System

Abstract

To provide a system.SOLUTION: A system includes: means for displaying a search box on the top of a website and receiving questions in natural language from a user through the search box; means for converting the questions into prompt sentences, sending it to generative AI and generating answers to the questions; means for analyzing an emotional state of the user using an emotion engine; means for adjusting how the answers are displayed based on the emotional state; and means for displaying the adjusted answers to the user.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] Many internal company websites suffer from information search problems, making it difficult for employees to efficiently find the information they need and verify its accuracy. [Means for solving the problem]

[0005] We will build a system in which all content is trained by generative AI and new information is automatically learned. A large search box will be placed at the top of the website, with existing information displayed at the bottom of the scroll. This will allow employees to search in natural language and return answers along with the location of the source of information that supports the answer. Employees will be able to check the accuracy of the answer using the supporting information. [Brief explanation of the drawings]

[0006] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 2 is a sequence diagram showing a flow of processing in the data processing system according to the first embodiment of the first form example. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Embodiment 1. [Figure 13] FIG. 10 is a sequence diagram showing a processing flow of a data processing system in a second embodiment of the second form example. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 of Embodiment Example 2. [Figure 15] FIG. 10 is a sequence diagram showing the flow of processing in a data processing system according to a third embodiment of the third embodiment. [Figure 16] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 3 of Embodiment 3. [Figure 17] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the first embodiment of the first form example when an emotion engine is combined. [Figure 18] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Form Example 1 when an emotion engine is combined. [Figure 19] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the second embodiment of the second form example when an emotion engine is combined. [Figure 20] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 of Form Example 2 when an emotion engine is combined. [Figure 21] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in the third embodiment of the third form example when an emotion engine is combined. [Figure 22] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 3 of Form Example 3 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0007] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0008] First, the terms used in the following description will be explained.

[0009] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, the processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), or a TPU (TENSOR PROCESSING UNIT (registered trademark)).

[0010] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0011] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0012] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0013] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0014] [First embodiment]

[0015] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0016] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0017] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0018] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0019] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0020] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0021] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0022] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0023] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0024] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0025] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0026] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0027] "Example 1"

[0028] This invention builds a system in which all content is learned by a generative AI and new information is automatically learned. Specifically, all content on an internal company website, including text, images, and videos, is learned by the generative AI. By automatically learning newly added information, it becomes possible to always provide the latest information.

[0029] "Example 2"

[0030] A large search box will be placed at the top of the website. This search box will allow employees to search in natural language. For example, if they enter a question such as "What is the latest sales strategy?", the generative AI will generate an answer from the information it has learned and display it.

[0031] "Example 3"

[0032] Existing information is displayed at the bottom of the scroll. This allows employees to see the location of the source of information that supports the answer along with the search results. For example, if the generative AI answers, "The latest sales strategy is ABC," the location of the source of information that supports this answer (for example, the "Latest Updates" section of the "Sales Strategy Page") will be displayed. This allows employees to check the accuracy of the answer for themselves.

[0033] The processing flow of each embodiment will be described below.

[0034] "Example 1"

[0035] Step 1: Have the generative AI learn all of the content. Specifically, have the generative AI learn all of the content on the company's internal website, including text, images, and videos.

[0036] Step 2: Automatically learns newly added information, making it possible to always provide the latest information.

[0037] "Example 2"

[0038] Step 1: Place a large search box at the top of your website.

[0039] Step 2: Allow employees to search in natural language, for example by entering a question such as "What is the latest sales strategy?"

[0040] Step 3: The generative AI generates an answer from the learned information and displays it.

[0041] "Example 3"

[0042] Step 1: Display existing information at the bottom of the scroll.

[0043] Step 2: Along with the search results, employees can see the location of the source that supports the answer.

[0044] Step 3: For example, if the generative AI answers, "The latest sales strategy is ABC," the location of the source of the answer (for example, the "Latest Updates" section of the "Sales Strategy Page") will be displayed. This allows employees to check the accuracy of the answer themselves.

[0045] Example 1

[0046] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0047] With conventional information management systems, it took time for the content of internal company websites to be updated with the latest information, making it difficult for users to quickly obtain the information they needed. Furthermore, there was a lack of means to verify the accuracy of the information, making it difficult to ensure the reliability of the information users obtained.

[0048] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means. In this invention, the server includes means for having the generative artificial intelligence learn all information and automatically learn new information, means for collecting information from websites, means for preprocessing the collected information, means for having the generative artificial intelligence learn the preprocessed information, means for periodically checking for newly added information and automatically collecting, preprocessing, and learning, means for the user to input a prompt sentence to the generative artificial intelligence to obtain the latest information, means for placing a large search box at the top of the website, and means for displaying existing information below the scroll bar. This allows users to always quickly obtain the latest information and confirm the accuracy of that information.

[0049] "Generative AI" is AI that has the ability to generate new information based on collected data.

[0050] "Automatic learning of new information" is a process of periodically checking for newly added information and automatically collecting, preprocessing, and learning from it.

[0051] "Means of collecting information from websites" refers to technologies used to obtain content such as text, images, and videos on websites.

[0052] "Means for preprocessing collected information" refers to techniques for analyzing collected data and performing processes such as tokenization, normalization, resizing, and noise removal.

[0053] "Means for training generative AI to learn preprocessed information" refers to a technique for inputting preprocessed data into generative AI and training a model.

[0054] A "prompt sentence" is text containing a question or request that a user inputs to a generative artificial intelligence.

[0055] A "search box" is an input field on a website into which a user enters a search query.

[0056] "Means for displaying below the scroll" refers to a technique for displaying existing information at the bottom of a web page.

[0057] "Means for verifying the accuracy of information" are techniques that provide the location of the source of information in order to verify the reliability of the information generated.

[0058] This invention is a system that uses generative artificial intelligence to learn all the content of an in-house website and automatically learns newly added information. Specific embodiments of this system are described below.

[0059] Data collection

[0060] The server collects all content, including text, images, and videos, from the company's internal website. Specifically, it uses Apache (registered trademark) HTTP Server to crawl each page of the website and obtain the HTML data. From the obtained HTML data, it uses BeautifulSoup to extract the URLs of the text, images, and videos.

[0061] Data Preprocessing

[0062] The server preprocesses the collected data. For text data, it uses a natural language processing library (e.g., NLTK or spaCy) to tokenize and normalize it. For image data, it uses OpenCV to resize and denoise it. For video data, it uses FFmpeg to extract frames and compress them.

[0063] Data Learning

[0064] The server inputs the preprocessed data into a generative artificial intelligence model for learning. Machine learning frameworks such as TENSORFLOW (registered trademark) and PyTorch are used for learning. The text data is input into GPT-4 (registered trademark), and the image data is input into DALL-E, and the model is optimized.

[0065] Automatic learning of new information

[0066] The server periodically checks for new information and automatically collects, pre-processes, and learns from it. Specifically, a Cron job is set up to periodically scan the website for new content.

[0067] Providing information

[0068] Users can obtain the latest information by inputting a prompt to the generative AI model. For example, if a user inputs "Please tell me the latest product information," the server passes the prompt to the generative AI model, and the model returns the latest product information to the user.

[0069] Examples of concrete examples and prompts

[0070] Specific examples

[0071] If a user wants to know the latest news on an internal company website, they input a prompt sentence to the generative artificial intelligence model: "Tell me the latest news."

[0072] The server generates the latest news from the learned data and provides it to users.

[0073] Prompt Sentence Examples

[0074] "Please tell me the latest product information"

[0075] "Tell me about this month's events"

[0076] "Explain the new policy."

[0077] In this way, the user can easily obtain the latest information at all times.

[0078] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0079] Step 1: Data collection

[0080] The server collects all content, including text, images, and videos, from the company's internal website. Specifically, it uses Apache HTTP Server to crawl each page of the website and obtains the HTML data. From the obtained HTML data, it uses BeautifulSoup to extract the URLs of the text, images, and videos.

[0081] Enter: Website URL

[0082] Output: List of text, image, and video URLs

[0083] Step 2: Data Preprocessing

[0084] The server pre-processes the collected data.

[0085] Preprocessing text data:

[0086] What it does: The server uses NLTK to tokenize the text data, removing unnecessary symbols and spaces, and spaCy to normalize the text (e.g., lowercasing, removing stop words).

[0087] Input: Text data

[0088] Output: Preprocessed text data

[0089] Image data preprocessing:

[0090] What happens: The server uses OpenCV to resize the image data, apply a noise reduction filter, and unify the image format (e.g., convert to JPEG) if necessary.

[0091] Input: Image data

[0092] Output: Preprocessed image data

[0093] Video data preprocessing:

[0094] What it does: The server uses FFmpeg to extract and compress the video data frame by frame, and then unifies the video resolution.

[0095] Input: Video data

[0096] Output: Preprocessed video data

[0097] Step 3: Data training

[0098] The server inputs the preprocessed data into a generative artificial intelligence model for training. Machine learning frameworks such as TensorFlow and PyTorch are used for training. The text data is input into GPT-4, and the image data is input into DALL-E, to optimize the model.

[0099] Input: Preprocessed text data, image data, and video data

[0100] Output: A trained generative artificial intelligence model

[0101] Step 4: Automatically learning new information

[0102] The server periodically checks for new information and automatically collects, pre-processes, and learns from it. Specifically, a Cron job is set up to periodically scan the website for new content.

[0103] Input: Newly added website content

[0104] Output: An updated trained generative artificial intelligence model

[0105] Step 5: Provide information

[0106] Users can obtain the latest information by inputting a prompt to the generative AI model. For example, if a user inputs "Please tell me the latest product information," the server passes the prompt to the generative AI model, and the model returns the latest product information to the user.

[0107] Input: prompt statement

[0108] Output: Latest information

[0109] In this way, the user can easily obtain the latest information at all times.

[0110] (Application example 1)

[0111] Next, a description will be given of Application Example 1 of Embodiment Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0112] Conventional content distribution systems have had difficulty providing users with the latest, personalized information. Furthermore, content generation based on user interests is performed manually, resulting in inefficiency and delays in providing the latest information. Furthermore, the generation of new articles based on news data is not automated, which means users are unable to receive consistent information.

[0113] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0114] In this invention, the server includes a means for having a generative AI learn all content and automatically learn new information, a means for placing a large search box at the top of the website, a means for displaying existing information below the scroll, a means for generating personalized content based on the user's interests, and a means for acquiring news data and having the generative AI generate new articles based on that data, making it possible to always provide users with the latest and personalized information.

[0115] "Generative AI" is artificial intelligence that learns content such as text, images, and videos, and automatically generates content based on new information.

[0116] "Automatic learning of new information" is a function that automatically learns newly added information and always provides the latest information.

[0117] A "search box" is an input field where users can enter keywords to search for information.

[0118] "Scroll Down" is a function that displays existing information at the bottom of a web page.

[0119] "Personalized content" is content that is individually optimized based on a user's interests and browsing history.

[0120] "News data acquisition" is the process of collecting the latest news information from external news sources.

[0121] "Article generation" is the process in which generative AI creates new articles based on collected news data.

[0122] A system for implementing this invention is configured as follows: The server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for generating personalized content based on user interests, and means for acquiring news data and having the generative AI generate new articles based on that data.

[0123] Hardware and software used

[0124] Hardware: Servers, smartphones

[0125] Software: Python, Transformers library (Hugging Face), Requests library

[0126] Data processing and calculation

[0127] 1. Data Acquisition: The server acquires the latest news data using the news API. For example, data is collected from the news API using a URL such as "https: / / newsapi.org / v2 / top-headlines?country=jp&apiKey=YOUR_API_KEY".

[0128] 2. Data training: The acquired news data is trained into a generative AI model (e.g., GPT-2). This gives the generative AI the ability to generate new content based on the latest news information.

[0129] 3. Content generation: Based on the learned data, new articles and content are generated to serve to users. The generated content is personalized based on the user's interests and browsing history.

[0130] Specific examples

[0131] For example, if a user is interested in "technology," the server will collect the latest technology-related news, and the generative AI will generate new articles based on that news. This allows users to always get the latest technology information.

[0132] Prompt Sentence Examples

[0133] "Generate new articles based on the latest technology news."

[0134] This way, users can stay up to date with the latest information and enjoy personalized content based on their interests.

[0135] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0136] Step 1:

[0137] The server uses a news API to obtain the latest news data.

[0138] Input: News API URL

[0139] Output: Latest news data (JSON format)

[0140] Specific operation: The server sends an HTTP request to the news API endpoint and receives the latest news data in response.

[0141] Step 2:

[0142] The server performs preprocessing on the acquired news data in order to input it into the generative AI model.

[0143] Input: Latest news data (JSON format)

[0144] Output: Preprocessed text data

[0145] Specific operation: The server extracts article titles and descriptions from the news data and formats them as text data.

[0146] Step 3:

[0147] The server inputs the preprocessed text data into a generative AI model to generate new articles.

[0148] Input: Preprocessed text data

[0149] Output: The new article generated.

[0150] How it works: The server inputs text data into a generative AI model (e.g., GPT-2), which then generates new articles.

[0151] Step 4:

[0152] The server personalizes the new articles generated based on the user's interests.

[0153] Input: Generated new articles, user interest data

[0154] Output: personalized article

[0155] Specific operation: The server filters or highlights generated articles based on the user's interest data (such as browsing history and past search keywords).

[0156] Step 5:

[0157] The server delivers personalized articles to the user's device.

[0158] Input: Personalized Article

[0159] Output: The article as it appears on the user's device

[0160] Specific operation: The server sends the personalized article to the user's smartphone or web browser for display.

[0161] Step 6:

[0162] Users can view the published article and use the search box to find additional information if needed.

[0163] Input: User's search keyword

[0164] Output: Search results

[0165] What it does: The user enters keywords into the search box, and the server searches for relevant information based on those keywords and returns the results to the user.

[0166] Example 2

[0167] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0168] With conventional information search systems, it was difficult for users to quickly and accurately obtain the information they needed. Furthermore, there was a lack of means to verify the basis of the information, making it difficult to obtain reliable information. Furthermore, the way existing information was displayed was inconvenient, impairing user convenience.

[0169] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes a means for having the generative AI model learn all information and automatically learn new information, a means for placing a large search box at the top of the website, a means for a user to input a question in natural language, a means for transmitting the input question to the generative AI model, a means for receiving and displaying an answer from the generative AI model, and a means for displaying existing information below the scroll. This allows a user to input a question in natural language and receive a quick and accurate answer from the generative AI model. In addition, since the basis information for the answer is also provided at the same time, the user can confirm reliable information. Furthermore, displaying existing information below the scroll improves user convenience.

[0170] A "generative AI model" is a type of artificial intelligence that uses natural language processing to generate answers to user questions.

[0171] A "prompt sentence" is a textual sentence that sends a question entered by a user to a generative AI model.

[0172] A "server" is a computer system that receives requests from a user, sends prompts to a generative AI model, and returns the generated answers to the user.

[0173] A "search box" is an input field placed on the top page of a website where users can enter questions in natural language.

[0174] "Natural language" refers to the language form that users use on a daily basis, and is free-flowing text that is not bound by any particular format or structure.

[0175] An "answer" is information that a generative AI model generates in response to a user's question.

[0176] "Evidence" refers to the location of the information source that the generative AI model referenced when generating the answer.

[0177] "Existing information" is information that already exists on the website and is displayed in a format that users can scroll through.

[0178] "Automatic learning" is the process by which a generative AI model continually learns new information, improving the model's accuracy and knowledge.

[0179] This invention is a system that allows users to input questions in natural language and obtain quick and accurate answers using a generative AI model. Specific embodiments of this system are described below.

[0180] Hardware and software used

[0181] server

[0182] The server receives requests from users, sends prompts to the generative AI model, and returns the generated answers to the users. The server uses the following hardware and software:

[0183] Hardware: A server machine with a powerful processor, sufficient memory, and a fast network connection

[0184] Software: web server software (e.g., Apache, Nginx), programming languages ​​(e.g., Python, JavaScript®), APIs for generative AI models (e.g., OpenAI®'s GPT-4)

[0185] Terminal

[0186] The device used by the user is a computer or smartphone with a web browser, which the user uses to access the website and enter a question into a search box.

[0187] Data processing and calculation

[0188] User Input

[0189] Users type a question into a large search box on the homepage of a website, for example, "What are the latest sales strategies?"

[0190] Receiving the server and generating a prompt

[0191] The server receives the question entered by the user and converts it into a prompt, which is simply "What is the latest sales strategy?"

[0192] Send to generative AI model and generate answer

[0193] The server sends the generated prompt to the generative AI model, which analyzes the received prompt and generates the optimal answer based on the information it has learned. For example, the following answer might be generated:

[0194] "The latest sales strategies include strengthening digital marketing, improving targeting accuracy through customer data analysis, and establishing a remote work environment."

[0195] Server display process

[0196] The server receives the answers returned by the generative AI model and displays them on the website's search results page, allowing users to see the answers to their questions.

[0197] Specific examples

[0198] For example, if a user types the question "What is the latest sales strategy?" into a search box, the following prompt will be sent to the generative AI model:

[0199] What are the latest sales strategies?

[0200] The generative AI model analyzes this prompt and generates the following answer:

[0201] The latest sales strategies include strengthening digital marketing, improving targeting accuracy through analysis of customer data, and establishing a remote work environment.

[0202] The server receives this answer and displays it on the website's search results page, where users can view it and use it to improve their business.

[0203] In this way, the invention allows users to enter questions in natural language and get fast and accurate answers using generative AI models.

[0204] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0205] Step 1:

[0206] A user accesses the homepage of a website and enters a question into a large search box, for example, "What is the latest sales strategy?" The entered question is sent from the user's device to the server.

[0207] Step 2:

[0208] The server receives questions sent by users. It saves the received questions in text format and prepares them to be sent to the generative AI model as prompts. The input is the user's question, and the output is the prompt.

[0209] Step 3:

[0210] The server sends the received question as a prompt to the generative AI model. The prompt is "What is the latest sales strategy?" The input is the prompt, and the output is a request to the generative AI model.

[0211] Step 4:

[0212] The generative AI model receives and analyzes the prompt sent from the server. The model uses an internal neural network to generate the optimal answer from the learned information. The input is the prompt, and the output is the generated answer.

[0213] Step 5:

[0214] The generative AI model sends the generated answer back to the server. For example, it might generate an answer like, "The latest sales strategies include strengthening digital marketing, improving targeting accuracy by analyzing customer data, and establishing a remote work environment." The input is the generated answer, and the output is the response to the server.

[0215] Step 6:

[0216] The server receives the answers returned by the generative AI model, embeds the received answers into an HTML template, and generates a search results page for the website. The input is the generated answers, and the output is the HTML search results page.

[0217] Step 7:

[0218] The server returns the generated search results page to the user's device. The user can then view the search results page in their browser and read the displayed answers. The input is the HTML search results page, and the output is the answer displayed in the user's browser.

[0219] In this way, users can enter questions in natural language and get fast and accurate answers using generative AI models.

[0220] (Application example 2)

[0221] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0222] There is a problem in that it is difficult for workers in factories to quickly and accurately obtain the information they need. In particular, when workers want to know the parts and procedures required for the next process, conventional methods can take time and reduce efficiency. In addition, there is a lack of means to confirm the accuracy of the information, so there is a risk that work will proceed based on incorrect information.

[0223] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for having a generative AI learn all information and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll; means for factory workers to input questions in natural language and for the generative AI to generate and display the optimal answer; means for inputting questions using a smartphone or tablet; and means for the generative AI to provide information about work in the factory. This enables workers to quickly and accurately obtain the information they need, improving work efficiency.

[0224] "Information" refers to data and knowledge related to operations within a factory.

[0225] "Generative AI" refers to artificial intelligence that uses natural language processing to generate answers to questions.

[0226] "Automatic learning" refers to the process of continually learning new data and information and updating its knowledge base.

[0227] A "search box" is an input field where a user can enter a question in natural language.

[0228] "Scroll Down" refers to the area at the bottom of a web page that becomes visible when the user scrolls down.

[0229] "Workers" means employees who perform work within a factory.

[0230] "Natural language" refers to text written in a language form that humans use on a daily basis.

[0231] A "smartphone" is a mobile phone that can connect to the Internet and use applications.

[0232] A "tablet" is a highly portable computing device that has a touchscreen.

[0233] A "best answer" is an answer that provides the most relevant and accurate information for a question.

[0234] A "source" is a place where the data or knowledge that forms the basis of an answer is stored.

[0235] "Accuracy" means that the information or answer is factual and free from errors.

[0236] The system for implementing this invention is configured as follows: First, the server uses generative AI to learn all information and automatically learns new information. This allows the latest information to be maintained at all times. Next, a large search box is placed at the top of the website, allowing users to enter questions in natural language. Furthermore, existing information is displayed at the bottom of the page, making it easy for users to access past information.

[0237] Factory workers use smartphones or tablets to input questions. For example, when a question such as "What parts are needed for the next process?" is input, the generative AI generates the optimal answer to the question and displays it to the worker. The generative AI generates answers based on data it has learned in advance to provide information about work within the factory.

[0238] The hardware used includes smartphones and tablets, and the software uses Python and the OpenAI API. Data processing and data calculation involves receiving questions in natural language from users and generating answers using OpenAI's generative AI model (GPT-3 (registered trademark)).

[0239] As a concrete example, if a worker in a factory asks, "What parts are needed for the next process?", the following prompt sentence is input into the generative AI model.

[0240] Example prompt sentence:

[0241] Factory operation questions: What part is needed for the next operation?

[0242] answer:

[0243] Based on this prompt, the generative AI generates a list of parts required for the next process and provides it to the worker, allowing the worker to quickly and accurately obtain the necessary information and improve work efficiency.

[0244] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0245] Step 1:

[0246] A user uses a smartphone or tablet to enter a question in natural language into a search box at the top of a website. The entered question is sent to the server. The input data is the user's question, and the output data is the question sent to the server.

[0247] Step 2:

[0248] The server sends the received question to the generative AI model. Specifically, the question is input to the generative AI model (GPT-3) as a prompt. The input data is the user's question, and the output data is the prompt sent to the generative AI model.

[0249] Step 3:

[0250] A generative AI model generates an optimal answer based on a prompt. Specifically, it generates an answer to a question and returns that answer to the server. The input data is the prompt, and the output data is the generated answer.

[0251] Step 4:

[0252] The server sends the answers received from the generative AI model to a smartphone or tablet for display to the user. The input data is the generated answer, and the output data is the answer sent to the user's device.

[0253] Step 5:

[0254] The user checks the response displayed on their smartphone or tablet and obtains the necessary information. Specifically, they check information such as the parts list and procedures required for the next process. The input data is the response sent from the server, and the output data is the information obtained by the user.

[0255] Through these processing steps, users can input questions in natural language and receive quick and accurate answers using generative AI models, thereby improving work efficiency within factories.

[0256] Example 3

[0257] Next, a description will be given of a third embodiment of the third embodiment. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0258] In conventional information retrieval systems, users lacked the means to verify the basis of the answers they received, making it difficult to determine the reliability of the answers. Another problem was that it took a lot of time and effort to identify the location of the information source. Furthermore, when new information was added, the system sometimes failed to automatically learn and provide the latest information.

[0259] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0260] In this invention, the server includes: means for training a generative AI model to learn all information and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll bar; means for sending a prompt entered by a user to the server; means for the server to send the prompt to the generative AI model and generate an answer; means for the server to search an internal database for information sources related to the answer; means for the server to send the search results and the answer to a terminal; and means for the terminal to display the search results and the answer to the user. This allows the user to easily check the basis for the generated answer, increasing the reliability of the answer. Furthermore, when new information is added, the system automatically learns and can provide the latest information.

[0261] A "generative AI model" is an artificial intelligence algorithm that generates answers based on prompts entered by the user.

[0262] A "prompt" is a natural language question or instruction that a user enters to retrieve information.

[0263] The "server" is a computer system that receives prompts from users, sends them to a generative AI model to generate answers, and then searches for relevant information sources and sends the results to the terminal.

[0264] A "terminal" is a device through which a user can enter prompts and display answers and related information sent by the server.

[0265] An "internal database" is a database that stores information sources held by a company or organization, and is used by the server to search for the basis of an answer.

[0266] A "search box" is an input field placed at the top of a website for users to enter a prompt phrase.

[0267] The "scroll bottom" is an area located at the bottom of a web page for displaying existing information.

[0268] An "answer" is information that a generative AI model generates based on a prompt.

[0269] A "source" is the data or document that forms the basis for the generated answer.

[0270] "Automatic learning" is the process by which a generative AI model continuously learns new information, enabling it to provide up-to-date information.

[0271] This invention is a system in which a generative AI model generates an answer based on a prompt sentence entered by a user and displays the information source that forms the basis of the answer. Specific embodiments of this system are described below.

[0272] First, the user inputs a prompt sentence into a search box installed on the web browser of the terminal. For example, the user inputs a prompt sentence such as "What is the latest sales strategy?"

[0273] The device sends the input prompt to the server using an HTTP POST request. The server then sends the received prompt as an API request to the generative AI model. For example, OpenAI's GPT-4 can be used as the generative AI model.

[0274] The generative AI model generates an answer based on the prompt sentence. For example, it generates the answer "The latest sales strategy is ABC." The server receives this generated answer and searches an internal database for information sources related to that answer. The internal database stores information sources held by companies and organizations, and searches are performed using SQL queries.

[0275] The server sends the search results and the generated answer in JSON format to the terminal. The terminal displays the received answer and the search results to the user. For example, along with the answer "The latest sales strategy is ABC," it displays a link to the "Latest Updates" section of the "Sales Strategy Page." The user can click this link to view more detailed information.

[0276] This system allows users to easily check the basis of the generated answers, increasing the reliability of the answers.In addition, when new information is added, the system automatically learns and can provide the latest information.

[0277] As a concrete example, if a user inputs the prompt sentence "What is the latest sales strategy?", the server uses the generative AI model to generate the answer "The latest sales strategy is ABC", and searches the internal database to identify the "Latest Updates" section of the "Sales Strategy Page". Ultimately, the answer "The latest sales strategy is ABC" is displayed on the user's device, along with a link to the "Latest Updates" section of the "Sales Strategy Page".

[0278] In this way, the user can easily check the basis of the information provided by the generative AI model. The flow of the identification process in the third embodiment will be described with reference to FIG.

[0279] Step 1:

[0280] The user enters a prompt statement.

[0281] The user inputs a prompt sentence, "What is the latest sales strategy?", into a search box installed on the web browser of the terminal. The input prompt sentence becomes the input data of the terminal.

[0282] Step 2:

[0283] The terminal sends the prompt to the server.

[0284] The terminal sends the input prompt text to the server using an HTTP POST request. The input data is the prompt text, and the output data is the prompt text sent to the server.

[0285] Step 3:

[0286] The server sends a prompt to the generative AI model, which generates an answer.

[0287] The server sends the received prompt sentence to the generative AI model as an API request. The generative AI model generates an answer based on the prompt sentence. The input data is the prompt sentence, and the output data is the generated answer. For example, the generated answer is "The latest sales strategy is ABC."

[0288] Step 4:

[0289] The server searches an internal database for sources relevant to the answer.

[0290] The server searches an internal database for information sources related to the generated answer. The internal database stores information sources owned by the company or organization, and performs a search using an SQL query. The input data is the generated answer, and the output data is the search results for the related information sources. For example, the "Latest Updates" section of the "Sales Strategy Page" is identified.

[0291] Step 5:

[0292] The server sends the search results and answers to the device.

[0293] The server sends the generated answers and search results to the terminal in JSON format. The input data are the generated answers and search results, and the output data is the JSON data sent to the terminal.

[0294] Step 6:

[0295] The device displays the search results and answers to the user.

[0296] The terminal displays the received answers and search results to the user. The input data is the JSON data received from the server, and the output data is the answer displayed to the user and a link to the information source. For example, along with the answer "The latest sales strategy is ABC," a link to the "Latest Updates" section of the "Sales Strategy Page" is displayed. The user can click this link to view more detailed information.

[0297] (Application example 3)

[0298] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0299] Conventional generative AI systems lacked sufficient means for users to verify the basis of the answers they received, making it difficult for factory workers, in particular, to obtain information quickly and accurately on-site. Furthermore, there was a lack of means to verify the reliability of the information provided by generative AI, making it difficult for users to verify the accuracy of the answers themselves.

[0300] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[0301] In this invention, the server includes means for having the generative AI learn all information and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for the generative AI to provide an answer when a user inputs a question and display the location of the source of the evidence, and means for the generative AI to provide an answer when a factory worker asks a question to the robot and display the location of the source of the evidence. This allows users to check the accuracy of the generative AI's answer using the evidence information, and enables factory workers to obtain information quickly and accurately on site.

[0302] "Information" is a general term for the data and knowledge that generative AI learns and provides to users.

[0303] "Generative AI" is an artificial intelligence system that generates answers in natural language to users' questions.

[0304] "Automatic learning" is the process by which generative AI automatically absorbs and learns new information.

[0305] A "search box" is an input field on a website where users can enter questions or keywords.

[0306] "Scroll Down" is the area at the bottom of a web page that becomes visible when the user scrolls down.

[0307] "Evidence" is a source of information that supports the accuracy of the answers provided by generative AI.

[0308] "Factory workers" refers to employees who work in factories.

[0309] A "robot" is an automated mechanical device that provides work assistance and information within a factory.

[0310] To implement this invention, all information must first be learned by a generative AI. A generative AI is an artificial intelligence system that generates answers in natural language to user questions and has the ability to automatically learn new information. This makes it possible to always provide the latest information.

[0311] Next, a large search box is placed on the homepage of the website. Users can make inquiries to the generative AI by entering questions or keywords into this search box. When a user enters a question, the generative AI generates an answer to that question and displays the location of the source of the answer.

[0312] Additionally, a new feature will be added that displays existing information at the bottom of the scroll, allowing users to easily refer to supporting information to verify the accuracy of answers provided by generative AI.

[0313] To enable factory workers to quickly and accurately obtain information on-site, this system will be installed on robots within the factory. When a factory worker asks the robot a question, generative AI will provide an answer and display the location of the source of the answer. This will enable workers to quickly obtain the information they need on-site and improve work efficiency.

[0314] The implementation of this system uses the following hardware and software. For hardware, we need factory robots and displays. For software, we use the OpenAI API, Python, and Tkinter (GUI library). We use the OpenAI API to obtain answers from a generative AI model (e.g., GPT-3), and we use Python and Tkinter to build the user interface.

[0315] For example, if a factory worker asks, "What is the latest quality control method?", the generative AI will respond, "The latest quality control method is XYZ." The location of the source of information that supports the answer (for example, the "Latest Updates" section of the "Quality Control Page") is displayed at the bottom of the scroll. This allows the worker to confirm the accuracy of the answer and improve their work efficiency.

[0316] An example prompt is:

[0317] Q: What are the latest quality control techniques?

[0318] In this way, users can verify the accuracy of generative AI's answers with supporting information, enabling factory workers to obtain information quickly and accurately on-site.

[0319] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[0320] Step 1:

[0321] A user types a question into a website search box.

[0322] Input: User question (e.g., "What are the latest quality control techniques?")

[0323] Output: The user's question is sent to the server.

[0324] What happens: A user types a question into the search box and clicks the submit button.

[0325] Step 2:

[0326] The server receives the user's question and sends it to the generative AI model.

[0327] Input: User question

[0328] Output: Prompt sentence for generative AI model

[0329] How it works: The server sends the user's question in text format to a generative AI model (e.g., GPT-3).

[0330] Step 3:

[0331] A generative AI model generates an answer based on the prompt.

[0332] Input: Prompt statement (e.g., "What are the latest quality control methods?")

[0333] Output: Answer from the generative AI model (e.g., "The most current quality control method is XYZ")

[0334] What it does: The generative AI model analyzes the prompt and generates an appropriate response.

[0335] Step 4:

[0336] The server receives the answer from the generative AI model and searches for supporting information.

[0337] Input: Answer from a generative AI model

[0338] Output: Location of answer and supporting information

[0339] Specific operation: The server receives the answer from the generative AI model and searches the database for the source of information that supports that answer.

[0340] Step 5:

[0341] The server returns the answer and supporting information to the user.

[0342] Input: Location of answer and supporting information

[0343] Output: Answer and rationale displayed on the user's device

[0344] Specific operation: The server sends the answer and supporting information to the user's device and displays it at the bottom of the scroll.

[0345] Step 6:

[0346] The user checks the answer and supporting information.

[0347] Input: Answer and rationale displayed on the user's device

[0348] Output: User understanding and confirmation

[0349] What happens: The user reviews the evidence displayed below the scroll bar and evaluates the accuracy of the answer.

[0350] Step 7:

[0351] A factory worker types a question into the robot.

[0352] Input: Factory worker question (e.g., "What are the latest quality control techniques?")

[0353] Output: The factory worker's question is sent to the robot.

[0354] How it works: A factory worker types a question into the robot's interface and clicks the send button.

[0355] Step 8:

[0356] The robot sends a question to the server and gets an answer based on a generative AI model.

[0357] Input: Factory worker question

[0358] Output: Answers and supporting information from the generative AI model

[0359] Specific operation: The robot sends a question to the server and obtains an answer and supporting information from the generative AI model.

[0360] Step 9:

[0361] The robot displays the answer and supporting information to the factory worker.

[0362] Input: Answer from the generative AI model and supporting information

[0363] Output: Answers and supporting information displayed on the factory worker's terminal

[0364] Specific operation: The robot displays the answer and supporting information on the factory worker's device, scrolling down below.

[0365] Step 10:

[0366] Factory workers will review the responses and supporting information.

[0367] Input: Answer and supporting information displayed on the factory worker's terminal

[0368] Output: Factory worker understanding and confirmation

[0369] What happens: The factory worker reviews the supporting information displayed at the bottom of the scroll and evaluates the accuracy of the answer.

[0370] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0371] "Example 1"

[0372] One embodiment of the present invention is a system that incorporates an emotion engine. This system includes a means for having a generative AI learn all content and automatically learn new information, a means for placing a large search box at the top of a website, and a means for displaying existing information below the scroll bar. It also includes an emotion engine that recognizes the user's emotions. Specifically, the emotion engine recognizes the user's emotional state when using a website, and the generative AI adjusts search results based on that emotional state. For example, if the user expresses joy, the generative AI will prioritize displaying information related to joy.

[0373] "Example 2"

[0374] The emotion engine also adjusts how information sources are displayed based on the user's emotions. For example, if the user expresses confusion, the generative AI will display the source's detailed explanation more clearly. This allows the user to receive the most appropriate information based on their emotional state.

[0375] "Example 3"

[0376] Furthermore, this system allows users to check the accuracy of answers using supporting information. Specifically, the location of the source of information that supports the answer provided by the generative AI is displayed, allowing users to check its accuracy for themselves. This allows users to receive the most appropriate information based on their emotional state and to confirm the reliability of that information.

[0377] The processing flow of each embodiment will be described below.

[0378] "Example 1"

[0379] Step 1: A user opens a website.

[0380] Step 2: The user enters a search query in natural language into the search box.

[0381] Step 3: The emotion engine recognizes the user's emotional state.

[0382] Step 4: The generative AI receives input from the emotion engine and refines the search results.

[0383] Step 5: The generative AI displays the adjusted search results to the user.

[0384] "Example 2"

[0385] Step 1: A user opens a website.

[0386] Step 2: The user enters a search query in natural language into the search box.

[0387] Step 3: The emotion engine recognizes the user's emotional state.

[0388] Step 4: The generative AI receives input from the emotion engine and adjusts how the source is presented.

[0389] Step 5: The generative AI displays the adjusted source information to the user.

[0390] "Example 3"

[0391] Step 1: A user opens a website.

[0392] Step 2: The user enters a search query in natural language into the search box.

[0393] Step 3: The generative AI generates search results and identifies the location of the source of information that supports them. Step 4: The generative AI displays the search results and the location of the source of information to the user.

[0394] Step 5: The user locates the displayed source and verifies the accuracy of the answer.

[0395] Example 1

[0396] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0397] On conventional websites, it was difficult for users to quickly and accurately obtain the information they needed. Furthermore, there was a problem in that the information provided did not correspond to the user's emotional state, resulting in a poor user experience. Furthermore, there was a lack of a system that could automatically learn new information and constantly provide the latest information.

[0398] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0399] In this invention, the server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means including an emotion engine that recognizes the user's emotional state, and means for the generative AI to adjust search results based on the user's emotional state. This allows users to quickly and accurately obtain the information they need, makes it possible to provide information according to their emotional state, and always provides the latest information.

[0400] "Generative AI" is artificial intelligence that generates new content based on collected data.

[0401] "Means for automatically learning new information" refers to a means that has the function of automatically collecting new added data and having the generative AI learn it.

[0402] "A means for placing a large search box at the top of a website" refers to a means for placing a large search box on the top page of a website so that users can easily perform searches.

[0403] The "means for displaying existing information below the scroll" is a means for displaying information below the web page so that the user can view the existing information by scrolling.

[0404] The "emotion engine that recognizes the user's emotional state" is an engine that analyzes emotions from the user's facial expressions and text input and recognizes their state.

[0405] "Means for the generative AI to adjust search results based on the emotional state" refers to means for the generative AI to adjust the search results displayed according to the user's emotional state.

[0406] MODE FOR CARRYING OUT THE INVENTION

[0407] This invention is a system that uses generative AI to learn all content and automatically learns new information. Specifically, the generative AI learns all content on an internal company website, including text, images, and videos. By automatically learning newly added information, it is possible to always provide the latest information.

[0408] Hardware and software used

[0409] The server uses the following hardware and software:

[0410] Web scraping tools: BeautifulSoup, Scrapy

[0411] Natural language processing tools: NLTK, SpaCy

[0412] Image processing libraries: OpenCV, PIL

[0413] Video processing library: FFmpeg

[0414] Generation AI: GPT-4, DALL-E

[0415] Emotion engine: Affectiva, IBM Watson® Tone Analyzer

[0416] Program processing

[0417] The server collects all content from the internal company website, including text, images, and videos. A web scraping tool is used to analyze the HTML structure of the web page and extract the required data. For example, text data is extracted from HTML tags, and image data is obtained by obtaining image URLs.

[0418] The collected data is tokenized, normalized, and stop words removed using natural language processing tools, image data is resized and formatted using an image processing library, and video data is frame-extracted and compressed using a video processing library.

[0419] The preprocessed data is then trained by a generative AI, which updates its model based on all collected content and automatically learns new information. For example, text data is fed into GPT-4, and image data is fed into DALL-E.

[0420] The device recognizes the user's emotional state when using a website. It uses an emotion engine to analyze emotions from the user's facial expressions and text input. For example, the device captures the user's facial expressions through a webcam, and the emotion engine analyzes them.

[0421] The server uses a generative AI to tailor search results based on the user's emotional state, obtained from the emotion engine. For example, if the user expresses joy, the generative AI will prioritize information related to joy.

[0422] Specific examples

[0423] A user visits an internal company website and types "new product reviews" into the search box. The device captures the user's facial expressions with a webcam, and the emotion engine recognizes the emotion of happiness from the user's facial expression. The server receives this emotion data and inputs a prompt to the generative AI: "If the user is expressing happiness, please prioritize displaying positive reviews." Based on this prompt, the generative AI prioritizes displaying positive reviews.

[0424] Example prompt sentence:

[0425] "Show reviews of new products. If users are expressing joy, prioritize positive reviews."

[0426] In this way, a system is constructed in which the server, terminals, and users work together to provide the latest information that is tailored to their emotions.

[0427] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0428] Step 1: Data collection

[0429] The server collects all content, including text, images, and videos, from internal company websites.

[0430] Input: Internal website URL

[0431] Specific operation: The server uses a web scraping tool (e.g., BeautifulSoup, Scrapy) to analyze the HTML structure of the web page and extract the necessary data. For example, text data is extracted from HTML tags, and image data is obtained by obtaining image URLs.

[0432] Output: Collected text, image, and video data

[0433] Step 2: Data Preprocessing

[0434] The server pre-processes the collected data.

[0435] Input: Collected text, image, and video data

[0436] Specific operation: The server uses natural language processing tools (e.g., NLTK, SpaCy) to tokenize, normalize, and remove stop words from text data. Image data is resized and formatted using an image processing library (e.g., OpenCV, PIL). Video data is extracted and compressed using a video processing library (e.g., FFmpeg).

[0437] Output: Preprocessed text, image, and video data

[0438] Step 3: Data training

[0439] The server trains the generative AI on the preprocessed data.

[0440] Input: Preprocessed text, image, and video data

[0441] Specific operation: The server inputs data into generative AI (e.g., GPT-4, DALL-E) and updates the model. For example, text data is input into GPT-4, and image data is input into DALL-E. When new information is added, the learning process is automatically restarted.

[0442] Output: An updated generative AI model

[0443] Step 4: Emotion Recognition

[0444] The terminal recognizes the emotional state of the user when using the website.

[0445] Input: User facial expression data, text input

[0446] Specific operation: The device uses an emotion engine (e.g., Affectiva, IBM Watson Tone Analyzer) to analyze emotions from the user's facial expressions and text input. For example, the device captures the user's facial expressions through a webcam, and the emotion engine analyzes them.

[0447] Output: User's emotional state data

[0448] Step 5: Refine your search results

[0449] The server uses generative AI to adjust search results based on the user's emotional state obtained from the emotion engine.

[0450] Input: User's emotional state data, user's search query

[0451] Specific operation: The server receives emotion data from the emotion engine and inputs prompts to the generative AI. For example, the server inputs a prompt such as "If the user is expressing joy, please prioritize displaying positive reviews" to adjust the search results.

[0452] Output: Refined search results

[0453] In this way, a system is constructed in which the server, terminals, and users work together to provide the latest information that is tailored to their emotions.

[0454] (Application example 1)

[0455] Next, a description will be given of Application Example 1 of Embodiment Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0456] Conventional online shopping sites recommend products without considering the user's emotional state, which can lead to low user satisfaction. Furthermore, it is difficult to provide the latest information, making it difficult for users to quickly access the information they need. This has led to a demand for improved user experience.

[0457] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0458] In this invention, the server includes: means for having a generative AI learn all content and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll; means including an emotion engine that recognizes the user's emotional state; means for the generative AI to adjust search results based on the user's emotional state; and means for recommending products based on the user's emotions. This enables product recommendations based on the user's emotional state, improving the user experience.

[0459] "Generative AI" is artificial intelligence that learns content such as text, images, and videos and generates new information.

[0460] "Automatic learning of new information" is a function that automatically learns newly added information and always provides the latest information.

[0461] A "search box" is an input field in which a user can enter keywords to search for information.

[0462] "Scroll Down" is a function that displays existing information at the bottom of a web page.

[0463] An "emotion engine" is an engine that recognizes the user's emotional state and adjusts the system's behavior based on that emotion.

[0464] An "emotional state" is an emotional state such as joy, sadness, or surprise expressed by a user.

[0465] "Product recommendation" is a function that suggests appropriate products to users based on their emotional state and search history.

[0466] "User experience" is a general term for the satisfaction and ease of use that users feel when using a system or service.

[0467] The system for implementing this invention is configured as follows: The server has a means for having generative AI learn all content and automatically learn new information. This makes it possible to always provide the latest information. Furthermore, it has a means for placing a large search box at the top of the website, allowing users to easily search for information. It also has a means for displaying existing information below the scroll bar, making it easy for users to access past information.

[0468] The server has a means including an emotion engine that recognizes the user's emotional state. This emotion engine analyzes the user's facial image and voice data to recognize the user's emotional state in real time. Emotional states include joy, sadness, surprise, etc. The emotion engine performs image processing using OpenCV and predicts emotions using an emotion recognition model built with Keras.

[0469] This also includes a means for generative AI to adjust search results based on the user's emotional state. The generative AI is built using the Transformers library and generates prompts based on the user's emotional state. For example, if the user expresses happiness, the prompt would be, "Recommend products that will make the user happy."

[0470] It also includes a method for recommending products based on the user's emotions. Generative AI recommends appropriate products based on prompt text. This enables product recommendations optimized for the user's emotional state, improving the user experience.

[0471] As a concrete example, consider a case where a user takes a picture of their face using a smartphone camera, and an application analyzes the image to recognize emotions. For example, if the user is smiling, the application generates a prompt such as "Please recommend products that will make the user happy," and the generative AI recommends products based on that prompt.

[0472] An example of a prompt is as follows:

[0473] "Recommend products that will bring joy to users."

[0474] "Recommend products that make users sad."

[0475] "Recommend products that will surprise users."

[0476] In this way, product recommendations can be made according to the user's emotional state, improving the user experience.

[0477] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0478] Step 1:

[0479] The user activates the smartphone camera and takes a picture of their face.

[0480] Input: User's face image

[0481] Output: Face image data

[0482] Specific operation: The user opens the camera app on their smartphone and takes a picture of their face. The captured image is sent to the application.

[0483] Step 2:

[0484] The device receives the facial image data and performs image processing using OpenCV.

[0485] Input: Facial image data

[0486] Output: Preprocessed image data

[0487] Specific operation: The device converts the facial image to grayscale, then resizes and normalizes it to a form suitable for the emotion recognition model.

[0488] Step 3:

[0489] The device uses an emotion recognition model built with Keras to predict the user's emotions from preprocessed image data.

[0490] Input: Preprocessed image data

[0491] Output: Emotional state (e.g., happy, sad, surprised)

[0492] Specific operation: The device inputs the preprocessed image data into the emotion recognition model to predict the emotional state, which is then sent to the next step.

[0493] Step 4:

[0494] The server generates an appropriate prompt sentence based on the emotional state.

[0495] Input: Emotional state

[0496] Output: prompt statement

[0497] Specific operation: The server generates a prompt sentence according to the emotional state, for example, "Please recommend a product that will bring joy to the user."

[0498] Step 5:

[0499] The server inputs prompts into a generative AI model (using the Transformers library) to generate product recommendations.

[0500] Input: prompt statement

[0501] Output: Product recommendation list

[0502] Specific operation: The server inputs a prompt sentence into the generative AI model and obtains the generated product recommendation list.

[0503] Step 6:

[0504] The server transmits the product recommendation list to the terminal, which displays it to the user.

[0505] Input: Product Recommendation List

[0506] Output: A list of product recommendations displayed to the user

[0507] Specific operation: The server sends the product recommendation list to the terminal, and the terminal displays it to the user, who can then check the recommended products.

[0508] In this way, product recommendations based on the user's emotional state are realized.

[0509] Example 2

[0510] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0511] In conventional information search systems, even if a user enters a question in natural language, it can be difficult to obtain an appropriate answer. Furthermore, since the information provided does not reflect the user's emotional state, the user may feel confused or anxious. Furthermore, since the source of the answer is not clearly stated, it is difficult to confirm the reliability of the answer.

[0512] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0513] In this invention, the server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for a user to input a question in natural language, means for sending the user's question as a prompt to the generative AI model, means for the generative AI model to generate an answer to the question and return it to the server, means for analyzing the user's emotional state using an emotion engine, means for adjusting the way the answer is displayed based on the user's emotional state, means for displaying the optimized answer to the user, and means for displaying existing information below the scroll. This allows a user to input a question in natural language and receive an appropriate answer from the generative AI model, and further enables the emotion engine to provide optimal information according to the user's emotional state.

[0514] "Generative AI" refers to artificial intelligence that generates new data and answers based on learned information.

[0515] A "search box" is an input field on a web page where a user can enter a question in natural language.

[0516] A "prompt" is an input sentence used to ask questions or give instructions to generative AI.

[0517] An "emotion engine" is software that analyzes and infers a user's emotional state from their input and behavior.

[0518] A "server" is a computer system that serves website content and processes user requests.

[0519] A "user" is someone who visits a website and types a question into a search box.

[0520] "Natural language" refers to text written in everyday language, and is free text that is not bound by any particular format or structure.

[0521] An "answer" is information that a generative AI generates in response to a user's question.

[0522] "Sources of information" refers to the data and literature that generative AI refers to when generating answers.

[0523] "Existing Information" refers to data or content that already exists on a website.

[0524] MODE FOR CARRYING OUT THE INVENTION

[0525] This invention is a system that allows a user to input a question in natural language and obtain an appropriate answer using a generative AI model. Specific embodiments for implementing this system are described below.

[0526] Hardware and software used

[0527] server

[0528] The server uses a high-performance cloud server. Specifically, an EC2 instance from Amazon Web Services (AWS (registered trademark)) can be used. The server provides website content and handles user requests.

[0529] Generative AI Model

[0530] The generative AI model used is OpenAI's GPT-4, which has been trained on a large amount of text data and is capable of generating appropriate answers to user questions.

[0531] Emotion Engine

[0532] The emotion engine uses Microsoft® Azure® Cognitive Services, which can analyze and infer emotional states from user input and behavior.

[0533] System Operation

[0534] Place a search box at the top of the website

[0535] The server displays a large search box on the homepage of the website, designed using HTML and CSS to allow users to enter questions in natural language.

[0536] Send user questions to generative AI models

[0537] When a user types a question into the search box and hits enter, the server receives the question and sends it to the generative AI model as a prompt, which is a properly formatted version of the user's question.

[0538] Answer generation using generative AI models

[0539] The generative AI model receives the prompt and generates an appropriate answer from the information it has learned, which is then returned to the server.

[0540] Sentiment analysis with emotion engine

[0541] The server uses an emotion engine to analyze the user's emotional state, inferring emotions such as confusion or anxiety from the user's input and past behavior.

[0542] Adjust how answers are displayed

[0543] The server adjusts how the generative AI model presents answers based on the emotion engine's analysis, for example, presenting answers in more detail if the user is confused.

[0544] Optimized answer display

[0545] The server displays the optimized answer to the user, allowing the user to receive information that is optimal for their emotional state.

[0546] Specific examples

[0547] For example, consider the case where a user types "What is the latest sales strategy?" into a search box. This question is sent to the server and passed as a prompt to the generative AI model. The generative AI model generates an answer based on the prompt, "Tell me about the latest sales strategy." The generated answer is, "The latest sales strategy revolves around strengthening digital marketing and analyzing customer data."

[0548] If the emotion engine analyzes the user's emotions and determines that they are confused, the server will display the generated answer in detail, for example, "Modern sales strategies revolve around strengthening digital marketing and analyzing customer data. For more information, please see this link."

[0549] In this way, users can input questions in natural language and receive appropriate answers from the generative AI model. The emotion engine also provides optimal information based on the user's emotional state.

[0550] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0551] Step 1:

[0552] A user accesses a website using a browser. The user enters the website's URL in the browser's address bar and presses the Enter key. The input is the website's URL, and the output is the website's homepage.

[0553] Step 2:

[0554] The server renders the website's homepage and displays a large search box. The server generates the homepage using HTML and CSS, centering the search box. The input is the website's content data, and the output is the homepage that is displayed in the user's browser.

[0555] Step 3:

[0556] A user enters a question in natural language into the search box. The user types "What is the latest sales strategy?" into the search box and presses the enter key. The input is the user's question, and the output is the question data sent to the server.

[0557] Step 4:

[0558] The server receives the user's question and sends it as a prompt to the generative AI model. The server converts the user's question into an appropriate prompt sentence and sends it to the generative AI model. The input is the user's question data, and the output is the prompt sentence sent to the generative AI model.

[0559] Step 5:

[0560] The generative AI model generates an answer to the question and returns it to the server. The generative AI model receives a prompt and generates an appropriate answer from the information it has learned. The input is the prompt, and the output is the generated answer data.

[0561] Step 6:

[0562] The server uses the emotion engine to analyze the user's emotional state. The server sends the user's input data to the emotion engine to analyze the emotional state. The input is the user's input data, and the output is the emotional state data returned from the emotion engine.

[0563] Step 7:

[0564] The server adjusts how the answer is displayed based on the user's emotional state. The server adjusts how the answer from the generative AI model is displayed based on the emotional state data. The input is the emotional state data and the generated answer data, and the output is the adjusted answer data.

[0565] Step 8:

[0566] The server displays the optimized answer to the user. The server sends the adjusted answer data to the user's browser for display. The input is the adjusted answer data, and the output is the answer displayed in the user's browser.

[0567] (Application example 2)

[0568] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0569] Conventional information search systems in factories have the problem that it is difficult for workers to quickly and accurately obtain the information they need. In addition, information provided does not reflect the emotional state of the workers, which often leaves them feeling confused and stressed. This can lead to reduced work efficiency and a negative impact on productivity.

[0570] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0571] In this invention, the server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for analyzing user emotions using an emotion recognition engine, means for adjusting the display method of information sources based on the user's emotions, means for being installed on robots in the factory, and means for the generative AI to generate and display answers when workers input questions in natural language. This allows workers to simply input questions in natural language, with the generative AI providing quick and accurate answers, and the emotion recognition engine displaying information according to the worker's emotions.

[0572] "Generative AI" is artificial intelligence that uses natural language processing technology to generate appropriate answers for input text.

[0573] An "emotion recognition engine" is software that analyzes emotions from user input, facial expressions, voice, etc., and identifies the user's emotional state.

[0574] A "search box" is an input field on a website where users can enter questions or keywords in natural language.

[0575] An "information source" is the source of information, such as data, literature, or web pages, that generative AI refers to when generating answers.

[0576] A "factory robot" is an automated mechanical device installed to assist with work in a factory, and may be equipped with a display and voice input functions.

[0577] "Natural language" refers to the language form used in everyday life, and is free text that does not depend on a specific format or structure.

[0578] "User emotion" refers to the psychological state that the user shows when making inputs or operations, and includes emotions such as confusion, joy, and anger.

[0579] "Source presentation method" refers to the format and style in which answers and information provided by generative AI are presented to users.

[0580] "Workers" are employees who work in factories and are responsible for searching for information and operating machines.

[0581] An "answer" is a response or explanation that a generative AI generates in response to a user's question.

[0582] A system for implementing the present invention includes an application installed on a robot in a factory. A specific embodiment of this system will be described below.

[0583] System configuration

[0584] The system consists of the following main components:

[0585] 1. Generative AI model: Uses natural language processing technology to generate appropriate answers to user questions.

[0586] 2. Emotion recognition engine: Analyzes emotions from user input, facial expressions, voice, etc., and identifies the user's emotional state.

[0587] 3. Search box: An input field on a website where users can enter questions or keywords in natural language.

[0588] 4. Information source: The source of information such as data, literature, and web pages that generative AI refers to when generating answers.

[0589] 5. Factory robots: These are automated machinery and devices installed to assist with work in factories, equipped with displays and voice input capabilities.

[0590] Program processing

[0591] The server processes and calculates data in the following steps:

[0592] 1. Question input: The worker types or speaks a question to the robot. The entered question is sent to the server via a search box.

[0593] 2. Generative AI answer generation: The server uses the OpenAI API to send the question to a generative AI model, which generates an appropriate answer.

[0594] 3. Emotion Recognition: The emotion recognition engine analyzes emotions from the worker's input and voice to identify their emotional state.

[0595] 4. Answer Adjustment: The generated answer is adjusted based on the emotional state identified by the emotion recognition engine. For example, if the worker is confused, a detailed explanation is added to the answer.

[0596] 5. Display Answer: The adjusted answer is displayed on the robot's display.

[0597] Hardware and software used

[0598] Hardware: Factory robot (with display), microphone (for voice input)

[0599] Software: OpenAI API (generative AI model), EmotionRecognizer (emotion recognition engine)

[0600] Specific examples

[0601] For example, if a worker asks, "How do I operate this new machine?", the generative AI will explain how to operate it. If the worker is confused, the emotion recognition engine will detect this and add more detailed explanations.

[0602] Prompt Sentence Examples

[0603] How do I operate the new machine?

[0604] In this way, the system can improve work efficiency by providing relevant answers to workers' questions and adjusting the way information is displayed based on their emotions.

[0605] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0606] Step 1:

[0607] The user enters a question in natural language into the robot's search box. The entered question is sent from the robot's terminal to the server. The input data is in text format, such as "Please tell me how to operate a new machine."

[0608] Step 2:

[0609] The server sends the received question to the generative AI model. Specifically, it uses the OpenAI API to pass the question as a prompt to the generative AI model. The input data is the text question, and the output data is the answer returned by the generative AI model.

[0610] Step 3:

[0611] The server receives the answer returned by the generative AI model in text format, such as "Here's how to operate the new machine..."

[0612] Step 4:

[0613] The server analyzes the user's emotions using an emotion recognition engine. Specifically, it passes the user's input text or voice data to the emotion recognition engine to identify the user's emotional state. The input data is in text or voice format, and the output data is the user's emotional state (e.g., confusion, joy, anger, etc.).

[0614] Step 5:

[0615] The server adjusts the answer from the generative AI model based on the identified emotional state. For example, if the user is confused, it adds a detailed explanation to the answer. The input data are the answer from the generative AI model and the emotional state, and the output data is the adjusted answer.

[0616] Step 6:

[0617] The server sends the adjusted answer to the robot's terminal, which displays the received answer on its display. The input data is the adjusted answer, and the output data is the text displayed on the display.

[0618] Step 7:

[0619] The user checks the answers displayed on the robot's display and enters additional questions if necessary, which restarts the process from step 1.

[0620] Example 3

[0621] Next, a description will be given of a third embodiment of the third embodiment. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0622] Conventional information search systems lack a means for users to confirm the reliability of the answers they receive, making it difficult for users to determine the accuracy of the information provided. Furthermore, since the basis information for the answers provided by generative AI models is not made clear, users are unable to confirm the reliability of the information.

[0623] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0624] In this invention, the server includes means for training all content in a generative AI model and automatically learning new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for the generative AI model to generate answers to user questions, means for searching and displaying information sources that serve as the basis for the generated answers, and means for allowing users to check the accuracy of the answers using the basis information. This makes it easier for users to check the reliability of the generative AI model's answers and obtain more accurate information.

[0625] A "generative AI model" is a type of artificial intelligence that generates answers in natural language to questions from users.

[0626] A "prompt" is text containing a question or instruction that is input to a generative AI model.

[0627] "Evidence" refers to sources of information that support the accuracy of the answers provided by the generative AI model.

[0628] A "server" is a computer system that runs a generative AI model, processes user questions, and provides answers and supporting information.

[0629] A "terminal" is a device on which a user displays answers and evidence information provided by a server.

[0630] A "user" is someone who uses the system to ask questions and review the answers and supporting information from the generative AI model.

[0631] A "search box" is an input field placed at the top of a website for users to enter questions.

[0632] "Scroll Down" is the area located at the bottom of the web page where existing information is displayed.

[0633] This invention is a system that uses a generative AI model to generate answers to questions from users and provides information on the basis of the answers. Specific embodiments of this system are described below.

[0634] Server Roles

[0635] The server runs the generative AI model and processes questions from users. The server operates as follows:

[0636] 1. Prompt generation

[0637] The server receives a question from the user. For example, the user asks, "What is the latest sales strategy?"

[0638] The server inputs this question as a prompt into the generative AI model.

[0639] 2. Answer Generation

[0640] The server sends a prompt to the generative AI model, which then generates an answer such as "The latest sales strategy is ABC."

[0641] The server receives the generated response.

[0642] 3. Obtaining evidence information

[0643] The server searches for the source of the generated answer, for example, identifying the "Latest Updates" section of the "Sales Strategy Page."

[0644] The server sends the answer and the basis information together to the terminal.

[0645] Device Role

[0646] The device receives the answer of the generative AI model and its basis information sent from the server and displays it to the user. Specifically, it works as follows.

[0647] 1. Displaying answers and supporting information

[0648] The terminal receives the response and the basis information sent from the server.

[0649] The terminal displays the answer and the basis information to the user. For example, it displays the answer "The latest sales strategy is ABC" and the "Latest Updates" section of the "Sales Strategy Page."

[0650] User Roles

[0651] The user checks the generated AI model's answer and the underlying information displayed on the device. Specifically, it works as follows:

[0652] 1. Check your answers

[0653] The user checks the answer and the basis information displayed on the terminal.

[0654] The user can then determine the accuracy of the answer based on the displayed evidence. For example, the user can check the "Latest Updates" section of the "Sales Strategy Page" to determine whether the answer is accurate.

[0655] Specific examples

[0656] Prompt Sentence Examples

[0657] What is the latest sales strategy?

[0658] This system makes it easier for users to check the reliability of the answers of the generative AI model, enabling them to obtain more accurate information. By having the server, terminal, and user fulfill their respective roles, users can efficiently obtain information and check its accuracy. The flow of the identification process in Example 3 will be described with reference to FIG. 21.

[0659] Step 1:

[0660] The server receives a question from the user. When the user inputs a question such as "What is the latest sales strategy?", the question is sent to the server. The server inputs this question as a prompt sentence into the generative AI model. The input data is the user's question, and the output data is the prompt sentence.

[0661] Step 2:

[0662] The server sends a prompt to the generative AI model. The generative AI model analyzes the prompt and generates an answer such as "The latest sales strategy is ABC." The input data is the prompt, and the output data is the generated answer.

[0663] Step 3:

[0664] The server receives the generated answer. Then, it searches the source of the answer. For example, it identifies the "Latest Updates" section of the "Sales Strategy Page." The input data is the generated answer, and the output data is the underlying information.

[0665] Step 4:

[0666] The server sends the answer and the basis information together to the terminal. The input data is the generated answer and the basis information, and the output data is the data sent to the terminal.

[0667] Step 5:

[0668] The terminal receives the answer and the supporting information sent from the server. The terminal displays the answer and the supporting information to the user. For example, it displays the answer "The latest sales strategy is ABC" and the "Latest Updates" section of the "Sales Strategy Page." The input data is the data sent from the server, and the output data is the information displayed to the user.

[0669] Step 6:

[0670] The user checks the answer and supporting information displayed on the device. The user then determines the accuracy of the answer based on the displayed supporting information. For example, the user checks the "Latest Updates" section of the "Sales Strategy Page" and determines whether the answer is accurate. The input data is the information displayed on the device, and the output data is the user's judgment.

[0671] (Application example 3)

[0672] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0673] Conventional generative AI systems lacked sufficient means for users to check the accuracy of the generated answers. In particular, when factory robots received work instructions, it was difficult to check the source of the information that served as the basis for those instructions. Furthermore, the system lacked the functionality to display the basis information for the generated answers, making it impossible for users to check the reliability of the answers. This resulted in problems such as reduced work efficiency and accuracy.

[0674] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means. In this invention, the server includes: means for having a generative AI learn all content and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll; means for displaying the information source that serves as the basis for a work instruction when the factory robot receives the instruction; means for sending a prompt to the generative AI model and generating an answer; and means for extracting and displaying the basis information for the generated answer from a database. This allows the user to check the accuracy of the generated answer and improve the reliability of the work instructions for the factory robot.

[0675] "Generative AI" is a system that uses artificial intelligence technology to generate new information and answers.

[0676] A "prompt sentence" is a question or instruction sentence that is input to a generative AI.

[0677] A "factory robot" is a mechanical device used to automate work in a factory.

[0678] "Work instructions" are information that instructs a factory robot on the specific work that it should perform.

[0679] "Evidence" is a source of information that supports the accuracy of the answers provided by generative AI.

[0680] A "database" is a system that systematically organizes and stores information, and allows it to be searched and extracted as needed.

[0681] A "search box" is an input field on a website into which a user enters a search query.

[0682] The "scroll bottom" is an area located at the bottom of a web page that is displayed when the user scrolls.

[0683] "Means of learning" are methods and techniques for enabling generative AI to take in and understand new information.

[0684] "Display means" refers to the method or technique for visually presenting information to a user.

[0685] The system for implementing the present invention uses the following hardware and software.

[0686] Hardware used:

[0687] Factory robots (e.g., general robotic arms)

[0688] Display device (e.g., a tablet attached to the robot)

[0689] Software used:

[0690] Generative AI models (e.g., OpenAI's GPT-4)

[0691] Database management system (e.g., MySQL®)

[0692] Web server (e.g. Apache)

[0693] System Overview:

[0694] The server has a means to train the generative AI on all content and automatically learn new information. A large search box is placed at the top of the website, and existing information is displayed at the bottom of the scroll. When a factory robot receives a work instruction, it also has a means to display the source of the information that supports that instruction. It also has a means to send prompts to the generative AI model to generate an answer, and a means to extract and display the information that supports the generated answer from a database.

[0695] The specific process flow:

[0696] 1. Prompt generation: When a factory robot receives a work instruction, the server sends a prompt to the generative AI model, which includes the work content and related information.

[0697] Example: "Please tell me the assembly steps for part A."

[0698] 2. Answer generation: The generative AI model generates an answer based on the prompt.

[0699] Example: "The assembly procedure for part A is XYZ."

[0700] 3. Extraction of evidence: The server extracts the sources of evidence for the generated answer from the database.

[0701] Example: "Part A Section" in "Assembly Manual"

[0702] 4. Display of source: The extracted source is displayed on the display device of the factory robot, allowing the user to check the accuracy of the generated answer.

[0703] Examples:

[0704] When a factory robot receives the work instruction "Assemble part A," the generative AI model responds, "The assembly procedure for part A is XYZ." The source of information that supports this response (e.g., "Part A's section" in the "Assembly Manual") is displayed on the screen. The user can review this information and verify the accuracy of the work instructions.

[0705] This system allows factory managers and engineers to check the accuracy of work instructions and make corrections or verifications as necessary, thereby improving work efficiency and accuracy.

[0706] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[0707] Step 1:

[0708] The server receives work instructions from factory robots. The input includes the content of the work instructions (e.g., "Assemble part A"). The server analyzes the work instructions and generates a prompt sentence to send to the generative AI model. The output is the generated prompt sentence (e.g., "Please tell me the assembly procedure for part A").

[0709] Step 2:

[0710] The server sends the generated prompt to the generative AI model. The input includes the prompt. The generative AI model generates an answer based on the prompt. The output is the generated answer (e.g., "The assembly procedure for part A is XYZ").

[0711] Step 3:

[0712] The server analyzes the answer received from the generative AI model and extracts the source of information that supports the answer from a database. The input includes the generated answer. The server uses a database management system (e.g., MySQL) to search and extract the source of information related to the answer. The output is the source of information (e.g., "Section for Part A" in the "Assembly Manual").

[0713] Step 4:

[0714] The server displays the extracted evidence information on the display device of the factory robot. The input includes the evidence information. The server uses a web server (e.g., Apache) to send the information to the display device and display it visually. The output is the evidence information displayed on the display device.

[0715] Step 5:

[0716] The user checks the basis information displayed on the display device of the factory robot. The basis information displayed on the display device is included as input. Based on this information, the user checks the accuracy of the work instructions and makes corrections or checks as necessary. The output is the user's confirmation result.

[0717] This series of processes improves the accuracy and reliability of work instructions for factory robots, increasing work efficiency.

[0718] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0719] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0720] Another example of generative AI is Gemini (registered trademark) (Internet search engine). <url: https: gemini.google.com ?hl="ja">) are mentioned.

[0721] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0722] [Second embodiment]

[0723] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0724] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

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

[0726] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0727] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

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

[0729] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0730] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

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

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

[0733] In the smart glasses 214, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0734] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0735] "Example 1"

[0736] This invention builds a system in which all content is learned by a generative AI and new information is automatically learned. Specifically, all content on an internal company website, including text, images, and videos, is learned by the generative AI. By automatically learning newly added information, it becomes possible to always provide the latest information.

[0737] "Example 2"

[0738] A large search box will be placed at the top of the website. This search box will allow employees to search in natural language. For example, if they enter a question such as "What is the latest sales strategy?", the generative AI will generate an answer from the information it has learned and display it.

[0739] "Example 3"

[0740] Existing information is displayed at the bottom of the scroll. This allows employees to see the location of the source of information that supports the answer along with the search results. For example, if the generative AI answers, "The latest sales strategy is ABC," the location of the source of information that supports this answer (for example, the "Latest Updates" section of the "Sales Strategy Page") will be displayed. This allows employees to check the accuracy of the answer for themselves.

[0741] The processing flow of each embodiment will be described below.

[0742] "Example 1"

[0743] Step 1: Have the generative AI learn all of the content. Specifically, have the generative AI learn all of the content on the company's internal website, including text, images, and videos.

[0744] Step 2: Automatically learns newly added information, making it possible to always provide the latest information.

[0745] "Example 2"

[0746] Step 1: Place a large search box at the top of your website.

[0747] Step 2: Allow employees to search in natural language, for example by entering a question such as "What is the latest sales strategy?"

[0748] Step 3: The generative AI generates an answer from the learned information and displays it.

[0749] "Example 3"

[0750] Step 1: Display existing information at the bottom of the scroll.

[0751] Step 2: Along with the search results, employees can see the location of the source that supports the answer.

[0752] Step 3: For example, if the generative AI answers, "The latest sales strategy is ABC," the location of the source of the answer (for example, the "Latest Updates" section of the "Sales Strategy Page") will be displayed. This allows employees to check the accuracy of the answer themselves.

[0753] Example 1

[0754] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0755] With conventional information management systems, it took time for the content of internal company websites to be updated with the latest information, making it difficult for users to quickly obtain the information they needed. Furthermore, there was a lack of means to verify the accuracy of the information, making it difficult to ensure the reliability of the information users obtained.

[0756] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means. In this invention, the server includes means for having the generative artificial intelligence learn all information and automatically learn new information, means for collecting information from websites, means for preprocessing the collected information, means for having the generative artificial intelligence learn the preprocessed information, means for periodically checking for newly added information and automatically collecting, preprocessing, and learning, means for the user to input a prompt sentence to the generative artificial intelligence to obtain the latest information, means for placing a large search box at the top of the website, and means for displaying existing information below the scroll bar. This allows users to always quickly obtain the latest information and confirm the accuracy of that information.

[0757] "Generative AI" is AI that has the ability to generate new information based on collected data.

[0758] "Automatic learning of new information" is a process of periodically checking for newly added information and automatically collecting, preprocessing, and learning from it.

[0759] "Means of collecting information from websites" refers to technologies used to obtain content such as text, images, and videos on websites.

[0760] "Means for preprocessing collected information" refers to techniques for analyzing collected data and performing processes such as tokenization, normalization, resizing, and noise removal.

[0761] "Means for training generative AI to learn preprocessed information" refers to a technique for inputting preprocessed data into generative AI and training a model.

[0762] A "prompt sentence" is text containing a question or request that a user inputs to a generative artificial intelligence.

[0763] A "search box" is an input field on a website into which a user enters a search query.

[0764] "Means for displaying below the scroll" refers to a technique for displaying existing information at the bottom of a web page.

[0765] "Means for verifying the accuracy of information" are techniques that provide the location of the source of information in order to verify the reliability of the information generated.

[0766] This invention is a system that uses generative artificial intelligence to learn all the content of an in-house website and automatically learns newly added information. Specific embodiments of this system are described below.

[0767] Data collection

[0768] The server collects all content, including text, images, and videos, from the company's internal website. Specifically, it uses Apache HTTP Server to crawl each page of the website and obtains the HTML data. From the obtained HTML data, it uses BeautifulSoup to extract the URLs of the text, images, and videos.

[0769] Data Preprocessing

[0770] The server preprocesses the collected data. For text data, it uses a natural language processing library (e.g., NLTK or spaCy) to tokenize and normalize it. For image data, it uses OpenCV to resize and denoise it. For video data, it uses FFmpeg to extract frames and compress them.

[0771] Data Learning

[0772] The server inputs the preprocessed data into a generative artificial intelligence model for training. Machine learning frameworks such as TensorFlow and PyTorch are used for training. The text data is input into GPT-4, and the image data is input into DALL-E, to optimize the model.

[0773] Automatic learning of new information

[0774] The server periodically checks for new information and automatically collects, pre-processes, and learns from it. Specifically, a Cron job is set up to periodically scan the website for new content.

[0775] Providing information

[0776] Users can obtain the latest information by inputting a prompt to the generative AI model. For example, if a user inputs "Please tell me the latest product information," the server passes the prompt to the generative AI model, and the model returns the latest product information to the user.

[0777] Examples of concrete examples and prompts

[0778] Specific examples

[0779] If a user wants to know the latest news on an internal company website, they input a prompt sentence to the generative artificial intelligence model: "Tell me the latest news."

[0780] The server generates the latest news from the learned data and provides it to users.

[0781] Prompt Sentence Examples

[0782] "Please tell me the latest product information"

[0783] "Tell me about this month's events"

[0784] "Explain the new policy."

[0785] In this way, the user can easily obtain the latest information at all times.

[0786] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0787] Step 1: Data collection

[0788] The server collects all content, including text, images, and videos, from the company's internal website. Specifically, it uses Apache HTTP Server to crawl each page of the website and obtains the HTML data. From the obtained HTML data, it uses BeautifulSoup to extract the URLs of the text, images, and videos.

[0789] Enter: Website URL

[0790] Output: List of text, image, and video URLs

[0791] Step 2: Data Preprocessing

[0792] The server pre-processes the collected data.

[0793] Preprocessing text data:

[0794] What it does: The server uses NLTK to tokenize the text data, removing unnecessary symbols and spaces, and spaCy to normalize the text (e.g., lowercasing, removing stop words).

[0795] Input: Text data

[0796] Output: Preprocessed text data

[0797] Image data preprocessing:

[0798] What happens: The server uses OpenCV to resize the image data, apply a noise reduction filter, and unify the image format (e.g., convert to JPEG) if necessary.

[0799] Input: Image data

[0800] Output: Preprocessed image data

[0801] Video data preprocessing:

[0802] What it does: The server uses FFmpeg to extract and compress the video data frame by frame, and then unifies the video resolution.

[0803] Input: Video data

[0804] Output: Preprocessed video data

[0805] Step 3: Data training

[0806] The server inputs the preprocessed data into a generative artificial intelligence model for training. Machine learning frameworks such as TensorFlow and PyTorch are used for training. The text data is input into GPT-4, and the image data is input into DALL-E, to optimize the model.

[0807] Input: Preprocessed text data, image data, and video data

[0808] Output: A trained generative artificial intelligence model

[0809] Step 4: Automatically learning new information

[0810] The server periodically checks for new information and automatically collects, pre-processes, and learns from it. Specifically, a Cron job is set up to periodically scan the website for new content.

[0811] Input: Newly added website content

[0812] Output: An updated trained generative artificial intelligence model

[0813] Step 5: Provide information

[0814] Users can obtain the latest information by inputting a prompt to the generative AI model. For example, if a user inputs "Please tell me the latest product information," the server passes the prompt to the generative AI model, and the model returns the latest product information to the user.

[0815] Input: prompt statement

[0816] Output: Latest information

[0817] In this way, the user can easily obtain the latest information at all times.

[0818] (Application example 1)

[0819] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0820] Conventional content distribution systems have had difficulty providing users with the latest, personalized information. Furthermore, content generation based on user interests is performed manually, resulting in inefficiency and delays in providing the latest information. Furthermore, the generation of new articles based on news data is not automated, which means users are unable to receive consistent information.

[0821] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0822] In this invention, the server includes a means for having a generative AI learn all content and automatically learn new information, a means for placing a large search box at the top of the website, a means for displaying existing information below the scroll, a means for generating personalized content based on the user's interests, and a means for acquiring news data and having the generative AI generate new articles based on that data, making it possible to always provide users with the latest and personalized information.

[0823] "Generative AI" is artificial intelligence that learns content such as text, images, and videos, and automatically generates content based on new information.

[0824] "Automatic learning of new information" is a function that automatically learns newly added information and always provides the latest information.

[0825] A "search box" is an input field where users can enter keywords to search for information.

[0826] "Scroll Down" is a function that displays existing information at the bottom of a web page.

[0827] "Personalized content" is content that is individually optimized based on a user's interests and browsing history.

[0828] "News data acquisition" is the process of collecting the latest news information from external news sources.

[0829] "Article generation" is the process in which generative AI creates new articles based on collected news data.

[0830] A system for implementing this invention is configured as follows: The server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for generating personalized content based on user interests, and means for acquiring news data and having the generative AI generate new articles based on that data.

[0831] Hardware and software used

[0832] Hardware: Servers, smartphones

[0833] Software: Python, Transformers library (Hugging Face), Requests library

[0834] Data processing and calculation

[0835] 1. Data Acquisition: The server acquires the latest news data using the news API. For example, data is collected from the news API using a URL such as "https: / / newsapi.org / v2 / top-headlines?country=jp&apiKey=YOUR_API_KEY".

[0836] 2. Data training: The acquired news data is trained into a generative AI model (e.g., GPT-2). This gives the generative AI the ability to generate new content based on the latest news information.

[0837] 3. Content generation: Based on the learned data, new articles and content are generated to serve to users. The generated content is personalized based on the user's interests and browsing history.

[0838] Specific examples

[0839] For example, if a user is interested in "technology," the server will collect the latest technology-related news, and the generative AI will generate new articles based on that news. This allows users to always get the latest technology information.

[0840] Prompt Sentence Examples

[0841] "Generate new articles based on the latest technology news."

[0842] This way, users can stay up to date with the latest information and enjoy personalized content based on their interests.

[0843] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0844] Step 1:

[0845] The server uses a news API to obtain the latest news data.

[0846] Input: News API URL

[0847] Output: Latest news data (JSON format)

[0848] Specific operation: The server sends an HTTP request to the news API endpoint and receives the latest news data in response.

[0849] Step 2:

[0850] The server performs preprocessing on the acquired news data in order to input it into the generative AI model.

[0851] Input: Latest news data (JSON format)

[0852] Output: Preprocessed text data

[0853] Specific operation: The server extracts article titles and descriptions from the news data and formats them as text data.

[0854] Step 3:

[0855] The server inputs the preprocessed text data into a generative AI model to generate new articles.

[0856] Input: Preprocessed text data

[0857] Output: The new article generated.

[0858] How it works: The server inputs text data into a generative AI model (e.g., GPT-2), which then generates new articles.

[0859] Step 4:

[0860] The server personalizes the new articles generated based on the user's interests.

[0861] Input: Generated new articles, user interest data

[0862] Output: personalized article

[0863] Specific operation: The server filters or highlights generated articles based on the user's interest data (such as browsing history and past search keywords).

[0864] Step 5:

[0865] The server delivers personalized articles to the user's device.

[0866] Input: Personalized Article

[0867] Output: The article as it appears on the user's device

[0868] Specific operation: The server sends the personalized article to the user's smartphone or web browser for display.

[0869] Step 6:

[0870] Users can view the published article and use the search box to find additional information if needed.

[0871] Input: User's search keyword

[0872] Output: Search results

[0873] What it does: The user enters keywords into the search box, and the server searches for relevant information based on those keywords and returns the results to the user.

[0874] Example 2

[0875] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0876] With conventional information search systems, it was difficult for users to quickly and accurately obtain the information they needed. Furthermore, there was a lack of means to verify the basis of the information, making it difficult to obtain reliable information. Furthermore, the way existing information was displayed was inconvenient, impairing user convenience.

[0877] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes a means for having the generative AI model learn all information and automatically learn new information, a means for placing a large search box at the top of the website, a means for a user to input a question in natural language, a means for transmitting the input question to the generative AI model, a means for receiving and displaying an answer from the generative AI model, and a means for displaying existing information below the scroll. This allows a user to input a question in natural language and receive a quick and accurate answer from the generative AI model. In addition, since the basis information for the answer is also provided at the same time, the user can confirm reliable information. Furthermore, displaying existing information below the scroll improves user convenience.

[0878] A "generative AI model" is a type of artificial intelligence that uses natural language processing to generate answers to user questions.

[0879] A "prompt sentence" is a textual sentence that sends a question entered by a user to a generative AI model.

[0880] A "server" is a computer system that receives requests from a user, sends prompts to a generative AI model, and returns the generated answers to the user.

[0881] A "search box" is an input field placed on the top page of a website where users can enter questions in natural language.

[0882] "Natural language" refers to the language form that users use on a daily basis, and is free-flowing text that is not bound by any particular format or structure.

[0883] An "answer" is information that a generative AI model generates in response to a user's question.

[0884] "Evidence" refers to the location of the information source that the generative AI model referenced when generating the answer.

[0885] "Existing information" is information that already exists on the website and is displayed in a format that users can scroll through.

[0886] "Automatic learning" is the process by which a generative AI model continually learns new information, improving the model's accuracy and knowledge.

[0887] This invention is a system that allows users to input questions in natural language and obtain quick and accurate answers using a generative AI model. Specific embodiments of this system are described below.

[0888] Hardware and software used

[0889] server

[0890] The server receives requests from users, sends prompts to the generative AI model, and returns the generated answers to the users. The server uses the following hardware and software:

[0891] Hardware: A server machine with a powerful processor, sufficient memory, and a fast network connection

[0892] Software: Web server software (e.g., Apache, Nginx), programming languages ​​(e.g., Python, JavaScript), APIs for generative AI models (e.g., OpenAI's GPT-4)

[0893] Terminal

[0894] The device used by the user is a computer or smartphone with a web browser, which the user uses to access the website and enter a question into a search box.

[0895] Data processing and calculation

[0896] User Input

[0897] Users type a question into a large search box on the homepage of a website, for example, "What are the latest sales strategies?"

[0898] Receiving the server and generating a prompt

[0899] The server receives the question entered by the user and converts it into a prompt, which is simply "What is the latest sales strategy?"

[0900] Send to generative AI model and generate answer

[0901] The server sends the generated prompt to the generative AI model, which analyzes the received prompt and generates the optimal answer based on the information it has learned. For example, the following answer might be generated:

[0902] "The latest sales strategies include strengthening digital marketing, improving targeting accuracy through customer data analysis, and establishing a remote work environment."

[0903] Server display process

[0904] The server receives the answers returned by the generative AI model and displays them on the website's search results page, allowing users to see the answers to their questions.

[0905] Specific examples

[0906] For example, if a user types the question "What is the latest sales strategy?" into a search box, the following prompt will be sent to the generative AI model:

[0907] What are the latest sales strategies?

[0908] The generative AI model analyzes this prompt and generates the following answer:

[0909] The latest sales strategies include strengthening digital marketing, improving targeting accuracy through analysis of customer data, and establishing a remote work environment.

[0910] The server receives this answer and displays it on the website's search results page, where users can view it and use it to improve their business.

[0911] In this way, the invention allows users to enter questions in natural language and get fast and accurate answers using generative AI models.

[0912] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0913] Step 1:

[0914] A user accesses the homepage of a website and enters a question into a large search box, for example, "What is the latest sales strategy?" The entered question is sent from the user's device to the server.

[0915] Step 2:

[0916] The server receives questions sent by users. It saves the received questions in text format and prepares them to be sent to the generative AI model as prompts. The input is the user's question, and the output is the prompt.

[0917] Step 3:

[0918] The server sends the received question as a prompt to the generative AI model. The prompt is "What is the latest sales strategy?" The input is the prompt, and the output is a request to the generative AI model.

[0919] Step 4:

[0920] The generative AI model receives and analyzes the prompt sent from the server. The model uses an internal neural network to generate the optimal answer from the learned information. The input is the prompt, and the output is the generated answer.

[0921] Step 5:

[0922] The generative AI model sends the generated answer back to the server. For example, it might generate an answer like, "The latest sales strategies include strengthening digital marketing, improving targeting accuracy by analyzing customer data, and establishing a remote work environment." The input is the generated answer, and the output is the response to the server.

[0923] Step 6:

[0924] The server receives the answers returned by the generative AI model, embeds the received answers into an HTML template, and generates a search results page for the website. The input is the generated answers, and the output is the HTML search results page.

[0925] Step 7:

[0926] The server returns the generated search results page to the user's device. The user can then view the search results page in their browser and read the displayed answers. The input is the HTML search results page, and the output is the answer displayed in the user's browser.

[0927] In this way, users can enter questions in natural language and get fast and accurate answers using generative AI models.

[0928] (Application example 2)

[0929] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0930] There is a problem in that it is difficult for workers in factories to quickly and accurately obtain the information they need. In particular, when workers want to know the parts and procedures required for the next process, conventional methods can take time and reduce efficiency. In addition, there is a lack of means to confirm the accuracy of the information, so there is a risk that work will proceed based on incorrect information.

[0931] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for having a generative AI learn all information and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll; means for factory workers to input questions in natural language and for the generative AI to generate and display the optimal answer; means for inputting questions using a smartphone or tablet; and means for the generative AI to provide information about work in the factory. This enables workers to quickly and accurately obtain the information they need, improving work efficiency.

[0932] "Information" refers to data and knowledge related to operations within a factory.

[0933] "Generative AI" refers to artificial intelligence that uses natural language processing to generate answers to questions.

[0934] "Automatic learning" refers to the process of continually learning new data and information and updating its knowledge base.

[0935] A "search box" is an input field where a user can enter a question in natural language.

[0936] "Scroll Down" refers to the area at the bottom of a web page that becomes visible when the user scrolls down.

[0937] "Workers" means employees who perform work within a factory.

[0938] "Natural language" refers to text written in a language form that humans use on a daily basis.

[0939] A "smartphone" is a mobile phone that can connect to the Internet and use applications.

[0940] A "tablet" is a highly portable computing device that has a touchscreen.

[0941] A "best answer" is an answer that provides the most relevant and accurate information for a question.

[0942] A "source" is a place where the data or knowledge that forms the basis of an answer is stored.

[0943] "Accuracy" means that the information or answer is factual and free from errors.

[0944] The system for implementing this invention is configured as follows: First, the server uses generative AI to learn all information and automatically learns new information. This allows the latest information to be maintained at all times. Next, a large search box is placed at the top of the website, allowing users to enter questions in natural language. Furthermore, existing information is displayed at the bottom of the page, making it easy for users to access past information.

[0945] Factory workers use smartphones or tablets to input questions. For example, when a question such as "What parts are needed for the next process?" is input, the generative AI generates the optimal answer to the question and displays it to the worker. The generative AI generates answers based on data it has learned in advance to provide information about work within the factory.

[0946] The hardware used includes smartphones and tablets, and the software uses Python and the OpenAI API. Data processing and data calculation involves receiving questions in natural language from users and generating answers using OpenAI's generative AI model (GPT-3).

[0947] As a concrete example, if a worker in a factory asks, "What parts are needed for the next process?", the following prompt sentence is input into the generative AI model.

[0948] Example prompt sentence:

[0949] Factory operation questions: What part is needed for the next operation?

[0950] answer:

[0951] Based on this prompt, the generative AI generates a list of parts required for the next process and provides it to the worker, allowing the worker to quickly and accurately obtain the necessary information and improve work efficiency.

[0952] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0953] Step 1:

[0954] A user uses a smartphone or tablet to enter a question in natural language into a search box at the top of a website. The entered question is sent to the server. The input data is the user's question, and the output data is the question sent to the server.

[0955] Step 2:

[0956] The server sends the received question to the generative AI model. Specifically, the question is input to the generative AI model (GPT-3) as a prompt. The input data is the user's question, and the output data is the prompt sent to the generative AI model.

[0957] Step 3:

[0958] A generative AI model generates an optimal answer based on a prompt. Specifically, it generates an answer to a question and returns that answer to the server. The input data is the prompt, and the output data is the generated answer.

[0959] Step 4:

[0960] The server sends the answers received from the generative AI model to a smartphone or tablet for display to the user. The input data is the generated answer, and the output data is the answer sent to the user's device.

[0961] Step 5:

[0962] The user checks the response displayed on their smartphone or tablet and obtains the necessary information. Specifically, they check information such as the parts list and procedures required for the next process. The input data is the response sent from the server, and the output data is the information obtained by the user.

[0963] Through these processing steps, users can input questions in natural language and receive quick and accurate answers using generative AI models, thereby improving work efficiency within factories.

[0964] Example 3

[0965] Next, a description will be given of Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0966] In conventional information retrieval systems, users lacked the means to verify the basis of the answers they received, making it difficult to determine the reliability of the answers. Another problem was that it took a lot of time and effort to identify the location of the information source. Furthermore, when new information was added, the system sometimes failed to automatically learn and provide the latest information.

[0967] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[0968] In this invention, the server includes: means for training a generative AI model to learn all information and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll bar; means for sending a prompt entered by a user to the server; means for the server to send the prompt to the generative AI model and generate an answer; means for the server to search an internal database for information sources related to the answer; means for the server to send the search results and the answer to a terminal; and means for the terminal to display the search results and the answer to the user. This allows the user to easily check the basis for the generated answer, increasing the reliability of the answer. Furthermore, when new information is added, the system automatically learns and can provide the latest information.

[0969] A "generative AI model" is an artificial intelligence algorithm that generates answers based on prompts entered by the user.

[0970] A "prompt" is a natural language question or instruction that a user enters to retrieve information.

[0971] The "server" is a computer system that receives prompts from users, sends them to a generative AI model to generate answers, and then searches for relevant information sources and sends the results to the terminal.

[0972] A "terminal" is a device through which a user can enter prompts and display answers and related information sent by the server.

[0973] An "internal database" is a database that stores information sources held by a company or organization, and is used by the server to search for the basis of an answer.

[0974] A "search box" is an input field placed at the top of a website for users to enter a prompt phrase.

[0975] The "scroll bottom" is an area located at the bottom of a web page for displaying existing information.

[0976] An "answer" is information that a generative AI model generates based on a prompt.

[0977] A "source" is the data or document that forms the basis for the generated answer.

[0978] "Automatic learning" is the process by which a generative AI model continuously learns new information, enabling it to provide up-to-date information.

[0979] This invention is a system in which a generative AI model generates an answer based on a prompt sentence entered by a user and displays the information source that forms the basis of the answer. Specific embodiments of this system are described below.

[0980] First, the user inputs a prompt sentence into a search box installed on the web browser of the terminal. For example, the user inputs a prompt sentence such as "What is the latest sales strategy?"

[0981] The device sends the input prompt to the server using an HTTP POST request. The server then sends the received prompt as an API request to the generative AI model. For example, OpenAI's GPT-4 can be used as the generative AI model.

[0982] The generative AI model generates an answer based on the prompt sentence. For example, it generates the answer "The latest sales strategy is ABC." The server receives this generated answer and searches an internal database for information sources related to that answer. The internal database stores information sources held by companies and organizations, and searches are performed using SQL queries.

[0983] The server sends the search results and the generated answer in JSON format to the terminal. The terminal displays the received answer and the search results to the user. For example, along with the answer "The latest sales strategy is ABC," it displays a link to the "Latest Updates" section of the "Sales Strategy Page." The user can click this link to view more detailed information.

[0984] This system allows users to easily check the basis of the generated answers, increasing the reliability of the answers.In addition, when new information is added, the system automatically learns and can provide the latest information.

[0985] As a concrete example, if a user inputs the prompt sentence "What is the latest sales strategy?", the server uses the generative AI model to generate the answer "The latest sales strategy is ABC", and searches the internal database to identify the "Latest Updates" section of the "Sales Strategy Page". Ultimately, the answer "The latest sales strategy is ABC" is displayed on the user's device, along with a link to the "Latest Updates" section of the "Sales Strategy Page".

[0986] In this way, the user can easily check the basis of the information provided by the generative AI model. The flow of the identification process in the third embodiment will be described with reference to FIG.

[0987] Step 1:

[0988] The user enters a prompt statement.

[0989] The user inputs a prompt sentence, "What is the latest sales strategy?", into a search box installed on the web browser of the terminal. The input prompt sentence becomes the input data of the terminal.

[0990] Step 2:

[0991] The terminal sends the prompt to the server.

[0992] The terminal sends the input prompt text to the server using an HTTP POST request. The input data is the prompt text, and the output data is the prompt text sent to the server.

[0993] Step 3:

[0994] The server sends a prompt to the generative AI model, which generates an answer.

[0995] The server sends the received prompt sentence to the generative AI model as an API request. The generative AI model generates an answer based on the prompt sentence. The input data is the prompt sentence, and the output data is the generated answer. For example, the generated answer is "The latest sales strategy is ABC."

[0996] Step 4:

[0997] The server searches an internal database for sources relevant to the answer.

[0998] The server searches an internal database for information sources related to the generated answer. The internal database stores information sources owned by the company or organization, and performs a search using an SQL query. The input data is the generated answer, and the output data is the search results for the related information sources. For example, the "Latest Updates" section of the "Sales Strategy Page" is identified.

[0999] Step 5:

[1000] The server sends the search results and answers to the device.

[1001] The server sends the generated answers and search results to the terminal in JSON format. The input data are the generated answers and search results, and the output data is the JSON data sent to the terminal.

[1002] Step 6:

[1003] The device displays the search results and answers to the user.

[1004] The terminal displays the received answers and search results to the user. The input data is the JSON data received from the server, and the output data is the answer displayed to the user and a link to the information source. For example, along with the answer "The latest sales strategy is ABC," a link to the "Latest Updates" section of the "Sales Strategy Page" is displayed. The user can click this link to view more detailed information.

[1005] (Application example 3)

[1006] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1007] Conventional generative AI systems lacked sufficient means for users to verify the basis of the answers they received, making it difficult for factory workers, in particular, to obtain information quickly and accurately on-site. Furthermore, there was a lack of means to verify the reliability of the information provided by generative AI, making it difficult for users to verify the accuracy of the answers themselves.

[1008] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[1009] In this invention, the server includes means for having the generative AI learn all information and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for the generative AI to provide an answer when a user inputs a question and display the location of the source of the evidence, and means for the generative AI to provide an answer when a factory worker asks a question to the robot and display the location of the source of the evidence. This allows users to check the accuracy of the generative AI's answer using the evidence information, and enables factory workers to obtain information quickly and accurately on site.

[1010] "Information" is a general term for the data and knowledge that generative AI learns and provides to users.

[1011] "Generative AI" is an artificial intelligence system that generates answers in natural language to users' questions.

[1012] "Automatic learning" is the process by which generative AI automatically absorbs and learns new information.

[1013] A "search box" is an input field on a website where users can enter questions or keywords.

[1014] "Scroll Down" is the area at the bottom of a web page that becomes visible when the user scrolls down.

[1015] "Evidence" is a source of information that supports the accuracy of the answers provided by generative AI.

[1016] "Factory workers" refers to employees who work in factories.

[1017] A "robot" is an automated mechanical device that provides work assistance and information within a factory.

[1018] To implement this invention, all information must first be learned by a generative AI. A generative AI is an artificial intelligence system that generates answers in natural language to user questions and has the ability to automatically learn new information. This makes it possible to always provide the latest information.

[1019] Next, a large search box is placed on the homepage of the website. Users can make inquiries to the generative AI by entering questions or keywords into this search box. When a user enters a question, the generative AI generates an answer to that question and displays the location of the source of the answer.

[1020] Additionally, a new feature will be added that displays existing information at the bottom of the scroll, allowing users to easily refer to supporting information to verify the accuracy of answers provided by generative AI.

[1021] To enable factory workers to quickly and accurately obtain information on-site, this system will be installed on robots within the factory. When a factory worker asks the robot a question, generative AI will provide an answer and display the location of the source of the answer. This will enable workers to quickly obtain the information they need on-site and improve work efficiency.

[1022] The implementation of this system uses the following hardware and software. For hardware, we need factory robots and displays. For software, we use the OpenAI API, Python, and Tkinter (GUI library). We use the OpenAI API to obtain answers from a generative AI model (e.g., GPT-3), and we use Python and Tkinter to build the user interface.

[1023] For example, if a factory worker asks, "What is the latest quality control method?", the generative AI will respond, "The latest quality control method is XYZ." The location of the source of information that supports the answer (for example, the "Latest Updates" section of the "Quality Control Page") is displayed at the bottom of the scroll. This allows the worker to confirm the accuracy of the answer and improve their work efficiency.

[1024] An example prompt is:

[1025] Q: What are the latest quality control techniques?

[1026] In this way, users can verify the accuracy of generative AI's answers with supporting information, enabling factory workers to obtain information quickly and accurately on-site.

[1027] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1028] Step 1:

[1029] A user types a question into a website search box.

[1030] Input: User question (e.g., "What are the latest quality control techniques?")

[1031] Output: The user's question is sent to the server.

[1032] What happens: A user types a question into the search box and clicks the submit button.

[1033] Step 2:

[1034] The server receives the user's question and sends it to the generative AI model.

[1035] Input: User question

[1036] Output: Prompt sentence for generative AI model

[1037] How it works: The server sends the user's question in text format to a generative AI model (e.g., GPT-3).

[1038] Step 3:

[1039] A generative AI model generates an answer based on the prompt.

[1040] Input: Prompt statement (e.g., "What are the latest quality control methods?")

[1041] Output: Answer from the generative AI model (e.g., "The most current quality control method is XYZ")

[1042] What it does: The generative AI model analyzes the prompt and generates an appropriate response.

[1043] Step 4:

[1044] The server receives the answer from the generative AI model and searches for supporting information.

[1045] Input: Answer from a generative AI model

[1046] Output: Location of answer and supporting information

[1047] Specific operation: The server receives the answer from the generative AI model and searches the database for the source of information that supports that answer.

[1048] Step 5:

[1049] The server returns the answer and supporting information to the user.

[1050] Input: Location of answer and supporting information

[1051] Output: Answer and rationale displayed on the user's device

[1052] Specific operation: The server sends the answer and supporting information to the user's device and displays it at the bottom of the scroll.

[1053] Step 6:

[1054] The user checks the answer and supporting information.

[1055] Input: Answer and rationale displayed on the user's device

[1056] Output: User understanding and confirmation

[1057] What happens: The user reviews the evidence displayed below the scroll bar and evaluates the accuracy of the answer.

[1058] Step 7:

[1059] A factory worker types a question into the robot.

[1060] Input: Factory worker question (e.g., "What are the latest quality control techniques?")

[1061] Output: The factory worker's question is sent to the robot.

[1062] How it works: A factory worker types a question into the robot's interface and clicks the send button.

[1063] Step 8:

[1064] The robot sends a question to the server and gets an answer based on a generative AI model.

[1065] Input: Factory worker question

[1066] Output: Answers and supporting information from the generative AI model

[1067] Specific operation: The robot sends a question to the server and obtains an answer and supporting information from the generative AI model.

[1068] Step 9:

[1069] The robot displays the answer and supporting information to the factory worker.

[1070] Input: Answer from the generative AI model and supporting information

[1071] Output: Answers and supporting information displayed on the factory worker's terminal

[1072] Specific operation: The robot displays the answer and supporting information on the factory worker's device, scrolling down below.

[1073] Step 10:

[1074] Factory workers will review the responses and supporting information.

[1075] Input: Answer and supporting information displayed on the factory worker's terminal

[1076] Output: Factory worker understanding and confirmation

[1077] What happens: The factory worker reviews the supporting information displayed at the bottom of the scroll and evaluates the accuracy of the answer.

[1078] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1079] "Example 1"

[1080] One embodiment of the present invention is a system that incorporates an emotion engine. This system includes a means for having a generative AI learn all content and automatically learn new information, a means for placing a large search box at the top of a website, and a means for displaying existing information below the scroll bar. It also includes an emotion engine that recognizes the user's emotions. Specifically, the emotion engine recognizes the user's emotional state when using a website, and the generative AI adjusts search results based on that emotional state. For example, if the user expresses joy, the generative AI will prioritize displaying information related to joy.

[1081] "Example 2"

[1082] The emotion engine also adjusts how information sources are displayed based on the user's emotions. For example, if the user expresses confusion, the generative AI will display the source's detailed explanation more clearly. This allows the user to receive the most appropriate information based on their emotional state.

[1083] "Example 3"

[1084] Furthermore, this system allows users to check the accuracy of answers using supporting information. Specifically, the location of the source of information that supports the answer provided by the generative AI is displayed, allowing users to check its accuracy for themselves. This allows users to receive the most appropriate information based on their emotional state and to confirm the reliability of that information.

[1085] The processing flow of each embodiment will be described below.

[1086] "Example 1"

[1087] Step 1: A user opens a website.

[1088] Step 2: The user enters a search query in natural language into the search box.

[1089] Step 3: The emotion engine recognizes the user's emotional state.

[1090] Step 4: The generative AI receives input from the emotion engine and refines the search results.

[1091] Step 5: The generative AI displays the adjusted search results to the user.

[1092] "Example 2"

[1093] Step 1: A user opens a website.

[1094] Step 2: The user enters a search query in natural language into the search box.

[1095] Step 3: The emotion engine recognizes the user's emotional state.

[1096] Step 4: The generative AI receives input from the emotion engine and adjusts how the source is presented.

[1097] Step 5: The generative AI displays the adjusted source information to the user.

[1098] "Example 3"

[1099] Step 1: A user opens a website.

[1100] Step 2: The user enters a search query in natural language into the search box.

[1101] Step 3: The generative AI generates search results and identifies the location of the source of information that supports them. Step 4: The generative AI displays the search results and the location of the source of information to the user.

[1102] Step 5: The user locates the displayed source and verifies the accuracy of the answer.

[1103] Example 1

[1104] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1105] On conventional websites, it was difficult for users to quickly and accurately obtain the information they needed. Furthermore, there was a problem in that the information provided did not correspond to the user's emotional state, resulting in a poor user experience. Furthermore, there was a lack of a system that could automatically learn new information and constantly provide the latest information.

[1106] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1107] In this invention, the server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means including an emotion engine that recognizes the user's emotional state, and means for the generative AI to adjust search results based on the user's emotional state. This allows users to quickly and accurately obtain the information they need, makes it possible to provide information according to their emotional state, and always provides the latest information.

[1108] "Generative AI" is artificial intelligence that generates new content based on collected data.

[1109] "Means for automatically learning new information" refers to a means that has the function of automatically collecting new added data and having the generative AI learn it.

[1110] "A means for placing a large search box at the top of a website" refers to a means for placing a large search box on the top page of a website so that users can easily perform searches.

[1111] The "means for displaying existing information below the scroll" is a means for displaying information below the web page so that the user can view the existing information by scrolling.

[1112] The "emotion engine that recognizes the user's emotional state" is an engine that analyzes emotions from the user's facial expressions and text input and recognizes their state.

[1113] "Means for the generative AI to adjust search results based on the emotional state" refers to means for the generative AI to adjust the search results displayed according to the user's emotional state.

[1114] MODE FOR CARRYING OUT THE INVENTION

[1115] This invention is a system that uses generative AI to learn all content and automatically learns new information. Specifically, the generative AI learns all content on an internal company website, including text, images, and videos. By automatically learning newly added information, it is possible to always provide the latest information.

[1116] Hardware and software used

[1117] The server uses the following hardware and software:

[1118] Web scraping tools: BeautifulSoup, Scrapy

[1119] Natural language processing tools: NLTK, SpaCy

[1120] Image processing libraries: OpenCV, PIL

[1121] Video processing library: FFmpeg

[1122] Generation AI: GPT-4, DALL-E

[1123] Emotion engine: Affectiva, IBM Watson Tone Analyzer

[1124] Program processing

[1125] The server collects all content from the internal company website, including text, images, and videos. A web scraping tool is used to analyze the HTML structure of the web page and extract the required data. For example, text data is extracted from HTML tags, and image data is obtained by obtaining image URLs.

[1126] The collected data is tokenized, normalized, and stop words removed using natural language processing tools, image data is resized and formatted using an image processing library, and video data is frame-extracted and compressed using a video processing library.

[1127] The preprocessed data is then trained by a generative AI, which updates its model based on all collected content and automatically learns new information. For example, text data is fed into GPT-4, and image data is fed into DALL-E.

[1128] The device recognizes the user's emotional state when using a website. It uses an emotion engine to analyze emotions from the user's facial expressions and text input. For example, the device captures the user's facial expressions through a webcam, and the emotion engine analyzes them.

[1129] The server uses a generative AI to tailor search results based on the user's emotional state, obtained from the emotion engine. For example, if the user expresses joy, the generative AI will prioritize information related to joy.

[1130] Specific examples

[1131] A user visits an internal company website and types "new product reviews" into the search box. The device captures the user's facial expressions with a webcam, and the emotion engine recognizes the emotion of happiness from the user's facial expression. The server receives this emotion data and inputs a prompt to the generative AI: "If the user is expressing happiness, please prioritize displaying positive reviews." Based on this prompt, the generative AI prioritizes displaying positive reviews.

[1132] Example prompt sentence:

[1133] "Show reviews of new products. If users are expressing joy, prioritize positive reviews."

[1134] In this way, a system is constructed in which the server, terminals, and users work together to provide the latest information that is tailored to their emotions.

[1135] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1136] Step 1: Data collection

[1137] The server collects all content, including text, images, and videos, from internal company websites.

[1138] Input: Internal website URL

[1139] Specific operation: The server uses a web scraping tool (e.g., BeautifulSoup, Scrapy) to analyze the HTML structure of the web page and extract the necessary data. For example, text data is extracted from HTML tags, and image data is obtained by obtaining image URLs.

[1140] Output: Collected text, image, and video data

[1141] Step 2: Data Preprocessing

[1142] The server pre-processes the collected data.

[1143] Input: Collected text, image, and video data

[1144] Specific operation: The server uses natural language processing tools (e.g., NLTK, SpaCy) to tokenize, normalize, and remove stop words from text data. Image data is resized and formatted using an image processing library (e.g., OpenCV, PIL). Video data is extracted and compressed using a video processing library (e.g., FFmpeg).

[1145] Output: Preprocessed text, image, and video data

[1146] Step 3: Data training

[1147] The server trains the generative AI on the preprocessed data.

[1148] Input: Preprocessed text, image, and video data

[1149] Specific operation: The server inputs data into generative AI (e.g., GPT-4, DALL-E) and updates the model. For example, text data is input into GPT-4, and image data is input into DALL-E. When new information is added, the learning process is automatically restarted.

[1150] Output: An updated generative AI model

[1151] Step 4: Emotion Recognition

[1152] The terminal recognizes the emotional state of the user when using the website.

[1153] Input: User facial expression data, text input

[1154] Specific operation: The device uses an emotion engine (e.g., Affectiva, IBM Watson Tone Analyzer) to analyze emotions from the user's facial expressions and text input. For example, the device captures the user's facial expressions through a webcam, and the emotion engine analyzes them.

[1155] Output: User's emotional state data

[1156] Step 5: Refine your search results

[1157] The server uses generative AI to adjust search results based on the user's emotional state obtained from the emotion engine.

[1158] Input: User's emotional state data, user's search query

[1159] Specific operation: The server receives emotion data from the emotion engine and inputs prompts to the generative AI. For example, the server inputs a prompt such as "If the user is expressing joy, please prioritize displaying positive reviews" to adjust the search results.

[1160] Output: Refined search results

[1161] In this way, a system is constructed in which the server, terminals, and users work together to provide the latest information that is tailored to their emotions.

[1162] (Application example 1)

[1163] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1164] Conventional online shopping sites recommend products without considering the user's emotional state, which can lead to low user satisfaction. Furthermore, it is difficult to provide the latest information, making it difficult for users to quickly access the information they need. This has led to a demand for improved user experience.

[1165] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1166] In this invention, the server includes: means for having a generative AI learn all content and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll; means including an emotion engine that recognizes the user's emotional state; means for the generative AI to adjust search results based on the user's emotional state; and means for recommending products based on the user's emotions. This enables product recommendations based on the user's emotional state, improving the user experience.

[1167] "Generative AI" is artificial intelligence that learns content such as text, images, and videos and generates new information.

[1168] "Automatic learning of new information" is a function that automatically learns newly added information and always provides the latest information.

[1169] A "search box" is an input field in which a user can enter keywords to search for information.

[1170] "Scroll Down" is a function that displays existing information at the bottom of a web page.

[1171] An "emotion engine" is an engine that recognizes the user's emotional state and adjusts the system's behavior based on that emotion.

[1172] An "emotional state" is an emotional state such as joy, sadness, or surprise expressed by a user.

[1173] "Product recommendation" is a function that suggests appropriate products to users based on their emotional state and search history.

[1174] "User experience" is a general term for the satisfaction and ease of use that users feel when using a system or service.

[1175] The system for implementing this invention is configured as follows: The server has a means for having generative AI learn all content and automatically learn new information. This makes it possible to always provide the latest information. Furthermore, it has a means for placing a large search box at the top of the website, allowing users to easily search for information. It also has a means for displaying existing information below the scroll bar, making it easy for users to access past information.

[1176] The server has a means including an emotion engine that recognizes the user's emotional state. This emotion engine analyzes the user's facial image and voice data to recognize the user's emotional state in real time. Emotional states include joy, sadness, surprise, etc. The emotion engine performs image processing using OpenCV and predicts emotions using an emotion recognition model built with Keras.

[1177] This also includes a means for generative AI to adjust search results based on the user's emotional state. The generative AI is built using the Transformers library and generates prompts based on the user's emotional state. For example, if the user expresses happiness, the prompt would be, "Recommend products that will make the user happy."

[1178] It also includes a method for recommending products based on the user's emotions. Generative AI recommends appropriate products based on prompt text. This enables product recommendations optimized for the user's emotional state, improving the user experience.

[1179] As a concrete example, consider a case where a user takes a picture of their face using a smartphone camera, and an application analyzes the image to recognize emotions. For example, if the user is smiling, the application generates a prompt such as "Please recommend products that will make the user happy," and the generative AI recommends products based on that prompt.

[1180] An example of a prompt is as follows:

[1181] "Recommend products that will bring joy to users."

[1182] "Recommend products that make users sad."

[1183] "Recommend products that will surprise users."

[1184] In this way, product recommendations can be made according to the user's emotional state, improving the user experience.

[1185] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1186] Step 1:

[1187] The user activates the smartphone camera and takes a picture of their face.

[1188] Input: User's face image

[1189] Output: Face image data

[1190] Specific operation: The user opens the camera app on their smartphone and takes a picture of their face. The captured image is sent to the application.

[1191] Step 2:

[1192] The device receives the facial image data and performs image processing using OpenCV.

[1193] Input: Facial image data

[1194] Output: Preprocessed image data

[1195] Specific operation: The device converts the facial image to grayscale, then resizes and normalizes it to a form suitable for the emotion recognition model.

[1196] Step 3:

[1197] The device uses an emotion recognition model built with Keras to predict the user's emotions from preprocessed image data.

[1198] Input: Preprocessed image data

[1199] Output: Emotional state (e.g., happy, sad, surprised)

[1200] Specific operation: The device inputs the preprocessed image data into the emotion recognition model to predict the emotional state, which is then sent to the next step.

[1201] Step 4:

[1202] The server generates an appropriate prompt sentence based on the emotional state.

[1203] Input: Emotional state

[1204] Output: prompt statement

[1205] Specific operation: The server generates a prompt sentence according to the emotional state, for example, "Please recommend a product that will bring joy to the user."

[1206] Step 5:

[1207] The server inputs prompts into a generative AI model (using the Transformers library) to generate product recommendations.

[1208] Input: prompt statement

[1209] Output: Product recommendation list

[1210] Specific operation: The server inputs a prompt sentence into the generative AI model and obtains the generated product recommendation list.

[1211] Step 6:

[1212] The server transmits the product recommendation list to the terminal, which displays it to the user.

[1213] Input: Product Recommendation List

[1214] Output: A list of product recommendations displayed to the user

[1215] Specific operation: The server sends the product recommendation list to the terminal, and the terminal displays it to the user, who can then check the recommended products.

[1216] In this way, product recommendations based on the user's emotional state are realized.

[1217] Example 2

[1218] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1219] In conventional information search systems, even if a user enters a question in natural language, it can be difficult to obtain an appropriate answer. Furthermore, since the information provided does not reflect the user's emotional state, the user may feel confused or anxious. Furthermore, since the source of the answer is not clearly stated, it is difficult to confirm the reliability of the answer.

[1220] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1221] In this invention, the server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for a user to input a question in natural language, means for sending the user's question as a prompt to the generative AI model, means for the generative AI model to generate an answer to the question and return it to the server, means for analyzing the user's emotional state using an emotion engine, means for adjusting the way the answer is displayed based on the user's emotional state, means for displaying the optimized answer to the user, and means for displaying existing information below the scroll. This allows a user to input a question in natural language and receive an appropriate answer from the generative AI model, and further enables the emotion engine to provide optimal information according to the user's emotional state.

[1222] "Generative AI" refers to artificial intelligence that generates new data and answers based on learned information.

[1223] A "search box" is an input field on a web page where a user can enter a question in natural language.

[1224] A "prompt" is an input sentence used to ask questions or give instructions to generative AI.

[1225] An "emotion engine" is software that analyzes and infers a user's emotional state from their input and behavior.

[1226] A "server" is a computer system that serves website content and processes user requests.

[1227] A "user" is someone who visits a website and types a question into a search box.

[1228] "Natural language" refers to text written in everyday language, and is free text that is not bound by any particular format or structure.

[1229] An "answer" is information that a generative AI generates in response to a user's question.

[1230] "Sources of information" refers to the data and literature that generative AI refers to when generating answers.

[1231] "Existing Information" refers to data or content that already exists on a website.

[1232] MODE FOR CARRYING OUT THE INVENTION

[1233] This invention is a system that allows a user to input a question in natural language and obtain an appropriate answer using a generative AI model. Specific embodiments for implementing this system are described below.

[1234] Hardware and software used

[1235] server

[1236] The server uses a high-performance cloud server, specifically an EC2 instance from Amazon Web Services (AWS), to provide website content and process user requests.

[1237] Generative AI Model

[1238] The generative AI model used is OpenAI's GPT-4, which has been trained on a large amount of text data and is capable of generating appropriate answers to user questions.

[1239] Emotion Engine

[1240] The emotion engine uses Microsoft's Azure Cognitive Services, which is able to analyze and infer emotional states from user input and behavior.

[1241] System Operation

[1242] Place a search box at the top of the website

[1243] The server displays a large search box on the homepage of the website, designed using HTML and CSS to allow users to enter questions in natural language.

[1244] Send user questions to generative AI models

[1245] When a user types a question into the search box and hits enter, the server receives the question and sends it to the generative AI model as a prompt, which is a properly formatted version of the user's question.

[1246] Answer generation using generative AI models

[1247] The generative AI model receives the prompt and generates an appropriate answer from the information it has learned, which is then returned to the server.

[1248] Sentiment analysis with emotion engine

[1249] The server uses an emotion engine to analyze the user's emotional state, inferring emotions such as confusion or anxiety from the user's input and past behavior.

[1250] Adjust how answers are displayed

[1251] The server adjusts how the generative AI model presents answers based on the emotion engine's analysis, for example, presenting answers in more detail if the user is confused.

[1252] Optimized answer display

[1253] The server displays the optimized answer to the user, allowing the user to receive information that is optimal for their emotional state.

[1254] Specific examples

[1255] For example, consider the case where a user types "What is the latest sales strategy?" into a search box. This question is sent to the server and passed as a prompt to the generative AI model. The generative AI model generates an answer based on the prompt, "Tell me about the latest sales strategy." The generated answer is, "The latest sales strategy revolves around strengthening digital marketing and analyzing customer data."

[1256] If the emotion engine analyzes the user's emotions and determines that they are confused, the server will display the generated answer in detail, for example, "Modern sales strategies revolve around strengthening digital marketing and analyzing customer data. For more information, please see this link."

[1257] In this way, users can input questions in natural language and receive appropriate answers from the generative AI model. The emotion engine also provides optimal information based on the user's emotional state.

[1258] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1259] Step 1:

[1260] A user accesses a website using a browser. The user enters the website's URL in the browser's address bar and presses the Enter key. The input is the website's URL, and the output is the website's homepage.

[1261] Step 2:

[1262] The server renders the website's homepage and displays a large search box. The server generates the homepage using HTML and CSS, centering the search box. The input is the website's content data, and the output is the homepage that is displayed in the user's browser.

[1263] Step 3:

[1264] A user enters a question in natural language into the search box. The user types "What is the latest sales strategy?" into the search box and presses the enter key. The input is the user's question, and the output is the question data sent to the server.

[1265] Step 4:

[1266] The server receives the user's question and sends it as a prompt to the generative AI model. The server converts the user's question into an appropriate prompt sentence and sends it to the generative AI model. The input is the user's question data, and the output is the prompt sentence sent to the generative AI model.

[1267] Step 5:

[1268] The generative AI model generates an answer to the question and returns it to the server. The generative AI model receives a prompt and generates an appropriate answer from the information it has learned. The input is the prompt, and the output is the generated answer data.

[1269] Step 6:

[1270] The server uses the emotion engine to analyze the user's emotional state. The server sends the user's input data to the emotion engine to analyze the emotional state. The input is the user's input data, and the output is the emotional state data returned from the emotion engine.

[1271] Step 7:

[1272] The server adjusts how the answer is displayed based on the user's emotional state. The server adjusts how the answer from the generative AI model is displayed based on the emotional state data. The input is the emotional state data and the generated answer data, and the output is the adjusted answer data.

[1273] Step 8:

[1274] The server displays the optimized answer to the user. The server sends the adjusted answer data to the user's browser for display. The input is the adjusted answer data, and the output is the answer displayed in the user's browser.

[1275] (Application example 2)

[1276] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1277] Conventional information search systems in factories have the problem that it is difficult for workers to quickly and accurately obtain the information they need. In addition, information provided does not reflect the emotional state of the workers, which often leaves them feeling confused and stressed. This can lead to reduced work efficiency and a negative impact on productivity.

[1278] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1279] In this invention, the server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for analyzing user emotions using an emotion recognition engine, means for adjusting the display method of information sources based on the user's emotions, means for being installed on robots in the factory, and means for the generative AI to generate and display answers when workers input questions in natural language. This allows workers to simply input questions in natural language, with the generative AI providing quick and accurate answers, and the emotion recognition engine displaying information according to the worker's emotions.

[1280] "Generative AI" is artificial intelligence that uses natural language processing technology to generate appropriate answers for input text.

[1281] An "emotion recognition engine" is software that analyzes emotions from user input, facial expressions, voice, etc., and identifies the user's emotional state.

[1282] A "search box" is an input field on a website where users can enter questions or keywords in natural language.

[1283] An "information source" is the source of information, such as data, literature, or web pages, that generative AI refers to when generating answers.

[1284] A "factory robot" is an automated mechanical device installed to assist with work in a factory, and may be equipped with a display and voice input functions.

[1285] "Natural language" refers to the language form used in everyday life, and is free text that does not depend on a specific format or structure.

[1286] "User emotion" refers to the psychological state that the user shows when making inputs or operations, and includes emotions such as confusion, joy, and anger.

[1287] "Source presentation method" refers to the format and style in which answers and information provided by generative AI are presented to users.

[1288] "Workers" are employees who work in factories and are responsible for searching for information and operating machines.

[1289] An "answer" is a response or explanation that a generative AI generates in response to a user's question.

[1290] A system for implementing the present invention includes an application installed on a robot in a factory. A specific embodiment of this system will be described below.

[1291] System configuration

[1292] The system consists of the following main components:

[1293] 1. Generative AI model: Uses natural language processing technology to generate appropriate answers to user questions.

[1294] 2. Emotion recognition engine: Analyzes emotions from user input, facial expressions, voice, etc., and identifies the user's emotional state.

[1295] 3. Search box: An input field on a website where users can enter questions or keywords in natural language.

[1296] 4. Information source: The source of information such as data, literature, and web pages that generative AI refers to when generating answers.

[1297] 5. Factory robots: These are automated machinery and devices installed to assist with work in factories, equipped with displays and voice input capabilities.

[1298] Program processing

[1299] The server processes and calculates data in the following steps:

[1300] 1. Question input: The worker types or speaks a question to the robot. The entered question is sent to the server via a search box.

[1301] 2. Generative AI answer generation: The server uses the OpenAI API to send the question to a generative AI model, which generates an appropriate answer.

[1302] 3. Emotion Recognition: The emotion recognition engine analyzes emotions from the worker's input and voice to identify their emotional state.

[1303] 4. Answer Adjustment: The generated answer is adjusted based on the emotional state identified by the emotion recognition engine. For example, if the worker is confused, a detailed explanation is added to the answer.

[1304] 5. Display Answer: The adjusted answer is displayed on the robot's display.

[1305] Hardware and software used

[1306] Hardware: Factory robot (with display), microphone (for voice input)

[1307] Software: OpenAI API (generative AI model), EmotionRecognizer (emotion recognition engine)

[1308] Specific examples

[1309] For example, if a worker asks, "How do I operate this new machine?", the generative AI will explain how to operate it. If the worker is confused, the emotion recognition engine will detect this and add more detailed explanations.

[1310] Prompt Sentence Examples

[1311] How do I operate the new machine?

[1312] In this way, the system can improve work efficiency by providing relevant answers to workers' questions and adjusting the way information is displayed based on their emotions.

[1313] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1314] Step 1:

[1315] The user enters a question in natural language into the robot's search box. The entered question is sent from the robot's terminal to the server. The input data is in text format, such as "Please tell me how to operate a new machine."

[1316] Step 2:

[1317] The server sends the received question to the generative AI model. Specifically, it uses the OpenAI API to pass the question as a prompt to the generative AI model. The input data is the text question, and the output data is the answer returned by the generative AI model.

[1318] Step 3:

[1319] The server receives the answer returned by the generative AI model in text format, such as "Here's how to operate the new machine..."

[1320] Step 4:

[1321] The server analyzes the user's emotions using an emotion recognition engine. Specifically, it passes the user's input text or voice data to the emotion recognition engine to identify the user's emotional state. The input data is in text or voice format, and the output data is the user's emotional state (e.g., confusion, joy, anger, etc.).

[1322] Step 5:

[1323] The server adjusts the answer from the generative AI model based on the identified emotional state. For example, if the user is confused, it adds a detailed explanation to the answer. The input data are the answer from the generative AI model and the emotional state, and the output data is the adjusted answer.

[1324] Step 6:

[1325] The server sends the adjusted answer to the robot's terminal, which displays the received answer on its display. The input data is the adjusted answer, and the output data is the text displayed on the display.

[1326] Step 7:

[1327] The user checks the answers displayed on the robot's display and enters additional questions if necessary, which restarts the process from step 1.

[1328] Example 3

[1329] Next, a description will be given of Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1330] Conventional information search systems lack a means for users to confirm the reliability of the answers they receive, making it difficult for users to determine the accuracy of the information provided. Furthermore, since the basis information for the answers provided by generative AI models is not made clear, users are unable to confirm the reliability of the information.

[1331] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1332] In this invention, the server includes means for training all content in a generative AI model and automatically learning new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for the generative AI model to generate answers to user questions, means for searching and displaying information sources that serve as the basis for the generated answers, and means for allowing users to check the accuracy of the answers using the basis information. This makes it easier for users to check the reliability of the generative AI model's answers and obtain more accurate information.

[1333] A "generative AI model" is a type of artificial intelligence that generates answers in natural language to questions from users.

[1334] A "prompt" is text containing a question or instruction that is input to a generative AI model.

[1335] "Evidence" refers to sources of information that support the accuracy of the answers provided by the generative AI model.

[1336] A "server" is a computer system that runs a generative AI model, processes user questions, and provides answers and supporting information.

[1337] A "terminal" is a device on which a user displays answers and evidence information provided by a server.

[1338] A "user" is someone who uses the system to ask questions and review the answers and supporting information from the generative AI model.

[1339] A "search box" is an input field placed at the top of a website for users to enter questions.

[1340] "Scroll Down" is the area located at the bottom of the web page where existing information is displayed.

[1341] This invention is a system that uses a generative AI model to generate answers to questions from users and provides information on the basis of the answers. Specific embodiments of this system are described below.

[1342] Server Roles

[1343] The server runs the generative AI model and processes questions from users. The server operates as follows:

[1344] 1. Prompt generation

[1345] The server receives a question from the user. For example, the user asks, "What is the latest sales strategy?"

[1346] The server inputs this question as a prompt into the generative AI model.

[1347] 2. Answer Generation

[1348] The server sends a prompt to the generative AI model, which then generates an answer such as "The latest sales strategy is ABC."

[1349] The server receives the generated response.

[1350] 3. Obtaining evidence information

[1351] The server searches for the source of the generated answer, for example, identifying the "Latest Updates" section of the "Sales Strategy Page."

[1352] The server sends the answer and the basis information together to the terminal.

[1353] Device Role

[1354] The device receives the answer of the generative AI model and its basis information sent from the server and displays it to the user. Specifically, it works as follows.

[1355] 1. Displaying answers and supporting information

[1356] The terminal receives the response and the basis information sent from the server.

[1357] The terminal displays the answer and the basis information to the user. For example, it displays the answer "The latest sales strategy is ABC" and the "Latest Updates" section of the "Sales Strategy Page."

[1358] User Roles

[1359] The user checks the generated AI model's answer and the underlying information displayed on the device. Specifically, it works as follows:

[1360] 1. Check your answers

[1361] The user checks the answer and the basis information displayed on the terminal.

[1362] The user can then determine the accuracy of the answer based on the displayed evidence. For example, the user can check the "Latest Updates" section of the "Sales Strategy Page" to determine whether the answer is accurate.

[1363] Specific examples

[1364] Prompt Sentence Examples

[1365] What is the latest sales strategy?

[1366] This system makes it easier for users to check the reliability of the answers of the generative AI model, enabling them to obtain more accurate information. By having the server, terminal, and user fulfill their respective roles, users can efficiently obtain information and check its accuracy. The flow of the identification process in Example 3 will be described with reference to FIG. 21.

[1367] Step 1:

[1368] The server receives a question from the user. When the user inputs a question such as "What is the latest sales strategy?", the question is sent to the server. The server inputs this question as a prompt sentence into the generative AI model. The input data is the user's question, and the output data is the prompt sentence.

[1369] Step 2:

[1370] The server sends a prompt to the generative AI model. The generative AI model analyzes the prompt and generates an answer such as "The latest sales strategy is ABC." The input data is the prompt, and the output data is the generated answer.

[1371] Step 3:

[1372] The server receives the generated answer. Then, it searches the source of the answer. For example, it identifies the "Latest Updates" section of the "Sales Strategy Page." The input data is the generated answer, and the output data is the underlying information.

[1373] Step 4:

[1374] The server sends the answer and the basis information together to the terminal. The input data is the generated answer and the basis information, and the output data is the data sent to the terminal.

[1375] Step 5:

[1376] The terminal receives the answer and the supporting information sent from the server. The terminal displays the answer and the supporting information to the user. For example, it displays the answer "The latest sales strategy is ABC" and the "Latest Updates" section of the "Sales Strategy Page." The input data is the data sent from the server, and the output data is the information displayed to the user.

[1377] Step 6:

[1378] The user checks the answer and supporting information displayed on the device. The user then determines the accuracy of the answer based on the displayed supporting information. For example, the user checks the "Latest Updates" section of the "Sales Strategy Page" and determines whether the answer is accurate. The input data is the information displayed on the device, and the output data is the user's judgment.

[1379] (Application example 3)

[1380] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1381] Conventional generative AI systems lacked sufficient means for users to check the accuracy of the generated answers. In particular, when factory robots received work instructions, it was difficult to check the source of the information that served as the basis for those instructions. Furthermore, the system lacked the functionality to display the basis information for the generated answers, making it impossible for users to check the reliability of the answers. This resulted in problems such as reduced work efficiency and accuracy.

[1382] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 3 is realized by the following means. In this invention, the server includes: means for having a generative AI learn all content and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll; means for displaying the information source that serves as the basis for a work instruction when the factory robot receives the instruction; means for sending a prompt to the generative AI model and generating an answer; and means for extracting and displaying the basis information for the generated answer from a database. This allows the user to check the accuracy of the generated answer and improve the reliability of the work instructions for the factory robot.

[1383] "Generative AI" is a system that uses artificial intelligence technology to generate new information and answers.

[1384] A "prompt sentence" is a question or instruction sentence that is input to a generative AI.

[1385] A "factory robot" is a mechanical device used to automate work in a factory.

[1386] "Work instructions" are information that instructs a factory robot on the specific work that it should perform.

[1387] "Evidence" is a source of information that supports the accuracy of the answers provided by generative AI.

[1388] A "database" is a system that systematically organizes and stores information, and allows it to be searched and extracted as needed.

[1389] A "search box" is an input field on a website into which a user enters a search query.

[1390] The "scroll bottom" is an area located at the bottom of a web page that is displayed when the user scrolls.

[1391] "Means of learning" are methods and techniques for enabling generative AI to take in and understand new information.

[1392] "Display means" refers to the method or technique for visually presenting information to a user.

[1393] The system for implementing the present invention uses the following hardware and software.

[1394] Hardware used:

[1395] Factory robots (e.g., general robotic arms)

[1396] Display device (e.g., a tablet attached to the robot)

[1397] Software used:

[1398] Generative AI models (e.g., OpenAI's GPT-4)

[1399] Database management system (e.g. MySQL)

[1400] Web server (e.g. Apache)

[1401] System Overview:

[1402] The server has a means to train the generative AI on all content and automatically learn new information. A large search box is placed at the top of the website, and existing information is displayed at the bottom of the scroll. When a factory robot receives a work instruction, it also has a means to display the source of the information that supports that instruction. It also has a means to send prompts to the generative AI model to generate an answer, and a means to extract and display the information that supports the generated answer from a database.

[1403] The specific process flow:

[1404] 1. Prompt generation: When a factory robot receives a work instruction, the server sends a prompt to the generative AI model, which includes the work content and related information.

[1405] Example: "Please tell me the assembly steps for part A."

[1406] 2. Answer generation: The generative AI model generates an answer based on the prompt.

[1407] Example: "The assembly procedure for part A is XYZ."

[1408] 3. Extraction of evidence: The server extracts the sources of evidence for the generated answer from the database.

[1409] Example: "Part A Section" in "Assembly Manual"

[1410] 4. Display of source: The extracted source is displayed on the display device of the factory robot, allowing the user to check the accuracy of the generated answer.

[1411] Examples:

[1412] When a factory robot receives the work instruction "Assemble part A," the generative AI model responds, "The assembly procedure for part A is XYZ." The source of information that supports this response (e.g., "Part A's section" in the "Assembly Manual") is displayed on the screen. The user can review this information and verify the accuracy of the work instructions.

[1413] This system allows factory managers and engineers to check the accuracy of work instructions and make corrections or verifications as necessary, thereby improving work efficiency and accuracy.

[1414] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1415] Step 1:

[1416] The server receives work instructions from factory robots. The input includes the content of the work instructions (e.g., "Assemble part A"). The server analyzes the work instructions and generates a prompt sentence to send to the generative AI model. The output is the generated prompt sentence (e.g., "Please tell me the assembly procedure for part A").

[1417] Step 2:

[1418] The server sends the generated prompt to the generative AI model. The input includes the prompt. The generative AI model generates an answer based on the prompt. The output is the generated answer (e.g., "The assembly procedure for part A is XYZ").

[1419] Step 3:

[1420] The server analyzes the answer received from the generative AI model and extracts the source of information that supports the answer from a database. The input includes the generated answer. The server uses a database management system (e.g., MySQL) to search and extract the source of information related to the answer. The output is the source of information (e.g., "Section for Part A" in the "Assembly Manual").

[1421] Step 4:

[1422] The server displays the extracted evidence information on the display device of the factory robot. The input includes the evidence information. The server uses a web server (e.g., Apache) to send the information to the display device and display it visually. The output is the evidence information displayed on the display device.

[1423] Step 5:

[1424] The user checks the basis information displayed on the display device of the factory robot. The basis information displayed on the display device is included as input. Based on this information, the user checks the accuracy of the work instructions and makes corrections or checks as necessary. The output is the user's confirmation result.

[1425] This series of processes improves the accuracy and reliability of work instructions for factory robots, increasing work efficiency.

[1426] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1427] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1428] Another example of generative AI is Gemini (internet search engine). <url: https: gemini.google.com ?hl="ja">) are mentioned.

[1429] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[1430] [Third embodiment]

[1431] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1432] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1433] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1434] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[1435] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1436] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1437] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1438] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1439] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1440] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1441] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1442] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[1443] "Example 1"

[1444] This invention builds a system in which all content is learned by a generative AI and new information is automatically learned. Specifically, all content on an internal company website, including text, images, and videos, is learned by the generative AI. By automatically learning newly added information, it becomes possible to always provide the latest information.

[1445] "Example 2"

[1446] A large search box will be placed at the top of the website. This search box will allow employees to search in natural language. For example, if they enter a question such as "What is the latest sales strategy?", the generative AI will generate an answer from the information it has learned and display it.

[1447] "Example 3"

[1448] Existing information is displayed at the bottom of the scroll. This allows employees to see the location of the source of information that supports the answer along with the search results. For example, if the generative AI answers, "The latest sales strategy is ABC," the location of the source of information that supports this answer (for example, the "Latest Updates" section of the "Sales Strategy Page") will be displayed. This allows employees to check the accuracy of the answer for themselves.

[1449] The processing flow of each embodiment will be described below.

[1450] "Example 1"

[1451] Step 1: Have the generative AI learn all of the content. Specifically, have the generative AI learn all of the content on the company's internal website, including text, images, and videos.

[1452] Step 2: Automatically learns newly added information, making it possible to always provide the latest information.

[1453] "Example 2"

[1454] Step 1: Place a large search box at the top of your website.

[1455] Step 2: Allow employees to search in natural language, for example by entering a question such as "What is the latest sales strategy?"

[1456] Step 3: The generative AI generates an answer from the learned information and displays it.

[1457] "Example 3"

[1458] Step 1: Display existing information at the bottom of the scroll.

[1459] Step 2: Along with the search results, employees can see the location of the source that supports the answer.

[1460] Step 3: For example, if the generative AI answers, "The latest sales strategy is ABC," the location of the source of the answer (for example, the "Latest Updates" section of the "Sales Strategy Page") will be displayed. This allows employees to check the accuracy of the answer themselves.

[1461] Example 1

[1462] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1463] With conventional information management systems, it took time for the content of internal company websites to be updated with the latest information, making it difficult for users to quickly obtain the information they needed. Furthermore, there was a lack of means to verify the accuracy of the information, making it difficult to ensure the reliability of the information users obtained.

[1464] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means. In this invention, the server includes means for having the generative artificial intelligence learn all information and automatically learn new information, means for collecting information from websites, means for preprocessing the collected information, means for having the generative artificial intelligence learn the preprocessed information, means for periodically checking for newly added information and automatically collecting, preprocessing, and learning, means for the user to input a prompt sentence to the generative artificial intelligence to obtain the latest information, means for placing a large search box at the top of the website, and means for displaying existing information below the scroll bar. This allows users to always quickly obtain the latest information and confirm the accuracy of that information.

[1465] "Generative AI" is AI that has the ability to generate new information based on collected data.

[1466] "Automatic learning of new information" is a process of periodically checking for newly added information and automatically collecting, preprocessing, and learning from it.

[1467] "Means of collecting information from websites" refers to technologies used to obtain content such as text, images, and videos on websites.

[1468] "Means for preprocessing collected information" refers to techniques for analyzing collected data and performing processes such as tokenization, normalization, resizing, and noise removal.

[1469] "Means for training generative AI to learn preprocessed information" refers to a technique for inputting preprocessed data into generative AI and training a model.

[1470] A "prompt sentence" is text containing a question or request that a user inputs to a generative artificial intelligence.

[1471] A "search box" is an input field on a website into which a user enters a search query.

[1472] "Means for displaying below the scroll" refers to a technique for displaying existing information at the bottom of a web page.

[1473] "Means for verifying the accuracy of information" are techniques that provide the location of the source of information in order to verify the reliability of the information generated.

[1474] This invention is a system that uses generative artificial intelligence to learn all the content of an in-house website and automatically learns newly added information. Specific embodiments of this system are described below.

[1475] Data collection

[1476] The server collects all content, including text, images, and videos, from the company's internal website. Specifically, it uses Apache HTTP Server to crawl each page of the website and obtains the HTML data. From the obtained HTML data, it uses BeautifulSoup to extract the URLs of the text, images, and videos.

[1477] Data Preprocessing

[1478] The server preprocesses the collected data. For text data, it uses a natural language processing library (e.g., NLTK or spaCy) to tokenize and normalize it. For image data, it uses OpenCV to resize and denoise it. For video data, it uses FFmpeg to extract frames and compress them.

[1479] Data Learning

[1480] The server inputs the preprocessed data into a generative artificial intelligence model for training. Machine learning frameworks such as TensorFlow and PyTorch are used for training. The text data is input into GPT-4, and the image data is input into DALL-E, to optimize the model.

[1481] Automatic learning of new information

[1482] The server periodically checks for new information and automatically collects, pre-processes, and learns from it. Specifically, a Cron job is set up to periodically scan the website for new content.

[1483] Providing information

[1484] Users can obtain the latest information by inputting a prompt to the generative AI model. For example, if a user inputs "Please tell me the latest product information," the server passes the prompt to the generative AI model, and the model returns the latest product information to the user.

[1485] Examples of concrete examples and prompts

[1486] Specific examples

[1487] If a user wants to know the latest news on an internal company website, they input a prompt sentence to the generative artificial intelligence model: "Tell me the latest news."

[1488] The server generates the latest news from the learned data and provides it to users.

[1489] Prompt Sentence Examples

[1490] "Please tell me the latest product information"

[1491] "Tell me about this month's events"

[1492] "Explain the new policy."

[1493] In this way, the user can easily obtain the latest information at all times.

[1494] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1495] Step 1: Data collection

[1496] The server collects all content, including text, images, and videos, from the company's internal website. Specifically, it uses Apache HTTP Server to crawl each page of the website and obtains the HTML data. From the obtained HTML data, it uses BeautifulSoup to extract the URLs of the text, images, and videos.

[1497] Enter: Website URL

[1498] Output: List of text, image, and video URLs

[1499] Step 2: Data Preprocessing

[1500] The server pre-processes the collected data.

[1501] Preprocessing text data:

[1502] What it does: The server uses NLTK to tokenize the text data, removing unnecessary symbols and spaces, and spaCy to normalize the text (e.g., lowercasing, removing stop words).

[1503] Input: Text data

[1504] Output: Preprocessed text data

[1505] Image data preprocessing:

[1506] What happens: The server uses OpenCV to resize the image data, apply a noise reduction filter, and unify the image format (e.g., convert to JPEG) if necessary.

[1507] Input: Image data

[1508] Output: Preprocessed image data

[1509] Video data preprocessing:

[1510] What it does: The server uses FFmpeg to extract and compress the video data frame by frame, and then unifies the video resolution.

[1511] Input: Video data

[1512] Output: Preprocessed video data

[1513] Step 3: Data training

[1514] The server inputs the preprocessed data into a generative artificial intelligence model for training. Machine learning frameworks such as TensorFlow and PyTorch are used for training. The text data is input into GPT-4, and the image data is input into DALL-E, to optimize the model.

[1515] Input: Preprocessed text data, image data, and video data

[1516] Output: A trained generative artificial intelligence model

[1517] Step 4: Automatically learning new information

[1518] The server periodically checks for new information and automatically collects, pre-processes, and learns from it. Specifically, a Cron job is set up to periodically scan the website for new content.

[1519] Input: Newly added website content

[1520] Output: An updated trained generative artificial intelligence model

[1521] Step 5: Provide information

[1522] Users can obtain the latest information by inputting a prompt to the generative AI model. For example, if a user inputs "Please tell me the latest product information," the server passes the prompt to the generative AI model, and the model returns the latest product information to the user.

[1523] Input: prompt statement

[1524] Output: Latest information

[1525] In this way, the user can easily obtain the latest information at all times.

[1526] (Application example 1)

[1527] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1528] Conventional content distribution systems have had difficulty providing users with the latest, personalized information. Furthermore, content generation based on user interests is performed manually, resulting in inefficiency and delays in providing the latest information. Furthermore, the generation of new articles based on news data is not automated, which means users are unable to receive consistent information.

[1529] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1530] In this invention, the server includes a means for having a generative AI learn all content and automatically learn new information, a means for placing a large search box at the top of the website, a means for displaying existing information below the scroll, a means for generating personalized content based on the user's interests, and a means for acquiring news data and having the generative AI generate new articles based on that data, making it possible to always provide users with the latest and personalized information.

[1531] "Generative AI" is artificial intelligence that learns content such as text, images, and videos, and automatically generates content based on new information.

[1532] "Automatic learning of new information" is a function that automatically learns newly added information and always provides the latest information.

[1533] A "search box" is an input field where users can enter keywords to search for information.

[1534] "Scroll Down" is a function that displays existing information at the bottom of a web page.

[1535] "Personalized content" is content that is individually optimized based on a user's interests and browsing history.

[1536] "News data acquisition" is the process of collecting the latest news information from external news sources.

[1537] "Article generation" is the process in which generative AI creates new articles based on collected news data.

[1538] A system for implementing this invention is configured as follows: The server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for generating personalized content based on user interests, and means for acquiring news data and having the generative AI generate new articles based on that data.

[1539] Hardware and software used

[1540] Hardware: Servers, smartphones

[1541] Software: Python, Transformers library (Hugging Face), Requests library

[1542] Data processing and calculation

[1543] 1. Data Acquisition: The server acquires the latest news data using the news API. For example, data is collected from the news API using a URL such as "https: / / newsapi.org / v2 / top-headlines?country=jp&apiKey=YOUR_API_KEY".

[1544] 2. Data training: The acquired news data is trained into a generative AI model (e.g., GPT-2). This gives the generative AI the ability to generate new content based on the latest news information.

[1545] 3. Content generation: Based on the learned data, new articles and content are generated to serve to users. The generated content is personalized based on the user's interests and browsing history.

[1546] Specific examples

[1547] For example, if a user is interested in "technology," the server will collect the latest technology-related news, and the generative AI will generate new articles based on that news. This allows users to always get the latest technology information.

[1548] Prompt Sentence Examples

[1549] "Generate new articles based on the latest technology news."

[1550] This way, users can stay up to date with the latest information and enjoy personalized content based on their interests.

[1551] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1552] Step 1:

[1553] The server uses a news API to obtain the latest news data.

[1554] Input: News API URL

[1555] Output: Latest news data (JSON format)

[1556] Specific operation: The server sends an HTTP request to the news API endpoint and receives the latest news data in response.

[1557] Step 2:

[1558] The server performs preprocessing on the acquired news data in order to input it into the generative AI model.

[1559] Input: Latest news data (JSON format)

[1560] Output: Preprocessed text data

[1561] Specific operation: The server extracts article titles and descriptions from the news data and formats them as text data.

[1562] Step 3:

[1563] The server inputs the preprocessed text data into a generative AI model to generate new articles.

[1564] Input: Preprocessed text data

[1565] Output: The new article generated.

[1566] How it works: The server inputs text data into a generative AI model (e.g., GPT-2), which then generates new articles.

[1567] Step 4:

[1568] The server personalizes the new articles generated based on the user's interests.

[1569] Input: Generated new articles, user interest data

[1570] Output: personalized article

[1571] Specific operation: The server filters or highlights generated articles based on the user's interest data (such as browsing history and past search keywords).

[1572] Step 5:

[1573] The server delivers personalized articles to the user's device.

[1574] Input: Personalized Article

[1575] Output: The article as it appears on the user's device

[1576] Specific operation: The server sends the personalized article to the user's smartphone or web browser for display.

[1577] Step 6:

[1578] Users can view the published article and use the search box to find additional information if needed.

[1579] Input: User's search keyword

[1580] Output: Search results

[1581] What it does: The user enters keywords into the search box, and the server searches for relevant information based on those keywords and returns the results to the user.

[1582] Example 2

[1583] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1584] With conventional information search systems, it was difficult for users to quickly and accurately obtain the information they needed. Furthermore, there was a lack of means to verify the basis of the information, making it difficult to obtain reliable information. Furthermore, the way existing information was displayed was inconvenient, impairing user convenience.

[1585] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes a means for having the generative AI model learn all information and automatically learn new information, a means for placing a large search box at the top of the website, a means for a user to input a question in natural language, a means for transmitting the input question to the generative AI model, a means for receiving and displaying an answer from the generative AI model, and a means for displaying existing information below the scroll. This allows a user to input a question in natural language and receive a quick and accurate answer from the generative AI model. In addition, since the basis information for the answer is also provided at the same time, the user can confirm reliable information. Furthermore, displaying existing information below the scroll improves user convenience.

[1586] A "generative AI model" is a type of artificial intelligence that uses natural language processing to generate answers to user questions.

[1587] A "prompt sentence" is a textual sentence that sends a question entered by a user to a generative AI model.

[1588] A "server" is a computer system that receives requests from a user, sends prompts to a generative AI model, and returns the generated answers to the user.

[1589] A "search box" is an input field placed on the top page of a website where users can enter questions in natural language.

[1590] "Natural language" refers to the language form that users use on a daily basis, and is free-flowing text that is not bound by any particular format or structure.

[1591] An "answer" is information that a generative AI model generates in response to a user's question.

[1592] "Evidence" refers to the location of the information source that the generative AI model referenced when generating the answer.

[1593] "Existing information" is information that already exists on the website and is displayed in a format that users can scroll through.

[1594] "Automatic learning" is the process by which a generative AI model continually learns new information, improving the model's accuracy and knowledge.

[1595] This invention is a system that allows users to input questions in natural language and obtain quick and accurate answers using a generative AI model. Specific embodiments of this system are described below.

[1596] Hardware and software used

[1597] server

[1598] The server receives requests from users, sends prompts to the generative AI model, and returns the generated answers to the users. The server uses the following hardware and software:

[1599] Hardware: A server machine with a powerful processor, sufficient memory, and a fast network connection

[1600] Software: Web server software (e.g., Apache, Nginx), programming languages ​​(e.g., Python, JavaScript), APIs for generative AI models (e.g., OpenAI's GPT-4)

[1601] Terminal

[1602] The device used by the user is a computer or smartphone with a web browser, which the user uses to access the website and enter a question into a search box.

[1603] Data processing and calculation

[1604] User Input

[1605] Users type a question into a large search box on the homepage of a website, for example, "What are the latest sales strategies?"

[1606] Receiving the server and generating a prompt

[1607] The server receives the question entered by the user and converts it into a prompt, which is simply "What is the latest sales strategy?"

[1608] Send to generative AI model and generate answer

[1609] The server sends the generated prompt to the generative AI model, which analyzes the received prompt and generates the optimal answer based on the information it has learned. For example, the following answer might be generated:

[1610] "The latest sales strategies include strengthening digital marketing, improving targeting accuracy through customer data analysis, and establishing a remote work environment."

[1611] Server display process

[1612] The server receives the answers returned by the generative AI model and displays them on the website's search results page, allowing users to see the answers to their questions.

[1613] Specific examples

[1614] For example, if a user types the question "What is the latest sales strategy?" into a search box, the following prompt will be sent to the generative AI model:

[1615] What are the latest sales strategies?

[1616] The generative AI model analyzes this prompt and generates the following answer:

[1617] The latest sales strategies include strengthening digital marketing, improving targeting accuracy through analysis of customer data, and establishing a remote work environment.

[1618] The server receives this answer and displays it on the website's search results page, where users can view it and use it to improve their business.

[1619] In this way, the invention allows users to enter questions in natural language and get fast and accurate answers using generative AI models.

[1620] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1621] Step 1:

[1622] A user accesses the homepage of a website and enters a question into a large search box, for example, "What is the latest sales strategy?" The entered question is sent from the user's device to the server.

[1623] Step 2:

[1624] The server receives questions sent by users. It saves the received questions in text format and prepares them to be sent to the generative AI model as prompts. The input is the user's question, and the output is the prompt.

[1625] Step 3:

[1626] The server sends the received question as a prompt to the generative AI model. The prompt is "What is the latest sales strategy?" The input is the prompt, and the output is a request to the generative AI model.

[1627] Step 4:

[1628] The generative AI model receives and analyzes the prompt sent from the server. The model uses an internal neural network to generate the optimal answer from the learned information. The input is the prompt, and the output is the generated answer.

[1629] Step 5:

[1630] The generative AI model sends the generated answer back to the server. For example, it might generate an answer like, "The latest sales strategies include strengthening digital marketing, improving targeting accuracy by analyzing customer data, and establishing a remote work environment." The input is the generated answer, and the output is the response to the server.

[1631] Step 6:

[1632] The server receives the answers returned by the generative AI model, embeds the received answers into an HTML template, and generates a search results page for the website. The input is the generated answers, and the output is the HTML search results page.

[1633] Step 7:

[1634] The server returns the generated search results page to the user's device. The user can then view the search results page in their browser and read the displayed answers. The input is the HTML search results page, and the output is the answer displayed in the user's browser.

[1635] In this way, users can enter questions in natural language and get fast and accurate answers using generative AI models.

[1636] (Application example 2)

[1637] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1638] There is a problem in that it is difficult for workers in factories to quickly and accurately obtain the information they need. In particular, when workers want to know the parts and procedures required for the next process, conventional methods can take time and reduce efficiency. In addition, there is a lack of means to confirm the accuracy of the information, so there is a risk that work will proceed based on incorrect information.

[1639] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for having a generative AI learn all information and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll; means for factory workers to input questions in natural language and for the generative AI to generate and display the optimal answer; means for inputting questions using a smartphone or tablet; and means for the generative AI to provide information about work in the factory. This enables workers to quickly and accurately obtain the information they need, improving work efficiency.

[1640] "Information" refers to data and knowledge related to operations within a factory.

[1641] "Generative AI" refers to artificial intelligence that uses natural language processing to generate answers to questions.

[1642] "Automatic learning" refers to the process of continually learning new data and information and updating its knowledge base.

[1643] A "search box" is an input field where a user can enter a question in natural language.

[1644] "Scroll Down" refers to the area at the bottom of a web page that becomes visible when the user scrolls down.

[1645] "Workers" means employees who perform work within a factory.

[1646] "Natural language" refers to text written in a language form that humans use on a daily basis.

[1647] A "smartphone" is a mobile phone that can connect to the Internet and use applications.

[1648] A "tablet" is a highly portable computing device that has a touchscreen.

[1649] A "best answer" is an answer that provides the most relevant and accurate information for a question.

[1650] A "source" is a place where the data or knowledge that forms the basis of an answer is stored.

[1651] "Accuracy" means that the information or answer is factual and free from errors.

[1652] The system for implementing this invention is configured as follows: First, the server uses generative AI to learn all information and automatically learns new information. This allows the latest information to be maintained at all times. Next, a large search box is placed at the top of the website, allowing users to enter questions in natural language. Furthermore, existing information is displayed at the bottom of the page, making it easy for users to access past information.

[1653] Factory workers use smartphones or tablets to input questions. For example, when a question such as "What parts are needed for the next process?" is input, the generative AI generates the optimal answer to the question and displays it to the worker. The generative AI generates answers based on data it has learned in advance to provide information about work within the factory.

[1654] The hardware used includes smartphones and tablets, and the software uses Python and the OpenAI API. Data processing and data calculation involves receiving questions in natural language from users and generating answers using OpenAI's generative AI model (GPT-3).

[1655] As a concrete example, if a worker in a factory asks, "What parts are needed for the next process?", the following prompt sentence is input into the generative AI model.

[1656] Example prompt sentence:

[1657] Factory operation questions: What part is needed for the next operation?

[1658] answer:

[1659] Based on this prompt, the generative AI generates a list of parts required for the next process and provides it to the worker, allowing the worker to quickly and accurately obtain the necessary information and improve work efficiency.

[1660] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1661] Step 1:

[1662] A user uses a smartphone or tablet to enter a question in natural language into a search box at the top of a website. The entered question is sent to the server. The input data is the user's question, and the output data is the question sent to the server.

[1663] Step 2:

[1664] The server sends the received question to the generative AI model. Specifically, the question is input to the generative AI model (GPT-3) as a prompt. The input data is the user's question, and the output data is the prompt sent to the generative AI model.

[1665] Step 3:

[1666] A generative AI model generates an optimal answer based on a prompt. Specifically, it generates an answer to a question and returns that answer to the server. The input data is the prompt, and the output data is the generated answer.

[1667] Step 4:

[1668] The server sends the answers received from the generative AI model to a smartphone or tablet for display to the user. The input data is the generated answer, and the output data is the answer sent to the user's device.

[1669] Step 5:

[1670] The user checks the response displayed on their smartphone or tablet and obtains the necessary information. Specifically, they check information such as the parts list and procedures required for the next process. The input data is the response sent from the server, and the output data is the information obtained by the user.

[1671] Through these processing steps, users can input questions in natural language and receive quick and accurate answers using generative AI models, thereby improving work efficiency within factories.

[1672] Example 3

[1673] Next, a third embodiment of the third embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1674] In conventional information retrieval systems, users lacked the means to verify the basis of the answers they received, making it difficult to determine the reliability of the answers. Another problem was that it took a lot of time and effort to identify the location of the information source. Furthermore, when new information was added, the system sometimes failed to automatically learn and provide the latest information.

[1675] The specific processing by the specific processing unit 290 of the data processing device 12 in the third embodiment is realized by the following means.

[1676] In this invention, the server includes: means for training a generative AI model to learn all information and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll bar; means for sending a prompt entered by a user to the server; means for the server to send the prompt to the generative AI model and generate an answer; means for the server to search an internal database for information sources related to the answer; means for the server to send the search results and the answer to a terminal; and means for the terminal to display the search results and the answer to the user. This allows the user to easily check the basis for the generated answer, increasing the reliability of the answer. Furthermore, when new information is added, the system automatically learns and can provide the latest information.

[1677] A "generative AI model" is an artificial intelligence algorithm that generates answers based on prompts entered by the user.

[1678] A "prompt" is a natural language question or instruction that a user enters to retrieve information.

[1679] The "server" is a computer system that receives prompts from users, sends them to a generative AI model to generate answers, and then searches for relevant information sources and sends the results to the terminal.

[1680] A "terminal" is a device through which a user can enter prompts and display answers and related information sent by the server.

[1681] An "internal database" is a database that stores information sources held by a company or organization, and is used by the server to search for the basis of an answer.

[1682] A "search box" is an input field placed at the top of a website for users to enter a prompt phrase.

[1683] The "scroll bottom" is an area located at the bottom of a web page for displaying existing information.

[1684] An "answer" is information that a generative AI model generates based on a prompt.

[1685] A "source" is the data or document that forms the basis for the generated answer.

[1686] "Automatic learning" is the process by which a generative AI model continuously learns new information, enabling it to provide up-to-date information.

[1687] This invention is a system in which a generative AI model generates an answer based on a prompt sentence entered by a user and displays the information source that forms the basis of the answer. Specific embodiments of this system are described below.

[1688] First, the user inputs a prompt sentence into a search box installed on the web browser of the terminal. For example, the user inputs a prompt sentence such as "What is the latest sales strategy?"

[1689] The device sends the input prompt to the server using an HTTP POST request. The server then sends the received prompt as an API request to the generative AI model. For example, OpenAI's GPT-4 can be used as the generative AI model.

[1690] The generative AI model generates an answer based on the prompt sentence. For example, it generates the answer "The latest sales strategy is ABC." The server receives this generated answer and searches an internal database for information sources related to that answer. The internal database stores information sources held by companies and organizations, and searches are performed using SQL queries.

[1691] The server sends the search results and the generated answer in JSON format to the terminal. The terminal displays the received answer and the search results to the user. For example, along with the answer "The latest sales strategy is ABC," it displays a link to the "Latest Updates" section of the "Sales Strategy Page." The user can click this link to view more detailed information.

[1692] This system allows users to easily check the basis of the generated answers, increasing the reliability of the answers.In addition, when new information is added, the system automatically learns and can provide the latest information.

[1693] As a concrete example, if a user inputs the prompt sentence "What is the latest sales strategy?", the server uses the generative AI model to generate the answer "The latest sales strategy is ABC", and searches the internal database to identify the "Latest Updates" section of the "Sales Strategy Page". Ultimately, the answer "The latest sales strategy is ABC" is displayed on the user's device, along with a link to the "Latest Updates" section of the "Sales Strategy Page".

[1694] In this way, the user can easily check the basis of the information provided by the generative AI model. The flow of the identification process in the third embodiment will be described with reference to FIG.

[1695] Step 1:

[1696] The user enters a prompt statement.

[1697] The user inputs a prompt sentence, "What is the latest sales strategy?", into a search box installed on the web browser of the terminal. The input prompt sentence becomes the input data of the terminal.

[1698] Step 2:

[1699] The terminal sends the prompt to the server.

[1700] The terminal sends the input prompt text to the server using an HTTP POST request. The input data is the prompt text, and the output data is the prompt text sent to the server.

[1701] Step 3:

[1702] The server sends a prompt to the generative AI model, which generates an answer.

[1703] The server sends the received prompt sentence to the generative AI model as an API request. The generative AI model generates an answer based on the prompt sentence. The input data is the prompt sentence, and the output data is the generated answer. For example, the generated answer is "The latest sales strategy is ABC."

[1704] Step 4:

[1705] The server searches an internal database for sources relevant to the answer.

[1706] The server searches an internal database for information sources related to the generated answer. The internal database stores information sources owned by the company or organization, and performs a search using an SQL query. The input data is the generated answer, and the output data is the search results for the related information sources. For example, the "Latest Updates" section of the "Sales Strategy Page" is identified.

[1707] Step 5:

[1708] The server sends the search results and answers to the device.

[1709] The server sends the generated answers and search results to the terminal in JSON format. The input data are the generated answers and search results, and the output data is the JSON data sent to the terminal.

[1710] Step 6:

[1711] The device displays the search results and answers to the user.

[1712] The terminal displays the received answers and search results to the user. The input data is the JSON data received from the server, and the output data is the answer displayed to the user and a link to the information source. For example, along with the answer "The latest sales strategy is ABC," a link to the "Latest Updates" section of the "Sales Strategy Page" is displayed. The user can click this link to view more detailed information.

[1713] (Application example 3)

[1714] Next, a description will be given of Application Example 3 of Form Example 3. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1715] Conventional generative AI systems lacked sufficient means for users to verify the basis of the answers they received, making it difficult for factory workers, in particular, to obtain information quickly and accurately on-site. Furthermore, there was a lack of means to verify the reliability of the information provided by generative AI, making it difficult for users to verify the accuracy of the answers themselves.

[1716] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 3 is realized by the following means.

[1717] In this invention, the server includes means for having the generative AI learn all information and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means for the generative AI to provide an answer when a user inputs a question and display the location of the source of the evidence, and means for the generative AI to provide an answer when a factory worker asks a question to the robot and display the location of the source of the evidence. This allows users to check the accuracy of the generative AI's answer using the evidence information, and enables factory workers to obtain information quickly and accurately on site.

[1718] "Information" is a general term for the data and knowledge that generative AI learns and provides to users.

[1719] "Generative AI" is an artificial intelligence system that generates answers in natural language to users' questions.

[1720] "Automatic learning" is the process by which generative AI automatically absorbs and learns new information.

[1721] A "search box" is an input field on a website where users can enter questions or keywords.

[1722] "Scroll Down" is the area at the bottom of a web page that becomes visible when the user scrolls down.

[1723] "Evidence" is a source of information that supports the accuracy of the answers provided by generative AI.

[1724] "Factory workers" refers to employees who work in factories.

[1725] A "robot" is an automated mechanical device that provides work assistance and information within a factory.

[1726] To implement this invention, all information must first be learned by a generative AI. A generative AI is an artificial intelligence system that generates answers in natural language to user questions and has the ability to automatically learn new information. This makes it possible to always provide the latest information.

[1727] Next, a large search box is placed on the homepage of the website. Users can make inquiries to the generative AI by entering questions or keywords into this search box. When a user enters a question, the generative AI generates an answer to that question and displays the location of the source of the answer.

[1728] Additionally, a new feature will be added that displays existing information at the bottom of the scroll, allowing users to easily refer to supporting information to verify the accuracy of answers provided by generative AI.

[1729] To enable factory workers to quickly and accurately obtain information on-site, this system will be installed on robots within the factory. When a factory worker asks the robot a question, generative AI will provide an answer and display the location of the source of the answer. This will enable workers to quickly obtain the information they need on-site and improve work efficiency.

[1730] The implementation of this system uses the following hardware and software. For hardware, we need factory robots and displays. For software, we use the OpenAI API, Python, and Tkinter (GUI library). We use the OpenAI API to obtain answers from a generative AI model (e.g., GPT-3), and we use Python and Tkinter to build the user interface.

[1731] For example, if a factory worker asks, "What is the latest quality control method?", the generative AI will respond, "The latest quality control method is XYZ." The location of the source of information that supports the answer (for example, the "Latest Updates" section of the "Quality Control Page") is displayed at the bottom of the scroll. This allows the worker to confirm the accuracy of the answer and improve their work efficiency.

[1732] An example prompt is:

[1733] Q: What are the latest quality control techniques?

[1734] In this way, users can verify the accuracy of generative AI's answers with supporting information, enabling factory workers to obtain information quickly and accurately on-site.

[1735] The flow of the specific processing in Application Example 3 will be described with reference to FIG.

[1736] Step 1:

[1737] A user types a question into a website search box.

[1738] Input: User question (e.g., "What are the latest quality control techniques?")

[1739] Output: The user's question is sent to the server.

[1740] What happens: A user types a question into the search box and clicks the submit button.

[1741] Step 2:

[1742] The server receives the user's question and sends it to the generative AI model.

[1743] Input: User question

[1744] Output: Prompt sentence for generative AI model

[1745] How it works: The server sends the user's question in text format to a generative AI model (e.g., GPT-3).

[1746] Step 3:

[1747] A generative AI model generates an answer based on the prompt.

[1748] Input: Prompt statement (e.g., "What are the latest quality control methods?")

[1749] Output: Answer from the generative AI model (e.g., "The most current quality control method is XYZ")

[1750] What it does: The generative AI model analyzes the prompt and generates an appropriate response.

[1751] Step 4:

[1752] The server receives the answer from the generative AI model and searches for supporting information.

[1753] Input: Answer from a generative AI model

[1754] Output: Location of answer and supporting information

[1755] Specific operation: The server receives the answer from the generative AI model and searches the database for the source of information that supports that answer.

[1756] Step 5:

[1757] The server returns the answer and supporting information to the user.

[1758] Input: Location of answer and supporting information

[1759] Output: Answer and rationale displayed on the user's device

[1760] Specific operation: The server sends the answer and supporting information to the user's device and displays it at the bottom of the scroll.

[1761] Step 6:

[1762] The user checks the answer and supporting information.

[1763] Input: Answer and rationale displayed on the user's device

[1764] Output: User understanding and confirmation

[1765] What happens: The user reviews the evidence displayed below the scroll bar and evaluates the accuracy of the answer.

[1766] Step 7:

[1767] A factory worker types a question into the robot.

[1768] Input: Factory worker question (e.g., "What are the latest quality control techniques?")

[1769] Output: The factory worker's question is sent to the robot.

[1770] How it works: A factory worker types a question into the robot's interface and clicks the send button.

[1771] Step 8:

[1772] The robot sends a question to the server and gets an answer based on a generative AI model.

[1773] Input: Factory worker question

[1774] Output: Answers and supporting information from the generative AI model

[1775] Specific operation: The robot sends a question to the server and obtains an answer and supporting information from the generative AI model.

[1776] Step 9:

[1777] The robot displays the answer and supporting information to the factory worker.

[1778] Input: Answer from the generative AI model and supporting information

[1779] Output: Answers and supporting information displayed on the factory worker's terminal

[1780] Specific operation: The robot displays the answer and supporting information on the factory worker's device, scrolling down below.

[1781] Step 10:

[1782] Factory workers will review the responses and supporting information.

[1783] Input: Answer and supporting information displayed on the factory worker's terminal

[1784] Output: Factory worker understanding and confirmation

[1785] What happens: The factory worker reviews the supporting information displayed at the bottom of the scroll and evaluates the accuracy of the answer.

[1786] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1787] "Example 1"

[1788] One embodiment of the present invention is a system that incorporates an emotion engine. This system includes a means for having a generative AI learn all content and automatically learn new information, a means for placing a large search box at the top of a website, and a means for displaying existing information below the scroll bar. It also includes an emotion engine that recognizes the user's emotions. Specifically, the emotion engine recognizes the user's emotional state when using a website, and the generative AI adjusts search results based on that emotional state. For example, if the user expresses joy, the generative AI will prioritize displaying information related to joy.

[1789] "Example 2"

[1790] The emotion engine also adjusts how information sources are displayed based on the user's emotions. For example, if the user expresses confusion, the generative AI will display the source's detailed explanation more clearly. This allows the user to receive the most appropriate information based on their emotional state.

[1791] "Example 3"

[1792] Furthermore, this system allows users to check the accuracy of answers using supporting information. Specifically, the location of the source of information that supports the answer provided by the generative AI is displayed, allowing users to check its accuracy for themselves. This allows users to receive the most appropriate information based on their emotional state and to confirm the reliability of that information.

[1793] The processing flow of each embodiment will be described below.

[1794] "Example 1"

[1795] Step 1: A user opens a website.

[1796] Step 2: The user enters a search query in natural language into the search box.

[1797] Step 3: The emotion engine recognizes the user's emotional state.

[1798] Step 4: The generative AI receives input from the emotion engine and refines the search results.

[1799] Step 5: The generative AI displays the adjusted search results to the user.

[1800] "Example 2"

[1801] Step 1: A user opens a website.

[1802] Step 2: The user enters a search query in natural language into the search box.

[1803] Step 3: The emotion engine recognizes the user's emotional state.

[1804] Step 4: The generative AI receives input from the emotion engine and adjusts how the source is presented.

[1805] Step 5: The generative AI displays the adjusted source information to the user.

[1806] "Example 3"

[1807] Step 1: A user opens a website.

[1808] Step 2: The user enters a search query in natural language into the search box.

[1809] Step 3: The generative AI generates search results and identifies the location of the source of information that supports them. Step 4: The generative AI displays the search results and the location of the source of information to the user.

[1810] Step 5: The user locates the displayed source and verifies the accuracy of the answer.

[1811] Example 1

[1812] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1813] On conventional websites, it was difficult for users to quickly and accurately obtain the information they needed. Furthermore, there was a problem in that the information provided did not correspond to the user's emotional state, resulting in a poor user experience. Furthermore, there was a lack of a system that could automatically learn new information and constantly provide the latest information.

[1814] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1815] In this invention, the server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for displaying existing information below the scroll, means including an emotion engine that recognizes the user's emotional state, and means for the generative AI to adjust search results based on the user's emotional state. This allows users to quickly and accurately obtain the information they need, makes it possible to provide information according to their emotional state, and always provides the latest information.

[1816] "Generative AI" is artificial intelligence that generates new content based on collected data.

[1817] "Means for automatically learning new information" refers to a means that has the function of automatically collecting new added data and having the generative AI learn it.

[1818] "A means for placing a large search box at the top of a website" refers to a means for placing a large search box on the top page of a website so that users can easily perform searches.

[1819] The "means for displaying existing information below the scroll" is a means for displaying information below the web page so that the user can view the existing information by scrolling.

[1820] The "emotion engine that recognizes the user's emotional state" is an engine that analyzes emotions from the user's facial expressions and text input and recognizes their state.

[1821] "Means for the generative AI to adjust search results based on the emotional state" refers to means for the generative AI to adjust the search results displayed according to the user's emotional state.

[1822] MODE FOR CARRYING OUT THE INVENTION

[1823] This invention is a system that uses generative AI to learn all content and automatically learns new information. Specifically, the generative AI learns all content on an internal company website, including text, images, and videos. By automatically learning newly added information, it is possible to always provide the latest information.

[1824] Hardware and software used

[1825] The server uses the following hardware and software:

[1826] Web scraping tools: BeautifulSoup, Scrapy

[1827] Natural language processing tools: NLTK, SpaCy

[1828] Image processing libraries: OpenCV, PIL

[1829] Video processing library: FFmpeg

[1830] Generation AI: GPT-4, DALL-E

[1831] Emotion engine: Affectiva, IBM Watson Tone Analyzer

[1832] Program processing

[1833] The server collects all content from the internal company website, including text, images, and videos. A web scraping tool is used to analyze the HTML structure of the web page and extract the required data. For example, text data is extracted from HTML tags, and image data is obtained by obtaining image URLs.

[1834] The collected data is tokenized, normalized, and stop words removed using natural language processing tools, image data is resized and formatted using an image processing library, and video data is frame-extracted and compressed using a video processing library.

[1835] The preprocessed data is then trained by a generative AI, which updates its model based on all collected content and automatically learns new information. For example, text data is fed into GPT-4, and image data is fed into DALL-E.

[1836] The device recognizes the user's emotional state when using a website. It uses an emotion engine to analyze emotions from the user's facial expressions and text input. For example, the device captures the user's facial expressions through a webcam, and the emotion engine analyzes them.

[1837] The server uses a generative AI to tailor search results based on the user's emotional state, obtained from the emotion engine. For example, if the user expresses joy, the generative AI will prioritize information related to joy.

[1838] Specific examples

[1839] A user visits an internal company website and types "new product reviews" into the search box. The device captures the user's facial expressions with a webcam, and the emotion engine recognizes the emotion of happiness from the user's facial expression. The server receives this emotion data and inputs a prompt to the generative AI: "If the user is expressing happiness, please prioritize displaying positive reviews." Based on this prompt, the generative AI prioritizes displaying positive reviews.

[1840] Example prompt sentence:

[1841] "Show reviews of new products. If users are expressing joy, prioritize positive reviews."

[1842] In this way, a system is constructed in which the server, terminals, and users work together to provide the latest information that is tailored to their emotions.

[1843] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1844] Step 1: Data collection

[1845] The server collects all content, including text, images, and videos, from internal company websites.

[1846] Input: Internal website URL

[1847] Specific operation: The server uses a web scraping tool (e.g., BeautifulSoup, Scrapy) to analyze the HTML structure of the web page and extract the necessary data. For example, text data is extracted from HTML tags, and image data is obtained by obtaining image URLs.

[1848] Output: Collected text, image, and video data

[1849] Step 2: Data Preprocessing

[1850] The server pre-processes the collected data.

[1851] Input: Collected text, image, and video data

[1852] Specific operation: The server uses natural language processing tools (e.g., NLTK, SpaCy) to tokenize, normalize, and remove stop words from text data. Image data is resized and formatted using an image processing library (e.g., OpenCV, PIL). Video data is extracted and compressed using a video processing library (e.g., FFmpeg).

[1853] Output: Preprocessed text, image, and video data

[1854] Step 3: Data training

[1855] The server trains the generative AI on the preprocessed data.

[1856] Input: Preprocessed text, image, and video data

[1857] Specific operation: The server inputs data into generative AI (e.g., GPT-4, DALL-E) and updates the model. For example, text data is input into GPT-4, and image data is input into DALL-E. When new information is added, the learning process is automatically restarted.

[1858] Output: An updated generative AI model

[1859] Step 4: Emotion Recognition

[1860] The terminal recognizes the emotional state of the user when using the website.

[1861] Input: User facial expression data, text input

[1862] Specific operation: The device uses an emotion engine (e.g., Affectiva, IBM Watson Tone Analyzer) to analyze emotions from the user's facial expressions and text input. For example, the device captures the user's facial expressions through a webcam, and the emotion engine analyzes them.

[1863] Output: User's emotional state data

[1864] Step 5: Refine your search results

[1865] The server uses generative AI to adjust search results based on the user's emotional state obtained from the emotion engine.

[1866] Input: User's emotional state data, user's search query

[1867] Specific operation: The server receives emotion data from the emotion engine and inputs prompts to the generative AI. For example, the server inputs a prompt such as "If the user is expressing joy, please prioritize displaying positive reviews" to adjust the search results.

[1868] Output: Refined search results

[1869] In this way, a system is constructed in which the server, terminals, and users work together to provide the latest information that is tailored to their emotions.

[1870] (Application example 1)

[1871] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1872] Conventional online shopping sites recommend products without considering the user's emotional state, which can lead to low user satisfaction. Furthermore, it is difficult to provide the latest information, making it difficult for users to quickly access the information they need. This has led to a demand for improved user experience.

[1873] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1874] In this invention, the server includes: means for having a generative AI learn all content and automatically learn new information; means for placing a large search box at the top of the website; means for displaying existing information below the scroll; means including an emotion engine that recognizes the user's emotional state; means for the generative AI to adjust search results based on the user's emotional state; and means for recommending products based on the user's emotions. This enables product recommendations based on the user's emotional state, improving the user experience.

[1875] "Generative AI" is artificial intelligence that learns content such as text, images, and videos and generates new information.

[1876] "Automatic learning of new information" is a function that automatically learns newly added information and always provides the latest information.

[1877] A "search box" is an input field in which a user can enter keywords to search for information.

[1878] "Scroll Down" is a function that displays existing information at the bottom of a web page.

[1879] An "emotion engine" is an engine that recognizes the user's emotional state and adjusts the system's behavior based on that emotion.

[1880] An "emotional state" is an emotional state such as joy, sadness, or surprise expressed by a user.

[1881] "Product recommendation" is a function that suggests appropriate products to users based on their emotional state and search history.

[1882] "User experience" is a general term for the satisfaction and ease of use that users feel when using a system or service.

[1883] The system for implementing this invention is configured as follows: The server has a means for having generative AI learn all content and automatically learn new information. This makes it possible to always provide the latest information. Furthermore, it has a means for placing a large search box at the top of the website, allowing users to easily search for information. It also has a means for displaying existing information below the scroll bar, making it easy for users to access past information.

[1884] The server has a means including an emotion engine that recognizes the user's emotional state. This emotion engine analyzes the user's facial image and voice data to recognize the user's emotional state in real time. Emotional states include joy, sadness, surprise, etc. The emotion engine performs image processing using OpenCV and predicts emotions using an emotion recognition model built with Keras.

[1885] This also includes a means for generative AI to adjust search results based on the user's emotional state. The generative AI is built using the Transformers library and generates prompts based on the user's emotional state. For example, if the user expresses happiness, the prompt would be, "Recommend products that will make the user happy."

[1886] It also includes a method for recommending products based on the user's emotions. Generative AI recommends appropriate products based on prompt text. This enables product recommendations optimized for the user's emotional state, improving the user experience.

[1887] As a concrete example, consider a case where a user takes a picture of their face using a smartphone camera, and an application analyzes the image to recognize emotions. For example, if the user is smiling, the application generates a prompt such as "Please recommend products that will make the user happy," and the generative AI recommends products based on that prompt.

[1888] An example of a prompt is as follows:

[1889] "Recommend products that will bring joy to users."

[1890] "Recommend products that make users sad."

[1891] "Recommend products that will surprise users."

[1892] In this way, product recommendations can be made according to the user's emotional state, improving the user experience.

[1893] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1894] Step 1:

[1895] The user activates the smartphone camera and takes a picture of their face.

[1896] Input: User's face image

[1897] Output: Face image data

[1898] Specific operation: The user opens the camera app on their smartphone and takes a picture of their face. The captured image is sent to the application.

[1899] Step 2:

[1900] The device receives the facial image data and performs image processing using OpenCV.

[1901] Input: Facial image data

[1902] Output: Preprocessed image data

[1903] Specific operation: The device converts the facial image to grayscale, then resizes and normalizes it to a form suitable for the emotion recognition model.

[1904] Step 3:

[1905] The device uses an emotion recognition model built with Keras to predict the user's emotions from preprocessed image data.

[1906] Input: Preprocessed image data

[1907] Output: Emotional state (e.g., happy, sad, surprised)

[1908] Specific operation: The device inputs the preprocessed image data into the emotion recognition model to predict the emotional state, which is then sent to the next step.

[1909] Step 4:

[1910] The server generates an appropriate prompt sentence based on the emotional state.

[1911] Input: Emotional state

[1912] Output: prompt statement

[1913] Specific operation: The server generates a prompt sentence according to the emotional state, for example, "Please recommend a product that will bring joy to the user."

[1914] Step 5:

[1915] The server inputs prompts into a generative AI model (using the Transformers library) to generate product recommendations.

[1916] Input: prompt statement

[1917] Output: Product recommendation list

[1918] Specific operation: The server inputs a prompt sentence into the generative AI model and obtains the generated product recommendation list.

[1919] Step 6:

[1920] The server transmits the product recommendation list to the terminal, which displays it to the user.

[1921] Input: Product Recommendation List

[1922] Output: A list of product recommendations displayed to the user

[1923] Specific operation: The server sends the product recommendation list to the terminal, and the terminal displays it to the user, who can then check the recommended products.

[1924] In this way, product recommendations based on the user's emotional state are realized.

[1925] Example 2

[1926] Next, a description will be given of Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1927] In conventional information search systems, even if a user enters a question in natural language, it can be difficult to obtain an appropriate answer. Furthermore, since the information provided does not reflect the user's emotional state, the user may feel confused or anxious. Furthermore, since the source of the answer is not clearly stated, it is difficult to confirm the reliability of the answer.

[1928] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1929] In this invention, the server includes means for having a generative AI learn all content and automatically learn new information, means for placing a large search box at the top of the website, means for a user to input a question in natural language, means for sending the user's question as a prompt to the generative AI model, means for the generative AI model to generate an answer to the question and return it to the server, means for analyzing the user's emotional state using an emotion engine, means for adjusting the way the answer is displayed based on the user's emotional state, means for displaying the optimized answer to the user, and means for displaying existing information below the scroll. This allows a user to input a question in natural language and receive an appropriate answer from the generative AI model, and further enables the emotion engine to provide optimal information according to the user's emotional state.

[1930] "Generative AI" refers to artificial intelligence that generates new data and answers based on learned information.

[1931] A "search box" is an input field on a web page where a user can enter a question in natural language.

[1932] A "prompt" is an input sentence used to ask questions or give instructions to generative AI.

[1933] An "emotion engine" is software that analyzes and infers a user's emotional state from their input and behavior.

[1934] A "server" is a computer system that serves website content and processes user requests.

[1935] A "user" is someone who visits a website and types a question into a search box.

[1936] "Natural language" refers to text written in everyday language, and is free text that is not bound by any particular format or structure.

[1937] An "answer" is information that a generative AI generates in response to a user's question.

[1938] "Sources of information" refers to the data and literature that generative AI refers to when generating answers.

[1939] "Existing Information" refers to data or content that already exists on a website.

[1940] MODE FOR CARRYING OUT THE INVENTION

[1941] This invention is a system that allows a user to input a question in natural language and obtain an appropriate answer using a generative AI model. Specific embodiments for implementing this system are described below.

[1942] Hardware and software used

[1943] server

[1944] The server uses a high-performance cloud server, specifically an EC2 instance from Amazon Web Services (AWS), to provide website content and process user requests.

[1945] Generative AI Model

[1946] The generative AI model used is OpenAI's GPT-4, which has been trained on a large amount of text data and is capable of generating appropriate answers to user questions.

[1947] Emotion Engine

[1948] The emotion engine uses Microsoft's Azure Cognitive Services, which is able to analyze and infer emotional states from user input and behavior.

[1949] System Operation

[1950] Place a search box at the top of the website

[1951] The server displays a large search box on the homepage of the website, designed using HTML and CSS to allow users to enter questions in natural language.

[1952] Send user questions to generative AI models

[1953] When a user types a question into the search box and hits enter, the server receives the question and sends it to the generative AI model as a prompt, which is a properly formatted version of the user's question.

[1954] Answer generation using generative AI models

[1955] The generative AI model receives the prompt and generates an appropriate answer from the information it has learned, which is then returned to the server.

[1956] Sentiment analysis with emotion engine

[1957] The server uses an emotion engine to analyze the user's emotional state, inferring emotions such as confusion or anxiety from the user's input and past behavior.

[1958] Adjust how answers are displayed

[1959] The server adjusts how the generative AI model presents answers based on the emotion engine's analysis, for example, presenting answers in more detail if the user is confused.

[1960] Optimized answer display

[1961] The server displays the optimized answer to the user, allowing the user to receive information that is optimal for their emotional state.

[1962] Specific examples

[1963] For example, consider the case where a user types "What is the latest sales strategy?" into a search box. This ...

Claims

1. a means for displaying a search box on the top of the website and accepting questions in natural language from users through said search box; A means for converting the question into a prompt sentence and sending the prompt sentence to a generation AI to generate an answer to the question; means for analyzing an emotional state of a user using an emotion engine; means for adjusting the display of answers based on said emotional state; means for displaying the adjusted answer to the user; A system including:

2. The system of claim 1 , wherein the generative AI returns the answer together with a location of a source on which the answer is based.

3. The system of claim 1 , wherein the system enables employees to verify the accuracy of answers with supporting information.

Images