Input: Generated text and user rating.

[0556] Specific actions: The user evaluates the generated text and inputs suggestions for improvement, such as detailed tactical descriptions and emotional expressions.

[0557] Output: Feedback provided by the user.

[0558] Step 9:

[0559] The server readjusts the generative AI model based on the feedback.

[0560] Input: User feedback and retraining data.

[0561] What it does: Based on the feedback, it retrains the generative AI model with additional training data and optimizes the model's parameters.

[0562] Output: An improved generative AI model.

[0563] Through these steps, the system can generate high-quality creative works tailored to the user's needs and continuously improve its accuracy and quality.

[0564] (Application example 1)

[0565] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0566] Previous systems using generative AI technology struggled to sufficiently improve the accuracy and quality of historical creations. They also lacked a process for effectively utilizing user feedback to optimize generative AI models. Furthermore, they lacked a means to visually display the generated information and improve the user experience.

[0567] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0568] In this invention, the server includes a data collection means, a means for cleaning the collected data, a means for analyzing the cleaned data and extracting important features, a means for training a generative AI model based on the extracted features, a means for generating information according to a user request using the trained model, a means for obtaining user feedback on the generated information, a means for readjusting the generative AI model based on the obtained feedback, and a means for visually displaying the generated information. This enables improvement in the accuracy and quality of historical creations using generative AI, optimization of the model based on effective feedback, and an improvement in user experience.

[0569] "Data collection methods" are methods for automatically collecting relevant literature, records, and papers from the Internet and databases.

[0570] "Data cleaning procedures" are procedures that convert collected data into a consistent format, remove duplicate content, and complete missing information.

[0571] "Data analysis tools" are tools for extracting important features from cleaned data and classifying the data by related topics.

[0572] The "generative AI model training means" is a means for training a generative AI model based on extracted features and optimizing the model parameters.

[0573] "Information generation means" refers to a means for generating information in response to a user request using a trained generative AI model.

[0574] The "feedback acquisition means" is a means for acquiring user feedback on the generated information.

[0575] The "readjustment means" is a means for readjusting the generative AI model based on the obtained feedback.

[0576] The "visual display means" is a means for visually displaying the generated information.

[0577] The present invention relates to a system for improving the accuracy and quality of historical fiction using generative AI, which is implemented using several means:

[0578] First, the server uses a data collection tool to automatically collect relevant literature, records, and papers from the Internet and databases. The software used for this is the Python requests library, which retrieves data through an API.

[0579] The collected data is then converted into a consistent format through data cleaning procedures, using tools such as Pandas to remove duplicate content and fill in missing information, for example by unifying descriptions of the same event from multiple documents and filling in missing date information.

[0580] The cleaned data is then analyzed using natural language processing techniques to extract key features, such as the date, time, location, and people involved in a particular historical event, using NLP libraries such as spaCy.

[0581] Based on the extracted features, the server trains a generative AI model using a machine learning framework such as TensorFlow or PyTorch, training the model to generate specific combat scenes or dialogues.

[0582] The server then uses the trained model to generate information based on the user's request. For example, if a user requests, "Generate a detailed description of the Gallipoli Landings on April 25, 1915," the generative AI model will generate a detailed description of the scene.

[0583] The feedback acquisition means acquires user feedback on the generated information, such as "I would like more detailed descriptions of tactics to be added."

[0584] Finally, the recalibration means recalibrates the generative AI model based on the obtained feedback. In this process, the training dataset is reconstructed based on the collected feedback and the model is retrained.

[0585] Using visual display means, the generated information is visually displayed on a user device. For example, it can be visually confirmed through a smartphone or head-mounted display. As a result, the accuracy and quality of historical creations using generative AI can be improved, models can be optimized based on effective feedback, and the user experience can be improved.

[0586] Example prompt sentence:

[0587] Generate a depiction of the Allied landing forces approaching the beaches of Gallipoli at sunrise on April 25, 1915. Detail specific soldier movements, the scenery, and emotions.

[0588] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0589] Step 1:

[0590] Data collection

[0591] The server uses data collection tools to automatically collect relevant literature, records, and papers from the Internet and databases, for example, using the Python requests library to retrieve data through APIs. The input is the relevant URL or API endpoint, and the output is the retrieved raw data.

[0592] Step 2:

[0593] Data Cleaning

[0594] The server uses data cleaning techniques to convert the collected data into a consistent format, remove duplicates, and fill in missing information. Specifically, it converts the data into a data frame using Pandas and removes duplicates using the .drop_duplicates() method. The input is raw data, and the output is cleaned data.

[0595] Step 3:

[0596] Data analysis

[0597] The server then uses data analysis tools and natural language processing techniques to extract important features from the cleaned data. For example, it uses an NLP library such as spaCy to extract proper nouns such as specific events or people. The input is the cleaned data, and the output is the data with important features extracted.

[0598] Step 4:

[0599] Training generative AI models

[0600] The server trains a generative AI model based on the extracted features. Using a machine learning framework such as TensorFlow or PyTorch, the model is trained to generate specific dialogues or scenes. The input is the extracted feature data, and the output is a trained generative AI model.

[0601] Step 5:

[0602] information generation

[0603] The server uses the trained model to generate information based on the user's request. The user inputs a specific prompt, and the generated information is a detailed description. The input is the user's prompt, and the output is the generated text content. For example, the prompt "Please generate a detailed description of the Gallipoli landings on April 25, 1915" is input, and a specific description is generated.

[0604] Step 6:

[0605] Get feedback

[0606] The server obtains user feedback on the generated information. It uses a feedback form or questionnaire to collect user opinions and correction requests. The input is the user's feedback content, and the output is the feedback data.

[0607] Step 7:

[0608] Model Rebalancing

[0609] The server retunes the generative AI model based on the feedback it receives. It reconstructs the training dataset based on the collected feedback and retrains the model. The input is the feedback data, and the output is the retuned generative AI model.

[0610] Step 8:

[0611] Visual Indication

[0612] The device visually displays the generated information using a visual display means. The user can view the generated detailed depiction through a smartphone or head-mounted display. The input is the generated text content, and the output is the visually displayed information.

[0613] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0614] The present invention further combines a system for improving the accuracy and quality of past creative works using generative AI with an emotion engine. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, a generative AI model readjustment means, and an emotion engine. An embodiment of this system is shown below.

[0615] 1. Data Collection

[0616] Users upload relevant documents and records to create their creative works. For example, if a user is writing a novel about World War I, they provide the system with digital files of historical documents they have on hand.

[0617] The server automatically collects relevant materials from the Internet and databases, for example, retrieves relevant data from online academic paper databases and encyclopedias, and incorporates them into the system.

[0618] 2. Data Cleaning

[0619] The server converts the collected data into a consistent format, removes duplicates, and fills in missing information, for example by unifying descriptions of the same event collected from multiple sources and filling in missing information such as dates and locations.

[0620] 3. Data Analysis

[0621] The server uses natural language processing techniques to extract key features from the cleaned data, such as the date, time, location, and people involved in a particular battle.

[0622] The server categorizes the data by relevant topic for later generation, for example by stage of a war or important event.

[0623] 4. Model training

[0624] The server trains a generative AI model based on the extracted data features. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[0625] The server optimizes the model's parameters to generate creations tailored to the user's needs, such as enhancing detailed tactical descriptions or emotional expression.

[0626] 5. Creation of creative works

[0627] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[0628] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0629] 6. User Feedback and Emotion Recognition

[0630] The user provides feedback on the generated creation, for example, "I would like more detailed descriptions of tactics added" or "I would like the expression of emotions improved."

[0631] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, it extracts positive, negative, and neutral emotions from the feedback content and further analyzes the user's detailed emotional state (e.g., satisfaction, dissatisfaction, surprise, etc.).

[0632] 7. Readjust based on feedback

[0633] The server readjusts the generative AI model based on user feedback and the results of emotion analysis by the emotion engine. For example, if the emotion engine detects the emotion "dissatisfaction," it identifies the cause of that emotion and optimizes the model parameters accordingly.

[0634] 8. Final generation and verification

[0635] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[0636] Check whether the user is satisfied with the final generated creation and provide further feedback if necessary to iterate and improve the quality of the process.

[0637] In this way, the system of the present invention not only uses generative AI to generate creative works about the past, but also uses an emotion engine to analyze user emotions and readjust the model based on feedback, thereby providing even more accurate and high-quality creative works. For example, when creating a novel or movie scenario about World War I, this system can generate detailed and accurate battle scenes and character dialogue, increasing the appeal to readers and viewers.

[0638] The processing flow will be explained below.

[0639] Step 1:

[0640] Users upload relevant documents and records to the system from their devices in order to create their creative works. For example, a user may submit digital files of historical documents they have on hand to write a novel about World War I.

[0641] Step 2:

[0642] The server automatically collects relevant materials from the Internet and databases, for example, retrieving relevant information from online academic paper databases and encyclopedias, and integrating it into the system.

[0643] Step 3:

[0644] The server cleans the collected data, which includes standardizing the data format, removing duplicate content, and completing missing information (for example, standardizing descriptions of the same event collected from multiple sources and completing missing information such as dates and locations).

[0645] Step 4:

[0646] The server then uses natural language processing techniques to extract key features from the cleaned data, such as the date, time, location, and people involved in a particular battle.

[0647] Step 5:

[0648] The server trains the generative AI model based on the analyzed data. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[0649] Step 6:

[0650] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[0651] Step 7:

[0652] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0653] Step 8:

[0654] The user provides feedback to the system about the generated creation, such as "I would like more detailed descriptions of tactics" or "I would like the expression of emotions to be improved."

[0655] Step 9:

[0656] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, it extracts positive, negative, and neutral emotions from the feedback content and analyzes more detailed emotional states (e.g., satisfaction, dissatisfaction, surprise, etc.).

[0657] Step 10:

[0658] The server readjusts the generative AI model based on user feedback and the results of emotion analysis. For example, if the emotion engine detects the emotion "dissatisfaction," it identifies the cause of that emotion and readjusts the model parameters accordingly.

[0659] Step 11:

[0660] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[0661] Step 12:

[0662] Check whether the user is satisfied with the final generated creation and provide further feedback if necessary to iterate and improve the quality of the process.

[0663] Through these steps, the system of the present invention can provide highly accurate and high-quality creative works by analyzing users' emotions in addition to the process of generating creative works related to the past using generative AI.

[0664] Example 2

[0665] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0666] Conventional creative creation systems using generative AI models lack the ability to analyze user feedback and readjust the model based on that feedback. As a result, the accuracy and quality of the created creations often do not meet user requirements. In particular, it is difficult to generate creations that appropriately reflect the user's emotions, and there is a need to improve user satisfaction.

[0667] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0668] In this invention, the server includes a data collection means, a means for cleaning the collected data into a consistent format, removing duplication, and completing missing information, a means for analyzing the cleaned data and extracting important features using natural language processing technology, a means for training a generative AI model based on the extracted features and optimizing model parameters, a means for generating information according to a user request using the trained model, a means for acquiring user feedback on the generated information, analyzing the feedback using an emotion engine, and recognizing the user's emotions, and a means for readjusting the generative AI model based on the acquired feedback and the emotion analysis results. This makes it possible to generate creative works with high accuracy and quality by appropriately readjusting the generative AI model based on the user's feedback and emotions.

[0669] "Data collection means" refers to the means by which user-provided documents and records are uploaded to the system and further related materials are automatically collected from the internet and databases.

[0670] "Data cleaning procedures" are procedures used to convert collected data into a consistent format, remove duplication, and complete missing information.

[0671] The "data analysis means" is a means for extracting important features from the cleaned data using natural language processing technology and classifying the data based on those features.

[0672] A "generative AI model" is an artificial intelligence model that is trained based on cleaned and analyzed data to generate information in response to user requests.

[0673] The "feedback acquisition means" is a means for collecting user feedback on the generated information or creation.

[0674] An "emotion engine" is a means for analyzing feedback provided by a user and recognizing the user's emotions based on that feedback.

[0675] "Model readjustment means" refers to a means for readjusting the generative AI model based on the obtained feedback and sentiment analysis results to improve its accuracy and quality.

[0676] The present invention provides a creative creation system that uses a generative AI model and an emotion engine to generate accurate and high-quality creative works based on user-provided documents and records. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a creative creation means, a feedback acquisition means, an emotion engine, and a model readjustment means.

[0677] An embodiment of this system is described below.

[0678] 1. Data Collection

[0679] A user uploads relevant documents and records related to their creative work to the system. For example, a user is writing a novel about World War I and provides the system with digital files (PDF, Word files, etc.) of historical documents.

[0680] The server collects relevant materials from internet resources and existing databases, for example, pulling the necessary information from Google Scholar or Wikipedia, and storing it in the system.

[0681] 2. Data Cleaning

[0682] The server converts the collected data into a consistent format, removes duplicates, and completes missing information, for example, unifying documents with different formats, removing duplicate content, and adding missing dates and other details.

[0683] 3. Data Analysis

[0684] The server uses natural language processing (NLP) techniques to extract key features from the cleaned data, specifically through text analysis, to extract information such as the date, time, location, and people involved in a particular battle.

[0685] The server uses the extracted features to categorize the data into related topics, for example, organizing data by phases of a war (start, major battles, intermediate ceasefires, etc.) or important events.

[0686] 4. Training the generative AI model

[0687] The server trains a generative AI model based on features extracted from the data, focusing on the scene the user wants to depict, for example, using data on the landing scene of the Gallipoli War.

[0688] The server optimizes the model parameters to generate a creation that meets the user's needs, for example by setting specific parameters to enhance tactical detail or emotional expression.

[0689] 5. Creation of creative works

[0690] The user inputs a specific request into the system, for example, a prompt such as "Please describe in detail the scenes from the Gallipoli landings on April 25, 1915."

[0691] The server uses a trained generative AI model to generate depictions based on user requests, such as detailed descriptions of soldier movements and scenery during a landing scene.

[0692] 6. User Feedback and Emotion Recognition

[0693] Users can provide feedback on the generated creation, for example by entering specific feedback in text, such as "I would like more detailed tactical descriptions."

[0694] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, the server can analyze the emotion of "dissatisfaction" extracted from the feedback, such as "I want more detailed tactical descriptions."

[0695] 7. Re-adjusting the model based on feedback

[0696] The server readjusts the generative AI model based on the feedback and emotion analysis results it receives. For example, if the emotion "dissatisfaction" is detected, the cause is identified and the model parameters are optimized.

[0697] 8. Final generation and verification

[0698] The server then uses the retuned model to generate the creation again and provides it to the user. Specifically, the adjusted model is used to generate a more detailed and emotionally rich Gallipoli War landing scene.

[0699] The user checks whether they are satisfied with the final generated creation and provides feedback again if necessary to iterate the process and improve the quality.

[0700] In this way, the system of the present invention can efficiently generate highly accurate and high-quality creative works by readjusting the generative AI model based on feedback provided by the user and the results of emotion analysis.

[0701] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0702] Step 1:

[0703] Data collection

[0704] Input: User-provided digital files of documents and records (e.g. PDF, Word files).

[0705] Specific operation: The user submits historical documents and record files they have on hand using the system's upload function.

[0706] Output: Literature and records stored in the system's database.

[0707] The server automatically collects relevant literature, records, and papers from the Internet and databases, specifically by accessing open data sources such as Google Scholar and Wikipedia, and stores the necessary information in the system.

[0708] Step 2:

[0709] Cleaning the data

[0710] Input: Collected literature and archival data.

[0711] What it does: It converts the data collected by the server into a consistent format (e.g., UTF-8 encoded text), detects and removes duplicate data, and fills in missing information (e.g., by filling in missing details like date or location from other resources).

[0712] Output: Consistent, deduplicated, and imputed bibliographic and archival data.

[0713] Step 3:

[0714] Data analysis

[0715] Input: Cleaned data.

[0716] What it does: The server uses natural language processing (NLP) techniques to extract important features from the data (e.g., battle dates, locations, and people involved), and then categorizes the data by related topics based on the extracted features. For example, it organizes data about each stage of a war or key events separately.

[0717] Output: Data with important features extracted and categorized by relevant topics.

[0718] Step 4:

[0719] Training generative AI models

[0720] Input: Extracted feature data.

[0721] How it works: The server trains a generative AI model based on the extracted feature data and optimizes it to generate information based on the user's request. For example, if a user wants a detailed depiction of the landing scene from the Gallipoli War, the server trains the model using data related to that scene.

[0722] Output: A trained generative AI model.

[0723] Step 5:

[0724] Creation

[0725] Input: The specific request (prompt) that the user provides to the system.

[0726] What it does: The user enters a prompt, such as "Please give me a detailed description of the Gallipoli Landings on April 25, 1915." The server uses a trained generative AI model to generate sentences and scene descriptions based on the request.

[0727] Output: The creative work generated based on the user's request.

[0728] Step 6:

[0729] User Feedback and Emotion Recognition

[0730] Input: Feedback provided by the user about the generated creation.

[0731] Specific operation: The user inputs feedback in text format, such as "I would like more detailed tactical descriptions." The server uses the emotion engine to analyze the feedback, extracts the user's emotion (e.g., "dissatisfied"), and analyzes it in detail.

[0732] Output: Parsed feedback and user sentiment data.

[0733] Step 7:

[0734] Retune the model based on feedback

[0735] Input: Feedback and sentiment analysis results.

[0736] Specific operation: The server readjusts the generative AI model based on the feedback and emotion analysis results it receives. For example, if the emotion "dissatisfaction" is detected, it identifies the cause of this and adjusts the model parameters.

[0737] Output: A retuned generative AI model.

[0738] Step 8:

[0739] Final generation and confirmation

[0740] Input: A retuned generative AI model and a new user request.

[0741] What happens: The server uses the refined model to generate new drawings and sentences, and serves them to the user. The server checks whether the user is satisfied with the final generated work, providing further feedback if necessary.

[0742] Output: High-quality creations that satisfy users.

[0743] (Application example 2)

[0744] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0745] Conventional generative AI models have limitations in their ability to generate content based on past data, making it particularly difficult to adjust the content to reflect emotional expressions and user feedback. As a result, the quality of the generated content varies and user feedback cannot be properly utilized. Furthermore, when users request specific scenes or historical events, it is difficult to respond to their requests accurately and in detail.

[0746] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a data collection means, a data cleaning means, and a means for analyzing the cleaned data and extracting important features. This enables the collected data to be organized into a consistent format and necessary features to be extracted. The server also includes a means for training a generative AI model, a means for generating information in response to a user request, a means for obtaining user feedback on the generated information, and a means for analyzing the user's emotional feedback using an emotion analysis means and reflecting the analysis results in readjusting the model. This allows the generated content to reflect user feedback and include emotional elements, making it possible to provide higher quality content.

[0747] "Data collection means" has the function of automatically collecting data provided by users and related materials from the Internet and databases.

[0748] "Data cleaning procedures" are procedures that convert collected data into a consistent format, remove duplicate content, and complete missing information.

[0749] "Data analysis means" extracts important features from the cleaned data so that they can be used in subsequent processes.

[0750] The "means for training a generative AI model" refers to optimizing the generative AI model based on the extracted features so that it can generate specific content.

[0751] An "information generation means" is a means that uses a trained generative AI model to generate information or content in response to a user's request.

[0752] The "user feedback acquisition means" has a function of collecting user opinions and impressions regarding the generated information.

[0753] The "means for readjusting the generative AI model" refers to the re-optimization of the generative AI model based on feedback obtained from users and the results of sentiment analysis, in order to improve its accuracy and quality.

[0754] The "emotion analysis means" analyzes the user's feedback from an emotional perspective and identifies the emotional state, such as positive, negative, or neutral.

[0755] This invention is a system for improving the accuracy and quality of past creative works using generative AI combined with an emotion engine. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, an information generation means, a user feedback acquisition means, a generative AI model readjustment means, and an emotion analysis means.

[0756] System Programs and Processing

[0757] Hardware and software used

[0758] Hardware: Server with high-performance CPU and GPU

[0759] Software: Python, TensorFlow / PyTorch, natural language processing libraries (spaCy, NLTK), sentiment analysis engine API

[0760] Data collection methods

[0761] The server collects documents and records uploaded by users and automatically retrieves related materials from the Internet and databases, thereby comprehensively obtaining the necessary data.

[0762] Data Cleaning Methods

[0763] The collected data is converted into a consistent format, duplicates are removed, and missing information is filled in. This cleaning process improves the quality of the data and makes it easier to analyze.

[0764] Data Analysis Methods

[0765] The server then uses natural language processing technology to analyze the cleaned data and extract important features, such as the date, time, and location of a particular event, as well as the people involved, which are then used in subsequent processes.

[0766] A means of training generative AI models

[0767] The server trains a generative AI model based on the extracted data, and the trained model is optimized to generate specific scenes and interactions requested by the user.

[0768] Information generation means

[0769] The server uses a trained generative AI model to generate information based on requests from users' devices. For example, a user might input a prompt such as, "Please describe the Normandy landings on June 6, 1944."

[0770] User feedback acquisition method

[0771] The user provides feedback on the generated information, which the server receives and uses for further improvement.

[0772] Emotion analysis means

[0773] The server analyzes the user feedback using an emotion engine to extract emotions such as positive, negative, or neutral, and passes the analysis results to the recalibration means.

[0774] Generative AI model readjustment method

[0775] Based on the information obtained from the user feedback acquisition means and the results of the sentiment analysis means, the server readjusts the generative AI model so that the next and subsequent generated content will better meet the user's needs.

[0776] Specific examples

[0777] For example, a user might enter a prompt request such as, "Please describe the Normandy landings on June 6, 1944." Based on this request, the server uses a trained generative AI model to generate a detailed and accurate scene description.

[0778] In this way, the system of the present invention is able to continually evolve and provide high quality creative works through user feedback and sentiment analysis.

[0779] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0780] Step 1: Data collection

[0781] The server uploads documents and records provided by users and automatically collects related materials from the Internet and databases.

[0782] Input: Digital files uploaded by users, data collected from the internet and databases.

[0783] Output: The collected dataset.

[0784] What it does: Users upload literature files about World War I, and the server also collects data from online academic paper databases.

[0785] Step 2: Data cleaning

[0786] The server cleans the collected data, standardizing the format, removing duplicates, and completing missing information.

[0787] Input: The collected dataset.

[0788] Output: A cleaned dataset.

[0789] What it does: The server unifies descriptions of the same event collected from multiple sources and completes missing information such as dates and locations.

[0790] Step 3: Data analysis

[0791] The server then analyzes the cleaned data using natural language processing techniques to extract key features.

[0792] Input: The cleaned dataset.

[0793] Output: Extracted feature dataset.

[0794] What it does: The server extracts data about specific fight scenes, such as the date, time, location, and people involved.

[0795] Step 4: Training the generative AI model

[0796] The server trains a generative AI model based on features extracted through data analysis.

[0797] Input: Extracted feature dataset.

[0798] Output: A trained generative AI model.

[0799] What it does: The server feeds the model with training data so that it can accurately depict scenes from the Gallipoli War.

[0800] Step 5: Information Generation

[0801] The user terminal inputs a prompt sentence, and the server generates information using a trained generative AI model.

[0802] Input: The prompt text entered by the user.

[0803] Output: The generated information (text or scene).

[0804] What it does: The user types in "Please describe the Normandy landings on June 6, 1944," and the server generates the scene.

[0805] Step 6: Get user feedback

[0806] The user provides feedback on the generated information, which is captured by the server.

[0807] Input: User feedback.

[0808] Output: Collected feedback data.

[0809] Specific action: The user enters feedback, such as "I need a detailed tactical description."

[0810] Step 7: Sentiment Analysis

[0811] The server uses an emotion engine to analyze the user feedback and extract positive, negative, or neutral emotions.

[0812] Input: Collected feedback data.

[0813] Output: Sentiment analysis result data.

[0814] Specific behavior: The server detects the emotion "dissatisfied" from the feedback.

[0815] Step 8: Refining the generative AI model

[0816] The server readjusts the generative AI model based on the emotion analysis results and feedback data.

[0817] Input: Sentiment analysis result data, feedback data.

[0818] Output: A retuned generative AI model.

[0819] Specific operation: The server adjusts parameters to "enhance tactical depiction."

[0820] Step 9: Final generation and verification

[0821] The prompt sentence is input again from the user terminal, and information is generated using the re-adjusted model and provided to the user.

[0822] Input: The prompt statement typed again.

[0823] Output: Improved generated information.

[0824] What it does: The user again types in "Describe the Normandy landings on June 6, 1944," generating a more detailed and emotive description of the scene.

[0825] The above are the specific processing steps for carrying out the invention.

[0826] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0827] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0828] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0829] [Third embodiment]

[0830] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0831] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0832] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0833] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0834] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0835] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0836] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0837] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0838] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0839] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0840] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0841] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0842] The present invention relates to a system for improving the accuracy and quality of past creative works using generative AI. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, and a generative AI model readjustment means. An embodiment of the system is described below.

[0843] 1. Data Collection

[0844] Users upload relevant documents and records to create their creative works, for example, providing relevant historical documents to write a novel about World War I.

[0845] The server automatically collects relevant materials from the Internet and databases, for example, retrieving relevant information from online academic journal databases and encyclopedias.

[0846] 2. Data Cleaning

[0847] The server converts the collected data into a consistent format, removes duplicates, and completes missing information, for example, by unifying descriptions of the same event from multiple sources and completing missing dates.

[0848] 3. Data Analysis

[0849] The server uses natural language processing techniques to extract key features from the cleaned data, such as identifying the date, time, location, and people involved in a particular battle.

[0850] The server categorizes the data by relevant topic for later generation, for example by stage of a war or important event.

[0851] 4. Model training

[0852] The server trains a generative AI model based on the extracted data features, for example, training the model to generate specific World War I battle scenes and soldier dialogue.

[0853] The server optimizes the model's parameters to generate creations tailored to the user's needs, such as enhancing detailed tactical descriptions or emotional expression.

[0854] 5. Creation of creative works

[0855] A user inputs a specific request into the system, for example, "Please depict the Gallipoli landings on April 25, 1915."

[0856] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0857] 6. User Feedback and Refinement

[0858] The user provides feedback on the generated creation, for example, "I would like more detailed descriptions of tactics added" or "I would like the expression of emotions improved."

[0859] The server readjusts the generative AI model based on user feedback to improve the accuracy and quality of the generated results, for example by adding training data to enhance emotional expression and readjusting the model parameters.

[0860] By using these methods, the system of the present invention can improve the accuracy and quality of creative works about the past and generate historically accurate and compelling content. For example, when creating a novel or movie scenario about World War I, this system can be used to generate detailed and accurate battle scenes and character dialogue, increasing the appeal to readers and viewers.

[0861] The processing flow will be explained below.

[0862] Step 1:

[0863] Users upload relevant documents and records to create their creative works. For example, if a user is writing a novel about World War I, they provide the system with digital files of historical documents they have on hand.

[0864] Step 2:

[0865] The server automatically collects relevant materials from the Internet and databases, for example, retrieves relevant data from online academic paper databases and encyclopedias, and incorporates them into the system.

[0866] Step 3:

[0867] The server cleans the collected data, which includes converting it into a consistent format, removing duplicates, and filling in missing information (for example, unifying descriptions of the same event collected from multiple sources and filling in missing information such as dates and locations).

[0868] Step 4:

[0869] The server then analyzes the cleaned data using natural language processing techniques to extract key features, such as the date, time, location, and people involved in a particular battle.

[0870] Step 5:

[0871] The server trains the generative AI model based on the analyzed data. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[0872] Step 6:

[0873] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[0874] Step 7:

[0875] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[0876] Step 8:

[0877] Users can provide feedback to the system about the generated creations, such as "I'd like more detailed descriptions of tactics" or "I'd like the expression of emotions to be improved."

[0878] Step 9:

[0879] The server readjusts the generative AI model based on user feedback. Based on the feedback, it readjusts the model parameters to improve the accuracy and quality of the generated results. For example, it adds training data to enhance emotional expression and retrains the model.

[0880] Step 10:

[0881] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[0882] Through this series of steps, the system of the present invention can effectively generate highly accurate and high-quality creative works about the past that meet the needs of the user.

[0883] Example 1

[0884] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0885] With conventional technologies, it is difficult to efficiently collect detailed historical information, analyze it in a unified manner, and automatically generate highly satisfying creative works using a highly accurate generative AI model. Furthermore, there is a lack of a process for incorporating user feedback on the generated content and continuously improving the generative AI model, which limits the improvement in the quality of the deliverables. Therefore, there is an urgent need to resolve these issues.

[0886] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0887] In this invention, the server includes a means for users to upload documents using their terminals, a means for the server to extract features from the cleaned data using natural language processing technology, a means for the server to classify the data by specific topic, and a means for the server to generate creative works using a Transformer model. This allows users to efficiently create high-quality creative works based on detailed historical information. Furthermore, by incorporating user feedback, the generative AI model can be continuously improved, thereby increasing the accuracy and quality of the generated content.

[0888] "Data collection means" means a means that has the function of uploading documents and records provided by users and automatically collecting related information from the Internet and databases.

[0889] "Means for cleaning collected data" means means for converting collected data into a consistent format, removing duplicate content, and completing missing information.

[0890] The "means for analyzing cleaned data and extracting important features" refers to a means having a function for extracting important features from cleaned data using natural language processing technology.

[0891] "Means for training a generative AI model" refers to means that have the function of training a generative AI model based on extracted feature data and improving the accuracy of information generation.

[0892] "Means for generating information in response to a user's request using a trained model" means means that have the function of generating information using a trained generative AI model based on a specific request from a user.

[0893] The "means for obtaining user feedback" refers to a means having a function for collecting user evaluations and opinions about the generated information.

[0894] "Means for readjusting the generative AI model" refers to means that have the function of retraining or adjusting the generative AI model based on obtained user feedback to improve the accuracy and quality of the generated content.

[0895] "Means for users to upload documents using their terminals" refers to means that allow users to send documents and records to the system from the devices they use.

[0896] "Natural language processing technology" is a technology that allows computers to process and understand natural language used by humans.

[0897] A "Transformer model" is a neural network architecture that has particularly powerful capabilities in natural language processing, and is a model for generating and translating text while taking context into account.

[0898] A "means for categorizing data by specific topics" is a means that has the ability to organize and group cleaned data based on specific themes or categories.

[0899] MODE FOR CARRYING OUT THE INVENTION

[0900] The present invention is a system for improving the accuracy and quality of past creative works using generative AI. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, and a generative AI model readjustment means. An embodiment of the present invention will be described in detail below.

[0901] Data collection

[0902] A user uploads relevant documents and records to the system to create a creative work using their terminal. This process is performed using the user's device (PC, smartphone, etc.). For example, a user may upload historical documents in PDF format to the system in order to write a novel about World War II.

[0903] The server then automatically collects relevant material from the internet and existing databases based on the uploaded documents, using Python libraries (e.g., BeautifulSoup) to scrape relevant information from websites and retrieve information from online academic paper databases and encyclopedias.

[0904] Data Cleaning

[0905] The server converts the collected data into a consistent format, removes duplicates, and completes missing information. It converts the data into a data frame using the Pandas library and standardizes it to a consistent date and time format. It also uses a natural language processing library (e.g., spaCy) to automatically complete missing information.

[0906] Data analysis

[0907] The server uses natural language processing techniques to extract important features from the cleaned data, such as topic modeling (e.g., LDA) to extract features related to specific events or people.

[0908] The server then classifies the data into relevant topics based on the extracted features, using a clustering algorithm (e.g., K-means) to categorize the data into categories such as battle scenes, diplomatic events, and daily life.

[0909] Model training

[0910] The server trains a generative AI model based on the cleaned feature data. Using Tensorflow, we train an RNN-based generative AI model to generate battle scenes and dialogue.

[0911] The server then adjusts and optimizes the model parameters, performing hyperparameter tuning to determine the optimal learning rate and number of epochs to improve generation accuracy.

[0912] Creation

[0913] A user enters a specific request into the system, such as "Please portray a scene from the Normandy landings in 1944," into a web form.

[0914] The server uses trained generative AI models to generate depictions based on user requests, using Transformer models to detail soldier movements and scenery from the Normandy landings.

[0915] User feedback and refinements

[0916] Users can provide feedback on the generated creations, such as "more detailed tactical descriptions are needed" or "improvement in emotional expression."

[0917] The server retunes the generative AI model based on user feedback, retraining the model with additional training data to improve the accuracy and quality of the generated content.

[0918] Prompt Sentence Examples

[0919] "Please describe the Gallipoli landings on 25 April 1915. Specifically, we are looking for a description of the landing boats approaching the shore, the tension among the soldiers, and the fighting that followed."

[0920] Through the system of the present invention, users can efficiently create high-quality creative works based on detailed historical information, and by incorporating user feedback, the generative AI model can be continuously improved, thereby increasing the accuracy and quality of the generated content.

[0921] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0922] Step 1:

[0923] Users upload documents and records to the system.

[0924] Input: Literature data such as PDFs or text files selected by the user from their device.

[0925] Specific operation: The user uses the system's upload function from their PC or smartphone to select historical documents and records and send them to the server.

[0926] Output: Uploaded bibliographic data stored on the server.

[0927] Step 2:

[0928] The server collects additional data from the internet and databases.

[0929] Input: A search query based on the uploaded literature data.

[0930] What it does: The server uses Python's BeautifulSoup library to scrape information from relevant websites and retrieve relevant data from sources like online academic paper databases.

[0931] Output: Additional data collected (e.g., text of web pages, abstracts of matching academic papers).

[0932] Step 3:

[0933] The server cleans the collected data.

[0934] Input: The raw text of the collected data.

[0935] What it does: The server uses the Pandas library to convert the data into a data frame, removes duplicates, converts to a consistent date and time format, and imputes missing data. It also uses a natural language processing library (e.g., spaCy) to auto-complete missing information.

[0936] Output: A cleaned and consistent dataset.

[0937] Step 4:

[0938] The server parses the cleaned data.

[0939] Input: The cleaned dataset.

[0940] Specific operations: Conduct topic modeling (e.g., LDA) to extract important features (e.g., involvement of specific events or people), and use machine learning algorithms to classify topics.

[0941] Output: Extracted features and classified topics.

[0942] Step 5:

[0943] The server trains the generative AI model.

[0944] Input: Extracted feature data.

[0945] What it does: Use Tensorflow to train an RNN-based generative AI model to generate specific scenarios and dialogues, and use hyperparameter tuning to select optimal model settings.

[0946] Output: A trained generative AI model.

[0947] Step 6:

[0948] A user inputs a specific request into the system.

[0949] Input: A prompt from the user (e.g., "Please describe a scene from the Normandy landings in 1944").

[0950] Specific action: The user enters a prompt sentence about a specific scene or situation into a web form and submits it to the system.

[0951] Output: The prompt received by the server.

[0952] Step 7:

[0953] The server generates information based on the user's request.

[0954] Input: A trained generative AI model and a user prompt.

[0955] Specific behavior: Use trained generative AI models to generate specific scenes and dialogues based on prompts, and use Transformer models to generate context-aware text.

[0956] Output: The generated text in response to the user's request.

[0957] Step 8:

[0958] The user provides feedback on the generated information.

[0959] Input: Generated text and user rating.

[0960] Specific actions: The user evaluates the generated text and inputs suggestions for improvement, such as detailed tactical descriptions and emotional expressions.

[0961] Output: Feedback provided by the user.

[0962] Step 9:

[0963] The server readjusts the generative AI model based on the feedback.

[0964] Input: User feedback and retraining data.

[0965] What it does: Based on the feedback, it retrains the generative AI model with additional training data and optimizes the model's parameters.

[0966] Output: An improved generative AI model.

[0967] Through these steps, the system can generate high-quality creative works tailored to the user's needs and continuously improve its accuracy and quality.

[0968] (Application example 1)

[0969] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0970] Previous systems using generative AI technology struggled to sufficiently improve the accuracy and quality of historical creations. They also lacked a process for effectively utilizing user feedback to optimize generative AI models. Furthermore, they lacked a means to visually display the generated information and improve the user experience.

[0971] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0972] In this invention, the server includes a data collection means, a means for cleaning the collected data, a means for analyzing the cleaned data and extracting important features, a means for training a generative AI model based on the extracted features, a means for generating information according to a user request using the trained model, a means for obtaining user feedback on the generated information, a means for readjusting the generative AI model based on the obtained feedback, and a means for visually displaying the generated information. This enables improvement in the accuracy and quality of historical creations using generative AI, optimization of the model based on effective feedback, and an improvement in user experience.

[0973] "Data collection methods" are methods for automatically collecting relevant literature, records, and papers from the Internet and databases.

[0974] "Data cleaning procedures" are procedures that convert collected data into a consistent format, remove duplicate content, and complete missing information.

[0975] "Data analysis tools" are tools for extracting important features from cleaned data and classifying the data by related topics.

[0976] The "generative AI model training means" is a means for training a generative AI model based on extracted features and optimizing the model parameters.

[0977] "Information generation means" refers to a means for generating information in response to a user request using a trained generative AI model.

[0978] The "feedback acquisition means" is a means for acquiring user feedback on the generated information.

[0979] The "readjustment means" is a means for readjusting the generative AI model based on the obtained feedback.

[0980] The "visual display means" is a means for visually displaying the generated information.

[0981] The present invention relates to a system for improving the accuracy and quality of historical fiction using generative AI, which is implemented using several means:

[0982] First, the server uses a data collection tool to automatically collect relevant literature, records, and papers from the Internet and databases. The software used for this is the Python requests library, which retrieves data through an API.

[0983] The collected data is then converted into a consistent format through data cleaning procedures, using tools such as Pandas to remove duplicate content and fill in missing information, for example by unifying descriptions of the same event from multiple documents and filling in missing date information.

[0984] The cleaned data is then analyzed using natural language processing techniques to extract key features, such as the date, time, location, and people involved in a particular historical event, using NLP libraries such as spaCy.

[0985] Based on the extracted features, the server trains a generative AI model using a machine learning framework such as TensorFlow or PyTorch, training the model to generate specific combat scenes or dialogues.

[0986] The server then uses the trained model to generate information based on the user's request. For example, if a user requests, "Generate a detailed description of the Gallipoli Landings on April 25, 1915," the generative AI model will generate a detailed description of the scene.

[0987] The feedback acquisition means acquires user feedback on the generated information, such as "I would like more detailed descriptions of tactics to be added."

[0988] Finally, the recalibration means recalibrates the generative AI model based on the obtained feedback. In this process, the training dataset is reconstructed based on the collected feedback and the model is retrained.

[0989] Using visual display means, the generated information is visually displayed on a user device. For example, it can be visually confirmed through a smartphone or head-mounted display. As a result, the accuracy and quality of historical creations using generative AI can be improved, models can be optimized based on effective feedback, and the user experience can be improved.

[0990] Example prompt sentence:

[0991] Generate a depiction of the Allied landing forces approaching the beaches of Gallipoli at sunrise on April 25, 1915. Detail specific soldier movements, the scenery, and emotions.

[0992] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0993] Step 1:

[0994] Data collection

[0995] The server uses data collection tools to automatically collect relevant literature, records, and papers from the Internet and databases, for example, using the Python requests library to retrieve data through APIs. The input is the relevant URL or API endpoint, and the output is the retrieved raw data.

[0996] Step 2:

[0997] Data Cleaning

[0998] The server uses data cleaning techniques to convert the collected data into a consistent format, remove duplicates, and fill in missing information. Specifically, it converts the data into a data frame using Pandas and removes duplicates using the .drop_duplicates() method. The input is raw data, and the output is cleaned data.

[0999] Step 3:

[1000] Data analysis

[1001] The server then uses data analysis tools and natural language processing techniques to extract important features from the cleaned data. For example, it uses an NLP library such as spaCy to extract proper nouns such as specific events or people. The input is the cleaned data, and the output is the data with important features extracted.

[1002] Step 4:

[1003] Training generative AI models

[1004] The server trains a generative AI model based on the extracted features. Using a machine learning framework such as TensorFlow or PyTorch, the model is trained to generate specific dialogues or scenes. The input is the extracted feature data, and the output is a trained generative AI model.

[1005] Step 5:

[1006] information generation

[1007] The server uses the trained model to generate information based on the user's request. The user inputs a specific prompt, and the generated information is a detailed description. The input is the user's prompt, and the output is the generated text content. For example, the prompt "Please generate a detailed description of the Gallipoli landings on April 25, 1915" is input, and a specific description is generated.

[1008] Step 6:

[1009] Get feedback

[1010] The server obtains user feedback on the generated information. It uses a feedback form or questionnaire to collect user opinions and correction requests. The input is the user's feedback content, and the output is the feedback data.

[1011] Step 7:

[1012] Model Rebalancing

[1013] The server retunes the generative AI model based on the feedback it receives. It reconstructs the training dataset based on the collected feedback and retrains the model. The input is the feedback data, and the output is the retuned generative AI model.

[1014] Step 8:

[1015] Visual Indication

[1016] The device visually displays the generated information using a visual display means. The user can view the generated detailed depiction through a smartphone or head-mounted display. The input is the generated text content, and the output is the visually displayed information.

[1017] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1018] The present invention further combines a system for improving the accuracy and quality of past creative works using generative AI with an emotion engine. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, a generative AI model readjustment means, and an emotion engine. An embodiment of this system is shown below.

[1019] 1. Data Collection

[1020] Users upload relevant documents and records to create their creative works. For example, if a user is writing a novel about World War I, they provide the system with digital files of historical documents they have on hand.

[1021] The server automatically collects relevant materials from the Internet and databases, for example, retrieves relevant data from online academic paper databases and encyclopedias, and incorporates them into the system.

[1022] 2. Data Cleaning

[1023] The server converts the collected data into a consistent format, removes duplicates, and fills in missing information, for example by unifying descriptions of the same event collected from multiple sources and filling in missing information such as dates and locations.

[1024] 3. Data Analysis

[1025] The server uses natural language processing techniques to extract key features from the cleaned data, such as the date, time, location, and people involved in a particular battle.

[1026] The server categorizes the data by relevant topic for later generation, for example by stage of a war or important event.

[1027] 4. Model training

[1028] The server trains a generative AI model based on the extracted data features. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[1029] The server optimizes the model's parameters to generate creations tailored to the user's needs, such as enhancing detailed tactical descriptions or emotional expression.

[1030] 5. Creation of creative works

[1031] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[1032] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[1033] 6. User Feedback and Emotion Recognition

[1034] The user provides feedback on the generated creation, for example, "I would like more detailed descriptions of tactics added" or "I would like the expression of emotions improved."

[1035] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, it extracts positive, negative, and neutral emotions from the feedback content and further analyzes the user's detailed emotional state (e.g., satisfaction, dissatisfaction, surprise, etc.).

[1036] 7. Readjust based on feedback

[1037] The server readjusts the generative AI model based on user feedback and the results of emotion analysis by the emotion engine. For example, if the emotion engine detects the emotion "dissatisfaction," it identifies the cause of that emotion and optimizes the model parameters accordingly.

[1038] 8. Final generation and verification

[1039] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[1040] Check whether the user is satisfied with the final generated creation and provide further feedback if necessary to iterate and improve the quality of the process.

[1041] In this way, the system of the present invention not only uses generative AI to generate creative works about the past, but also uses an emotion engine to analyze user emotions and readjust the model based on feedback, thereby providing even more accurate and high-quality creative works. For example, when creating a novel or movie scenario about World War I, this system can generate detailed and accurate battle scenes and character dialogue, increasing the appeal to readers and viewers.

[1042] The processing flow will be explained below.

[1043] Step 1:

[1044] Users upload relevant documents and records to the system from their devices in order to create their creative works. For example, a user may submit digital files of historical documents they have on hand to write a novel about World War I.

[1045] Step 2:

[1046] The server automatically collects relevant materials from the Internet and databases, for example, retrieving relevant information from online academic paper databases and encyclopedias, and integrating it into the system.

[1047] Step 3:

[1048] The server cleans the collected data, which includes standardizing the data format, removing duplicate content, and completing missing information (for example, standardizing descriptions of the same event collected from multiple sources and completing missing information such as dates and locations).

[1049] Step 4:

[1050] The server then uses natural language processing techniques to extract key features from the cleaned data, such as the date, time, location, and people involved in a particular battle.

[1051] Step 5:

[1052] The server trains the generative AI model based on the analyzed data. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[1053] Step 6:

[1054] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[1055] Step 7:

[1056] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[1057] Step 8:

[1058] The user provides feedback to the system about the generated creation, such as "I would like more detailed descriptions of tactics" or "I would like the expression of emotions to be improved."

[1059] Step 9:

[1060] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, it extracts positive, negative, and neutral emotions from the feedback content and analyzes more detailed emotional states (e.g., satisfaction, dissatisfaction, surprise, etc.).

[1061] Step 10:

[1062] The server readjusts the generative AI model based on user feedback and the results of emotion analysis. For example, if the emotion engine detects the emotion "dissatisfaction," it identifies the cause of that emotion and readjusts the model parameters accordingly.

[1063] Step 11:

[1064] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[1065] Step 12:

[1066] Check whether the user is satisfied with the final generated creation and provide further feedback if necessary to iterate and improve the quality of the process.

[1067] Through these steps, the system of the present invention can provide highly accurate and high-quality creative works by analyzing users' emotions in addition to the process of generating creative works related to the past using generative AI.

[1068] Example 2

[1069] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1070] Conventional creative creation systems using generative AI models lack the ability to analyze user feedback and readjust the model based on that feedback. As a result, the accuracy and quality of the created creations often do not meet user requirements. In particular, it is difficult to generate creations that appropriately reflect the user's emotions, and there is a need to improve user satisfaction.

[1071] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1072] In this invention, the server includes a data collection means, a means for cleaning the collected data into a consistent format, removing duplication, and completing missing information, a means for analyzing the cleaned data and extracting important features using natural language processing technology, a means for training a generative AI model based on the extracted features and optimizing model parameters, a means for generating information according to a user request using the trained model, a means for acquiring user feedback on the generated information, analyzing the feedback using an emotion engine, and recognizing the user's emotions, and a means for readjusting the generative AI model based on the acquired feedback and the emotion analysis results. This makes it possible to generate creative works with high accuracy and quality by appropriately readjusting the generative AI model based on the user's feedback and emotions.

[1073] "Data collection means" refers to the means by which user-provided documents and records are uploaded to the system and further related materials are automatically collected from the internet and databases.

[1074] "Data cleaning procedures" are procedures used to convert collected data into a consistent format, remove duplication, and complete missing information.

[1075] The "data analysis means" is a means for extracting important features from the cleaned data using natural language processing technology and classifying the data based on those features.

[1076] A "generative AI model" is an artificial intelligence model that is trained based on cleaned and analyzed data to generate information in response to user requests.

[1077] The "feedback acquisition means" is a means for collecting user feedback on the generated information or creation.

[1078] An "emotion engine" is a means for analyzing feedback provided by a user and recognizing the user's emotions based on that feedback.

[1079] "Model readjustment means" refers to a means for readjusting the generative AI model based on the obtained feedback and sentiment analysis results to improve its accuracy and quality.

[1080] The present invention provides a creative creation system that uses a generative AI model and an emotion engine to generate accurate and high-quality creative works based on user-provided documents and records. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a creative creation means, a feedback acquisition means, an emotion engine, and a model readjustment means.

[1081] An embodiment of this system is described below.

[1082] 1. Data Collection

[1083] A user uploads relevant documents and records related to their creative work to the system. For example, a user is writing a novel about World War I and provides the system with digital files (PDF, Word files, etc.) of historical documents.

[1084] The server collects relevant materials from internet resources and existing databases, for example, pulling the necessary information from Google Scholar or Wikipedia, and storing it in the system.

[1085] 2. Data Cleaning

[1086] The server converts the collected data into a consistent format, removes duplicates, and completes missing information, for example, unifying documents with different formats, removing duplicate content, and adding missing dates and other details.

[1087] 3. Data Analysis

[1088] The server uses natural language processing (NLP) techniques to extract key features from the cleaned data, specifically through text analysis, to extract information such as the date, time, location, and people involved in a particular battle.

[1089] The server uses the extracted features to categorize the data into related topics, for example, organizing data by phases of a war (start, major battles, intermediate ceasefires, etc.) or important events.

[1090] 4. Training the generative AI model

[1091] The server trains a generative AI model based on features extracted from the data, focusing on the scene the user wants to depict, for example, using data on the landing scene of the Gallipoli War.

[1092] The server optimizes the model parameters to generate a creation that meets the user's needs, for example by setting specific parameters to enhance tactical detail or emotional expression.

[1093] 5. Creation of creative works

[1094] The user inputs a specific request into the system, for example, a prompt such as "Please describe in detail the scenes from the Gallipoli landings on April 25, 1915."

[1095] The server uses a trained generative AI model to generate depictions based on user requests, such as detailed descriptions of soldier movements and scenery during a landing scene.

[1096] 6. User Feedback and Emotion Recognition

[1097] Users can provide feedback on the generated creation, for example by entering specific feedback in text, such as "I would like more detailed tactical descriptions."

[1098] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, the server can analyze the emotion of "dissatisfaction" extracted from the feedback, such as "I want more detailed tactical descriptions."

[1099] 7. Re-adjusting the model based on feedback

[1100] The server readjusts the generative AI model based on the feedback and emotion analysis results it receives. For example, if the emotion "dissatisfaction" is detected, the cause is identified and the model parameters are optimized.

[1101] 8. Final generation and verification

[1102] The server then uses the retuned model to generate the creation again and provides it to the user. Specifically, the adjusted model is used to generate a more detailed and emotionally rich Gallipoli War landing scene.

[1103] The user checks whether they are satisfied with the final generated creation and provides feedback again if necessary to iterate the process and improve the quality.

[1104] In this way, the system of the present invention can efficiently generate highly accurate and high-quality creative works by readjusting the generative AI model based on feedback provided by the user and the results of emotion analysis.

[1105] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1106] Step 1:

[1107] Data collection

[1108] Input: User-provided digital files of documents and records (e.g. PDF, Word files).

[1109] Specific operation: The user submits historical documents and record files they have on hand using the system's upload function.

[1110] Output: Literature and records stored in the system's database.

[1111] The server automatically collects relevant literature, records, and papers from the Internet and databases, specifically by accessing open data sources such as Google Scholar and Wikipedia, and stores the necessary information in the system.

[1112] Step 2:

[1113] Cleaning the data

[1114] Input: Collected literature and archival data.

[1115] What it does: It converts the data collected by the server into a consistent format (e.g., UTF-8 encoded text), detects and removes duplicate data, and fills in missing information (e.g., by filling in missing details like date or location from other resources).

[1116] Output: Consistent, deduplicated, and imputed bibliographic and archival data.

[1117] Step 3:

[1118] Data analysis

[1119] Input: Cleaned data.

[1120] What it does: The server uses natural language processing (NLP) techniques to extract important features from the data (e.g., battle dates, locations, and people involved), and then categorizes the data by related topics based on the extracted features. For example, it organizes data about each stage of a war or key events separately.

[1121] Output: Data with important features extracted and categorized by relevant topics.

[1122] Step 4:

[1123] Training generative AI models

[1124] Input: Extracted feature data.

[1125] How it works: The server trains a generative AI model based on the extracted feature data and optimizes it to generate information based on the user's request. For example, if a user wants a detailed depiction of the landing scene from the Gallipoli War, the server trains the model using data related to that scene.

[1126] Output: A trained generative AI model.

[1127] Step 5:

[1128] Creation

[1129] Input: The specific request (prompt) that the user provides to the system.

[1130] What it does: The user enters a prompt, such as "Please give me a detailed description of the Gallipoli Landings on April 25, 1915." The server uses a trained generative AI model to generate sentences and scene descriptions based on the request.

[1131] Output: The creative work generated based on the user's request.

[1132] Step 6:

[1133] User Feedback and Emotion Recognition

[1134] Input: Feedback provided by the user about the generated creation.

[1135] Specific operation: The user inputs feedback in text format, such as "I would like more detailed tactical descriptions." The server uses the emotion engine to analyze the feedback, extracts the user's emotion (e.g., "dissatisfied"), and analyzes it in detail.

[1136] Output: Parsed feedback and user sentiment data.

[1137] Step 7:

[1138] Retune the model based on feedback

[1139] Input: Feedback and sentiment analysis results.

[1140] Specific operation: The server readjusts the generative AI model based on the feedback and emotion analysis results it receives. For example, if the emotion "dissatisfaction" is detected, it identifies the cause of this and adjusts the model parameters.

[1141] Output: A retuned generative AI model.

[1142] Step 8:

[1143] Final generation and confirmation

[1144] Input: A retuned generative AI model and a new user request.

[1145] What happens: The server uses the refined model to generate new drawings and sentences, and serves them to the user. The server checks whether the user is satisfied with the final generated work, providing further feedback if necessary.

[1146] Output: High-quality creations that satisfy users.

[1147] (Application example 2)

[1148] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1149] Conventional generative AI models have limitations in their ability to generate content based on past data, making it particularly difficult to adjust the content to reflect emotional expressions and user feedback. As a result, the quality of the generated content varies and user feedback cannot be properly utilized. Furthermore, when users request specific scenes or historical events, it is difficult to respond to their requests accurately and in detail.

[1150] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a data collection means, a data cleaning means, and a means for analyzing the cleaned data and extracting important features. This enables the collected data to be organized into a consistent format and necessary features to be extracted. The server also includes a means for training a generative AI model, a means for generating information in response to a user request, a means for obtaining user feedback on the generated information, and a means for analyzing the user's emotional feedback using an emotion analysis means and reflecting the analysis results in readjusting the model. This allows the generated content to reflect user feedback and include emotional elements, making it possible to provide higher quality content.

[1151] "Data collection means" has the function of automatically collecting data provided by users and related materials from the Internet and databases.

[1152] "Data cleaning procedures" are procedures that convert collected data into a consistent format, remove duplicate content, and complete missing information.

[1153] "Data analysis means" extracts important features from the cleaned data so that they can be used in subsequent processes.

[1154] The "means for training a generative AI model" refers to optimizing the generative AI model based on the extracted features so that it can generate specific content.

[1155] An "information generation means" is a means that uses a trained generative AI model to generate information or content in response to a user's request.

[1156] The "user feedback acquisition means" has a function of collecting user opinions and impressions regarding the generated information.

[1157] The "means for readjusting the generative AI model" refers to the re-optimization of the generative AI model based on feedback obtained from users and the results of sentiment analysis, in order to improve its accuracy and quality.

[1158] The "emotion analysis means" analyzes the user's feedback from an emotional perspective and identifies the emotional state, such as positive, negative, or neutral.

[1159] This invention is a system for improving the accuracy and quality of past creative works using generative AI combined with an emotion engine. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, an information generation means, a user feedback acquisition means, a generative AI model readjustment means, and an emotion analysis means.

[1160] System Programs and Processing

[1161] Hardware and software used

[1162] Hardware: Server with high-performance CPU and GPU

[1163] Software: Python, TensorFlow / PyTorch, natural language processing libraries (spaCy, NLTK), sentiment analysis engine API

[1164] Data collection methods

[1165] The server collects documents and records uploaded by users and automatically retrieves related materials from the Internet and databases, thereby comprehensively obtaining the necessary data.

[1166] Data Cleaning Methods

[1167] The collected data is converted into a consistent format, duplicates are removed, and missing information is filled in. This cleaning process improves the quality of the data and makes it easier to analyze.

[1168] Data Analysis Methods

[1169] The server then uses natural language processing technology to analyze the cleaned data and extract important features, such as the date, time, and location of a particular event, as well as the people involved, which are then used in subsequent processes.

[1170] A means of training generative AI models

[1171] The server trains a generative AI model based on the extracted data, and the trained model is optimized to generate specific scenes and interactions requested by the user.

[1172] Information generation means

[1173] The server uses a trained generative AI model to generate information based on requests from users' devices. For example, a user might input a prompt such as, "Please describe the Normandy landings on June 6, 1944."

[1174] User feedback acquisition method

[1175] The user provides feedback on the generated information, which the server receives and uses for further improvement.

[1176] Emotion analysis means

[1177] The server analyzes the user feedback using an emotion engine to extract emotions such as positive, negative, or neutral, and passes the analysis results to the recalibration means.

[1178] Generative AI model readjustment method

[1179] Based on the information obtained from the user feedback acquisition means and the results of the sentiment analysis means, the server readjusts the generative AI model so that the next and subsequent generated content will better meet the user's needs.

[1180] Specific examples

[1181] For example, a user might enter a prompt request such as, "Please describe the Normandy landings on June 6, 1944." Based on this request, the server uses a trained generative AI model to generate a detailed and accurate scene description.

[1182] In this way, the system of the present invention is able to continually evolve and provide high quality creative works through user feedback and sentiment analysis.

[1183] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1184] Step 1: Data collection

[1185] The server uploads documents and records provided by users and automatically collects related materials from the Internet and databases.

[1186] Input: Digital files uploaded by users, data collected from the internet and databases.

[1187] Output: The collected dataset.

[1188] What it does: Users upload literature files about World War I, and the server also collects data from online academic paper databases.

[1189] Step 2: Data cleaning

[1190] The server cleans the collected data, standardizing the format, removing duplicates, and completing missing information.

[1191] Input: The collected dataset.

[1192] Output: A cleaned dataset.

[1193] What it does: The server unifies descriptions of the same event collected from multiple sources and completes missing information such as dates and locations.

[1194] Step 3: Data analysis

[1195] The server then analyzes the cleaned data using natural language processing techniques to extract key features.

[1196] Input: The cleaned dataset.

[1197] Output: Extracted feature dataset.

[1198] What it does: The server extracts data about specific fight scenes, such as the date, time, location, and people involved.

[1199] Step 4: Training the generative AI model

[1200] The server trains a generative AI model based on features extracted through data analysis.

[1201] Input: Extracted feature dataset.

[1202] Output: A trained generative AI model.

[1203] What it does: The server feeds the model with training data so that it can accurately depict scenes from the Gallipoli War.

[1204] Step 5: Information Generation

[1205] The user terminal inputs a prompt sentence, and the server generates information using a trained generative AI model.

[1206] Input: The prompt text entered by the user.

[1207] Output: The generated information (text or scene).

[1208] What it does: The user types in "Please describe the Normandy landings on June 6, 1944," and the server generates the scene.

[1209] Step 6: Get user feedback

[1210] The user provides feedback on the generated information, which is captured by the server.

[1211] Input: User feedback.

[1212] Output: Collected feedback data.

[1213] Specific action: The user enters feedback, such as "I need a detailed tactical description."

[1214] Step 7: Sentiment Analysis

[1215] The server uses an emotion engine to analyze the user feedback and extract positive, negative, or neutral emotions.

[1216] Input: Collected feedback data.

[1217] Output: Sentiment analysis result data.

[1218] Specific behavior: The server detects the emotion "dissatisfied" from the feedback.

[1219] Step 8: Refining the generative AI model

[1220] The server readjusts the generative AI model based on the emotion analysis results and feedback data.

[1221] Input: Sentiment analysis result data, feedback data.

[1222] Output: A retuned generative AI model.

[1223] Specific operation: The server adjusts parameters to "enhance tactical depiction."

[1224] Step 9: Final generation and verification

[1225] The prompt sentence is input again from the user terminal, and information is generated using the re-adjusted model and provided to the user.

[1226] Input: The prompt statement typed again.

[1227] Output: Improved generated information.

[1228] What it does: The user again types in "Describe the Normandy landings on June 6, 1944," generating a more detailed and emotive description of the scene.

[1229] The above are the specific processing steps for carrying out the invention.

[1230] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1231] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1232] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1233] [Fourth embodiment]

[1234] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1235] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1236] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1237] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1238] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1239] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1240] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1241] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1242] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1243] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1244] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1245] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1246] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1247] The present invention relates to a system for improving the accuracy and quality of past creative works using generative AI. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, and a generative AI model readjustment means. An embodiment of the system is described below.

[1248] 1. Data Collection

[1249] Users upload relevant documents and records to create their creative works, for example, providing relevant historical documents to write a novel about World War I.

[1250] The server automatically collects relevant materials from the Internet and databases, for example, retrieving relevant information from online academic journal databases and encyclopedias.

[1251] 2. Data Cleaning

[1252] The server converts the collected data into a consistent format, removes duplicates, and completes missing information, for example, by unifying descriptions of the same event from multiple sources and completing missing dates.

[1253] 3. Data Analysis

[1254] The server uses natural language processing techniques to extract key features from the cleaned data, such as identifying the date, time, location, and people involved in a particular battle.

[1255] The server categorizes the data by relevant topic for later generation, for example by stage of a war or important event.

[1256] 4. Model training

[1257] The server trains a generative AI model based on the extracted data features, for example, training the model to generate specific World War I battle scenes and soldier dialogue.

[1258] The server optimizes the model's parameters to generate creations tailored to the user's needs, such as enhancing detailed tactical descriptions or emotional expression.

[1259] 5. Creation of creative works

[1260] A user inputs a specific request into the system, for example, "Please depict the Gallipoli landings on April 25, 1915."

[1261] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[1262] 6. User Feedback and Refinement

[1263] The user provides feedback on the generated creation, for example, "I would like more detailed descriptions of tactics added" or "I would like the expression of emotions improved."

[1264] The server readjusts the generative AI model based on user feedback to improve the accuracy and quality of the generated results, for example by adding training data to enhance emotional expression and readjusting the model parameters.

[1265] By using these methods, the system of the present invention can improve the accuracy and quality of creative works about the past and generate historically accurate and compelling content. For example, when creating a novel or movie scenario about World War I, this system can be used to generate detailed and accurate battle scenes and character dialogue, increasing the appeal to readers and viewers.

[1266] The processing flow will be explained below.

[1267] Step 1:

[1268] Users upload relevant documents and records to create their creative works. For example, if a user is writing a novel about World War I, they provide the system with digital files of historical documents they have on hand.

[1269] Step 2:

[1270] The server automatically collects relevant materials from the Internet and databases, for example, retrieves relevant data from online academic paper databases and encyclopedias, and incorporates them into the system.

[1271] Step 3:

[1272] The server cleans the collected data, which includes converting it into a consistent format, removing duplicates, and filling in missing information (for example, unifying descriptions of the same event collected from multiple sources and filling in missing information such as dates and locations).

[1273] Step 4:

[1274] The server then analyzes the cleaned data using natural language processing techniques to extract key features, such as the date, time, location, and people involved in a particular battle.

[1275] Step 5:

[1276] The server trains the generative AI model based on the analyzed data. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[1277] Step 6:

[1278] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[1279] Step 7:

[1280] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[1281] Step 8:

[1282] Users can provide feedback to the system about the generated creations, such as "I'd like more detailed descriptions of tactics" or "I'd like the expression of emotions to be improved."

[1283] Step 9:

[1284] The server readjusts the generative AI model based on user feedback. Based on the feedback, it readjusts the model parameters to improve the accuracy and quality of the generated results. For example, it adds training data to enhance emotional expression and retrains the model.

[1285] Step 10:

[1286] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[1287] Through this series of steps, the system of the present invention can effectively generate highly accurate and high-quality creative works about the past that meet the needs of the user.

[1288] Example 1

[1289] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1290] With conventional technologies, it is difficult to efficiently collect detailed historical information, analyze it in a unified manner, and automatically generate highly satisfying creative works using a highly accurate generative AI model. Furthermore, there is a lack of a process for incorporating user feedback on the generated content and continuously improving the generative AI model, which limits the improvement in the quality of the deliverables. Therefore, there is an urgent need to resolve these issues.

[1291] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1292] In this invention, the server includes a means for users to upload documents using their terminals, a means for the server to extract features from the cleaned data using natural language processing technology, a means for the server to classify the data by specific topic, and a means for the server to generate creative works using a Transformer model. This allows users to efficiently create high-quality creative works based on detailed historical information. Furthermore, by incorporating user feedback, the generative AI model can be continuously improved, thereby increasing the accuracy and quality of the generated content.

[1293] "Data collection means" means a means that has the function of uploading documents and records provided by users and automatically collecting related information from the Internet and databases.

[1294] "Means for cleaning collected data" means means for converting collected data into a consistent format, removing duplicate content, and completing missing information.

[1295] The "means for analyzing cleaned data and extracting important features" refers to a means having a function for extracting important features from cleaned data using natural language processing technology.

[1296] "Means for training a generative AI model" refers to means that trains a generative AI model based on extracted feature data and has the function of improving the accuracy of information generation.

[1297] "Means for generating information in response to a user's request using a trained model" means means that have the function of generating information using a trained generative AI model based on a specific request from a user.

[1298] The "means for obtaining user feedback" refers to a means having a function for collecting user evaluations and opinions about the generated information.

[1299] "Means for readjusting the generative AI model" refers to means that have the function of retraining or adjusting the generative AI model based on obtained user feedback to improve the accuracy and quality of the generated content.

[1300] "Means for users to upload documents using their terminals" refers to means that allow users to send documents and records to the system from the devices they use.

[1301] "Natural language processing technology" is a technology that allows computers to process and understand natural language used by humans.

[1302] A "Transformer model" is a neural network architecture that has particularly powerful capabilities in natural language processing, and is a model for context-aware text generation and translation.

[1303] A "means for categorizing data by specific topics" is a means that has the ability to organize and group cleaned data based on specific themes or categories.

[1304] MODE FOR CARRYING OUT THE INVENTION

[1305] The present invention is a system for improving the accuracy and quality of past creative works using generative AI. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, and a generative AI model readjustment means. An embodiment of the present invention will be described in detail below.

[1306] Data collection

[1307] A user uploads relevant documents and records to the system to create a creative work using their terminal. This process is performed using the user's device (PC, smartphone, etc.). For example, a user may upload historical documents in PDF format to the system in order to write a novel about World War II.

[1308] The server then automatically collects relevant material from the internet and existing databases based on the uploaded documents, using Python libraries (e.g., BeautifulSoup) to scrape relevant information from websites and retrieve information from online academic paper databases and encyclopedias.

[1309] Data Cleaning

[1310] The server converts the collected data into a consistent format, removes duplicates, and completes missing information. It converts the data into a data frame using the Pandas library and standardizes it to a consistent date and time format. It also uses a natural language processing library (e.g., spaCy) to automatically complete missing information.

[1311] Data analysis

[1312] The server uses natural language processing techniques to extract important features from the cleaned data, such as topic modeling (e.g., LDA) to extract features related to specific events or people.

[1313] The server then classifies the data into relevant topics based on the extracted features, using a clustering algorithm (e.g., K-means) to categorize the data into categories such as battle scenes, diplomatic events, and daily life.

[1314] Model training

[1315] The server trains a generative AI model based on the cleaned feature data. Using Tensorflow, we train an RNN-based generative AI model to generate battle scenes and dialogue.

[1316] The server then adjusts and optimizes the model parameters, performing hyperparameter tuning to determine the optimal learning rate and number of epochs to improve generation accuracy.

[1317] Creation

[1318] A user enters a specific request into the system, such as "Please portray a scene from the Normandy landings in 1944," into a web form.

[1319] The server uses trained generative AI models to generate depictions based on user requests, using Transformer models to detail soldier movements and scenery from the Normandy landings.

[1320] User feedback and refinements

[1321] Users can provide feedback on the generated creations, such as "more detailed tactical descriptions are needed" or "improvement in emotional expression."

[1322] The server retunes the generative AI model based on user feedback, retraining the model with additional training data to improve the accuracy and quality of the generated content.

[1323] Prompt Sentence Examples

[1324] "Please describe the Gallipoli landings on 25 April 1915. Specifically, we are looking for a description of the landing boats approaching the shore, the tension among the soldiers, and the fighting that followed."

[1325] Through the system of the present invention, users can efficiently create high-quality creative works based on detailed historical information, and by incorporating user feedback, the generative AI model can be continuously improved, thereby increasing the accuracy and quality of the generated content.

[1326] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1327] Step 1:

[1328] Users upload documents and records to the system.

[1329] Input: Literature data such as PDFs or text files selected by the user from their device.

[1330] Specific operation: The user uses the system's upload function from their PC or smartphone to select historical documents and records and send them to the server.

[1331] Output: Uploaded bibliographic data stored on the server.

[1332] Step 2:

[1333] The server collects additional data from the internet and databases.

[1334] Input: A search query based on the uploaded literature data.

[1335] What it does: The server uses Python's BeautifulSoup library to scrape information from relevant websites and retrieve relevant data from sources like online academic paper databases.

[1336] Output: Additional data collected (e.g., text of web pages, abstracts of matching academic papers).

[1337] Step 3:

[1338] The server cleans the collected data.

[1339] Input: The raw text of the collected data.

[1340] What it does: The server uses the Pandas library to convert the data into a data frame, removes duplicates, converts to a consistent date and time format, and imputes missing data. It also uses a natural language processing library (e.g., spaCy) to auto-complete missing information.

[1341] Output: A cleaned and consistent dataset.

[1342] Step 4:

[1343] The server parses the cleaned data.

[1344] Input: The cleaned dataset.

[1345] Specific operations: Conduct topic modeling (e.g., LDA) to extract important features (e.g., involvement of specific events or people), and use machine learning algorithms to classify topics.

[1346] Output: Extracted features and classified topics.

[1347] Step 5:

[1348] The server trains the generative AI model.

[1349] Input: Extracted feature data.

[1350] What it does: Use Tensorflow to train an RNN-based generative AI model to generate specific scenarios and dialogues, and use hyperparameter tuning to select optimal model settings.

[1351] Output: A trained generative AI model.

[1352] Step 6:

[1353] A user inputs a specific request into the system.

[1354] Input: A prompt from the user (e.g., "Please describe a scene from the Normandy landings in 1944").

[1355] Specific action: The user enters a prompt sentence about a specific scene or situation into a web form and submits it to the system.

[1356] Output: The prompt received by the server.

[1357] Step 7:

[1358] The server generates information based on the user's request.

[1359] Input: A trained generative AI model and a user prompt.

[1360] Specific behavior: Use trained generative AI models to generate specific scenes and dialogues based on prompts, and use Transformer models to generate context-aware text.

[1361] Output: The generated text in response to the user's request.

[1362] Step 8:

[1363] The user provides feedback on the generated information.

[1364] Input: Generated text and user rating.

[1365] Specific actions: The user evaluates the generated text and inputs suggestions for improvement, such as detailed tactical descriptions and emotional expressions.

[1366] Output: Feedback provided by the user.

[1367] Step 9:

[1368] The server readjusts the generative AI model based on the feedback.

[1369] Input: User feedback and retraining data.

[1370] What it does: Based on the feedback, it retrains the generative AI model with additional training data and optimizes the model's parameters.

[1371] Output: An improved generative AI model.

[1372] Through these steps, the system can generate high-quality creative works tailored to the user's needs and continuously improve its accuracy and quality.

[1373] (Application example 1)

[1374] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1375] Previous systems using generative AI technology struggled to sufficiently improve the accuracy and quality of historical creations. They also lacked a process for effectively utilizing user feedback to optimize generative AI models. Furthermore, they lacked a means to visually display the generated information and improve the user experience.

[1376] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1377] In this invention, the server includes a data collection means, a means for cleaning the collected data, a means for analyzing the cleaned data and extracting important features, a means for training a generative AI model based on the extracted features, a means for generating information according to a user request using the trained model, a means for obtaining user feedback on the generated information, a means for readjusting the generative AI model based on the obtained feedback, and a means for visually displaying the generated information. This enables improvement in the accuracy and quality of historical creations using generative AI, optimization of the model based on effective feedback, and an improvement in user experience.

[1378] "Data collection methods" are methods for automatically collecting relevant literature, records, and papers from the Internet and databases.

[1379] "Data cleaning procedures" are procedures that convert collected data into a consistent format, remove duplicate content, and complete missing information.

[1380] "Data analysis tools" are tools for extracting important features from cleaned data and classifying the data by related topics.

[1381] The "generative AI model training means" is a means for training a generative AI model based on extracted features and optimizing the model parameters.

[1382] "Information generation means" refers to a means for generating information in response to a user request using a trained generative AI model.

[1383] The "feedback acquisition means" is a means for acquiring user feedback on the generated information.

[1384] The "readjustment means" is a means for readjusting the generative AI model based on the obtained feedback.

[1385] The "visual display means" is a means for visually displaying the generated information.

[1386] The present invention relates to a system for improving the accuracy and quality of historical fiction using generative AI, which is implemented using several means:

[1387] First, the server uses a data collection tool to automatically collect relevant literature, records, and papers from the Internet and databases. The software used for this is the Python requests library, which retrieves data through an API.

[1388] The collected data is then converted into a consistent format through data cleaning procedures, using tools such as Pandas to remove duplicate content and fill in missing information, for example by unifying descriptions of the same event from multiple documents and filling in missing date information.

[1389] The cleaned data is then analyzed using natural language processing techniques to extract key features, such as the date, time, location, and people involved in a particular historical event, using NLP libraries such as spaCy.

[1390] Based on the extracted features, the server trains a generative AI model using a machine learning framework such as TensorFlow or PyTorch, training the model to generate specific combat scenes or dialogues.

[1391] The server then uses the trained model to generate information based on the user's request. For example, if a user requests, "Generate a detailed description of the Gallipoli Landings on April 25, 1915," the generative AI model will generate a detailed description of the scene.

[1392] The feedback acquisition means acquires user feedback on the generated information, such as "I would like more detailed descriptions of tactics to be added."

[1393] Finally, the recalibration means recalibrates the generative AI model based on the obtained feedback. In this process, the training dataset is reconstructed based on the collected feedback and the model is retrained.

[1394] Using visual display means, the generated information is visually displayed on a user device. For example, it can be visually confirmed through a smartphone or head-mounted display. As a result, the accuracy and quality of historical creations using generative AI can be improved, models can be optimized based on effective feedback, and the user experience can be improved.

[1395] Example prompt sentence:

[1396] Generate a depiction of the Allied landing forces approaching the beaches of Gallipoli at sunrise on April 25, 1915. Detail specific soldier movements, the scenery, and emotions.

[1397] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1398] Step 1:

[1399] Data collection

[1400] The server uses data collection tools to automatically collect relevant literature, records, and papers from the Internet and databases, for example, using the Python requests library to retrieve data through APIs. The input is the relevant URL or API endpoint, and the output is the retrieved raw data.

[1401] Step 2:

[1402] Data Cleaning

[1403] The server uses data cleaning techniques to convert the collected data into a consistent format, remove duplicates, and fill in missing information. Specifically, it converts the data into a data frame using Pandas and removes duplicates using the .drop_duplicates() method. The input is raw data, and the output is cleaned data.

[1404] Step 3:

[1405] Data analysis

[1406] The server then uses data analysis tools and natural language processing techniques to extract important features from the cleaned data. For example, it uses an NLP library such as spaCy to extract proper nouns such as specific events or people. The input is the cleaned data, and the output is the data with important features extracted.

[1407] Step 4:

[1408] Training generative AI models

[1409] The server trains a generative AI model based on the extracted features. Using a machine learning framework such as TensorFlow or PyTorch, the model is trained to generate specific dialogues or scenes. The input is the extracted feature data, and the output is a trained generative AI model.

[1410] Step 5:

[1411] information generation

[1412] The server uses the trained model to generate information based on the user's request. The user inputs a specific prompt, and the generated information is a detailed description. The input is the user's prompt, and the output is the generated text content. For example, the prompt "Please generate a detailed description of the Gallipoli landings on April 25, 1915" is input, and a specific description is generated.

[1413] Step 6:

[1414] Get feedback

[1415] The server obtains user feedback on the generated information. It uses a feedback form or questionnaire to collect user opinions and correction requests. The input is the user's feedback content, and the output is the feedback data.

[1416] Step 7:

[1417] Model Rebalancing

[1418] The server retunes the generative AI model based on the feedback it receives. It reconstructs the training dataset based on the collected feedback and retrains the model. The input is the feedback data, and the output is the retuned generative AI model.

[1419] Step 8:

[1420] Visual Indication

[1421] The device visually displays the generated information using a visual display means. The user can view the generated detailed depiction through a smartphone or head-mounted display. The input is the generated text content, and the output is the visually displayed information.

[1422] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1423] The present invention further combines a system for improving the accuracy and quality of past creative works using generative AI with an emotion engine. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a generation means, a feedback acquisition means, a generative AI model readjustment means, and an emotion engine. An embodiment of this system is shown below.

[1424] 1. Data Collection

[1425] Users upload relevant documents and records to create their creative works. For example, if a user is writing a novel about World War I, they provide the system with digital files of historical documents they have on hand.

[1426] The server automatically collects relevant materials from the Internet and databases, for example, retrieves relevant data from online academic paper databases and encyclopedias, and incorporates them into the system.

[1427] 2. Data Cleaning

[1428] The server converts the collected data into a consistent format, removes duplicates, and fills in missing information, for example by unifying descriptions of the same event collected from multiple sources and filling in missing information such as dates and locations.

[1429] 3. Data Analysis

[1430] The server uses natural language processing techniques to extract key features from the cleaned data, such as the date, time, location, and people involved in a particular battle.

[1431] The server categorizes the data by relevant topic for later generation, for example by stage of a war or important event.

[1432] 4. Model training

[1433] The server trains a generative AI model based on the extracted data features. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[1434] The server optimizes the model's parameters to generate creations tailored to the user's needs, such as enhancing detailed tactical descriptions or emotional expression.

[1435] 5. Creation of creative works

[1436] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[1437] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[1438] 6. User Feedback and Emotion Recognition

[1439] The user provides feedback on the generated creation, for example, "I would like more detailed descriptions of tactics added" or "I would like the expression of emotions improved."

[1440] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, it extracts positive, negative, and neutral emotions from the feedback content and further analyzes the user's detailed emotional state (e.g., satisfaction, dissatisfaction, surprise, etc.).

[1441] 7. Readjust based on feedback

[1442] The server readjusts the generative AI model based on user feedback and the results of emotion analysis by the emotion engine. For example, if the emotion engine detects the emotion "dissatisfaction," it identifies the cause of that emotion and optimizes the model parameters accordingly.

[1443] 8. Final generation and verification

[1444] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[1445] Check whether the user is satisfied with the final generated creation and provide further feedback if necessary to iterate and improve the quality of the process.

[1446] In this way, the system of the present invention not only uses generative AI to generate creative works about the past, but also uses an emotion engine to analyze user emotions and readjust the model based on feedback, thereby providing even more accurate and high-quality creative works. For example, when creating a novel or movie scenario about World War I, this system can generate detailed and accurate battle scenes and character dialogue, increasing the appeal to readers and viewers.

[1447] The processing flow will be explained below.

[1448] Step 1:

[1449] Users upload relevant documents and records to the system from their devices in order to create their creative works. For example, a user may submit digital files of historical documents they have on hand to write a novel about World War I.

[1450] Step 2:

[1451] The server automatically collects relevant materials from the Internet and databases, for example, retrieving relevant information from online academic paper databases and encyclopedias, and integrating it into the system.

[1452] Step 3:

[1453] The server cleans the collected data, which includes standardizing the data format, removing duplicate content, and completing missing information (for example, standardizing descriptions of the same event collected from multiple sources and completing missing information such as dates and locations).

[1454] Step 4:

[1455] The server then uses natural language processing techniques to extract key features from the cleaned data, such as the date, time, location, and people involved in a particular battle.

[1456] Step 5:

[1457] The server trains the generative AI model based on the analyzed data. Training involves optimizing the model to generate specific content that users want to generate, such as specific battle scenes or soldier dialogue. For example, training the model to be able to depict the detailed landing scene of the Gallipoli War.

[1458] Step 6:

[1459] A user inputs a specific request into the system, for example, "Please describe the Gallipoli landings on April 25, 1915."

[1460] Step 7:

[1461] The server uses a trained generative AI model to generate depictions based on user requests, such as specific soldier movements and landscape depictions in a landing scene during the Gallipoli War.

[1462] Step 8:

[1463] The user provides feedback to the system about the generated creation, such as "I would like more detailed descriptions of tactics" or "I would like the expression of emotions to be improved."

[1464] Step 9:

[1465] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, it extracts positive, negative, and neutral emotions from the feedback content and analyzes more detailed emotional states (e.g., satisfaction, dissatisfaction, surprise, etc.).

[1466] Step 10:

[1467] The server readjusts the generative AI model based on user feedback and the results of emotion analysis. For example, if the emotion engine detects the emotion "dissatisfaction," it identifies the cause of that emotion and readjusts the model parameters accordingly.

[1468] Step 11:

[1469] The server uses the retuned model to generate content based on the user's request and provides it to the user. For example, the retuned model is used to generate a more detailed and emotionally rich landing scene from the Gallipoli War.

[1470] Step 12:

[1471] Check whether the user is satisfied with the final generated creation and provide further feedback if necessary to iterate and improve the quality of the process.

[1472] Through these steps, the system of the present invention can provide highly accurate and high-quality creative works by analyzing users' emotions in addition to the process of generating creative works related to the past using generative AI.

[1473] Example 2

[1474] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1475] Conventional creative creation systems using generative AI models lack the ability to analyze user feedback and readjust the model based on that feedback. As a result, the accuracy and quality of the created creations often do not meet user requirements. In particular, it is difficult to generate creations that appropriately reflect the user's emotions, and there is a need to improve user satisfaction.

[1476] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1477] In this invention, the server includes a data collection means, a means for cleaning the collected data into a consistent format, removing duplication, and completing missing information, a means for analyzing the cleaned data and extracting important features using natural language processing technology, a means for training a generative AI model based on the extracted features and optimizing model parameters, a means for generating information according to a user request using the trained model, a means for acquiring user feedback on the generated information, analyzing the feedback using an emotion engine, and recognizing the user's emotions, and a means for readjusting the generative AI model based on the acquired feedback and the emotion analysis results. This makes it possible to generate creative works with high accuracy and quality by appropriately readjusting the generative AI model based on the user's feedback and emotions.

[1478] "Data collection means" refers to the means by which user-provided documents and records are uploaded to the system and further related materials are automatically collected from the internet and databases.

[1479] "Data cleaning procedures" are procedures used to convert collected data into a consistent format, remove duplication, and complete missing information.

[1480] The "data analysis means" is a means for extracting important features from the cleaned data using natural language processing technology and classifying the data based on those features.

[1481] A "generative AI model" is an artificial intelligence model that is trained based on cleaned and analyzed data to generate information in response to user requests.

[1482] The "feedback acquisition means" is a means for collecting user feedback on the generated information or creation.

[1483] An "emotion engine" is a means for analyzing feedback provided by a user and recognizing the user's emotions based on that feedback.

[1484] "Model readjustment means" refers to a means for readjusting the generative AI model based on the obtained feedback and sentiment analysis results to improve its accuracy and quality.

[1485] The present invention provides a creative creation system that uses a generative AI model and an emotion engine to generate accurate and high-quality creative works based on user-provided documents and records. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, a creative creation means, a feedback acquisition means, an emotion engine, and a model readjustment means.

[1486] An embodiment of this system is described below.

[1487] 1. Data Collection

[1488] A user uploads relevant documents and records related to their creative work to the system. For example, a user is writing a novel about World War I and provides the system with digital files (PDF, Word files, etc.) of historical documents.

[1489] The server collects relevant materials from internet resources and existing databases, for example, pulling the necessary information from Google Scholar or Wikipedia, and storing it in the system.

[1490] 2. Data Cleaning

[1491] The server converts the collected data into a consistent format, removes duplicates, and completes missing information, for example, unifying documents with different formats, removing duplicate content, and adding missing dates and other details.

[1492] 3. Data Analysis

[1493] The server uses natural language processing (NLP) techniques to extract key features from the cleaned data, specifically through text analysis, to extract information such as the date, time, location, and people involved in a particular battle.

[1494] The server uses the extracted features to categorize the data into related topics, for example, organizing data by phases of a war (start, major battles, intermediate ceasefires, etc.) or important events.

[1495] 4. Training the generative AI model

[1496] The server trains a generative AI model based on features extracted from the data, focusing on the scene the user wants to depict, for example, using data on the landing scene of the Gallipoli War.

[1497] The server optimizes the model parameters to generate a creation that meets the user's needs, for example by setting specific parameters to enhance tactical detail or emotional expression.

[1498] 5. Creation of creative works

[1499] The user inputs a specific request into the system, for example, a prompt such as "Please describe in detail the scenes from the Gallipoli landings on April 25, 1915."

[1500] The server uses a trained generative AI model to generate depictions based on user requests, such as detailed descriptions of soldier movements and scenery during a landing scene.

[1501] 6. User Feedback and Emotion Recognition

[1502] Users can provide feedback on the generated creation, for example by entering specific feedback in text, such as "I would like more detailed tactical descriptions."

[1503] The server uses an emotion engine to analyze the feedback and recognize the user's emotions. For example, the server can analyze the emotion of "dissatisfaction" extracted from the feedback, such as "I want more detailed tactical descriptions."

[1504] 7. Re-adjusting the model based on feedback

[1505] The server readjusts the generative AI model based on the feedback and emotion analysis results it receives. For example, if the emotion "dissatisfaction" is detected, the cause is identified and the model parameters are optimized.

[1506] 8. Final generation and verification

[1507] The server then uses the retuned model to generate the creation again and provides it to the user. Specifically, the adjusted model is used to generate a more detailed and emotionally rich Gallipoli War landing scene.

[1508] The user checks whether they are satisfied with the final generated creation and provides feedback again if necessary to iterate the process and improve the quality.

[1509] In this way, the system of the present invention can efficiently generate highly accurate and high-quality creative works by readjusting the generative AI model based on feedback provided by the user and the results of emotion analysis.

[1510] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1511] Step 1:

[1512] Data collection

[1513] Input: User-provided digital files of documents and records (e.g. PDF, Word files).

[1514] Specific operation: The user submits historical documents and record files they have on hand using the system's upload function.

[1515] Output: Literature and records stored in the system's database.

[1516] The server automatically collects relevant literature, records, and papers from the Internet and databases, specifically by accessing open data sources such as Google Scholar and Wikipedia, and stores the necessary information in the system.

[1517] Step 2:

[1518] Cleaning the data

[1519] Input: Collected literature and archival data.

[1520] What it does: It converts the data collected by the server into a consistent format (e.g., UTF-8 encoded text), detects and removes duplicate data, and fills in missing information (e.g., by filling in missing details like date or location from other resources).

[1521] Output: Consistent, deduplicated, and imputed bibliographic and archival data.

[1522] Step 3:

[1523] Data analysis

[1524] Input: Cleaned data.

[1525] What it does: The server uses natural language processing (NLP) techniques to extract important features from the data (e.g., battle dates, locations, and people involved), and then categorizes the data by related topics based on the extracted features. For example, it organizes data about each stage of a war or key events separately.

[1526] Output: Data with important features extracted and categorized by relevant topics.

[1527] Step 4:

[1528] Training generative AI models

[1529] Input: Extracted feature data.

[1530] How it works: The server trains a generative AI model based on the extracted feature data and optimizes it to generate information based on the user's request. For example, if a user wants a detailed depiction of the landing scene from the Gallipoli War, the server trains the model using data related to that scene.

[1531] Output: A trained generative AI model.

[1532] Step 5:

[1533] Creation

[1534] Input: The specific request (prompt) that the user provides to the system.

[1535] What it does: The user enters a prompt, such as "Please give me a detailed description of the Gallipoli Landings on April 25, 1915." The server uses a trained generative AI model to generate sentences and scene descriptions based on the request.

[1536] Output: The creative work generated based on the user's request.

[1537] Step 6:

[1538] User Feedback and Emotion Recognition

[1539] Input: Feedback provided by the user about the generated creation.

[1540] Specific operation: The user inputs feedback in text format, such as "I would like more detailed tactical descriptions." The server uses the emotion engine to analyze the feedback, extracts the user's emotion (e.g., "dissatisfied"), and analyzes it in detail.

[1541] Output: Parsed feedback and user sentiment data.

[1542] Step 7:

[1543] Retune the model based on feedback

[1544] Input: Feedback and sentiment analysis results.

[1545] Specific operation: The server readjusts the generative AI model based on the feedback and emotion analysis results it receives. For example, if the emotion "dissatisfaction" is detected, it identifies the cause of this and adjusts the model parameters.

[1546] Output: A retuned generative AI model.

[1547] Step 8:

[1548] Final generation and confirmation

[1549] Input: A retuned generative AI model and a new user request.

[1550] What happens: The server uses the refined model to generate new drawings and sentences, and serves them to the user. The server checks whether the user is satisfied with the final generated work, providing further feedback if necessary.

[1551] Output: High-quality creations that satisfy users.

[1552] (Application example 2)

[1553] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1554] Conventional generative AI models have limitations in their ability to generate content based on past data, making it particularly difficult to adjust the content to reflect emotional expressions and user feedback. As a result, the quality of the generated content varies and user feedback cannot be properly utilized. Furthermore, when users request specific scenes or historical events, it is difficult to respond to their requests accurately and in detail.

[1555] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a data collection means, a data cleaning means, and a means for analyzing the cleaned data and extracting important features. This enables the collected data to be organized into a consistent format and necessary features to be extracted. The server also includes a means for training a generative AI model, a means for generating information in response to a user request, a means for obtaining user feedback on the generated information, and a means for analyzing the user's emotional feedback using an emotion analysis means and reflecting the analysis results in readjusting the model. This allows the generated content to reflect user feedback and include emotional elements, making it possible to provide higher quality content.

[1556] "Data collection means" has the function of automatically collecting data provided by users and related materials from the Internet and databases.

[1557] "Data cleaning procedures" are procedures that convert collected data into a consistent format, remove duplicate content, and complete missing information.

[1558] "Data analysis means" extracts important features from the cleaned data so that they can be used in subsequent processes.

[1559] The "means for training a generative AI model" refers to optimizing the generative AI model based on the extracted features so that it can generate specific content.

[1560] An "information generation means" is a means that uses a trained generative AI model to generate information or content in response to a user's request.

[1561] The "user feedback acquisition means" has a function of collecting user opinions and impressions regarding the generated information.

[1562] The "means for readjusting the generative AI model" refers to the re-optimization of the generative AI model based on feedback obtained from users and the results of sentiment analysis, in order to improve its accuracy and quality.

[1563] The "emotion analysis means" analyzes the user's feedback from an emotional perspective and identifies the emotional state, such as positive, negative, or neutral.

[1564] This invention is a system for improving the accuracy and quality of past creative works using a generative AI combined with an emotion engine. The system includes a data collection means, a data cleaning means, a data analysis means, a generative AI model training means, an information generation means, a user feedback acquisition means, a generative AI model readjustment means, and an emotion analysis means.

[1565] System Programs and Processing

[1566] Hardware and software used

[1567] Hardware: Server with high-performance CPU and GPU

[1568] Software: Python, TensorFlow / PyTorch, natural language processing libraries (spaCy, NLTK), sentiment analysis engine API

[1569] Data collection methods

[1570] The server collects documents and records uploaded by users and automatically retrieves related materials from the Internet and databases, thereby comprehensively obtaining the necessary data.

[1571] Data Cleaning Methods

[1572] The collected data is converted into a consistent format, duplicates are removed, and missing information is filled in. This cleaning process improves the quality of the data and makes it easier to analyze.

[1573] Data Analysis Methods

[1574] The server then uses natural language processing technology to analyze the cleaned data and extract important features, such as the date, time, and location of a particular event, as well as the people involved, which are then used in subsequent processes.

[1575] A means of training generative AI models

[1576] The server trains a generative AI model based on the extracted data, and the trained model is optimized to generate specific scenes and interactions requested by the user.

[1577] Information generation means

[1578] The server uses a trained generative AI model to generate information based on requests from users' devices. For example, a user might input a prompt such as, "Please describe the Normandy landings on June 6, 1944."

[1579] User feedback acquisition method

[1580] The user provides feedback on the generated information, which the server receives and uses for further improvement.

[1581] Emotion analysis means

[1582] The server analyzes the user feedback using an emotion engine to extract emotions such as positive, negative, or neutral, and passes the analysis results to the recalibration means.

[1583] Generative AI model readjustment method

[1584] Based on the information obtained from the user feedback acquisition means and the results of the sentiment analysis means, the server readjusts the generative AI model so that the next and subsequent generated content will better meet the user's needs.

[1585] Specific examples

[1586] For example, a user might enter a prompt request such as, "Please describe the Normandy landings on June 6, 1944." Based on this request, the server uses a trained generative AI model to generate a detailed and accurate scene description.

[1587] In this way, the system of the present invention is able to continually evolve and provide high quality creative works through user feedback and sentiment analysis.

[1588] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1589] Step 1: Data collection

[1590] The server uploads documents and records provided by users and automatically collects related materials from the Internet and databases.

[1591] Input: Digital files uploaded by users, data collected from the internet and databases.

[1592] Output: The collected dataset.

[1593] What it does: Users upload literature files about World War I, and the server also collects data from online academic paper databases.

[1594] Step 2: Data cleaning

[1595] The server cleans the collected data, standardizing the format, removing duplicates, and completing missing information.

[1596] Input: The collected dataset.

[1597] Output: A cleaned dataset.

[1598] What it does: The server unifies descriptions of the same event collected from multiple sources and completes missing information such as dates and locations.

[1599] Step 3: Data analysis

[1600] The server then analyzes the cleaned data using natural language processing techniques to extract key features.

[1601] Input: The cleaned dataset.

[1602] Output: Extracted feature dataset.

[1603] What it does: The server extracts data about specific fight scenes, such as the date, time, location, and people involved.

[1604] Step 4: Training the generative AI model

[1605] The server trains a generative AI model based on features extracted through data analysis.

[1606] Input: Extracted feature dataset.

[1607] Output: A trained generative AI model.

[1608] What it does: The server feeds the model with training data so that it can accurately depict scenes from the Gallipoli War.

[1609] Step 5: Information Generation

[1610] The user terminal inputs a prompt sentence, and the server generates information using a trained generative AI model.

[1611] Input: The prompt text entered by the user.

[1612] Output: The generated information (text or scene).

[1613] What it does: The user types in "Please describe the Normandy landings on June 6, 1944," and the server generates the scene.

[1614] Step 6: Get user feedback

[1615] The user provides feedback on the generated information, which is captured by the server.

[1616] Input: User feedback.

[1617] Output: Collected feedback data.

[1618] Specific action: The user enters feedback, such as "I need a detailed tactical description."

[1619] Step 7: Sentiment Analysis

[1620] The server uses an emotion engine to analyze the user feedback and extract positive, negative, or neutral emotions.

[1621] Input: Collected feedback data.

[1622] Output: Sentiment analysis result data.

[1623] Specific behavior: The server detects the emotion "dissatisfied" from the feedback.

[1624] Step 8: Refining the generative AI model

[1625] The server readjusts the generative AI model based on the emotion analysis results and feedback data.

[1626] Input: Sentiment analysis result data, feedback data.

[1627] Output: A retuned generative AI model.

[1628] Specific operation: The server adjusts parameters to "enhance tactical depiction."

[1629] Step 9: Final generation and verification

[1630] The prompt sentence is input again from the user terminal, and information is generated using the re-adjusted model and provided to the user.

[1631] Input: The prompt statement typed again.

[1632] Output: Improved generated information.

[1633] What it does: The user again types in "Describe the Normandy landings on June 6, 1944," generating a more detailed and emotive description of the scene.

[1634] The above are the specific processing steps for carrying out the invention.

[1635] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1636] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1637] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1638] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1639] FIG. 9 illustrates an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and behaviors arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1640] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1641] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1642] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1643] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1644] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1645] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1646] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1647] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1648] 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.

[1649] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1650] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1651] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1652] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1653] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1654] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1655] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1656] The following is further disclosed regarding the above embodiment.

[1657] (Claim 1)

[1658] data collection means;

[1659] a means for cleaning the collected data;

[1660] A means of analyzing the cleaned data and extracting important features;

[1661] a means for training a generative AI model based on the extracted features; and

[1662] means for generating information in response to a user request using the trained model;

[1663] means for obtaining user feedback on the generated information;

[1664] A means to retune the generative AI model based on the feedback obtained; and

[1665] A system including:

[1666] (Claim 2)

[1667] 2. The system according to claim 1, wherein the data collection means is a means for automatically collecting relevant literature, records, and papers from the Internet and databases.

[1668] (Claim 3)

[1669] 10. The system of claim 1, wherein the generative AI model is trained using natural language processing techniques.

[1670] "Example 1"

[1671] (Claim 1)

[1672] data collection means;

[1673] a means for cleaning the collected data;

[1674] A means of analyzing the cleaned data and extracting important features;

[1675] a means for training a generative AI model based on the extracted features; and

[1676] means for generating information in response to a user request using the trained model;

[1677] means for obtaining user feedback on the generated information;

[1678] A means to retune the generative AI model based on the feedback obtained; and

[1679] a means for a user to upload documents using a terminal;

[1680] A means for the server to extract features from the cleaned data using natural language processing techniques;

[1681] a means for the server to categorize the data by specific topics;

[1682] a means for the server to generate a creation using the Transformer model;

[1683] A system including:

[1684] (Claim 2)

[1685] 2. The system according to claim 1, wherein the data collection means is a means for automatically collecting relevant literature, records, and papers from the Internet and databases.

[1686] (Claim 3)

[1687] 10. The system of claim 1, wherein the generative AI model is trained using natural language processing techniques.

[1688] "Application Example 1"

[1689] (Claim 1)

[1690] data collection means;

[1691] a means for cleaning the collected data;

[1692] A means of analyzing the cleaned data and extracting important features;

[1693] a means for training a generative AI model based on the extracted features; and

[1694] means for generating information in response to a user request using the trained model;

[1695] means for obtaining user feedback on the generated information;

[1696] A means to retune the generative AI model based on the feedback obtained; and

[1697] a means for visually displaying the generated information;

[1698] A system including:

[1699] (Claim 2)

[1700] 2. The system according to claim 1, wherein the data collection means is a means for automatically collecting relevant literature, records, and papers from the Internet and databases.

[1701] (Claim 3)

[1702] 10. The system of claim 1, wherein the generative AI model is trained using natural language processing techniques.

[1703] "Example 2: Combining Emotion Engines"

[1704] (Claim 1)

[1705] data collection means;

[1706] A means of cleaning the collected data into a consistent format, removing duplicates and completing missing information;

[1707] A means for analyzing the cleaned data and extracting important features using natural language processing techniques;

[1708] A means to train a generative AI model based on the extracted features and optimize the model parameters; and

[1709] means for generating information in response to a user request using the trained model;

[1710] a means for acquiring a user's feedback on the generated information, analyzing the feedback using an emotion engine, and recognizing the user's emotion;

[1711] A means to readjust the generative AI model based on the feedback and sentiment analysis results obtained;

[1712] A system including:

[1713] (Claim 2)

[1714] 2. The system according to claim 1, wherein the data collection means is a means for automatically collecting relevant literature, records, and papers from the Internet and databases.

[1715] (Claim 3)

[1716] 10. The system of claim 1, wherein the generative AI model is trained using natural language processing techniques.

[1717] "Application example 2 when combining emotion engines"

[1718] (Claim 1)

[1719] data collection means;

[1720] a means for cleaning the collected data;

[1721] A means of analyzing the cleaned data and extracting important features;

[1722] a means for training a generative AI model based on the extracted features; and

[1723] means for generating information in response to a user request using the trained model;

[1724] means for obtaining user feedback on the generated information;

[1725] A means to retune the generative AI model based on the feedback obtained; and

[1726] A means for analyzing user's emotional feedback using an emotion analysis means and reflecting the feedback in readjusting the model;

[1727] A system including:

[1728] (Claim 2)

[1729] 2. The system according to claim 1, wherein the data collection means is a means for automatically collecting relevant literature, records, and papers from the Internet and databases.

[1730] (Claim 3)

[1731] 10. The system of claim 1, wherein the generative AI model is trained using natural language processing techniques. [Explanation of symbols]

[1732] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. data collection means; a means for cleaning the collected data; A means of analyzing the cleaned data and extracting important features; a means for training a generative AI model based on the extracted features; and means for generating information in response to a user request using the trained model; means for obtaining user feedback on the generated information; A means to retune the generative AI model based on the feedback obtained; and A system including:

2. 2. The system according to claim 1, wherein the data collection means is a means for automatically collecting related documents, records, and papers from the Internet and databases.

3. 10. The system of claim 1, wherein the generative AI model is trained using natural language processing techniques.

Images