system

Abstract

We provide the system. [Solution] A means of collecting data to generate images related to a specific culture, A means for detecting cultural bias in generated images, A means for evaluating the discomfort index of an image based on detected bias, A means of updating the model using the evaluation results, A means of improving image representation using an updated model, A system that includes this.

Description

Technical Field

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

Background Art

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

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the current image generation technology, there are biases against specific cultures and ethnic groups, resulting in inaccurate cultural expressions and discomfort, which is an issue. In particular, there are problems with prejudices and misunderstandings against Asian cultures including Japanese culture in many generated images. Furthermore, due to the lack of a mechanism to automatically detect and improve such biases and inaccuracies, the reliability and ethics of generated images are impaired.

Means for Solving the Problems

[0005] To address this challenge, the present invention provides means for collecting data to generate images related to a specific culture and for detecting cultural bias in the generated images. Furthermore, by providing a combination of means for evaluating the discomfort index of an image based on the detected bias and means for updating the model using the obtained evaluation results, it is possible to improve the representation of the image with the updated model. This series of processes makes it possible to ensure that the generated images are accurate to a specific culture and do not cause discomfort, thereby improving technical and ethical issues.

[0006] "Specific culture" refers to a collection of traditions, customs, values, and social behaviors that are unique to a particular region or ethnic group.

[0007] "Image generation" is the process of creating visual images based on specific themes or conditions using computer algorithms.

[0008] "Data collection" refers to the activity of systematically gathering and organizing information and image data related to a specific culture or theme.

[0009] "Cultural bias" refers to misunderstandings and discrimination that arise from inaccurate or prejudiced portrayals of particular cultures or ethnic groups.

[0010] The "discomfort index" is a numerical indicator that quantifies the degree to which a generated image is likely to evoke discomfort or resentment in a particular culture or audience.

[0011] "Model updating" is the process of retraining and adjusting the algorithms and models responsible for the generation process, based on feedback and new data, in order to improve their accuracy. [Brief explanation of the drawing]

[0012] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11] This is a sequence diagram showing the processing flow of the data processing system in Example 1. [Figure 12] This is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13] This is a sequence diagram showing the processing flow of the data processing system in Example 2, which incorporates an emotion engine. [Figure 14] This is a sequence diagram showing the processing flow of the data processing system in Application Example 2, which combines an emotion engine. [Modes for carrying out the invention]

[0013] Hereinafter, an example of an embodiment of the system relating to the technology of this disclosure will be described with reference to the attached drawings.

[0014] First, the terms used in the following description will be explained.

[0015] In the following embodiments, a processor with a reference numeral (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), and the like.

[0016] In the following embodiments, a RAM (Random Access Memory) with a reference numeral is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0017] In the following embodiments, a storage with a reference numeral is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0018] In the following embodiments, a communication I / F (Interface) with a reference numeral is an interface including a communication processor and an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applicable to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark), etc.

[0019] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0020] [First Embodiment]

[0021] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0022] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0023] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0024] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0025] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0026] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0027] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0028] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0029] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0030] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0031] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0032] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0033] One embodiment of the present invention is to provide an image generation system that accurately reflects a specific culture. The entire system consists of an image generation server, a terminal capable of bias detection, and a user that provides evaluation and feedback.

[0034] Image generation and bias detection

[0035] The server collects large amounts of image data related to a specific culture from the internet and existing databases. This data is stored in the database as a foundational dataset for learning that reflects the target culture. The collected data is tagged with metadata.

[0036] The device generates images using datasets provided by the server, based on specific conditions and themes. This image generation process is optimized to ensure appropriate representation, taking into account how the generated images will be received in a particular culture.

[0037] The device performs cultural bias detection on the generated images. This is a process that analyzes whether the images contain cultural inaccuracies or negative stereotypes and evaluates the discomfort index. The detection algorithm operates based on specific bias detection criteria and issues a warning for images with a discomfort index that exceeds the criteria.

[0038] Feedback and model improvement

[0039] Users receive prior training to minimize personal bias and provide appropriate feedback on the generated images. The training is designed to enable evaluation from diverse perspectives.

[0040] The server analyzes feedback collected from users and generates a new training dataset. This allows the model to be retrained, improving its accuracy so that it can generate images that more accurately reflect specific cultures.

[0041] As a concrete example, consider the case of generating images related to traditional Japanese festivals. In this scenario, the server collects diverse images related to festivals and builds a dataset. The terminal uses this dataset to generate festival-themed images, detecting and correcting cultural biases in advance. The user evaluates the generated images and provides feedback, aiming to generate more accurate images of Japanese culture.

[0042] The following describes the processing flow.

[0043] Step 1:

[0044] The server collects a vast amount of image data related to a specific culture from the internet and existing databases. These images are tagged with metadata and are appropriately culturally tagged. This constructs the dataset used for generation.

[0045] Step 2:

[0046] The device generates images using collected datasets based on specific conditions or themes. The generated images are optimized based on algorithms designed to faithfully reproduce the characteristics of a particular culture.

[0047] Step 3:

[0048] The terminal sends the generated image to a bias detection module, which automatically analyzes whether the image contains bias or inaccuracies regarding a particular culture or ethnicity.

[0049] Step 4:

[0050] The device evaluates the discomfort index of an image and displays a warning if the index is high, according to the set criteria. This process can prevent inaccurate representations of specific cultures.

[0051] Step 5:

[0052] Users complete pre-training and provide expert feedback on the generated images. By evaluating images from diverse perspectives and providing feedback, users can mitigate personal bias.

[0053] Step 6:

[0054] The server collects and analyzes the feedback received from users. Based on the results of this analysis, a new dataset is generated, and the model is retrained.

[0055] Step 7:

[0056] The server incorporates the results of retraining, enabling more accurate image generation. The goal is to generate unbiased images that accurately reflect a specific culture.

[0057] (Example 1)

[0058] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0059] When generating images that accurately reflect a specific culture, there is a risk of including cultural biases or inappropriate representations. This can lead to misunderstandings and cultural friction, highlighting the need to improve the quality and cultural appropriateness of the generated images.

[0060] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0061] In this invention, the server includes means for extracting visual data related to a specific culture, means for creating images in response to input instructions using a generative AI model, and means for detecting cultural bias within the created images. This enables the generation of high-quality images that are sensitive to culture.

[0062] "Visual data" refers to data that includes visual information such as images and videos, and is used as an element that reflects a particular culture.

[0063] A "generative AI model" is an algorithm that uses artificial intelligence to create new images and designs, and is a technology that generates images based on specific conditions.

[0064] "Input instructions" refer to instructions that explicitly state the request for a specific image generation, and mean prompts or commands given to the generation AI model.

[0065] "Cultural bias" refers to a state in which misunderstandings or negative stereotypes about a particular culture are included in expressions, and biased views are a factor that hinders a correct understanding of a culture.

[0066] "Inappropriateness" is an index that evaluates how culturally appropriate a generated image is, and indicates that corrections are necessary if the image does not meet the standard.

[0067] A "warning" is a notification issued when the level of cultural bias or inappropriateness within an image exceeds a certain standard, and it is an important sign that action should be taken.

[0068] "Evaluation information" refers to the feedback and evaluation data provided for the generated images, and is used to improve the model.

[0069] A "dataset" is a collection of data used for training and retraining, and forms the basis for generating new images.

[0070] This invention provides an image generation system that accurately reflects a specific culture. The system includes the management of visual data, the creation of images using a generative AI model, the detection and warning of cultural biases, and the continuous improvement of the model.

[0071] The server first collects visual data related to a specific culture from the internet and existing databases. This is done using web scraping techniques to retrieve images through APIs of public databases. The collected data is then stored in the database with metadata added.

[0072] The device generates images based on a dataset provided by the server, using a generative AI model. This involves using generative models such as GANs (Generative Adversarial Networks) or VQ-VAE, and providing the model with prompts such as "Please depict a traditional Japanese festival." The generated images then undergo a cultural bias detection process using a combination of natural language processing and image recognition techniques. If the degree of inappropriateness exceeds a certain threshold, the device issues a warning.

[0073] Users evaluate the generated images and provide feedback. This feedback includes specific comments and numerical ratings regarding image quality and cultural accuracy. This evaluation information is analyzed by the server, used to update the data set, and to retrain the generation AI model. This ensures that the generated images are closer to the expectations of customers and target audiences, and are more culturally appropriate.

[0074] As a concrete example, consider image generation for a traditional Japanese festival. In this case, the server aggregates various image data related to the festival, and the terminal generates images based on this data. Based on user feedback, the server optimizes the model and continuously improves it to provide even higher quality images in the next generation.

[0075] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0076] Step 1:

[0077] The server collects visual data from the internet and existing databases. During this collection process, it uses web scraping techniques and APIs to retrieve images, adds metadata, and stores them. The input is the URL of the source website or database, and the output is a dataset of images with metadata attached.

[0078] Step 2:

[0079] The terminal receives a dataset provided by the server and uses a generative AI model to generate images in response to input prompts. These prompts are given as specific instructions, such as "Please draw a scene from a traditional Japanese festival." The input consists of the prompt text and the dataset, and the output is the generated image.

[0080] Step 3:

[0081] The device applies an algorithm to detect cultural bias in the generated image. Using natural language processing and image recognition technologies, it determines whether elements within the image are culturally appropriate. The input is the generated image, and it outputs an evaluation of its degree of inappropriateness. If the degree of inappropriateness exceeds a set threshold, it issues a warning.

[0082] Step 4:

[0083] Users review the generated images and provide evaluation information. This includes not only numerical ratings but also text comments and suggestions for improvement. The input is the generated image and the user's perspective, and the output is the evaluation information.

[0084] Step 5:

[0085] The server analyzes user feedback and generates a new dataset. Based on this dataset, the generative AI model is retrained to improve its accuracy and cultural applicability. The input is evaluation information and existing datasets, and the output is the improved generative AI model.

[0086] (Application Example 1)

[0087] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0088] When generating content related to a specific culture, there is a risk of creating visual information that contains cultural bias or inaccurate representations. This invention aims to eliminate such offensive content and ensure that visual information has accurate and appropriate representations that reflect diverse cultural backgrounds. It also aims to improve the accuracy of the generation model by effectively utilizing user feedback.

[0089] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0090] In this invention, the server includes means for collecting data to generate visual information related to a specific culture, means for detecting cultural bias in the generated visual information, and means for collecting user feedback and enhancing model learning. This enables the generation of accurate and unbiased visual content related to a specific culture.

[0091] "Specific culture" refers to the unique traditions, customs, and values ​​shared within a particular region, ethnic group, religion, or social group.

[0092] "Visual information" refers to information that is perceived visually, such as images and videos.

[0093] "Cultural bias" refers to biased expressions that contain stereotypes or misleading content associated with a particular culture.

[0094] The "discomfort index" refers to an evaluation scale that quantifies the level of discomfort felt in generated visual information.

[0095] A "model" refers to an algorithm that generates visual information by learning from data.

[0096] "User evaluations" refer to information such as impressions, opinions, and feedback provided by users regarding the generated visual information.

[0097] As an embodiment of the present invention, a system is provided that generates visual information accurately based on a specific culture. This system consists of three main components: a server, a terminal, and a user.

[0098] First, the server collects visual data related to a specific culture through a large database. The collected data is constructed as a base dataset for generating visual information and used to train generative AI models. This data is then managed with appropriate metadata. The server can utilize image generation models such as OpenAI's DALL-E or Stable Diffusion.

[0099] Next, the device generates visual information based on a dataset provided by the server. A bias detection algorithm verifies whether the generated visual information contains cultural bias, and makes corrections as needed. Analysis tools such as Google® Cloud Vision API can be used to check for bias.

[0100] Finally, users play a role in providing feedback on the generated visual information. This user feedback is used to retrain and improve the model. Users evaluate the generated visual information from diverse perspectives to ensure its cultural appropriateness. The training used in this feedback process includes educational content to support fair evaluation.

[0101] As a concrete example, consider a case where a user requests an image of a "Japanese spring festival." In this case, the server collects visual data related to Japanese spring festivals and uses it to generate an image on the terminal. The generated image is checked for cultural bias and, if appropriate, is provided to the user. After reviewing the image, the user provides feedback based on the prompt "Generate a vibrant image of a traditional Japanese spring festival including cherry blossoms and traditional costumes," and this data is used to further improve the system.

[0102] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0103] Step 1:

[0104] The server collects visual data related to a specific culture. The data is extracted from the internet and existing databases, and the collected visual data, along with metadata, is stored in the database. The input is raw image data from the internet, and the output is a dataset organized for training. The collected data undergoes data processing such as categorization and tagging before being used to train a generative AI model.

[0105] Step 2:

[0106] The terminal receives a dataset from the server and generates visual information based on user instructions. The input consists of the user's prompt text and associated dataset, while the output is the generated visual information. A generative AI model is used to output culturally appropriate images corresponding to the prompt text.

[0107] Step 3:

[0108] The device detects whether the generated visual information contains cultural bias. The input is the generated visual information, and the output is the bias detection result. The image is analyzed using the Google Cloud Vision API, and if bias is found, it is corrected.

[0109] Step 4:

[0110] The server sends the user the visual information after bias detection is complete. The input is the corrected visual information, and the output is the image presented to the user. The server verifies the corrected content and transfers it to the user's device.

[0111] Step 5:

[0112] Users provide feedback on the visual information they are given. The input is the presented visual information, and the output is the feedback information. They evaluate the information from various perspectives and send their opinions back to the server.

[0113] Step 6:

[0114] The server analyzes user feedback and retrains the generative AI model. The input is feedback data, and the output is the improved trained model. The feedback is then incorporated into new training data to improve the model's accuracy.

[0115] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0116] This invention combines a system that generates images related to a specific culture and detects and evaluates cultural bias in the generation process with an emotion engine that recognizes the user's emotions. This system consists of a server, a terminal, and a user interface equipped with the emotion engine.

[0117] Image generation and bias detection

[0118] The server collects data related to a specific culture, tags it, and adds metadata to it, then stores the resulting dataset in a database. This data provides a foundation for accurately representing that particular culture.

[0119] The device executes an image generation process based on the collected data, and generates images with cultural characteristics using an algorithm optimized based on specific conditions.

[0120] The device automatically analyzes the generated images to detect bias, thereby verifying their cultural validity. It also evaluates the discomfort index and displays a warning if it exceeds the set threshold.

[0121] User emotion recognition by an emotion engine

[0122] The user interacts with an interface to evaluate the provided images. During this process, an emotion engine recognizes the user's emotional state in real time and records their emotional response to the images.

[0123] The emotion engine collects recognized user emotion data and sends it to the server as part of the feedback based on the emotional review of the image. This information is used to evaluate the image's discomfort index and improve the model.

[0124] Model update and improvement process

[0125] The server analyzes the sentiment data and feedback obtained from users and generates a new dataset based on this information.

[0126] The server retrains its model by incorporating feedback, including sentiment data, with the aim of more accurately reflecting specific cultures. This process improves the quality and cultural appropriateness of the generated images.

[0127] As a concrete example, consider the case of image generation themed around the Japanese tea ceremony. The server collects high-precision image data related to the tea ceremony, and the terminal uses this data to generate images that recreate the tea ceremony ritual. The user evaluates the generated images through an emotional interface, and the emotional responses are also used in the analysis. This allows the improved model to provide more culturally accurate and user-friendly results in future image generation.

[0128] The following describes the processing flow.

[0129] Step 1:

[0130] The server collects data related to a specific culture from the internet and dedicated databases. This data includes information such as the history, tools, and procedures of the tea ceremony, and is tagged with detailed metadata. The server creates a dataset based on this data and stores it in the database as the basis for image generation.

[0131] Step 2:

[0132] The device generates images themed around the Japanese tea ceremony. Using a dataset supplied from the server, it generates images based on an algorithm that faithfully reproduces specific cultural elements. In this process, the generation algorithm aims to create culturally accurate images by making maximum use of information about the culture.

[0133] Step 3:

[0134] The device sends the generated image to a bias detection module. Here, it automatically analyzes for cultural bias and inaccuracies and calculates an discomfort index. If the discomfort index exceeds a set threshold, the device displays a warning and requests image adjustments.

[0135] Step 4:

[0136] The user evaluates the generated image using their device. During this process, the emotion engine analyzes the user's emotions in real time and records their emotional response to the image as emotion data. The user then provides specific feedback based on this evaluation.

[0137] Step 5:

[0138] The emotion engine uses the acquired emotion data to send it to the server along with user feedback. The emotion data is used as supplementary information for evaluating the discomfort index of images.

[0139] Step 6:

[0140] The server collects and analyzes user feedback and sentiment data. Based on this, a new training dataset is created, and the model is retrained to improve the accuracy and cultural relevance of the generated images.

[0141] Step 7:

[0142] The server applies the retrained model to the system. This updated model enables future image generation to be more accurate and culturally appropriate. This process is continuous, enhancing the overall system performance and ethical aspects.

[0143] (Example 2)

[0144] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0145] In generating images that reflect specific cultures, the evaluation of cultural bias and discomfort indices is insufficient, potentially leading to unpleasant experiences for users. Furthermore, the lack of mechanisms to effectively utilize user feedback and sentiment data to improve the model makes it difficult to improve the quality and cultural appropriateness of generated images.

[0146] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0147] In this invention, the server includes means for automatically collecting and organizing information related to a specific culture, means for analyzing whether the generated images are free from cultural bias, and means for collecting emotional data from users and generating feedback based on that data. This makes it possible to improve image generation that accurately reflects the characteristics of a specific culture while responding to the user's emotions.

[0148] "Specific culture" refers to the characteristics of a culture that include customs, values, and aesthetics unique to a particular region or society.

[0149] "Information" refers to materials such as text, images, audio, and video that are stored and processed as digital data.

[0150] A "prompt statement" is an instruction given to a generative AI model, providing guidelines for generating images with a specific theme or style.

[0151] "Image generation" refers to the process of creating visual representations from text and data using computer algorithms.

[0152] "Cultural bias" refers to the phenomenon where prejudices and stereotypes about a particular culture or society are reflected in digital content.

[0153] The "Discomfort Index" refers to a criterion for evaluating digital content that may cause stress or discomfort to users.

[0154] "Emotional data" refers to information that indicates a user's emotional state, including data such as facial expressions, tone of voice, and other biometric information.

[0155] "Feedback" refers to responses and opinions generated based on user evaluations, which are used to improve systems and processes.

[0156] "Retraining" refers to the process of training an existing machine learning model again using a new dataset, which improves its accuracy and performance.

[0157] The embodiments for carrying out the invention are described below.

[0158] This invention provides a system that generates images reflecting a specific culture, detects and evaluates cultural bias, and incorporates user emotional data as feedback. Specifically, it performs the following processing using a server, terminal, and a user interface equipped with an emotion engine.

[0159] Server Processing

[0160] The server automatically collects information related to a specific culture from the internet and existing datasets. This information includes images, text, audio, and video. This information is tagged and stored in a database. This database provides the foundation necessary for generating images that accurately reflect the characteristics of that particular culture.

[0161] Terminal processing

[0162] The terminal generates images using a generative AI model based on information obtained from the server. Specifically, it takes a prompt message as input and executes the image generation algorithm. For example, using the prompt "Generate an image that expresses the spirit of the Japanese tea ceremony," it will generate an image related to the Japanese tea ceremony. The generated image is then analyzed to detect cultural bias.

[0163] Emotion engine and user processing

[0164] Users evaluate images generated using a user interface that includes an emotion engine. During this process, emotional data is collected from the user's facial expressions, tone of voice, and other factors. The emotion engine analyzes this data and generates feedback based on the user's emotions.

[0165] This feedback is sent to the server and used to retrain the model. Retraining makes the generated images more culturally appropriate and appealing to users.

[0166] The implementation of this system will enable continuous improvement in image generation that accurately reflects specific cultures, thereby enhancing the user experience.

[0167] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0168] Step 1:

[0169] The server automatically collects information related to a specific culture from the internet and existing datasets. Specifically, the server uses a crawling algorithm to extract images, text, and other data based on specific keywords and tags. The input to this process is keywords related to the specific culture, and the output is a dataset of tagged information. The server automatically tags the collected information and stores it in a database.

[0170] Step 2:

[0171] The device generates an image by inputting a prompt message into a generation AI model based on a dataset obtained from the server. An example of a prompt message might be, "Please generate an image that expresses the spirit of the Japanese tea ceremony." Based on this input, the device runs the generation AI model and generates an image that possesses the characteristics of a specific culture. The output is an image related to the target culture.

[0172] Step 3:

[0173] The device analyzes the generated image for cultural bias. Specifically, it uses an image analysis algorithm to evaluate the image and calculate an discomfort index. The input for this step is the generated image, and the output is the bias analysis result and the discomfort index evaluation. If the discomfort index exceeds a set threshold, the device displays a warning message to the user.

[0174] Step 4:

[0175] The user reviews the images generated through the interface. During the interaction, the emotion engine collects the user's emotional data. Specific actions include detecting the user's facial expressions and voice tone, and receiving feedback in the form of a questionnaire. The input for this step is the user's emotional expressions and feedback, and the output is stored in the system as emotional data.

[0176] Step 5:

[0177] The server analyzes the collected user sentiment data and generates new feedback. This data is then used to retrain the model to improve system performance. Specifically, this involves creating a new dataset that reflects the sentiment data and feedback, and retraining the machine learning model. The input for this step is the sentiment data and feedback obtained from the user, and the output is the retuned generative AI model.

[0178] (Application Example 2)

[0179] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0180] In creating content related to a specific culture, there is a risk of inaccurate cultural representation due to cultural bias, and a problem exists where content that is not optimized for individual users is provided due to a failure to consider users' emotional responses, resulting in decreased user satisfaction. It is necessary to resolve these issues and provide users with culturally accurate and personalized visual experiences.

[0181] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0182] In this invention, the server includes means for acquiring information related to a specific culture, means for detecting cultural bias in the generated visual data, and means for recognizing the user's emotional state and recording their emotional response. This makes it possible to generate and provide culturally accurate and personalized content to the user.

[0183] A "specific culture" refers to a cultural system that includes customs, traditions, values, or artistic expressions unique to a particular region, ethnic group, or country.

[0184] An "image" is a representation of visual information as digital data, displayed on the screen of a computer or electronic device.

[0185] "Bias" refers to any inclination or preconception that influences data or results, whether intentionally or unintentionally.

[0186] The "Discomfort Index" is a numerical indicator that quantifies the degree of discomfort or inappropriateness experienced by users.

[0187] An "algorithmic model" refers to a mathematical or computational method designed to process data and produce an output for a specific purpose.

[0188] "Emotional state" refers to the user's psychological response, including emotions such as happiness, anger, and sadness.

[0189] "Emotional response" refers to the emotional changes a person exhibits in response to a specific stimulus.

[0190] "Content personalization" refers to customizing the information and services provided based on the individual user's preferences and needs.

[0191] In embodiments of this invention, the system includes a server, a terminal, and a user interface. The server acquires a wide range of information related to a specific culture and stores this information in an organized database. The data used serves as a foundation for accurately reflecting the specified culture. The collected data is tagged and metadata is added to enable efficient searching and access.

[0192] The device receives data from the server and generates culturally relevant visual data using a generative AI model. During the generation process, the algorithmic model creates images with culturally appropriate features. Simultaneously, the device detects cultural bias in the generated visual data and evaluates whether there are any elements that might be offensive to the user. If the discomfort index exceeds a certain threshold, the device displays a warning.

[0193] Users evaluate images and visual content through the provided interface. The device uses an emotion engine to recognize the user's emotional state in real time and record their emotional responses. This emotion data is sent to a server, where the content is personalized based on the emotional responses and reflected in future image generation. This ensures that culturally precise and engaging content tailored to each individual user is provided.

[0194] For example, when a user selects "French culture" for the system, the server uses data including historical buildings, art, and food culture related to France, and the device generates visual content about France that the user might find interesting. The user views and evaluates the images, and the content generated next time is improved based on their more personal experience.

[0195] An example of a prompt for a generative AI model is, "Generate an image that evokes the atmosphere of France, using historical French buildings and culinary styles." In this way, the system continuously learns and improves its accuracy, resulting in richer cultural expression.

[0196] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0197] Step 1:

[0198] The server collects information related to a specific culture from the internet and existing databases. This information includes various formats such as text, images, and audio. The input is a prompt about the culture to be collected. The server organizes this information, adds metadata, and stores it. The output is a well-organized dataset associated with the specific culture.

[0199] Step 2:

[0200] The terminal receives a dataset sent from the server and generates visual data using a generative AI model based on a specific prompt (for example, "Generate an image that evokes the atmosphere of France using historical French buildings and culinary styles"). The prompt becomes the input, the generative AI model analyzes the data, and outputs an image with cultural characteristics. In operation, the AI ​​model learns patterns in the data and creates images.

[0201] Step 3:

[0202] The device automatically analyzes the generated visual data to detect whether it contains cultural bias. The generated image is used as input, and its cultural features are analyzed by a bias detection algorithm. The output provides the presence or absence of bias and an evaluation of the discomfort index. If the bias is high, a warning is displayed.

[0203] Step 4:

[0204] The user uses an interface to evaluate images and visual content provided on the device. During this process, an emotion engine recognizes the user's facial expressions and reactions in real time and records their emotional state. The input is the user's visual reactions, and the output is emotion data. The process involves a camera capturing facial expressions, and the emotion engine analyzing that data.

[0205] Step 5:

[0206] The server receives user emotional responses and traditional feedback data, and generates a new dataset based on this. The input is emotional data from the user, and the output is a new training dataset. This dataset is used in subsequent generation processes, enabling the provision of more personalized visual data. The server updates the database to ensure the entire system evolves continuously.

[0207] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0208] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0209] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0210] [Second Embodiment]

[0211] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0212] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0213] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0214] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0215] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0216] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0217] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0218] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0219] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0220] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0221] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0222] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0223] One embodiment of the present invention is to provide an image generation system that accurately reflects a specific culture. The entire system consists of an image generation server, a terminal capable of bias detection, and a user that provides evaluation and feedback.

[0224] Image generation and bias detection

[0225] The server collects large amounts of image data related to a specific culture from the internet and existing databases. This data is stored in the database as a foundational dataset for learning that reflects the target culture. The collected data is tagged with metadata.

[0226] The device generates images using datasets provided by the server, based on specific conditions and themes. This image generation process is optimized to ensure appropriate representation, taking into account how the generated images will be received in a particular culture.

[0227] The device performs cultural bias detection on the generated images. This is a process that analyzes whether the images contain cultural inaccuracies or negative stereotypes and evaluates the discomfort index. The detection algorithm operates based on specific bias detection criteria and issues a warning for images with a discomfort index that exceeds the criteria.

[0228] Feedback and model improvement

[0229] Users receive prior training to minimize personal bias and provide appropriate feedback on the generated images. The training is designed to enable evaluation from diverse perspectives.

[0230] The server analyzes feedback collected from users and generates a new training dataset. This allows the model to be retrained, improving its accuracy so that it can generate images that more accurately reflect specific cultures.

[0231] As a concrete example, consider the case of generating images related to traditional Japanese festivals. In this scenario, the server collects diverse images related to festivals and builds a dataset. The terminal uses this dataset to generate festival-themed images, detecting and correcting cultural biases in advance. The user evaluates the generated images and provides feedback, aiming to generate more accurate images of Japanese culture.

[0232] The following describes the processing flow.

[0233] Step 1:

[0234] The server collects a vast amount of image data related to a specific culture from the internet and existing databases. These images are tagged with metadata and are appropriately culturally tagged. This constructs the dataset used for generation.

[0235] Step 2:

[0236] The device generates images using collected datasets based on specific conditions or themes. The generated images are optimized based on algorithms designed to faithfully reproduce the characteristics of a particular culture.

[0237] Step 3:

[0238] The terminal sends the generated image to a bias detection module, which automatically analyzes whether the image contains bias or inaccuracies regarding a particular culture or ethnicity.

[0239] Step 4:

[0240] The device evaluates the discomfort index of an image and displays a warning if the index is high, according to the set criteria. This process can prevent inaccurate representations of specific cultures.

[0241] Step 5:

[0242] Users complete pre-training and provide expert feedback on the generated images. By evaluating images from diverse perspectives and providing feedback, users can mitigate personal bias.

[0243] Step 6:

[0244] The server collects and analyzes the feedback received from users. Based on the results of this analysis, a new dataset is generated, and the model is retrained.

[0245] Step 7:

[0246] The server incorporates the results of retraining, enabling more accurate image generation. The goal is to generate unbiased images that accurately reflect a specific culture.

[0247] (Example 1)

[0248] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0249] When generating images that accurately reflect a specific culture, there is a risk of including cultural biases or inappropriate representations. This can lead to misunderstandings and cultural friction, highlighting the need to improve the quality and cultural appropriateness of the generated images.

[0250] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0251] In this invention, the server includes means for extracting visual data related to a specific culture, means for creating images in response to input instructions using a generative AI model, and means for detecting cultural bias within the created images. This enables the generation of high-quality images that are sensitive to culture.

[0252] "Visual data" refers to data that includes visual information such as images and videos, and is used as an element that reflects a particular culture.

[0253] A "generative AI model" is an algorithm that uses artificial intelligence to create new images and designs, and is a technology that generates images based on specific conditions.

[0254] "Input instructions" refer to instructions that explicitly state the request for a specific image generation, and mean prompts or commands given to the generation AI model.

[0255] "Cultural bias" refers to a state in which misunderstandings or negative stereotypes about a particular culture are included in expressions, and biased views are a factor that hinders a correct understanding of a culture.

[0256] "Inappropriateness" is an index that evaluates how culturally appropriate a generated image is, and indicates that corrections are necessary if the image does not meet the standard.

[0257] A "warning" is a notification issued when the level of cultural bias or inappropriateness within an image exceeds a certain standard, and it is an important sign that action should be taken.

[0258] "Evaluation information" refers to the feedback and evaluation data provided for the generated images, and is used to improve the model.

[0259] A "dataset" is a collection of data used for training and retraining, and forms the basis for generating new images.

[0260] This invention provides an image generation system that accurately reflects a specific culture. The system includes the management of visual data, the creation of images using a generative AI model, the detection and warning of cultural biases, and the continuous improvement of the model.

[0261] The server first collects visual data related to a specific culture from the internet and existing databases. This is done using web scraping techniques to retrieve images through APIs of public databases. The collected data is then stored in the database with metadata added.

[0262] The device generates images based on a dataset provided by the server, using a generative AI model. This involves using generative models such as GANs (Generative Adversarial Networks) or VQ-VAE, and providing the model with prompts such as "Please depict a traditional Japanese festival." The generated images then undergo a cultural bias detection process using a combination of natural language processing and image recognition techniques. If the degree of inappropriateness exceeds a certain threshold, the device issues a warning.

[0263] Users evaluate the generated images and provide feedback. This feedback includes specific comments and numerical ratings regarding image quality and cultural accuracy. This evaluation information is analyzed by the server, used to update the data set, and to retrain the generation AI model. This ensures that the generated images are closer to the expectations of customers and target audiences, and are more culturally appropriate.

[0264] As a concrete example, consider image generation for a traditional Japanese festival. In this case, the server aggregates various image data related to the festival, and the terminal generates images based on this data. Based on user feedback, the server optimizes the model and continuously improves it to provide even higher quality images in the next generation.

[0265] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0266] Step 1:

[0267] The server collects visual data from the internet and existing databases. During this collection process, it uses web scraping techniques and APIs to retrieve images, adds metadata, and stores them. The input is the URL of the source website or database, and the output is a dataset of images with metadata attached.

[0268] Step 2:

[0269] The terminal receives a dataset provided by the server and uses a generative AI model to generate images in response to input prompts. These prompts are given as specific instructions, such as "Please draw a scene from a traditional Japanese festival." The input consists of the prompt text and the dataset, and the output is the generated image.

[0270] Step 3:

[0271] The device applies an algorithm to detect cultural bias in the generated image. Using natural language processing and image recognition technologies, it determines whether elements within the image are culturally appropriate. The input is the generated image, and it outputs an evaluation of its degree of inappropriateness. If the degree of inappropriateness exceeds a set threshold, it issues a warning.

[0272] Step 4:

[0273] Users review the generated images and provide evaluation information. This includes not only numerical ratings but also text comments and suggestions for improvement. The input is the generated image and the user's perspective, and the output is the evaluation information.

[0274] Step 5:

[0275] The server analyzes user feedback and generates a new dataset. Based on this dataset, the generative AI model is retrained to improve its accuracy and cultural applicability. The input is evaluation information and existing datasets, and the output is the improved generative AI model.

[0276] (Application Example 1)

[0277] Next, Application Example 1 will be described. In the following description, the data processing device 12 is referred to as a "server", and the smart glasses 214 are referred to as a "terminal".

[0278] In the generation of content related to a specific culture, there is a risk that visual information including cultural biases and inaccurate expressions will be created. The present invention aims to eliminate the resulting unpleasant content and to ensure that visual information has accurate and appropriate expressions according to various cultural backgrounds. It is also an object to effectively utilize feedback from users to improve the accuracy of the generation model.

[0279] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0280] In this invention, the server includes means for collecting data for generating visual information related to a specific culture, means for detecting cultural biases in the generated visual information, and means for collecting evaluations from users and strengthening model learning. This enables the generation of accurate and unbiased visual content for a specific culture.

[0281] "Specific culture" refers to unique traditions, customs, values, etc. shared within a specific region, ethnic group, religion, or social group.

[0282] "Visual information" refers to information that can be visually recognized, such as images and videos.

[0283] "Cultural bias" refers to a biased expression that includes stereotypes and misunderstandings related to a specific culture.

[0284] "Unpleasant index" refers to an evaluation scale that quantifies the unpleasantness in the generated visual information.

[0285] The "model" refers to an algorithm that generates visual information by learning based on data.

[0286] The "evaluation from the user" refers to information such as impressions, opinions, and feedback provided by the user for the generated visual information.

[0287] As a form for implementing the present invention, a system for generating visual information accurately based on a specific culture is provided. This system consists of three main components: a server, a terminal, and a user.

[0288] First, the server collects visual data related to a specific culture through a large database. The collected data is constructed as a basic dataset for generating visual information and is used for the learning of the generation AI model. Appropriate metadata is added to and managed for this data. The server can utilize, for example, image generation models such as OpenAI's DALL-E or Stable Diffusion.

[0289] Next, the terminal generates visual information based on the dataset provided by the server. It verifies whether the generated visual information contains cultural biases using a bias detection algorithm and makes corrections if necessary. Analysis tools such as Google Cloud Vision API can be used to confirm the biases.

[0290] Finally, the user plays the role of providing feedback on the generated visual information. The user's feedback is reflected in the re-learning and improvement of the model. The user evaluates from various perspectives to confirm whether the generated visual information is culturally appropriate. The training utilized in this feedback process includes educational content to support fair evaluation.

[0291] As a concrete example, consider a case where a user requests an image of a "Japanese spring festival." In this case, the server collects visual data related to Japanese spring festivals and uses it to generate an image on the terminal. The generated image is checked for cultural bias and, if appropriate, is provided to the user. After reviewing the image, the user provides feedback based on the prompt "Generate a vibrant image of a traditional Japanese spring festival including cherry blossoms and traditional costumes," and this data is used to further improve the system.

[0292] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0293] Step 1:

[0294] The server collects visual data related to a specific culture. The data is extracted from the internet and existing databases, and the collected visual data, along with metadata, is stored in the database. The input is raw image data from the internet, and the output is a dataset organized for training. The collected data undergoes data processing such as categorization and tagging before being used to train a generative AI model.

[0295] Step 2:

[0296] The terminal receives a dataset from the server and generates visual information based on user instructions. The input consists of the user's prompt text and associated dataset, while the output is the generated visual information. A generative AI model is used to output culturally appropriate images corresponding to the prompt text.

[0297] Step 3:

[0298] The device detects whether the generated visual information contains cultural bias. The input is the generated visual information, and the output is the bias detection result. The image is analyzed using the Google Cloud Vision API, and if bias is found, it is corrected.

[0299] Step 4:

[0300] The server sends the user the visual information after bias detection is complete. The input is the corrected visual information, and the output is the image presented to the user. The server verifies the corrected content and transfers it to the user's device.

[0301] Step 5:

[0302] Users provide feedback on the visual information they are given. The input is the presented visual information, and the output is the feedback information. They evaluate the information from various perspectives and send their opinions back to the server.

[0303] Step 6:

[0304] The server analyzes user feedback and retrains the generative AI model. The input is feedback data, and the output is the improved trained model. The feedback is then incorporated into new training data to improve the model's accuracy.

[0305] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0306] This invention combines a system that generates images related to a specific culture and detects and evaluates cultural bias in the generation process with an emotion engine that recognizes the user's emotions. This system consists of a server, a terminal, and a user interface equipped with the emotion engine.

[0307] Image Generation and Bias Detection

[0308] The server collects data related to a specific culture, tags and attaches metadata to it, and stores the dataset in a database. This data provides a basis for accurately representing the specific culture.

[0309] The terminal executes an image generation process based on the collected data and uses an optimized algorithm based on specific conditions to generate images with cultural characteristics.

[0310] The terminal automatically analyzes the generated image for the presence or absence of bias, thereby confirming the cultural validity of the image. It evaluates the discomfort index and displays a warning if it exceeds the set criteria.

[0311] User Emotion Recognition by the Emotion Engine

[0312] The interface for evaluating the images provided by the user is operated. At this time, the emotion engine recognizes the user's emotional state in real time and records the emotional reaction to the images.

[0313] The emotion engine collects the recognized user emotion data and sends it to the server as part of the feedback based on the emotional review of the images. This information is utilized for evaluating the discomfort index of the images and improving the model.

[0314] Model Update and Improvement Process

[0315] The server analyzes the emotion data and feedback obtained from the user and generates a new dataset based on this information.

[0316] The server incorporates the feedback including the emotion data to retrain the model, aiming to more accurately reflect the specific culture. Through this process, the quality and cultural appropriateness of the generated images are improved.

[0317] As a concrete example, consider the case of image generation themed around the Japanese tea ceremony. The server collects high-precision image data related to the tea ceremony, and the terminal uses this data to generate images that recreate the tea ceremony ritual. The user evaluates the generated images through an emotional interface, and the emotional responses are also used in the analysis. This allows the improved model to provide more culturally accurate and user-friendly results in future image generation.

[0318] The following describes the processing flow.

[0319] Step 1:

[0320] The server collects data related to a specific culture from the internet and dedicated databases. This data includes information such as the history, tools, and procedures of the tea ceremony, and is tagged with detailed metadata. The server creates a dataset based on this data and stores it in the database as the basis for image generation.

[0321] Step 2:

[0322] The device generates images themed around the Japanese tea ceremony. Using a dataset supplied from the server, it generates images based on an algorithm that faithfully reproduces specific cultural elements. In this process, the generation algorithm aims to create culturally accurate images by making maximum use of information about the culture.

[0323] Step 3:

[0324] The device sends the generated image to a bias detection module. Here, it automatically analyzes for cultural bias and inaccuracies and calculates an discomfort index. If the discomfort index exceeds a set threshold, the device displays a warning and requests image adjustments.

[0325] Step 4:

[0326] The user evaluates the generated image using their device. During this process, the emotion engine analyzes the user's emotions in real time and records their emotional response to the image as emotion data. The user then provides specific feedback based on this evaluation.

[0327] Step 5:

[0328] The emotion engine uses the acquired emotion data to send it to the server along with user feedback. The emotion data is used as supplementary information for evaluating the discomfort index of images.

[0329] Step 6:

[0330] The server collects and analyzes user feedback and sentiment data. Based on this, a new training dataset is created, and the model is retrained to improve the accuracy and cultural relevance of the generated images.

[0331] Step 7:

[0332] The server applies the retrained model to the system. This updated model enables future image generation to be more accurate and culturally appropriate. This process is continuous, enhancing the overall system performance and ethical aspects.

[0333] (Example 2)

[0334] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0335] In generating images that reflect specific cultures, the evaluation of cultural bias and discomfort indices is insufficient, potentially leading to unpleasant experiences for users. Furthermore, the lack of mechanisms to effectively utilize user feedback and sentiment data to improve the model makes it difficult to improve the quality and cultural appropriateness of generated images.

[0336] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0337] In this invention, the server includes means for automatically collecting and organizing information related to a specific culture, means for analyzing whether the generated images are free from cultural bias, and means for collecting emotional data from users and generating feedback based on that data. This makes it possible to improve image generation that accurately reflects the characteristics of a specific culture while responding to the user's emotions.

[0338] "Specific culture" refers to the characteristics of a culture that include customs, values, and aesthetics unique to a particular region or society.

[0339] "Information" refers to materials such as text, images, audio, and video that are stored and processed as digital data.

[0340] A "prompt statement" is an instruction given to a generative AI model, providing guidelines for generating images with a specific theme or style.

[0341] "Image generation" refers to the process of creating visual representations from text and data using computer algorithms.

[0342] "Cultural bias" refers to the phenomenon where prejudices and stereotypes about a particular culture or society are reflected in digital content.

[0343] The "Discomfort Index" refers to a criterion for evaluating digital content that may cause stress or discomfort to users.

[0344] "Emotional data" refers to information that indicates a user's emotional state, including data such as facial expressions, tone of voice, and other biometric information.

[0345] "Feedback" refers to responses and opinions generated based on user evaluations, which are used to improve systems and processes.

[0346] "Retraining" refers to the process of training an existing machine learning model again using a new dataset, which improves its accuracy and performance.

[0347] The embodiments for carrying out the invention are described below.

[0348] This invention provides a system that generates images reflecting a specific culture, detects and evaluates cultural bias, and incorporates user emotional data as feedback. Specifically, it performs the following processing using a server, terminal, and a user interface equipped with an emotion engine.

[0349] Server Processing

[0350] The server automatically collects information related to a specific culture from the internet and existing datasets. This information includes images, text, audio, and video. This information is tagged and stored in a database. This database provides the foundation necessary for generating images that accurately reflect the characteristics of that particular culture.

[0351] Terminal processing

[0352] The terminal generates images using a generative AI model based on information obtained from the server. Specifically, it takes a prompt message as input and executes the image generation algorithm. For example, using the prompt "Generate an image that expresses the spirit of the Japanese tea ceremony," it will generate an image related to the Japanese tea ceremony. The generated image is then analyzed to detect cultural bias.

[0353] Emotion engine and user processing

[0354] Users evaluate images generated using a user interface that includes an emotion engine. During this process, emotional data is collected from the user's facial expressions, tone of voice, and other factors. The emotion engine analyzes this data and generates feedback based on the user's emotions.

[0355] This feedback is sent to the server and used to retrain the model. Retraining makes the generated images more culturally appropriate and appealing to users.

[0356] The implementation of this system will enable continuous improvement in image generation that accurately reflects specific cultures, thereby enhancing the user experience.

[0357] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0358] Step 1:

[0359] The server automatically collects information related to a specific culture from the internet and existing datasets. Specifically, the server uses a crawling algorithm to extract images, text, and other data based on specific keywords and tags. The input to this process is keywords related to the specific culture, and the output is a dataset of tagged information. The server automatically tags the collected information and stores it in a database.

[0360] Step 2:

[0361] The device generates an image by inputting a prompt message into a generation AI model based on a dataset obtained from the server. An example of a prompt message might be, "Please generate an image that expresses the spirit of the Japanese tea ceremony." Based on this input, the device runs the generation AI model and generates an image that possesses the characteristics of a specific culture. The output is an image related to the target culture.

[0362] Step 3:

[0363] The device analyzes the generated image for cultural bias. Specifically, it uses an image analysis algorithm to evaluate the image and calculate an discomfort index. The input for this step is the generated image, and the output is the bias analysis result and the discomfort index evaluation. If the discomfort index exceeds a set threshold, the device displays a warning message to the user.

[0364] Step 4:

[0365] The user reviews the images generated through the interface. During the interaction, the emotion engine collects the user's emotional data. Specific actions include detecting the user's facial expressions and voice tone, and receiving feedback in the form of a questionnaire. The input for this step is the user's emotional expressions and feedback, and the output is stored in the system as emotional data.

[0366] Step 5:

[0367] The server analyzes the collected user sentiment data and generates new feedback. This data is then used to retrain the model to improve system performance. Specifically, this involves creating a new dataset that reflects the sentiment data and feedback, and retraining the machine learning model. The input for this step is the sentiment data and feedback obtained from the user, and the output is the retuned generative AI model.

[0368] (Application Example 2)

[0369] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0370] In creating content related to a specific culture, there is a risk of inaccurate cultural representation due to cultural bias, and a problem exists where content that is not optimized for individual users is provided due to a failure to consider users' emotional responses, resulting in decreased user satisfaction. It is necessary to resolve these issues and provide users with culturally accurate and personalized visual experiences.

[0371] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0372] In this invention, the server includes means for acquiring information related to a specific culture, means for detecting cultural bias in the generated visual data, and means for recognizing the user's emotional state and recording their emotional response. This makes it possible to generate and provide culturally accurate and personalized content to the user.

[0373] A "specific culture" refers to a cultural system that includes customs, traditions, values, or artistic expressions unique to a particular region, ethnic group, or country.

[0374] An "image" is a representation of visual information as digital data, displayed on the screen of a computer or electronic device.

[0375] "Bias" refers to any inclination or preconception that influences data or results, whether intentionally or unintentionally.

[0376] The "Discomfort Index" is a numerical indicator that quantifies the degree of discomfort or inappropriateness experienced by users.

[0377] An "algorithmic model" refers to a mathematical or computational method designed to process data and produce an output for a specific purpose.

[0378] "Emotional state" refers to the user's psychological response, including emotions such as happiness, anger, and sadness.

[0379] "Emotional response" refers to the emotional changes a person exhibits in response to a specific stimulus.

[0380] "Content personalization" refers to customizing the information and services provided based on the individual user's preferences and needs.

[0381] In embodiments of this invention, the system includes a server, a terminal, and a user interface. The server acquires a wide range of information related to a specific culture and stores this information in an organized database. The data used serves as a foundation for accurately reflecting the specified culture. The collected data is tagged and metadata is added to enable efficient searching and access.

[0382] The device receives data from the server and generates culturally relevant visual data using a generative AI model. During the generation process, the algorithmic model creates images with culturally appropriate features. Simultaneously, the device detects cultural bias in the generated visual data and evaluates whether there are any elements that might be offensive to the user. If the discomfort index exceeds a certain threshold, the device displays a warning.

[0383] Users evaluate images and visual content through the provided interface. The device uses an emotion engine to recognize the user's emotional state in real time and record their emotional responses. This emotion data is sent to a server, where the content is personalized based on the emotional responses and reflected in future image generation. This ensures that culturally precise and engaging content tailored to each individual user is provided.

[0384] For example, when a user selects "French culture" for the system, the server uses data including historical buildings, art, and food culture related to France, and the device generates visual content about France that the user might find interesting. The user views and evaluates the images, and the content generated next time is improved based on their more personal experience.

[0385] An example of a prompt for a generative AI model is, "Generate an image that evokes the atmosphere of France, using historical French buildings and culinary styles." In this way, the system continuously learns and improves its accuracy, resulting in richer cultural expression.

[0386] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0387] Step 1:

[0388] The server collects information related to a specific culture from the internet and existing databases. This information includes various formats such as text, images, and audio. The input is a prompt about the culture to be collected. The server organizes this information, adds metadata, and stores it. The output is a well-organized dataset associated with the specific culture.

[0389] Step 2:

[0390] The terminal receives a dataset sent from the server and generates visual data using a generative AI model based on a specific prompt (for example, "Generate an image that evokes the atmosphere of France using historical French buildings and culinary styles"). The prompt becomes the input, the generative AI model analyzes the data, and outputs an image with cultural characteristics. In operation, the AI ​​model learns patterns in the data and creates images.

[0391] Step 3:

[0392] The device automatically analyzes the generated visual data to detect whether it contains cultural bias. The generated image is used as input, and its cultural features are analyzed by a bias detection algorithm. The output provides the presence or absence of bias and an evaluation of the discomfort index. If the bias is high, a warning is displayed.

[0393] Step 4:

[0394] The user uses an interface to evaluate images and visual content provided on the device. During this process, an emotion engine recognizes the user's facial expressions and reactions in real time and records their emotional state. The input is the user's visual reactions, and the output is emotion data. The process involves a camera capturing facial expressions, and the emotion engine analyzing that data.

[0395] Step 5:

[0396] The server receives user emotional responses and traditional feedback data, and generates a new dataset based on this. The input is emotional data from the user, and the output is a new training dataset. This dataset is used in subsequent generation processes, enabling the provision of more personalized visual data. The server updates the database to ensure the entire system evolves continuously.

[0397] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0398] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0399] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0400] [Third Embodiment]

[0401] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0402] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0403] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0404] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0405] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0406] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0407] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0408] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0409] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0410] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0411] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0412] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0413] One embodiment of the present invention is to provide an image generation system that accurately reflects a specific culture. The entire system consists of an image generation server, a terminal capable of bias detection, and a user that provides evaluation and feedback.

[0414] Image generation and bias detection

[0415] The server collects large amounts of image data related to a specific culture from the internet and existing databases. This data is stored in the database as a foundational dataset for learning that reflects the target culture. The collected data is tagged with metadata.

[0416] The device generates images using datasets provided by the server, based on specific conditions and themes. This image generation process is optimized to ensure appropriate representation, taking into account how the generated images will be received in a particular culture.

[0417] The device performs cultural bias detection on the generated images. This is a process that analyzes whether the images contain cultural inaccuracies or negative stereotypes and evaluates the discomfort index. The detection algorithm operates based on specific bias detection criteria and issues a warning for images with a discomfort index that exceeds the criteria.

[0418] Feedback and model improvement

[0419] Users receive prior training to minimize personal bias and provide appropriate feedback on the generated images. The training is designed to enable evaluation from diverse perspectives.

[0420] The server analyzes feedback collected from users and generates a new training dataset. This allows the model to be retrained, improving its accuracy so that it can generate images that more accurately reflect specific cultures.

[0421] As a concrete example, consider the case of generating images related to traditional Japanese festivals. In this scenario, the server collects diverse images related to festivals and builds a dataset. The terminal uses this dataset to generate festival-themed images, detecting and correcting cultural biases in advance. The user evaluates the generated images and provides feedback, aiming to generate more accurate images of Japanese culture.

[0422] The following describes the processing flow.

[0423] Step 1:

[0424] The server collects a vast amount of image data related to a specific culture from the internet and existing databases. These images are tagged with metadata and are appropriately culturally tagged. This constructs the dataset used for generation.

[0425] Step 2:

[0426] The device generates images using collected datasets based on specific conditions or themes. The generated images are optimized based on algorithms designed to faithfully reproduce the characteristics of a particular culture.

[0427] Step 3:

[0428] The terminal sends the generated image to a bias detection module, which automatically analyzes whether the image contains bias or inaccuracies regarding a particular culture or ethnicity.

[0429] Step 4:

[0430] The device evaluates the discomfort index of an image and displays a warning if the index is high, according to the set criteria. This process can prevent inaccurate representations of specific cultures.

[0431] Step 5:

[0432] Users complete pre-training and provide expert feedback on the generated images. By evaluating images from diverse perspectives and providing feedback, users can mitigate personal bias.

[0433] Step 6:

[0434] The server collects and analyzes the feedback received from users. Based on the results of this analysis, a new dataset is generated, and the model is retrained.

[0435] Step 7:

[0436] The server incorporates the results of retraining, enabling more accurate image generation. The goal is to generate unbiased images that accurately reflect a specific culture.

[0437] (Example 1)

[0438] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0439] When generating images that accurately reflect a specific culture, there is a risk of including cultural biases or inappropriate representations. This can lead to misunderstandings and cultural friction, highlighting the need to improve the quality and cultural appropriateness of the generated images.

[0440] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0441] In this invention, the server includes means for extracting visual data related to a specific culture, means for creating images in response to input instructions using a generative AI model, and means for detecting cultural bias within the created images. This enables the generation of high-quality images that are sensitive to culture.

[0442] "Visual data" refers to data that includes visual information such as images and videos, and is used as an element that reflects a particular culture.

[0443] A "generative AI model" is an algorithm that uses artificial intelligence to create new images and designs, and is a technology that generates images based on specific conditions.

[0444] "Input instructions" refer to instructions that explicitly state the request for a specific image generation, and mean prompts or commands given to the generation AI model.

[0445] "Cultural bias" refers to a state in which misunderstandings or negative stereotypes about a particular culture are included in expressions, and biased views are a factor that hinders a correct understanding of a culture.

[0446] "Inappropriateness" is an index that evaluates how culturally appropriate a generated image is, and indicates that corrections are necessary if the image does not meet the standard.

[0447] A "warning" is a notification issued when the level of cultural bias or inappropriateness within an image exceeds a certain standard, and it is an important sign that action should be taken.

[0448] "Evaluation information" refers to the feedback and evaluation data provided for the generated images, and is used to improve the model.

[0449] A "dataset" is a collection of data used for training and retraining, and forms the basis for generating new images.

[0450] This invention provides an image generation system that accurately reflects a specific culture. The system includes the management of visual data, the creation of images using a generative AI model, the detection and warning of cultural biases, and the continuous improvement of the model.

[0451] The server first collects visual data related to a specific culture from the internet and existing databases. This is done using web scraping techniques to retrieve images through APIs of public databases. The collected data is then stored in the database with metadata added.

[0452] The device generates images based on a dataset provided by the server, using a generative AI model. This involves using generative models such as GANs (Generative Adversarial Networks) or VQ-VAE, and providing the model with prompts such as "Please depict a traditional Japanese festival." The generated images then undergo a cultural bias detection process using a combination of natural language processing and image recognition techniques. If the degree of inappropriateness exceeds a certain threshold, the device issues a warning.

[0453] Users evaluate the generated images and provide feedback. This feedback includes specific comments and numerical ratings regarding image quality and cultural accuracy. This evaluation information is analyzed by the server, used to update the data set, and to retrain the generation AI model. This ensures that the generated images are closer to the expectations of customers and target audiences, and are more culturally appropriate.

[0454] As a concrete example, consider image generation for a traditional Japanese festival. In this case, the server aggregates various image data related to the festival, and the terminal generates images based on this data. Based on user feedback, the server optimizes the model and continuously improves it to provide even higher quality images in the next generation.

[0455] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0456] Step 1:

[0457] The server collects visual data from the internet and existing databases. During this collection process, it uses web scraping techniques and APIs to retrieve images, adds metadata, and stores them. The input is the URL of the source website or database, and the output is a dataset of images with metadata attached.

[0458] Step 2:

[0459] The terminal receives a dataset provided by the server and uses a generative AI model to generate images in response to input prompts. These prompts are given as specific instructions, such as "Please draw a scene from a traditional Japanese festival." The input consists of the prompt text and the dataset, and the output is the generated image.

[0460] Step 3:

[0461] The device applies an algorithm to detect cultural bias in the generated image. Using natural language processing and image recognition technologies, it determines whether elements within the image are culturally appropriate. The input is the generated image, and it outputs an evaluation of its degree of inappropriateness. If the degree of inappropriateness exceeds a set threshold, it issues a warning.

[0462] Step 4:

[0463] Users review the generated images and provide evaluation information. This includes not only numerical ratings but also text comments and suggestions for improvement. The input is the generated image and the user's perspective, and the output is the evaluation information.

[0464] Step 5:

[0465] The server analyzes user feedback and generates a new dataset. Based on this dataset, the generative AI model is retrained to improve its accuracy and cultural applicability. The input is evaluation information and existing datasets, and the output is the improved generative AI model.

[0466] (Application Example 1)

[0467] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0468] When generating content related to a specific culture, there is a risk of creating visual information that contains cultural bias or inaccurate representations. This invention aims to eliminate such offensive content and ensure that visual information has accurate and appropriate representations that reflect diverse cultural backgrounds. It also aims to improve the accuracy of the generation model by effectively utilizing user feedback.

[0469] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0470] In this invention, the server includes means for collecting data to generate visual information related to a specific culture, means for detecting cultural bias in the generated visual information, and means for collecting user feedback and enhancing model learning. This enables the generation of accurate and unbiased visual content related to a specific culture.

[0471] "Specific culture" refers to the unique traditions, customs, and values ​​shared within a particular region, ethnic group, religion, or social group.

[0472] "Visual information" refers to information that is perceived visually, such as images and videos.

[0473] "Cultural bias" refers to biased expressions that contain stereotypes or misleading content associated with a particular culture.

[0474] The "discomfort index" refers to an evaluation scale that quantifies the level of discomfort felt in generated visual information.

[0475] A "model" refers to an algorithm that generates visual information by learning from data.

[0476] "User evaluations" refer to information such as impressions, opinions, and feedback provided by users regarding the generated visual information.

[0477] As an embodiment of the present invention, a system is provided that generates visual information accurately based on a specific culture. This system consists of three main components: a server, a terminal, and a user.

[0478] First, the server collects visual data related to a specific culture through a large database. The collected data is constructed as a foundational dataset for generating visual information and used to train generative AI models. This data is then managed with appropriate metadata. The server can utilize image generation models such as OpenAI's DALL-E or Stable Diffusion.

[0479] Next, the device generates visual information based on a dataset provided by the server. A bias detection algorithm verifies whether the generated visual information contains cultural bias, and corrects it as needed. Analysis tools such as the Google Cloud Vision API can be used to check for bias.

[0480] Finally, users play a role in providing feedback on the generated visual information. This user feedback is used to retrain and improve the model. Users evaluate the generated visual information from diverse perspectives to ensure its cultural appropriateness. The training used in this feedback process includes educational content to support fair evaluation.

[0481] As a concrete example, consider a case where a user requests an image of a "Japanese spring festival." In this case, the server collects visual data related to Japanese spring festivals and uses it to generate an image on the terminal. The generated image is checked for cultural bias and, if appropriate, is provided to the user. After reviewing the image, the user provides feedback based on the prompt "Generate a vibrant image of a traditional Japanese spring festival including cherry blossoms and traditional costumes," and this data is used to further improve the system.

[0482] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0483] Step 1:

[0484] The server collects visual data related to a specific culture. The data is extracted from the internet and existing databases, and the collected visual data, along with metadata, is stored in the database. The input is raw image data from the internet, and the output is a dataset organized for training. The collected data undergoes data processing such as categorization and tagging before being used to train a generative AI model.

[0485] Step 2:

[0486] The terminal receives a dataset from the server and generates visual information based on user instructions. The input consists of the user's prompt text and associated dataset, while the output is the generated visual information. A generative AI model is used to output culturally appropriate images corresponding to the prompt text.

[0487] Step 3:

[0488] The device detects whether the generated visual information contains cultural bias. The input is the generated visual information, and the output is the bias detection result. The image is analyzed using the Google Cloud Vision API, and if bias is found, it is corrected.

[0489] Step 4:

[0490] The server sends the user the visual information after bias detection is complete. The input is the corrected visual information, and the output is the image presented to the user. The server verifies the corrected content and transfers it to the user's device.

[0491] Step 5:

[0492] Users provide feedback on the visual information they are given. The input is the presented visual information, and the output is the feedback information. They evaluate the information from various perspectives and send their opinions back to the server.

[0493] Step 6:

[0494] The server analyzes user feedback and retrains the generative AI model. The input is feedback data, and the output is the improved trained model. The feedback is then incorporated into new training data to improve the model's accuracy.

[0495] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0496] This invention combines a system that generates images related to a specific culture and detects and evaluates cultural bias in the generation process with an emotion engine that recognizes the user's emotions. This system consists of a server, a terminal, and a user interface equipped with the emotion engine.

[0497] Image generation and bias detection

[0498] The server collects data related to a specific culture, tags it, and adds metadata to it, then stores the resulting dataset in a database. This data provides a foundation for accurately representing that particular culture.

[0499] The device executes an image generation process based on the collected data, and generates images with cultural characteristics using an algorithm optimized based on specific conditions.

[0500] The device automatically analyzes the generated images to detect bias, thereby verifying their cultural validity. It also evaluates the discomfort index and displays a warning if it exceeds the set threshold.

[0501] User emotion recognition by an emotion engine

[0502] The user interacts with an interface to evaluate the provided images. During this process, an emotion engine recognizes the user's emotional state in real time and records their emotional response to the images.

[0503] The emotion engine collects recognized user emotion data and sends it to the server as part of the feedback based on the emotional review of the image. This information is used to evaluate the image's discomfort index and improve the model.

[0504] Model update and improvement process

[0505] The server analyzes the sentiment data and feedback obtained from users and generates a new dataset based on this information.

[0506] The server retrains its model by incorporating feedback, including sentiment data, with the aim of more accurately reflecting specific cultures. This process improves the quality and cultural appropriateness of the generated images.

[0507] As a concrete example, consider the case of image generation themed around the Japanese tea ceremony. The server collects high-precision image data related to the tea ceremony, and the terminal uses this data to generate images that recreate the tea ceremony ritual. The user evaluates the generated images through an emotional interface, and the emotional responses are also used in the analysis. This allows the improved model to provide more culturally accurate and user-friendly results in future image generation.

[0508] The following describes the processing flow.

[0509] Step 1:

[0510] The server collects data related to a specific culture from the internet and dedicated databases. This data includes information such as the history, tools, and procedures of the tea ceremony, and is tagged with detailed metadata. The server creates a dataset based on this data and stores it in the database as the basis for image generation.

[0511] Step 2:

[0512] The device generates images themed around the Japanese tea ceremony. Using a dataset supplied from the server, it generates images based on an algorithm that faithfully reproduces specific cultural elements. In this process, the generation algorithm aims to create culturally accurate images by making maximum use of information about the culture.

[0513] Step 3:

[0514] The device sends the generated image to a bias detection module. Here, it automatically analyzes for cultural bias and inaccuracies and calculates an discomfort index. If the discomfort index exceeds a set threshold, the device displays a warning and requests image adjustments.

[0515] Step 4:

[0516] The user evaluates the generated image using their device. During this process, the emotion engine analyzes the user's emotions in real time and records their emotional response to the image as emotion data. The user then provides specific feedback based on this evaluation.

[0517] Step 5:

[0518] The emotion engine uses the acquired emotion data to send it to the server along with user feedback. The emotion data is used as supplementary information for evaluating the discomfort index of images.

[0519] Step 6:

[0520] The server collects and analyzes user feedback and sentiment data. Based on this, a new training dataset is created, and the model is retrained to improve the accuracy and cultural relevance of the generated images.

[0521] Step 7:

[0522] The server applies the retrained model to the system. This updated model enables future image generation to be more accurate and culturally appropriate. This process is continuous, enhancing the overall system performance and ethical aspects.

[0523] (Example 2)

[0524] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0525] In generating images that reflect specific cultures, the evaluation of cultural bias and discomfort indices is insufficient, potentially leading to unpleasant experiences for users. Furthermore, the lack of mechanisms to effectively utilize user feedback and sentiment data to improve the model makes it difficult to improve the quality and cultural appropriateness of generated images.

[0526] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0527] In this invention, the server includes means for automatically collecting and organizing information related to a specific culture, means for analyzing whether the generated images are free from cultural bias, and means for collecting emotional data from users and generating feedback based on that data. This makes it possible to improve image generation that accurately reflects the characteristics of a specific culture while responding to the user's emotions.

[0528] "Specific culture" refers to the characteristics of a culture that include customs, values, and aesthetics unique to a particular region or society.

[0529] "Information" refers to materials such as text, images, audio, and video that are stored and processed as digital data.

[0530] A "prompt statement" is an instruction given to a generative AI model, providing guidelines for generating images with a specific theme or style.

[0531] "Image generation" refers to the process of creating visual representations from text and data using computer algorithms.

[0532] "Cultural bias" refers to the phenomenon where prejudices and stereotypes about a particular culture or society are reflected in digital content.

[0533] The "Discomfort Index" refers to a criterion for evaluating digital content that may cause stress or discomfort to users.

[0534] "Emotional data" refers to information that indicates a user's emotional state, including data such as facial expressions, tone of voice, and other biometric information.

[0535] "Feedback" refers to responses and opinions generated based on user evaluations, which are used to improve systems and processes.

[0536] "Retraining" refers to the process of training an existing machine learning model again using a new dataset, which improves its accuracy and performance.

[0537] The embodiments for carrying out the invention are described below.

[0538] This invention provides a system that generates images reflecting a specific culture, detects and evaluates cultural bias, and incorporates user emotional data as feedback. Specifically, it performs the following processing using a server, terminal, and a user interface equipped with an emotion engine.

[0539] Server Processing

[0540] The server automatically collects information related to a specific culture from the internet and existing datasets. This information includes images, text, audio, and video. This information is tagged and stored in a database. This database provides the foundation necessary for generating images that accurately reflect the characteristics of that particular culture.

[0541] Terminal processing

[0542] The terminal generates images using a generative AI model based on information obtained from the server. Specifically, it takes a prompt message as input and executes the image generation algorithm. For example, using the prompt "Generate an image that expresses the spirit of the Japanese tea ceremony," it will generate an image related to the Japanese tea ceremony. The generated image is then analyzed to detect cultural bias.

[0543] Emotion engine and user processing

[0544] Users evaluate images generated using a user interface that includes an emotion engine. During this process, emotional data is collected from the user's facial expressions, tone of voice, and other factors. The emotion engine analyzes this data and generates feedback based on the user's emotions.

[0545] This feedback is sent to the server and used to retrain the model. Retraining makes the generated images more culturally appropriate and appealing to users.

[0546] The implementation of this system will enable continuous improvement in image generation that accurately reflects specific cultures, thereby enhancing the user experience.

[0547] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0548] Step 1:

[0549] The server automatically collects information related to a specific culture from the internet and existing datasets. Specifically, the server uses a crawling algorithm to extract images, text, and other data based on specific keywords and tags. The input to this process is keywords related to the specific culture, and the output is a dataset of tagged information. The server automatically tags the collected information and stores it in a database.

[0550] Step 2:

[0551] The device generates an image by inputting a prompt message into a generation AI model based on a dataset obtained from the server. An example of a prompt message might be, "Please generate an image that expresses the spirit of the Japanese tea ceremony." Based on this input, the device runs the generation AI model and generates an image that possesses the characteristics of a specific culture. The output is an image related to the target culture.

[0552] Step 3:

[0553] The device analyzes the generated image for cultural bias. Specifically, it uses an image analysis algorithm to evaluate the image and calculate an discomfort index. The input for this step is the generated image, and the output is the bias analysis result and the discomfort index evaluation. If the discomfort index exceeds a set threshold, the device displays a warning message to the user.

[0554] Step 4:

[0555] The user reviews the images generated through the interface. During the interaction, the emotion engine collects the user's emotional data. Specific actions include detecting the user's facial expressions and voice tone, and receiving feedback in the form of a questionnaire. The input for this step is the user's emotional expressions and feedback, and the output is stored in the system as emotional data.

[0556] Step 5:

[0557] The server analyzes the collected user sentiment data and generates new feedback. This data is then used to retrain the model to improve system performance. Specifically, this involves creating a new dataset that reflects the sentiment data and feedback, and retraining the machine learning model. The input for this step is the sentiment data and feedback obtained from the user, and the output is the retuned generative AI model.

[0558] (Application Example 2)

[0559] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0560] In creating content related to a specific culture, there is a risk of inaccurate cultural representation due to cultural bias, and a problem exists where content that is not optimized for individual users is provided due to a failure to consider users' emotional responses, resulting in decreased user satisfaction. It is necessary to resolve these issues and provide users with culturally accurate and personalized visual experiences.

[0561] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0562] In this invention, the server includes means for acquiring information related to a specific culture, means for detecting cultural bias in the generated visual data, and means for recognizing the user's emotional state and recording their emotional response. This makes it possible to generate and provide culturally accurate and personalized content to the user.

[0563] A "specific culture" refers to a cultural system that includes customs, traditions, values, or artistic expressions unique to a particular region, ethnic group, or country.

[0564] An "image" is a representation of visual information as digital data, displayed on the screen of a computer or electronic device.

[0565] "Bias" refers to any inclination or preconception that influences data or results, whether intentionally or unintentionally.

[0566] The "Discomfort Index" is a numerical indicator that quantifies the degree of discomfort or inappropriateness experienced by users.

[0567] An "algorithmic model" refers to a mathematical or computational method designed to process data and produce an output for a specific purpose.

[0568] "Emotional state" refers to the user's psychological response, including emotions such as happiness, anger, and sadness.

[0569] "Emotional response" refers to the emotional changes a person exhibits in response to a specific stimulus.

[0570] "Content personalization" refers to customizing the information and services provided based on the individual user's preferences and needs.

[0571] In embodiments of this invention, the system includes a server, a terminal, and a user interface. The server acquires a wide range of information related to a specific culture and stores this information in an organized database. The data used serves as a foundation for accurately reflecting the specified culture. The collected data is tagged and metadata is added to enable efficient searching and access.

[0572] The device receives data from the server and generates culturally relevant visual data using a generative AI model. During the generation process, the algorithmic model creates images with culturally appropriate features. Simultaneously, the device detects cultural bias in the generated visual data and evaluates whether there are any elements that might be offensive to the user. If the discomfort index exceeds a certain threshold, the device displays a warning.

[0573] Users evaluate images and visual content through the provided interface. The device uses an emotion engine to recognize the user's emotional state in real time and record their emotional responses. This emotion data is sent to a server, where the content is personalized based on the emotional responses and reflected in future image generation. This ensures that culturally precise and engaging content tailored to each individual user is provided.

[0574] For example, when a user selects "French culture" for the system, the server uses data including historical buildings, art, and food culture related to France, and the device generates visual content about France that the user might find interesting. The user views and evaluates the images, and the content generated next time is improved based on their more personal experience.

[0575] An example of a prompt for a generative AI model is, "Generate an image that evokes the atmosphere of France, using historical French buildings and culinary styles." In this way, the system continuously learns and improves its accuracy, resulting in richer cultural expression.

[0576] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0577] Step 1:

[0578] The server collects information related to a specific culture from the internet and existing databases. This information includes various formats such as text, images, and audio. The input is a prompt about the culture to be collected. The server organizes this information, adds metadata, and stores it. The output is a well-organized dataset associated with the specific culture.

[0579] Step 2:

[0580] The terminal receives a dataset sent from the server and generates visual data using a generative AI model based on a specific prompt (for example, "Generate an image that evokes the atmosphere of France using historical French buildings and culinary styles"). The prompt becomes the input, the generative AI model analyzes the data, and outputs an image with cultural characteristics. In operation, the AI ​​model learns patterns in the data and creates images.

[0581] Step 3:

[0582] The device automatically analyzes the generated visual data to detect whether it contains cultural bias. The generated image is used as input, and its cultural features are analyzed by a bias detection algorithm. The output provides the presence or absence of bias and an evaluation of the discomfort index. If the bias is high, a warning is displayed.

[0583] Step 4:

[0584] The user uses an interface to evaluate images and visual content provided on the device. During this process, an emotion engine recognizes the user's facial expressions and reactions in real time and records their emotional state. The input is the user's visual reactions, and the output is emotion data. The process involves a camera capturing facial expressions, and the emotion engine analyzing that data.

[0585] Step 5:

[0586] The server receives user emotional responses and traditional feedback data, and generates a new dataset based on this. The input is emotional data from the user, and the output is a new training dataset. This dataset is used in subsequent generation processes, enabling the provision of more personalized visual data. The server updates the database to ensure the entire system evolves continuously.

[0587] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0588] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0589] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[0590] [Fourth Embodiment]

[0591] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0592] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0593] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0594] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[0595] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0596] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0597] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0598] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[0599] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0600] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0601] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0602] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0603] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0604] One embodiment of the present invention is to provide an image generation system that accurately reflects a specific culture. The entire system consists of an image generation server, a terminal capable of bias detection, and a user that provides evaluation and feedback.

[0605] Image generation and bias detection

[0606] The server collects large amounts of image data related to a specific culture from the internet and existing databases. This data is stored in the database as a foundational dataset for learning that reflects the target culture. The collected data is tagged with metadata.

[0607] The device generates images using datasets provided by the server, based on specific conditions and themes. This image generation process is optimized to ensure appropriate representation, taking into account how the generated images will be received in a particular culture.

[0608] The device performs cultural bias detection on the generated images. This is a process that analyzes whether the images contain cultural inaccuracies or negative stereotypes and evaluates the discomfort index. The detection algorithm operates based on specific bias detection criteria and issues a warning for images with a discomfort index that exceeds the criteria.

[0609] Feedback and model improvement

[0610] Users receive prior training to minimize personal bias and provide appropriate feedback on the generated images. The training is designed to enable evaluation from diverse perspectives.

[0611] The server analyzes feedback collected from users and generates a new training dataset. This allows the model to be retrained, improving its accuracy so that it can generate images that more accurately reflect specific cultures.

[0612] As a concrete example, consider the case of generating images related to traditional Japanese festivals. In this scenario, the server collects diverse images related to festivals and builds a dataset. The terminal uses this dataset to generate festival-themed images, detecting and correcting cultural biases in advance. The user evaluates the generated images and provides feedback, aiming to generate more accurate images of Japanese culture.

[0613] The following describes the processing flow.

[0614] Step 1:

[0615] The server collects a vast amount of image data related to a specific culture from the internet and existing databases. These images are tagged with metadata and are appropriately culturally tagged. This constructs the dataset used for generation.

[0616] Step 2:

[0617] The device generates images using collected datasets based on specific conditions or themes. The generated images are optimized based on algorithms designed to faithfully reproduce the characteristics of a particular culture.

[0618] Step 3:

[0619] The terminal sends the generated image to a bias detection module, which automatically analyzes whether the image contains bias or inaccuracies regarding a particular culture or ethnicity.

[0620] Step 4:

[0621] The device evaluates the discomfort index of an image and displays a warning if the index is high, according to the set criteria. This process can prevent inaccurate representations of specific cultures.

[0622] Step 5:

[0623] Users complete pre-training and provide expert feedback on the generated images. By evaluating images from diverse perspectives and providing feedback, users can mitigate personal bias.

[0624] Step 6:

[0625] The server collects and analyzes the feedback received from users. Based on the results of this analysis, a new dataset is generated, and the model is retrained.

[0626] Step 7:

[0627] The server incorporates the results of retraining, enabling more accurate image generation. The goal is to generate unbiased images that accurately reflect a specific culture.

[0628] (Example 1)

[0629] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0630] When generating images that accurately reflect a specific culture, there is a risk of including cultural biases or inappropriate representations. This can lead to misunderstandings and cultural friction, highlighting the need to improve the quality and cultural appropriateness of the generated images.

[0631] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0632] In this invention, the server includes means for extracting visual data related to a specific culture, means for creating images in response to input instructions using a generative AI model, and means for detecting cultural bias within the created images. This enables the generation of high-quality images that are sensitive to culture.

[0633] "Visual data" refers to data that includes visual information such as images and videos, and is used as an element that reflects a particular culture.

[0634] A "generative AI model" is an algorithm that uses artificial intelligence to create new images and designs, and is a technology that generates images based on specific conditions.

[0635] "Input instructions" refer to instructions that explicitly state the request for a specific image generation, and mean prompts or commands given to the generation AI model.

[0636] "Cultural bias" refers to a state in which misunderstandings or negative stereotypes about a particular culture are included in expressions, and biased views are a factor that hinders a correct understanding of a culture.

[0637] "Inappropriateness" is an index that evaluates how culturally appropriate a generated image is, and indicates that corrections are necessary if the image does not meet the standard.

[0638] A "warning" is a notification issued when the level of cultural bias or inappropriateness within an image exceeds a certain standard, and it is an important sign that action should be taken.

[0639] "Evaluation information" refers to the feedback and evaluation data provided for the generated images, and is used to improve the model.

[0640] A "dataset" is a collection of data used for training and retraining, and forms the basis for generating new images.

[0641] This invention provides an image generation system that accurately reflects a specific culture. The system includes the management of visual data, the creation of images using a generative AI model, the detection and warning of cultural biases, and the continuous improvement of the model.

[0642] The server first collects visual data related to a specific culture from the internet and existing databases. This is done using web scraping techniques to retrieve images through APIs of public databases. The collected data is then stored in the database with metadata added.

[0643] The device generates images based on a dataset provided by the server, using a generative AI model. This involves using generative models such as GANs (Generative Adversarial Networks) or VQ-VAE, and providing the model with prompts such as "Please depict a traditional Japanese festival." The generated images then undergo a cultural bias detection process using a combination of natural language processing and image recognition techniques. If the degree of inappropriateness exceeds a certain threshold, the device issues a warning.

[0644] Users evaluate the generated images and provide feedback. This feedback includes specific comments and numerical ratings regarding image quality and cultural accuracy. This evaluation information is analyzed by the server, used to update the data set, and to retrain the generation AI model. This ensures that the generated images are closer to the expectations of customers and target audiences, and are more culturally appropriate.

[0645] As a concrete example, consider image generation for a traditional Japanese festival. In this case, the server aggregates various image data related to the festival, and the terminal generates images based on this data. Based on user feedback, the server optimizes the model and continuously improves it to provide even higher quality images in the next generation.

[0646] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0647] Step 1:

[0648] The server collects visual data from the internet and existing databases. During this collection process, it uses web scraping techniques and APIs to retrieve images, adds metadata, and stores them. The input is the URL of the source website or database, and the output is a dataset of images with metadata attached.

[0649] Step 2:

[0650] The terminal receives a dataset provided by the server and uses a generative AI model to generate images in response to input prompts. These prompts are given as specific instructions, such as "Please draw a scene from a traditional Japanese festival." The input consists of the prompt text and the dataset, and the output is the generated image.

[0651] Step 3:

[0652] The device applies an algorithm to detect cultural bias in the generated image. Using natural language processing and image recognition technologies, it determines whether elements within the image are culturally appropriate. The input is the generated image, and it outputs an evaluation of its degree of inappropriateness. If the degree of inappropriateness exceeds a set threshold, it issues a warning.

[0653] Step 4:

[0654] Users review the generated images and provide evaluation information. This includes not only numerical ratings but also text comments and suggestions for improvement. The input is the generated image and the user's perspective, and the output is the evaluation information.

[0655] Step 5:

[0656] The server analyzes user feedback and generates a new dataset. Based on this dataset, the generative AI model is retrained to improve its accuracy and cultural applicability. The input is evaluation information and existing datasets, and the output is the improved generative AI model.

[0657] (Application Example 1)

[0658] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0659] When generating content related to a specific culture, there is a risk of creating visual information that contains cultural bias or inaccurate representations. This invention aims to eliminate such offensive content and ensure that visual information has accurate and appropriate representations that reflect diverse cultural backgrounds. It also aims to improve the accuracy of the generation model by effectively utilizing user feedback.

[0660] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0661] In this invention, the server includes means for collecting data to generate visual information related to a specific culture, means for detecting cultural bias in the generated visual information, and means for collecting user feedback and enhancing model learning. This enables the generation of accurate and unbiased visual content related to a specific culture.

[0662] "Specific culture" refers to the unique traditions, customs, and values ​​shared within a particular region, ethnic group, religion, or social group.

[0663] "Visual information" refers to information that is perceived visually, such as images and videos.

[0664] "Cultural bias" refers to biased expressions that contain stereotypes or misleading content associated with a particular culture.

[0665] The "discomfort index" refers to an evaluation scale that quantifies the level of discomfort felt in generated visual information.

[0666] A "model" refers to an algorithm that generates visual information by learning from data.

[0667] "User evaluations" refer to information such as impressions, opinions, and feedback provided by users regarding the generated visual information.

[0668] As an embodiment of the present invention, a system is provided that generates visual information accurately based on a specific culture. This system consists of three main components: a server, a terminal, and a user.

[0669] First, the server collects visual data related to a specific culture through a large database. The collected data is constructed as a foundational dataset for generating visual information and used to train generative AI models. This data is then managed with appropriate metadata. The server can utilize image generation models such as OpenAI's DALL-E or Stable Diffusion.

[0670] Next, the device generates visual information based on a dataset provided by the server. A bias detection algorithm verifies whether the generated visual information contains cultural bias, and corrects it as needed. Analysis tools such as the Google Cloud Vision API can be used to check for bias.

[0671] Finally, users play a role in providing feedback on the generated visual information. This user feedback is used to retrain and improve the model. Users evaluate the generated visual information from diverse perspectives to ensure its cultural appropriateness. The training used in this feedback process includes educational content to support fair evaluation.

[0672] As a concrete example, consider a case where a user requests an image of a "Japanese spring festival." In this case, the server collects visual data related to Japanese spring festivals and uses it to generate an image on the terminal. The generated image is checked for cultural bias and, if appropriate, is provided to the user. After reviewing the image, the user provides feedback based on the prompt "Generate a vibrant image of a traditional Japanese spring festival including cherry blossoms and traditional costumes," and this data is used to further improve the system.

[0673] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0674] Step 1:

[0675] The server collects visual data related to a specific culture. The data is extracted from the internet and existing databases, and the collected visual data, along with metadata, is stored in the database. The input is raw image data from the internet, and the output is a dataset organized for training. The collected data undergoes data processing such as categorization and tagging before being used to train a generative AI model.

[0676] Step 2:

[0677] The terminal receives a dataset from the server and generates visual information based on user instructions. The input consists of the user's prompt text and associated dataset, while the output is the generated visual information. A generative AI model is used to output culturally appropriate images corresponding to the prompt text.

[0678] Step 3:

[0679] The device detects whether the generated visual information contains cultural bias. The input is the generated visual information, and the output is the bias detection result. The image is analyzed using the Google Cloud Vision API, and if bias is found, it is corrected.

[0680] Step 4:

[0681] The server sends the user the visual information after bias detection is complete. The input is the corrected visual information, and the output is the image presented to the user. The server verifies the corrected content and transfers it to the user's device.

[0682] Step 5:

[0683] Users provide feedback on the visual information they are given. The input is the presented visual information, and the output is the feedback information. They evaluate the information from various perspectives and send their opinions back to the server.

[0684] Step 6:

[0685] The server analyzes user feedback and retrains the generative AI model. The input is feedback data, and the output is the improved trained model. The feedback is then incorporated into new training data to improve the model's accuracy.

[0686] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0687] This invention combines a system that generates images related to a specific culture and detects and evaluates cultural bias in the generation process with an emotion engine that recognizes the user's emotions. This system consists of a server, a terminal, and a user interface equipped with the emotion engine.

[0688] Image generation and bias detection

[0689] The server collects data related to a specific culture, tags it, and adds metadata to it, then stores the resulting dataset in a database. This data provides a foundation for accurately representing that particular culture.

[0690] The device executes an image generation process based on the collected data, and generates images with cultural characteristics using an algorithm optimized based on specific conditions.

[0691] The device automatically analyzes the generated images to detect bias, thereby verifying their cultural validity. It also evaluates the discomfort index and displays a warning if it exceeds the set threshold.

[0692] User emotion recognition by an emotion engine

[0693] The user interacts with an interface to evaluate the provided images. During this process, an emotion engine recognizes the user's emotional state in real time and records their emotional response to the images.

[0694] The emotion engine collects recognized user emotion data and sends it to the server as part of the feedback based on the emotional review of the image. This information is used to evaluate the image's discomfort index and improve the model.

[0695] Model update and improvement process

[0696] The server analyzes the sentiment data and feedback obtained from users and generates a new dataset based on this information.

[0697] The server retrains its model by incorporating feedback, including sentiment data, with the aim of more accurately reflecting specific cultures. This process improves the quality and cultural appropriateness of the generated images.

[0698] As a concrete example, consider the case of image generation themed around the Japanese tea ceremony. The server collects high-precision image data related to the tea ceremony, and the terminal uses this data to generate images that recreate the tea ceremony ritual. The user evaluates the generated images through an emotional interface, and the emotional responses are also used in the analysis. This allows the improved model to provide more culturally accurate and user-friendly results in future image generation.

[0699] The following describes the processing flow.

[0700] Step 1:

[0701] The server collects data related to a specific culture from the internet and dedicated databases. This data includes information such as the history, tools, and procedures of the tea ceremony, and is tagged with detailed metadata. The server creates a dataset based on this data and stores it in the database as the basis for image generation.

[0702] Step 2:

[0703] The device generates images themed around the Japanese tea ceremony. Using a dataset supplied from the server, it generates images based on an algorithm that faithfully reproduces specific cultural elements. In this process, the generation algorithm aims to create culturally accurate images by making maximum use of information about the culture.

[0704] Step 3:

[0705] The device sends the generated image to a bias detection module. Here, it automatically analyzes for cultural bias and inaccuracies and calculates an discomfort index. If the discomfort index exceeds a set threshold, the device displays a warning and requests image adjustments.

[0706] Step 4:

[0707] The user evaluates the generated image using their device. During this process, the emotion engine analyzes the user's emotions in real time and records their emotional response to the image as emotion data. The user then provides specific feedback based on this evaluation.

[0708] Step 5:

[0709] The emotion engine uses the acquired emotion data to send it to the server along with user feedback. The emotion data is used as supplementary information for evaluating the discomfort index of images.

[0710] Step 6:

[0711] The server collects and analyzes user feedback and sentiment data. Based on this, a new training dataset is created, and the model is retrained to improve the accuracy and cultural relevance of the generated images.

[0712] Step 7:

[0713] The server applies the retrained model to the system. This updated model enables future image generation to be more accurate and culturally appropriate. This process is continuous, enhancing the overall system performance and ethical aspects.

[0714] (Example 2)

[0715] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0716] In generating images that reflect specific cultures, the evaluation of cultural bias and discomfort indices is insufficient, potentially leading to unpleasant experiences for users. Furthermore, the lack of mechanisms to effectively utilize user feedback and sentiment data to improve the model makes it difficult to improve the quality and cultural appropriateness of generated images.

[0717] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0718] In this invention, the server includes means for automatically collecting and organizing information related to a specific culture, means for analyzing whether the generated images are free from cultural bias, and means for collecting emotional data from users and generating feedback based on that data. This makes it possible to improve image generation that accurately reflects the characteristics of a specific culture while responding to the user's emotions.

[0719] "Specific culture" refers to the characteristics of a culture that include customs, values, and aesthetics unique to a particular region or society.

[0720] "Information" refers to materials such as text, images, audio, and video that are stored and processed as digital data.

[0721] A "prompt statement" is an instruction given to a generative AI model, providing guidelines for generating images with a specific theme or style.

[0722] "Image generation" refers to the process of creating visual representations from text and data using computer algorithms.

[0723] "Cultural bias" refers to the phenomenon where prejudices and stereotypes about a particular culture or society are reflected in digital content.

[0724] The "Discomfort Index" refers to a criterion for evaluating digital content that may cause stress or discomfort to users.

[0725] "Emotional data" refers to information that indicates a user's emotional state, including data such as facial expressions, tone of voice, and other biometric information.

[0726] "Feedback" refers to responses and opinions generated based on user evaluations, which are used to improve systems and processes.

[0727] "Retraining" refers to the process of training an existing machine learning model again using a new dataset, which improves its accuracy and performance.

[0728] The embodiments for carrying out the invention are described below.

[0729] This invention provides a system that generates images reflecting a specific culture, detects and evaluates cultural bias, and incorporates user emotional data as feedback. Specifically, it performs the following processing using a server, terminal, and a user interface equipped with an emotion engine.

[0730] Server Processing

[0731] The server automatically collects information related to a specific culture from the internet and existing datasets. This information includes images, text, audio, and video. This information is tagged and stored in a database. This database provides the foundation necessary for generating images that accurately reflect the characteristics of that particular culture.

[0732] Terminal processing

[0733] The terminal generates images using a generative AI model based on information obtained from the server. Specifically, it takes a prompt message as input and executes the image generation algorithm. For example, using the prompt "Generate an image that expresses the spirit of the Japanese tea ceremony," it will generate an image related to the Japanese tea ceremony. The generated image is then analyzed to detect cultural bias.

[0734] Emotion engine and user processing

[0735] Users evaluate images generated using a user interface that includes an emotion engine. During this process, emotional data is collected from the user's facial expressions, tone of voice, and other factors. The emotion engine analyzes this data and generates feedback based on the user's emotions.

[0736] This feedback is sent to the server and used to retrain the model. Retraining makes the generated images more culturally appropriate and appealing to users.

[0737] The implementation of this system will enable continuous improvement in image generation that accurately reflects specific cultures, thereby enhancing the user experience.

[0738] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0739] Step 1:

[0740] The server automatically collects information related to a specific culture from the internet and existing datasets. Specifically, the server uses a crawling algorithm to extract images, text, and other data based on specific keywords and tags. The input to this process is keywords related to the specific culture, and the output is a dataset of tagged information. The server automatically tags the collected information and stores it in a database.

[0741] Step 2:

[0742] The device generates an image by inputting a prompt message into a generation AI model based on a dataset obtained from the server. An example of a prompt message might be, "Please generate an image that expresses the spirit of the Japanese tea ceremony." Based on this input, the device runs the generation AI model and generates an image that possesses the characteristics of a specific culture. The output is an image related to the target culture.

[0743] Step 3:

[0744] The device analyzes the generated image for cultural bias. Specifically, it uses an image analysis algorithm to evaluate the image and calculate an discomfort index. The input for this step is the generated image, and the output is the bias analysis result and the discomfort index evaluation. If the discomfort index exceeds a set threshold, the device displays a warning message to the user.

[0745] Step 4:

[0746] The user reviews the images generated through the interface. During the interaction, the emotion engine collects the user's emotional data. Specific actions include detecting the user's facial expressions and voice tone, and receiving feedback in the form of a questionnaire. The input for this step is the user's emotional expressions and feedback, and the output is stored in the system as emotional data.

[0747] Step 5:

[0748] The server analyzes the collected user sentiment data and generates new feedback. This data is then used to retrain the model to improve system performance. Specifically, this involves creating a new dataset that reflects the sentiment data and feedback, and retraining the machine learning model. The input for this step is the sentiment data and feedback obtained from the user, and the output is the retuned generative AI model.

[0749] (Application Example 2)

[0750] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0751] In creating content related to a specific culture, there is a risk of inaccurate cultural representation due to cultural bias, and a problem exists where content that is not optimized for individual users is provided due to a failure to consider users' emotional responses, resulting in decreased user satisfaction. It is necessary to resolve these issues and provide users with culturally accurate and personalized visual experiences.

[0752] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0753] In this invention, the server includes means for acquiring information related to a specific culture, means for detecting cultural bias in the generated visual data, and means for recognizing the user's emotional state and recording their emotional response. This makes it possible to generate and provide culturally accurate and personalized content to the user.

[0754] A "specific culture" refers to a cultural system that includes customs, traditions, values, or artistic expressions unique to a particular region, ethnic group, or country.

[0755] An "image" is a representation of visual information as digital data, displayed on the screen of a computer or electronic device.

[0756] "Bias" refers to any inclination or preconception that influences data or results, whether intentionally or unintentionally.

[0757] The "Discomfort Index" is a numerical indicator that quantifies the degree of discomfort or inappropriateness experienced by users.

[0758] An "algorithmic model" refers to a mathematical or computational method designed to process data and produce an output for a specific purpose.

[0759] "Emotional state" refers to the user's psychological response, including emotions such as happiness, anger, and sadness.

[0760] "Emotional response" refers to the emotional changes a person exhibits in response to a specific stimulus.

[0761] "Content personalization" refers to customizing the information and services provided based on the individual user's preferences and needs.

[0762] In embodiments of this invention, the system includes a server, a terminal, and a user interface. The server acquires a wide range of information related to a specific culture and stores this information in an organized database. The data used serves as a foundation for accurately reflecting the specified culture. The collected data is tagged and metadata is added to enable efficient searching and access.

[0763] The device receives data from the server and generates culturally relevant visual data using a generative AI model. During the generation process, the algorithmic model creates images with culturally appropriate features. Simultaneously, the device detects cultural bias in the generated visual data and evaluates whether there are any elements that might be offensive to the user. If the discomfort index exceeds a certain threshold, the device displays a warning.

[0764] Users evaluate images and visual content through the provided interface. The device uses an emotion engine to recognize the user's emotional state in real time and record their emotional responses. This emotion data is sent to a server, where the content is personalized based on the emotional responses and reflected in future image generation. This ensures that culturally precise and engaging content tailored to each individual user is provided.

[0765] For example, when a user selects "French culture" for the system, the server uses data including historical buildings, art, and food culture related to France, and the device generates visual content about France that the user might find interesting. The user views and evaluates the images, and the content generated next time is improved based on their more personal experience.

[0766] An example of a prompt for a generative AI model is, "Generate an image that evokes the atmosphere of France, using historical French buildings and culinary styles." In this way, the system continuously learns and improves its accuracy, resulting in richer cultural expression.

[0767] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0768] Step 1:

[0769] The server collects information related to a specific culture from the internet and existing databases. This information includes various formats such as text, images, and audio. The input is a prompt about the culture to be collected. The server organizes this information, adds metadata, and stores it. The output is a well-organized dataset associated with the specific culture.

[0770] Step 2:

[0771] The terminal receives a dataset sent from the server and generates visual data using a generative AI model based on a specific prompt (for example, "Generate an image that evokes the atmosphere of France using historical French buildings and culinary styles"). The prompt becomes the input, the generative AI model analyzes the data, and outputs an image with cultural characteristics. In operation, the AI ​​model learns patterns in the data and creates images.

[0772] Step 3:

[0773] The device automatically analyzes the generated visual data to detect whether it contains cultural bias. The generated image is used as input, and its cultural features are analyzed by a bias detection algorithm. The output provides the presence or absence of bias and an evaluation of the discomfort index. If the bias is high, a warning is displayed.

[0774] Step 4:

[0775] The user uses an interface to evaluate images and visual content provided on the device. During this process, an emotion engine recognizes the user's facial expressions and reactions in real time and records their emotional state. The input is the user's visual reactions, and the output is emotion data. The process involves a camera capturing facial expressions, and the emotion engine analyzing that data.

[0776] Step 5:

[0777] The server receives user emotional responses and traditional feedback data, and generates a new dataset based on this. The input is emotional data from the user, and the output is a new training dataset. This dataset is used in subsequent generation processes, enabling the provision of more personalized visual data. The server updates the database to ensure the entire system evolves continuously.

[0778] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0779] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0780] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[0781] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[0782] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[0783] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[0784] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[0785] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[0786] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[0787] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[0788] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[0789] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[0790] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[0791] 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.

[0792] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[0793] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[0794] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[0795] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[0796] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[0797] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[0798] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[0799] The following is further disclosed regarding the embodiments described above.

[0800] (Claim 1)

[0801] A means of collecting data to generate images related to a specific culture,

[0802] A means for detecting cultural bias in generated images,

[0803] A means for evaluating the discomfort index of an image based on detected bias,

[0804] A means of updating the model using the evaluation results,

[0805] A means of improving image representation using an updated model,

[0806] A system that includes this.

[0807] (Claim 2)

[0808] The system according to claim 1, characterized in that it issues a warning when the image discomfort index exceeds a certain level.

[0809] (Claim 3)

[0810] The system according to claim 1, characterized by generating a new dataset based on the received feedback and performing retraining.

[0811] "Example 1"

[0812] (Claim 1)

[0813] A means of extracting visual data related to a specific culture,

[0814] A means of creating images in response to input instructions using a generative AI model,

[0815] A means for detecting cultural bias within the created image,

[0816] A means of measuring the inappropriateness of an image based on cultural bias,

[0817] A means of issuing warnings based on the evaluation results of the degree of inappropriateness,

[0818] A means of analyzing the received evaluation information and updating the data set,

[0819] A means of improving image quality based on a retrained generative model,

[0820] A system that includes this.

[0821] (Claim 2)

[0822] The system according to claim 1, characterized in that it issues a warning when the degree of inappropriateness of an image exceeds a certain threshold.

[0823] (Claim 3)

[0824] The system according to claim 1, characterized by creating a new set of data based on feedback obtained from users and performing retraining.

[0825] "Application Example 1"

[0826] (Claim 1)

[0827] A means of collecting data to generate visual information related to a specific culture,

[0828] A means for detecting cultural bias in generated visual information,

[0829] A means for evaluating the discomfort index of visual information based on detected bias,

[0830] A means of updating the model using the evaluation results,

[0831] A means of improving the representation of visual information using an updated model,

[0832] A means of collecting user feedback and strengthening model learning,

[0833] A system that includes this.

[0834] (Claim 2)

[0835] The system according to claim 1, characterized in that it issues a warning when the visual information discomfort index exceeds a certain level.

[0836] (Claim 3)

[0837] The system according to claim 1, characterized by generating a new set of data based on evaluations received from users and performing retraining.

[0838] "Example 2 of combining an emotion engine"

[0839] (Claim 1)

[0840] A means of automatically collecting and organizing information related to a specific culture,

[0841] A means of generating images using prompt statements,

[0842] A means of analyzing whether the generated images are free from cultural bias,

[0843] A means for evaluating the discomfort index of an image based on the analysis results,

[0844] A means of collecting emotional data from users and generating feedback based on that data,

[0845] A means to retrain the model based on the generated feedback and improve image generation,

[0846] A system that includes this.

[0847] (Claim 2)

[0848] The system according to claim 1, characterized in that it issues a warning when the discomfort index of an image exceeds a set standard.

[0849] (Claim 3)

[0850] The system according to claim 1, characterized in that it organizes new information based on received emotion data and performs retraining.

[0851] "Application example 2 when combining with an emotional engine"

[0852] (Claim 1)

[0853] A means of obtaining information for generating images related to a specific culture,

[0854] A means for detecting cultural bias in generated visual data,

[0855] A means for evaluating the discomfort index of an image based on detected bias,

[0856] A means of updating the algorithm model using the evaluation results,

[0857] A means of improving image representation using an updated algorithm model,

[0858] A means of recognizing the user's emotional state and recording their emotional response,

[0859] A means for personalizing content based on the recorded emotional responses,

[0860] A system that includes this.

[0861] (Claim 2)

[0862] The system according to claim 1, characterized in that it displays a warning when the image discomfort index exceeds a certain level and records the user's emotional response.

[0863] (Claim 3)

[0864] The system according to claim 1, characterized by generating a new information set and performing retraining based on the feedback received and the user's emotional response. [Explanation of symbols]

[0865] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. A means of collecting data to generate images related to a specific culture, A means for detecting cultural bias in generated images, A means for evaluating the discomfort index of an image based on detected bias, A means of updating the model using the evaluation results, A means of improving image representation using an updated model, A system that includes this.

2. The system according to claim 1, characterized in that it issues a warning when the image discomfort index exceeds a certain level.

3. The system according to claim 1, characterized by generating a new dataset based on the received feedback and performing retraining.

Images