System

Abstract

A system is provided.SOLUTION: The system includes a voice recognition means for recognizing a speech during a conference in real time, a conversion means for converting the contents of the speech recognized by the voice recognition means into text data, a retrieval means for retrieving related information from a database on the basis of a keyword extracted by the conversion means, a display means for displaying the related information acquired by the retrieval means, and a documentation means for automatically documenting the conference contents.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] During meetings, participants often have difficulty recalling documents or past cases quickly and accurately, which reduces meeting efficiency and hinders productivity. To address this issue, a system is needed that can analyze statements made during meetings in real time and instantly provide relevant information. [Means for solving the problem]

[0005] The present invention solves the above problem by providing a system including a voice recognition means for recognizing statements made during a meeting in real time, a conversion means for converting the content of the statements recognized by the voice recognition means into text data, an analysis means for analyzing the text data converted by the conversion means, a search means for searching a database for related information based on keywords extracted by the analysis means, a display means for displaying the related information obtained by the search means, and a documentation means for automatically documenting the contents of the meeting.

[0006] The "voice recognition means" is a device or software that captures voices spoken during a conference in real time and obtains the voice data.

[0007] The "conversion means" is a processing device or software for converting the voice data acquired by the voice recognition means into text data.

[0008] The "analysis means" is a processing device or software for analyzing the text data generated by the conversion means and extracting important keywords and phrases.

[0009] The "search means" refers to a processing device or software for searching the database based on the keywords and phrases extracted by the analysis means and obtaining related information.

[0010] The "display means" is a device or software for visually presenting the related information obtained by the search means to the user.

[0011] "Documentation means" refers to a processing device or software that automatically organizes and saves meeting contents as text documents and provides them as minutes.

[0012] A "large-scale language model" is an artificial intelligence model trained on large amounts of text data, and is an algorithm for natural language analysis and keyword extraction.

[0013] A "database" is a storage device or software that organizes and stores information such as documents, past conversation history, and emails held by users. [Brief explanation of the drawings]

[0014] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0015] 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.

[0016] First, the terms used in the following description will be explained.

[0017] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0018] 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.

[0019] 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.

[0020] 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.

[0021] 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."

[0022] [First embodiment]

[0023] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0024] 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.

[0025] 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).

[0026] 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.

[0027] 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.

[0028] 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.

[0029] 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.

[0030] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0031] 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.

[0032] 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.

[0033] 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.

[0034] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0035] The system according to the present invention is designed to improve the efficiency of meetings by recognizing and analyzing comments made during meetings in real time and providing relevant information. Specific embodiments of the present invention are described below.

[0036] Building a database

[0037] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[0038] Speech Recognition and Analysis

[0039] The device captures the audio during the meeting in real time. The captured audio data is sent to a speech recognition API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM). The LLM extracts important keywords and phrases from the text data.

[0040] Real-time search and information provision

[0041] The server searches the database for relevant information based on the keywords extracted by the analysis means. The retrieved information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is sent to the terminal and displayed to the user.

[0042] Automatic documentation of meeting content

[0043] The server automatically documents what is said during the meeting and the information obtained as a result. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The user receives a notification and can review and edit the minutes.

[0044] Specific examples

[0045] Interaction during the meeting

[0046] 1. User: "Oh, what's last month's sales data?"

[0047] 2. The device captures this utterance and sends it to the speech recognition API.

[0048] 3. The speech recognition API converts the speech into text ("What was last month's sales data?") and sends it to the server.

[0049] 4. The server inputs the text into LLM and extracts related keywords such as "last month" and "sales data."

[0050] 5. The server searches the database based on these keywords to retrieve relevant sales data.

[0051] 6. The server sends the acquired sales data to the terminal, which displays it to the user.

[0052] 7. User: "That's the data, thank you," and continues the meeting.

[0053] Post-conference report materials

[0054] 1. The server generates minutes of the meeting based on the speech and speech-recognized text during the meeting.

[0055] 2. The generated minutes are linked to the user's account by the server.

[0056] 3. The server sends the user a notification after the meeting ends, including a link to the minutes.

[0057] 4. The user confirms the notification and opens the generated minutes.

[0058] 5. If necessary, the user can edit the minutes and use them for the next meeting or report.

[0059] In this way, the present invention not only quickly provides necessary materials and information during a meeting, but also automates the creation of minutes after the meeting, thereby significantly improving work efficiency.

[0060] The processing flow will be explained below.

[0061] Step 1:

[0062] The server receives user-provided materials, past conversation history, and email data, imports them into a database, checks that the data is properly formatted, and reports errors if there are any anomalies.

[0063] Step 2:

[0064] The server analyzes the content of the imported data and automatically generates related keywords and tags to add them as metadata. Based on this metadata, it builds an index to improve search performance.

[0065] Step 3:

[0066] The device continues to capture audio during the meeting in real time, and the captured audio data is sent to the speech recognition API at predetermined intervals.

[0067] Step 4:

[0068] The speech recognition API converts the received speech data into text data and sends the converted text data to the server.

[0069] Step 5:

[0070] The server then inputs the received text data into a large-scale language model (LLM) to analyze the conversation, extracting important keywords and phrases from the text.

[0071] Step 6:

[0072] The server searches for relevant information from an indexed database based on the extracted keywords or phrases, retrieves the search results, and ranks them based on relevance and priority.

[0073] Step 7:

[0074] The server selects the most relevant information from the ranked search results and sends it to the device, based on criteria such as keyword match and importance of the material.

[0075] Step 8:

[0076] The device visually displays the received relevant information and materials to the user, and can display the information in a pop-up display or a new window on the user's screen.

[0077] Step 9:

[0078] The server automatically generates meeting minutes based on the voice data captured during the meeting and the corresponding voice recognition results, and the generated minutes are automatically associated with the user's account after the meeting ends.

[0079] Step 10:

[0080] The server notifies the user of the generated minutes and sends a link so that the user can view and edit the minutes.

[0081] Step 11:

[0082] The user opens the notification, reviews the automatically generated minutes, and, if necessary, completes the minutes by checking for errors and adding additional comments or corrections.

[0083] In this way, by performing specific operations for each processing step, the system according to the present invention can efficiently provide information during a meeting and generate minutes after the meeting.

[0084] Example 1

[0085] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0086] Conventional conferencing systems lack the ability to recognize and analyze what is said during a meeting in real time and provide relevant information, resulting in reduced meeting efficiency. Meeting minutes also need to be created manually after the meeting, which also requires time and effort. Furthermore, there are insufficient search and ranking methods to ensure the accuracy and relevance of the information provided in real time. To solve these problems, a more efficient and accurate meeting support system is needed.

[0087] 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.

[0088] In this invention, the server includes a speech recognition unit that recognizes speech during a conference in real time, a conversion unit that converts speech content recognized by the speech recognition unit into text data, an analysis unit that analyzes the text data converted by the conversion unit, a search unit that searches a database for related information based on keywords extracted by the analysis unit, a display unit that displays the related information acquired by the search unit, a documentation unit that automatically documents the conference content, a link unit that links the document generated by the documentation unit to a user account, a notification unit that sends a notification to the user after the conference ends, a speech capture unit that captures speech using a microphone, a transmission unit that transmits the captured speech data to a speech recognition API, and a ranking unit that ranks the results of the database search. This allows speech during a conference to be recognized in real time and related information to be quickly provided, thereby improving the efficiency of the conference and automatically generating and notifying minutes after the conference ends.

[0089] "Speech recognition means" refers to a device or program that recognizes speech during a meeting in real time and acquires the speech data.

[0090] The "conversion means" refers to a device or program that converts the speech content recognized by the speech recognition means into text data.

[0091] The "analysis means" refers to a device or program that analyzes the text data converted by the conversion means and extracts important keywords and phrases.

[0092] The "search means" refers to a device or program that searches a database for related information based on the keywords extracted by the analysis means.

[0093] The "display means" refers to a device or program that visually displays the related information obtained by the search means to the user.

[0094] "Documentation means" refers to a device or program that automatically documents the contents of a meeting and generates meeting minutes.

[0095] "Linking means" refers to a device or program that associates the document generated by the documentation means with a user account.

[0096] "Notification means" refers to a device or program that sends a notification to the user after the conference has ended.

[0097] "Audio capture means" refers to a device or program that uses a microphone to capture audio during a meeting.

[0098] "Transmission means" refers to the device or program that transmits the captured voice data to the voice recognition API.

[0099] "Ranking means" refers to a device or program that ranks the results of a database search based on relevance or priority.

[0100] A "large-scale language model" is a natural language processing model generated based on massive amounts of data, and has the ability to extract important keywords and phrases from text data.

[0101] The system according to the present invention is designed to improve the efficiency of meetings by recognizing and analyzing comments made during meetings in real time and providing relevant information. Specific embodiments of the present invention will be described below.

[0102] Building a database

[0103] The server imports users' documents, past conversation history, and email data. This data is stored in a database such as MongoDB. The server then analyzes the imported data using Python's natural language processing (NLP) library (e.g., NLTK or spaCy) and assigns relevant keywords and tags based on the content of each document. This enables relevant information to be provided quickly and accurately when searching.

[0104] Speech Recognition and Analysis

[0105] The device uses a microphone to capture audio during the meeting in real time. The captured audio data is sent to the Google® Speech-to-Text API and converted into text data. This text data is then sent to a server. The server then analyzes the text data using a large-scale language model (e.g., OpenAI®'s GPT-3®) to extract important keywords and phrases.

[0106] Real-time search and information provision

[0107] The server searches the database based on the extracted keywords to retrieve relevant information. This information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is then sent to the terminal and displayed to the user.

[0108] Automatic documentation of meeting content

[0109] The server automatically documents what is said and the information obtained during the meeting. This process uses a Python document generation tool (e.g., ReportLab). After the meeting ends, the generated minutes are linked to the user's account, and a notification is sent via email or a chat tool (e.g., Slack). The user receives a notification and can review and edit the minutes.

[0110] Specific examples

[0111] For example, if a user says "I want to know last month's sales data" during a meeting, the device captures this utterance and sends the audio data to the Google Speech-to-Text API. The speech recognition API converts the speech into text "I want to know last month's sales data" and sends it to the server. The server inputs this text into GPT-3 and extracts the keywords "last month" and "sales data." The server then searches the database based on these keywords to retrieve relevant sales data. The retrieved data is sent to the device and displayed to the user. The user can then say, "That's the data, thank you," and continue the meeting.

[0112] In this invention, examples of prompt sentences include:

[0113] "I want to know last month's sales data."

[0114] "Tell me about the latest market trends"

[0115] "What's the next step for this project?"

[0116] In this way, the present invention not only quickly provides necessary materials and information during a meeting, but also automates the creation of minutes after the meeting, thereby significantly improving work efficiency.

[0117] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0118] Step 1:

[0119] The terminal uses a microphone to capture audio during the meeting in real time.

[0120] Input: Speech during the meeting (audio data)

[0121] Action: Capture audio data with a microphone

[0122] Output: Captured audio data

[0123] Step 2:

[0124] The device sends the captured audio data to the Google Speech-to-Text API.

[0125] Input: Captured audio data

[0126] What it does: Sends audio data to the Google Speech-to-Text API

[0127] Output: Audio data is sent to the API

[0128] Step 3:

[0129] The speech recognition API converts the speech data into text data and sends the text data to the server.

[0130] Input: Transmitted audio data

[0131] Operation: Converts voice data into text data

[0132] Output: Converted text data

[0133] Step 4:

[0134] The server inputs text data sent from the Google Speech-to-Text API into GPT-3.

[0135] Input: Converted text data

[0136] Action: Input text data into GPT-3

[0137] Output: Data analyzed by GPT-3

[0138] Step 5:

[0139] The server extracts important keywords and phrases from the data analyzed by GPT-3.

[0140] Input: Parsed data

[0141] What it does: Extracts important keywords and phrases

[0142] Output: Extracted keywords and phrases

[0143] Step 6:

[0144] The server searches the MongoDB database based on the extracted keywords.

[0145] Input: Extracted keywords or phrases

[0146] How it works: Generating and executing database search queries

[0147] Output: Related search results

[0148] Step 7:

[0149] The server ranks the search results based on relevance and priority to select the most suitable results.

[0150] Input: Search results

[0151] Behavior: Ranking based on relevance and priority

[0152] Output: Best results

[0153] Step 8:

[0154] The server transmits the selected information to the terminal.

[0155] Input: Best results

[0156] Action: Send information to device

[0157] Output: Sending information to a terminal

[0158] Step 9:

[0159] The terminal displays the received information to the user.

[0160] Input: Submitted information

[0161] Action: Display information

[0162] Output: Display data that the user can see

[0163] Step 10:

[0164] The server generates meeting minutes using Python's ReportLab based on comments made during the meeting and the information obtained.

[0165] Input: What was said during the meeting and what information was captured

[0166] Action: Generate minutes

[0167] Output: Generated meeting minutes

[0168] Step 11:

[0169] The server links the generated minutes to the user's account and sends a notification.

[0170] Input: Generated minutes

[0171] Action: Send meeting minutes link and notification

[0172] Output: User notification

[0173] Step 12:

[0174] Users receive notifications and can review and edit the minutes.

[0175] Input: Notified link

[0176] Action: Check and edit minutes

[0177] Output: Edited transcript

[0178] In this way, the system of the present invention can recognize and analyze statements made during a meeting in real time, quickly provide relevant information, and automatically generate and notify minutes after the meeting has ended.

[0179] (Application example 1)

[0180] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0181] Modern conference systems provide functions such as speech recognition, information search, and meeting minutes generation, but they lack the ability to provide product information in physical stores or real-time reference to inventory information. Therefore, there is a need for systems that not only provide effective information during meetings but also improve everyday shopping experiences. Furthermore, by creating an environment where customers can obtain product information in real time, it is necessary to improve shopping efficiency and satisfaction.

[0182] 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.

[0183] In this invention, the server includes a speech recognition means for recognizing utterances made during a conference in real time, a conversion means for converting the utterances recognized by the speech recognition means into text data, an analysis means for analyzing the text data converted by the conversion means, a display means for providing detailed product information and inventory information in a physical store, and a documentation means for automatically documenting the conference contents. This not only enables efficient information provision during a conference, but also improves the customer experience in a physical store and makes it possible to provide product information and inventory information in real time.

[0184] "Speech recognition means" refers to a device or software that recognizes speech during a meeting in real time.

[0185] The "conversion means" is a device or software that has the function of converting the speech content recognized by the speech recognition means into text data.

[0186] The "analysis means" is a device or software that has the function of analyzing the text data converted by the conversion means and extracting related keywords and phrases.

[0187] The "search means" is a device or software that has the function of searching a database for related information based on the keywords extracted by the analysis means.

[0188] The "display means" is a device or software for displaying the related information obtained by the search means to the user.

[0189] "Documentation means" refers to a device or software that has the function of automatically documenting the contents of a meeting.

[0190] A "large-scale language model" is an advanced natural language processing model used to analyze text data, and is trained on large amounts of text data.

[0191] "Display means for providing detailed product information and inventory information in a physical store" refers to a device or software for displaying detailed product information and inventory information in real time in a physical store based on customer requests.

[0192] The system of the present invention recognizes comments made during a meeting in real time, provides relevant information, and improves meeting efficiency. Furthermore, by incorporating a function to provide real-time product information and inventory information in physical stores, the customer shopping experience can be improved. Specific embodiments are described below.

[0193] System configuration

[0194] The system consists of the following components:

[0195] Voice recognition means

[0196] Recognizes what is said in a meeting in real time. This feature uses a microphone to capture what the user is saying and speech recognition software to process the audio data.

[0197] Conversion Method

[0198] The speech recognized by the speech recognition means is converted into text data. In this process, a speech recognition API (e.g., Google Cloud Speech-to-Text API) is used.

[0199] Analysis means

[0200] The text data converted by the conversion method is analyzed to extract related keywords and phrases. Here, a large-scale language model (a generative AI model such as GPT-2) is used.

[0201] Search methods

[0202] Based on the keywords extracted by the analysis means, related information is searched from a database, which may include meeting materials, sales data, product inventory information, etc.

[0203] Display means

[0204] Related information obtained by the search tool is displayed to the user using a smartphone, tablet, smart glasses, or other device.

[0205] Documentation Methods

[0206] Automatically document meeting content. This function automatically generates meeting minutes based on speech recognition and text data.

[0207] How to operate the system

[0208] The specific operation method is to follow the steps below.

[0209] 1. Speech recognition during meetings

[0210] When the user says, "What other colors do you have for this item?", the voice recognition means captures this.

[0211] 2. Audio data conversion

[0212] The voice data is converted into text data via a voice recognition API.

[0213] 3. Text Data Analysis

[0214] The converted text data is analyzed by an analysis means, and important keywords (in this case, "product" and "color") are extracted.

[0215] 4. Information Search

[0216] Based on the extracted keywords, the search means searches the database for related product information and inventory information.

[0217] 5. Information display

[0218] The searched information is provided to the user by a display means. For example, a message such as "This product is available in red, blue, and green" is displayed on the screen of a smartphone.

[0219] Examples of concrete examples and prompts

[0220] As a specific example, the response when a user asks a question about a product will be shown.

[0221] Specific examples

[0222] User: "What other colors does this product come in?"

[0223] App: "This item comes in red, blue, and green."

[0224] Prompt Sentence Examples

[0225] "Generate a response when the user requests a different color."

[0226] In this way, the system of the present invention allows users to quickly and accurately obtain necessary information not only during meetings but also in physical stores, which not only improves the efficiency of meetings but also the customer's shopping experience.

[0227] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0228] Step 1:

[0229] Capture speech

[0230] The user says, "What other colors do you have for this product?" The device (such as a smartphone or smart glasses) captures this speech through a microphone. The input is speech data, which is collected for conversion into text data in a later step.

[0231] Step 2:

[0232] Voice Recognition

[0233] The captured voice data is sent from the device to a voice recognition API (for example, Google Cloud Speech-to-Text API). This API converts the voice data into text data and returns the conversion result to the device. The input is voice data and the output is text data.

[0234] Step 3:

[0235] Sending text data

[0236] The converted text data is sent from the terminal to the server. The server receives this text data and proceeds to the next analysis process. The input is text data, and the output is data transfer to the server.

[0237] Step 4:

[0238] Text data analysis

[0239] The server inputs the received text data into an analysis means (a generative AI model, such as GPT-2). The analysis means analyzes the text data and extracts important keywords and phrases (in this case, "product" and "color"). The input is the text data, and the output is the extracted keywords.

[0240] Step 5:

[0241] Information Search

[0242] The server uses a search engine to search for related product information and inventory information from a database based on the extracted keywords. The database contains past conversation history, product information, inventory information, etc. The input is the keywords, and the output is related information.

[0243] Step 6:

[0244] Information ranking

[0245] The server ranks the retrieved information based on relevance and priority. The most appropriate information is selected and sent to the device in the next step. The input is the relevant information, and the output is the ranked information.

[0246] Step 7:

[0247] Providing information

[0248] The server sends the ranking information to the terminal. The terminal receives it and displays it to the user. The user sees a message on the screen of their smartphone saying, "This product is available in red, blue, and green." The input is the ranking information, and the output is the display content.

[0249] Step 8:

[0250] Documenting meeting content

[0251] The server automatically generates meeting minutes based on comments made during the meeting and related information. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The input is comments and related information, and the output is the minutes and notification.

[0252] 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.

[0253] The system of the present invention recognizes and analyzes comments made during a meeting in real time, provides relevant information, and recognizes the user's emotions, presents information and creates minutes accordingly, thereby improving the efficiency and effectiveness of the meeting. Specific embodiments of the present invention are described below.

[0254] Building a database

[0255] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[0256] Speech Recognition and Analysis

[0257] The device captures the audio during the meeting in real time. The captured audio data is sent to a speech recognition API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM). The LLM extracts important keywords and phrases from the text data.

[0258] Real-time search and information provision

[0259] The server searches the database for relevant information based on the keywords extracted by the analysis means. The retrieved information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is sent to the terminal and displayed to the user.

[0260] Automatic documentation of meeting content

[0261] The server automatically documents what is said during the meeting and the information obtained as a result. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The user receives a notification and can review and edit the minutes.

[0262] Introducing emotion recognition

[0263] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to the emotion engine, which analyzes and recognizes the user's emotions (e.g., excitement, confusion, interest, etc.).

[0264] Coordination of information provision

[0265] Based on the emotional data recognized by the emotion engine, the server automatically adjusts how it presents relevant information: for example, if the user is confused, it will provide additional explanations or detailed materials, while if the user expresses interest, it will provide relevant, in-depth information.

[0266] Emotion tagging

[0267] The recognized emotional information is also reflected in the minutes of the meeting. The server assigns emotional tags to each part of the minutes, allowing for a more detailed understanding of the progress of the meeting.

[0268] Specific examples

[0269] Interaction during the meeting

[0270] 1. User: "I'm not sure if this data is correct."

[0271] 2. The device captures this utterance and sends it to the speech recognition API.

[0272] 3. The speech recognition API converts the speech into text "It is unclear whether this data is correct." and sends it to the server.

[0273] 4. The server analyzes the text using a large-scale language model and extracts important keywords: "data," "correct," and "unknown."

[0274] 5. The server searches the database based on these keywords and retrieves the relevant verification procedure materials.

[0275] 6. The server ranks the retrieved materials and sends the most relevant information to the terminal.

[0276] 7. The terminal displays the relevant information to the user.

[0277] 8. At the same time, the device sends data to the emotion engine to analyze emotions from the user's facial expressions and voice.

[0278] 9. The emotion engine recognizes that the user is "confused" and issues an alert.

[0279] 10. The server sends the terminal again materials containing more detailed explanations based on the emotion recognition.

[0280] Post-conference report materials

[0281] 1. The server automatically generates meeting minutes based on the voice data captured during the meeting and the voice recognition results.

[0282] 2. The generated minutes are tagged with the recognized emotion tags along with the content of each part of the speech.

[0283] 3. The minutes are linked to the user's account by the server.

[0284] 4. The server sends a notification to the user after the meeting ends, including a link to the minutes.

[0285] 5. The user confirms the notification and opens the automatically generated minutes.

[0286] 6. If necessary, the user can edit the minutes and use them for the next meeting or report.

[0287] In this way, the system of the present invention, which combines an emotion engine, can further improve the efficiency and effectiveness of meetings by providing real-time information during the meeting and generating detailed minutes after the meeting, as well as recognizing the user's emotions and presenting information accordingly.

[0288] The processing flow will be explained below.

[0289] Step 1:

[0290] The server receives user-provided materials, past conversation history, and email data, imports them into a database, checks that the data is properly formatted, and reports errors if there are any anomalies.

[0291] Step 2:

[0292] The server analyzes the content of the imported data and automatically generates related keywords and tags to add them as metadata. Based on this metadata, it builds an index to improve search performance.

[0293] Step 3:

[0294] The device continues to capture audio during the meeting in real time, and the captured audio data is sent to the speech recognition API at predetermined intervals.

[0295] Step 4:

[0296] The speech recognition API converts the received speech data into text data and sends the converted text data to the server.

[0297] Step 5:

[0298] The server then inputs the received text data into a large-scale language model (LLM) to analyze the conversation, extracting important keywords and phrases from the text.

[0299] Step 6:

[0300] The device captures the user's voice, facial expressions, and physical movements in real time and transmits this data to the emotion engine.

[0301] Step 7:

[0302] The emotion engine analyzes the received data and recognizes and classifies the user's emotions (e.g., excitement, confusion, interest, etc.).

[0303] Step 8:

[0304] The server searches the database for relevant information based on the emotion data recognized by the emotion engine and the keywords extracted by the LLM, and then retrieves the search results and ranks them based on relevance and priority.

[0305] Step 9:

[0306] The server selects the most relevant information from the ranked search results and automatically adjusts the way the information is presented based on the user's emotions.

[0307] Step 10:

[0308] The server sends tailored information to the terminal, for example, sending additional explanations or detailed materials if the user is confused.

[0309] Step 11:

[0310] The device visually displays the received relevant information and materials to the user, and can display the information in a pop-up display or a new window on the user's screen.

[0311] Step 12:

[0312] The server automatically generates minutes of the meeting based on the voice data captured during the meeting and the corresponding voice recognition results.

[0313] Step 13:

[0314] The server assigns emotion tags to the generated minutes and automatically associates them with the user's account after the meeting ends.

[0315] Step 14:

[0316] After the meeting, the server sends a notification to the user with a link to the minutes, allowing the user to view and edit the minutes.

[0317] Step 15:

[0318] The user opens the notification, reviews the automatically generated minutes, and, if necessary, completes the minutes by checking for errors and adding additional comments or corrections.

[0319] In this way, by performing specific operations for each processing step, the system of the present invention can efficiently provide information during a meeting and generate minutes after the meeting. In addition, by recognizing the user's emotions and adjusting the information presentation method accordingly, the quality of the meeting can be further improved.

[0320] Example 2

[0321] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0322] Conventional meeting support systems could recognize comments in real time, present relevant information, and create minutes, but they had limitations in providing information that took user emotions into account and automatically documenting meeting content. In addition, there were cases where users were unable to provide appropriate support because they did not understand the emotions behind their comments. Furthermore, because the emotions of participants were not reflected in post-meeting minutes, it was difficult to accurately grasp the progress and effectiveness of the meeting.

[0323] 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.

[0324] In this invention, the server includes speech recognition means for recognizing utterances made during a conference in real time, conversion means for converting the utterances recognized by the speech recognition means into text data, analysis means for analyzing the text data converted by the conversion means, search means for searching a database for related information based on keywords extracted by the analysis means, display means for displaying the related information acquired by the search means, emotion recognition means for recognizing emotion data of users during the conference, means for adjusting the method of presenting information based on the emotion data acquired by the emotion recognition means, documentation means for automatically documenting the conference content, and means for reflecting the emotion information in documents automatically generated by the documentation means. This makes it possible to grasp the progress of the conference and the emotions of the participants in real time, and to provide more effective support during and after the conference.

[0325] "Speech recognition means" refers to a device or software for recognizing speech during a meeting in real time.

[0326] The "conversion means" refers to a device or software that has the function of converting the speech content recognized by the speech recognition means into text data.

[0327] The "analysis means" refers to a device or software for analyzing the text data converted by the conversion means and extracting important keywords and phrases.

[0328] The "search means" refers to a device or software for searching a database for related information based on the keywords extracted by the analysis means.

[0329] The "display means" refers to a device or software that visually presents the related information obtained by the search means to the user.

[0330] "Emotion recognition means" refers to a device or software for analyzing a user's voice, facial expression, or physical movement during a meeting to recognize emotional data.

[0331] "Emotional data" refers to data that expresses a user's emotional state (e.g., excitement, confusion, interest, etc.).

[0332] "Documentation tool" refers to a device or software that automatically documents what is said and related information during a meeting.

[0333] "Emotion information" refers to information indicating the emotional state of the user acquired by emotion recognition means.

[0334] "Large-scale language model" refers to a highly trained machine learning model used for natural language processing analysis.

[0335] The conference support system according to the present invention recognizes comments made during a conference in real time, provides related information, creates minutes, and recognizes the user's emotions and presents information based on those emotions.

[0336] Building a database

[0337] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[0338] For example, when a user uploads relevant materials before a meeting, the server updates the database based on those materials and automatically assigns necessary keywords and tags.

[0339] Speech Recognition and Analysis

[0340] The device captures audio during the meeting in real time. The captured audio data is sent to a speech recognition API such as Google Cloud Speech-to-Text API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM) such as OpenAI's GPT-4 (registered trademark). The LLM extracts important keywords and phrases from the text data.

[0341] For example, if a user says, "I'm not sure if this data is correct," the device captures it and sends it to a speech recognition API, which generates text data. This text data is then sent to a server and analyzed using a large-scale language model.

[0342] Real-time search and information provision

[0343] The server searches the database based on the keywords extracted by the analysis means. The search results are ranked based on relevance and priority, and the most relevant information is selected. The selected information is sent to the terminal and displayed to the user.

[0344] As a specific example, the server searches for relevant verification procedure documents based on keywords such as "data," "correct," and "unknown," and sends the most relevant information to the user's terminal.

[0345] Automatic documentation of meeting content

[0346] All statements made during the meeting and related information automatically acquired are documented by the server. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. Users can view and edit the minutes through the notification.

[0347] Specifically, after the meeting ends, the server automatically generates minutes and creates a link to the user's account. The contents of the meeting and related information are all organized in text format.

[0348] Introducing emotion recognition

[0349] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to a pre-installed emotion recognition engine, which analyzes and recognizes the user's emotions (e.g., excitement, confusion, interest, etc.).

[0350] For example, the device's camera and microphone capture the user's reactions and send the real-time data to the emotion engine, which then analyzes the user's emotions.

[0351] Coordination of information provision

[0352] The server adjusts the content and method of information presented based on the emotional data recognized by the emotion engine. For example, if the emotion engine determines that the user is confused, the server will present additional explanations or detailed materials. If the user shows interest, the server will provide related in-depth information.

[0353] As a specific example, if the emotion engine detects that the user is confused, the server retransmits detailed information to the terminal and displays it to the user.

[0354] Emotion tagging

[0355] The server reflects the emotional information recognized during the meeting in the minutes. By assigning appropriate emotional tags to each part of the minutes, the progress of the meeting can be understood in more detail.

[0356] For example, the server automatically generates emotion tags for each comment and inserts them into the minutes, recording not only the content of the meeting but also changes in participants' emotions.

[0357] By combining these functions, a system can provide real-time information during a meeting, generate detailed minutes after the meeting, and present information according to the user's emotions. Such a system will further improve the efficiency and effectiveness of meetings.

[0358] Prompt Sentence Examples

[0359] For example, you can check the system's operation by using the following prompt sentence:

[0360] It is unclear whether this data is correct.

[0361] The server analyzes the text data to search for relevant information, and also recognizes the user's emotions to adjust the way information is presented.

[0362] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0363] Step 1: Importing data and building the database

[0364] The server imports documents uploaded by users, past conversation history, and email data, and builds a database based on these. It analyzes the content of each document in the imported data and assigns relevant keywords and tags. The input is file data in various formats uploaded by users, and the output is an indexed database. Specifically, the server scans the documents and automatically extracts relevant keywords using NLP (natural language processing) technology, updating the database accordingly.

[0365] Step 2: Capture and convert audio

[0366] The device captures speech during the meeting in real time. The captured voice data is sent to a speech recognition API such as the Google Cloud Speech-to-Text API and converted into text data. The input is the real-time captured voice data, and the output is the recognized text data. Specifically, the microphone captures the voice and sends it to the speech recognition API, which then converts the voice into text and sends it back to the server.

[0367] Step 3: Analyzing the text data

[0368] The server receives the converted text data and analyzes it using an internal large-scale language model (for example, OpenAI's generative AI model). The analysis means extracts important keywords and phrases from the text data. The input is the text data received from the speech recognition API, and the output is the extracted keywords and phrases. Specifically, the large-scale language model analyzes the text and automatically extracts specific keywords and phrases from it.

[0369] Step 4: Find related information

[0370] The server searches for relevant information from a database based on the extracted keywords. The search results are ranked based on relevance and priority. The input is the extracted keywords, and the output is the ranked search results. Specifically, the server executes a database query to retrieve relevant materials and data, and ranks them based on priority.

[0371] Step 5: Viewing search results

[0372] The device displays the search results sent from the server to the user, allowing the user to see the most relevant information in real time. The input is the ranked search results sent from the server, and the output is the specific information displayed on the user's screen. In specific operation, the screen display portion of the device displays the relevant information through a user interface.

[0373] Step 6: Capturing and analyzing user emotions

[0374] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to an emotion recognition engine. The emotion recognition engine analyzes the acquired emotion data and recognizes the user's emotions. The input is the captured real-time user data, and the output is analyzed emotion data. Specifically, the camera and microphone capture the user's voice and facial movements and send this data to the emotion engine.

[0375] Step 7: Tailor your information based on emotions

[0376] The server adjusts the content and format of the information it presents based on the emotional data analyzed by the emotion recognition engine. For example, if it recognizes that the user is confused, the server will provide additional explanations or detailed materials. The input is the emotional data received from the emotion engine, and the output is the adjusted presentation of information. Specifically, the server dynamically changes the content and format of the information it presents based on the emotional data and sends it to the device.

[0377] Step 8: Automatically document meeting content

[0378] The server documents what was said during the meeting and any related information that was automatically acquired. After the meeting ends, this document is linked to the user's account as minutes, and a notification is sent. The input is the analyzed text data and related information, and the output is the automatically generated minutes. Specifically, the server collects, organizes, and documents the data from the meeting, saves the results in the user's account, and sends a notification.

[0379] Step 9: Sentiment tagging of meeting notes

[0380] The server also reflects the emotional information recognized during the meeting in the minutes. This allows each part of the minutes to be assigned an emotional tag. The input is the emotional data and the generated minutes, and the output is a detailed minutes with the emotional tags. Specifically, the server automatically generates an emotional tag corresponding to each utterance and adds it to the corresponding part of the minutes.

[0381] (Application example 2)

[0382] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0383] While conventional conferencing systems can recognize comments made during meetings in real time and provide relevant information, they provide information without considering the emotions of workers working on-site, making it difficult to provide efficient work support. There is also a need for a system that can appropriately recognize workers' emotions and stress levels and adjust and provide information accordingly.

[0384] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a voice recognition means for recognizing utterances made during a meeting in real time; a conversion means for converting the utterances recognized by the voice recognition means into text data; an analysis means for analyzing the text data converted by the conversion means; a search means for searching a database for related information based on keywords extracted by the analysis means; a display means for displaying the related information acquired by the search means; a documentation means for automatically documenting the meeting contents; an expression recognition means for capturing the facial expressions of workers at the workplace and recognizing their emotions in real time; and an emotion analysis means for adjusting the presentation method of related information based on the emotion data obtained by the expression recognition means. This enables efficient provision of information and work support in accordance with the emotions of workers.

[0385] The "voice recognition means" is a means for recognizing speech during a meeting in real time and converting the speech data into text.

[0386] The "conversion means" is a means for converting the speech content recognized by the speech recognition means into text data.

[0387] The "analysis means" is a means for analyzing the text data converted by the conversion means and extracting important keywords and phrases.

[0388] The "search means" is a means for searching the database for related information based on the keywords extracted by the analysis means.

[0389] The "display means" is a means for visually displaying the related information obtained by the search means to the user.

[0390] The "documentation means" is a means for automatically recording the contents of a meeting in a document format and notifying the user of the minutes generated after the meeting has ended.

[0391] The "facial expression recognition means" is a means for capturing the facial expressions of workers at the workplace and recognizing their emotions in real time.

[0392] The "emotion analysis means" is a means for adjusting the presentation method of related information based on the emotion data obtained by the facial expression recognition means.

[0393] The system of the present invention recognizes and analyzes worker conversations and emotions in real time during work meetings and daily tasks in a factory, and provides relevant work procedures, manuals, and past history information. Furthermore, by analyzing worker emotions and providing appropriate work instructions and supplementary explanations accordingly, it is possible to improve work efficiency and safety. This system includes the following hardware and software configuration:

[0394] First, the Google Cloud Speech-to-Text API is used as a speech recognition means to recognize statements made during meetings in real time. This converts the voice data into text data. The converted text data is sent to a server and analyzed using a large-scale language model (for example, OpenAI GPT). The analysis means extracts important keywords and phrases, and a search means operates to search a database for related information based on those keywords. The searched information is ranked based on relevance and priority, and the most appropriate information is sent from the server to the terminal and visually presented to the user via the terminal's display means. The display means can be a standard monitor, smartphone, or tablet.

[0395] Furthermore, a camera is used as an expression recognition means to capture the facial expressions of workers on the workplace. The captured facial expression data is analyzed using OpenCV to recognize the worker's emotions (e.g., confusion, excitement, interest, etc.) in real time. This emotion data is processed by the emotion analysis means, and the way related information is presented is automatically adjusted. For example, if the worker is confused, additional explanations or detailed materials are presented. On the other hand, if the worker shows interest, relevant in-depth information is provided.

[0396] The meeting content is automatically documented by the server, and meeting minutes are generated based on the captured audio data and its analysis. The generated minutes are linked to the user's account, and a notification is sent after the meeting ends.

[0397] Specific examples

[0398] Example 1: Providing information in response to a worker's confusion

[0399] 1. Worker: "I don't know how to set up this machine."

[0400] 2. The server captures this utterance and sends it to the speech recognition API.

[0401] 3. The speech recognition API converts the speech into text "I don't know how to set up this machine" and sends it to the server.

[0402] 4. The server analyzes the text using a large-scale language model and extracts important keywords such as "machine," "settings," and "don't know."

[0403] 5. The server searches the database based on these keywords and retrieves the relevant configuration instructions.

[0404] 6. At the same time, the facial expression recognition means captures the worker's facial expression and recognizes that he is confused.

[0405] 7. The server will provide further detailed configuration instructions and additional instructions.

[0406] Prompt Sentence Examples

[0407] User input: "I don't know how to configure this machine."

[0408] Prompt for generative AI model: "Please provide relevant instructions or manuals based on the following sentence: 'I don't know how to set up this machine.'"

[0409] In this way, the system according to the present invention recognizes the emotions of workers in real time and provides efficient and appropriate information and support accordingly.

[0410] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0411] Step 1:

[0412] A user speaks while working or in a meeting. This speech is captured in real time by the microphone of the terminal. The input is the user's voice, and the output is the captured voice data.

[0413] Step 2:

[0414] The device sends the captured voice data to the Google Cloud Speech-to-Text API, where the data is converted into text data. The output from the API is text data in string format.

[0415] Step 3:

[0416] The server receives the text data. This text data is then input into a large-scale language model such as OpenAI GPT to extract important keywords and phrases. This involves data calculations such as TEKST data analysis and keyword extraction, and the output is a list of keywords.

[0417] Step 4:

[0418] The server then searches a database based on the extracted keyword list to retrieve related information. The input is the keyword list, and the database search outputs related materials, procedure manuals, etc.

[0419] Step 5:

[0420] The server transmits the acquired related information to the terminal, where the data transfer operation is performed and the information is provided to the user by the display means. The information visually presented using the terminal's display is the output.

[0421] Step 6:

[0422] At the same time, the terminal uses a camera to capture the worker's facial expression, whose input is the worker's facial image data and whose output is the captured facial data.

[0423] Step 7:

[0424] The device uses OpenCV to recognize facial expressions and analyzes the facial expressions of the worker from the captured facial data. The input is facial data, and emotion analysis is performed as data calculation. The analysis results are output as emotional data such as confusion, interest, and excitement.

[0425] Step 8:

[0426] The server receives the emotion data and adjusts the presentation of related information based on it. For example, if the worker is confused, it will send a document with additional detailed explanations to the terminal again. The input data is emotion data, and the presentation information is adjusted as a data calculation. The output is the readjusted related information.

[0427] Step 9:

[0428] The server automatically documents the meeting contents. It generates meeting minutes using the captured audio data and the analysis results based on it. The input data is audio data and text data, and the data is processed into documentation. The output is the generated minutes.

[0429] Step 10:

[0430] The generated minutes are linked to the user's account by the server and a notification is sent after the meeting ends. The output is a notification with a link to the minutes.

[0431] Through the above steps, the system according to the present invention recognizes the emotions of workers in real time and provides efficient and appropriate information and support accordingly.

[0432] 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.

[0433] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0434] 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.

[0435] [Second embodiment]

[0436] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0437] 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.

[0438] 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).

[0439] 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.

[0440] 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.

[0441] 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).

[0442] 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.

[0443] 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.

[0444] 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.

[0445] 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.

[0446] In the smart glasses 214, the 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.

[0447] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0448] The system according to the present invention is designed to improve the efficiency of meetings by recognizing and analyzing comments made during meetings in real time and providing relevant information. Specific embodiments of the present invention are described below.

[0449] Building a database

[0450] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[0451] Speech Recognition and Analysis

[0452] The device captures the audio during the meeting in real time. The captured audio data is sent to a speech recognition API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM). The LLM extracts important keywords and phrases from the text data.

[0453] Real-time search and information provision

[0454] The server searches the database for relevant information based on the keywords extracted by the analysis means. The retrieved information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is sent to the terminal and displayed to the user.

[0455] Automatic documentation of meeting content

[0456] The server automatically documents what is said during the meeting and the information obtained as a result. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The user receives a notification and can review and edit the minutes.

[0457] Specific examples

[0458] Interaction during the meeting

[0459] 1. User: "Oh, what's last month's sales data?"

[0460] 2. The device captures this utterance and sends it to the speech recognition API.

[0461] 3. The speech recognition API converts the speech into text ("What was last month's sales data?") and sends it to the server.

[0462] 4. The server inputs the text into LLM and extracts related keywords such as "last month" and "sales data."

[0463] 5. The server searches the database based on these keywords to retrieve relevant sales data.

[0464] 6. The server sends the acquired sales data to the terminal, which displays it to the user.

[0465] 7. User: "That's the data, thank you," and continues the meeting.

[0466] Post-conference report materials

[0467] 1. The server generates minutes of the meeting based on the speech and speech-recognized text during the meeting.

[0468] 2. The generated minutes are linked to the user's account by the server.

[0469] 3. The server sends the user a notification after the meeting ends, including a link to the minutes.

[0470] 4. The user confirms the notification and opens the generated minutes.

[0471] 5. If necessary, the user can edit the minutes and use them for the next meeting or report.

[0472] In this way, the present invention not only quickly provides necessary materials and information during a meeting, but also automates the creation of minutes after the meeting, thereby significantly improving work efficiency.

[0473] The processing flow will be explained below.

[0474] Step 1:

[0475] The server receives user-provided materials, past conversation history, and email data, imports them into a database, checks that the data is properly formatted, and reports errors if there are any anomalies.

[0476] Step 2:

[0477] The server analyzes the content of the imported data and automatically generates related keywords and tags to add them as metadata. Based on this metadata, it builds an index to improve search performance.

[0478] Step 3:

[0479] The device continues to capture audio during the meeting in real time, and the captured audio data is sent to the speech recognition API at predetermined intervals.

[0480] Step 4:

[0481] The speech recognition API converts the received speech data into text data and sends the converted text data to the server.

[0482] Step 5:

[0483] The server then inputs the received text data into a large-scale language model (LLM) to analyze the conversation, extracting important keywords and phrases from the text.

[0484] Step 6:

[0485] The server searches for relevant information from an indexed database based on the extracted keywords or phrases, retrieves the search results, and ranks them based on relevance and priority.

[0486] Step 7:

[0487] The server selects the most relevant information from the ranked search results and sends it to the device, based on criteria such as keyword match and importance of the material.

[0488] Step 8:

[0489] The device visually displays the received relevant information and materials to the user, and can display the information in a pop-up display or a new window on the user's screen.

[0490] Step 9:

[0491] The server automatically generates meeting minutes based on the voice data captured during the meeting and the corresponding voice recognition results, and the generated minutes are automatically associated with the user's account after the meeting ends.

[0492] Step 10:

[0493] The server notifies the user of the generated minutes and sends a link so that the user can view and edit the minutes.

[0494] Step 11:

[0495] The user opens the notification, reviews the automatically generated minutes, and, if necessary, completes the minutes by checking for errors and adding additional comments or corrections.

[0496] In this way, by performing specific operations for each processing step, the system according to the present invention can efficiently provide information during a meeting and generate minutes after the meeting.

[0497] Example 1

[0498] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0499] Conventional conferencing systems lack the ability to recognize and analyze what is said during a meeting in real time and provide relevant information, resulting in reduced meeting efficiency. Meeting minutes also need to be created manually after the meeting, which also requires time and effort. Furthermore, there are insufficient search and ranking methods to ensure the accuracy and relevance of the information provided in real time. To solve these problems, a more efficient and accurate meeting support system is needed.

[0500] 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.

[0501] In this invention, the server includes a speech recognition unit that recognizes speech during a conference in real time, a conversion unit that converts speech content recognized by the speech recognition unit into text data, an analysis unit that analyzes the text data converted by the conversion unit, a search unit that searches a database for related information based on keywords extracted by the analysis unit, a display unit that displays the related information acquired by the search unit, a documentation unit that automatically documents the conference content, a link unit that links the document generated by the documentation unit to a user account, a notification unit that sends a notification to the user after the conference ends, a speech capture unit that captures speech using a microphone, a transmission unit that transmits the captured speech data to a speech recognition API, and a ranking unit that ranks the results of the database search. This allows speech during a conference to be recognized in real time and related information to be quickly provided, thereby improving the efficiency of the conference and automatically generating and notifying minutes after the conference ends.

[0502] "Speech recognition means" refers to a device or program that recognizes speech during a meeting in real time and acquires the speech data.

[0503] The "conversion means" refers to a device or program that converts the speech content recognized by the speech recognition means into text data.

[0504] The "analysis means" refers to a device or program that analyzes the text data converted by the conversion means and extracts important keywords and phrases.

[0505] The "search means" refers to a device or program that searches a database for related information based on the keywords extracted by the analysis means.

[0506] The "display means" refers to a device or program that visually displays the related information obtained by the search means to the user.

[0507] "Documentation means" refers to a device or program that automatically documents the contents of a meeting and generates meeting minutes.

[0508] "Linking means" refers to a device or program that associates the document generated by the documentation means with a user account.

[0509] "Notification means" refers to a device or program that sends a notification to the user after the conference has ended.

[0510] "Audio capture means" refers to a device or program that uses a microphone to capture audio during a meeting.

[0511] "Transmission means" refers to the device or program that transmits the captured voice data to the voice recognition API.

[0512] "Ranking means" refers to a device or program that ranks the results of a database search based on relevance or priority.

[0513] A "large-scale language model" is a natural language processing model generated based on massive amounts of data, and has the ability to extract important keywords and phrases from text data.

[0514] The system according to the present invention is designed to improve the efficiency of meetings by recognizing and analyzing comments made during meetings in real time and providing relevant information. Specific embodiments of the present invention will be described below.

[0515] Building a database

[0516] The server imports users' documents, past conversation history, and email data. This data is stored in a database such as MongoDB. The server then analyzes the imported data using Python's natural language processing (NLP) library (e.g., NLTK or spaCy) and assigns relevant keywords and tags based on the content of each document. This enables relevant information to be provided quickly and accurately when searching.

[0517] Speech Recognition and Analysis

[0518] The device uses a microphone to capture audio during the meeting in real time. The captured audio data is sent to the Google Speech-to-Text API and converted into text data. This text data is then sent to a server. The server then analyzes the text data using a large-scale language model (e.g., OpenAI's GPT-3) to extract important keywords and phrases.

[0519] Real-time search and information provision

[0520] The server searches the database based on the extracted keywords to retrieve relevant information. This information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is then sent to the terminal and displayed to the user.

[0521] Automatic documentation of meeting content

[0522] The server automatically documents what is said and the information obtained during the meeting. This process uses a Python document generation tool (e.g., ReportLab). After the meeting ends, the generated minutes are linked to the user's account, and a notification is sent via email or a chat tool (e.g., Slack). The user receives a notification and can review and edit the minutes.

[0523] Specific examples

[0524] For example, if a user says "I want to know last month's sales data" during a meeting, the device captures this utterance and sends the audio data to the Google Speech-to-Text API. The speech recognition API converts the speech into text "I want to know last month's sales data" and sends it to the server. The server inputs this text into GPT-3 and extracts the keywords "last month" and "sales data." The server then searches the database based on these keywords to retrieve relevant sales data. The retrieved data is sent to the device and displayed to the user. The user can then say, "That's the data, thank you," and continue the meeting.

[0525] In this invention, examples of prompt sentences include:

[0526] "I want to know last month's sales data."

[0527] "Tell me about the latest market trends"

[0528] "What's the next step for this project?"

[0529] In this way, the present invention not only quickly provides necessary materials and information during a meeting, but also automates the creation of minutes after the meeting, thereby significantly improving work efficiency.

[0530] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0531] Step 1:

[0532] The terminal uses a microphone to capture audio during the meeting in real time.

[0533] Input: Speech during the meeting (audio data)

[0534] Action: Capture audio data with a microphone

[0535] Output: Captured audio data

[0536] Step 2:

[0537] The device sends the captured audio data to the Google Speech-to-Text API.

[0538] Input: Captured audio data

[0539] What it does: Sends audio data to the Google Speech-to-Text API

[0540] Output: Audio data is sent to the API

[0541] Step 3:

[0542] The speech recognition API converts the speech data into text data and sends the text data to the server.

[0543] Input: Transmitted audio data

[0544] Operation: Converts voice data into text data

[0545] Output: Converted text data

[0546] Step 4:

[0547] The server inputs text data sent from the Google Speech-to-Text API into GPT-3.

[0548] Input: Converted text data

[0549] Action: Input text data into GPT-3

[0550] Output: Data analyzed by GPT-3

[0551] Step 5:

[0552] The server extracts important keywords and phrases from the data analyzed by GPT-3.

[0553] Input: Parsed data

[0554] What it does: Extracts important keywords and phrases

[0555] Output: Extracted keywords and phrases

[0556] Step 6:

[0557] The server searches the MongoDB database based on the extracted keywords.

[0558] Input: Extracted keywords or phrases

[0559] How it works: Generating and executing database search queries

[0560] Output: Related search results

[0561] Step 7:

[0562] The server ranks the search results based on relevance and priority to select the most suitable results.

[0563] Input: Search results

[0564] Behavior: Ranking based on relevance and priority

[0565] Output: Best results

[0566] Step 8:

[0567] The server transmits the selected information to the terminal.

[0568] Input: Best results

[0569] Action: Send information to device

[0570] Output: Sending information to a terminal

[0571] Step 9:

[0572] The terminal displays the received information to the user.

[0573] Input: Submitted information

[0574] Action: Display information

[0575] Output: Display data that the user can see

[0576] Step 10:

[0577] The server generates meeting minutes using Python's ReportLab based on comments made during the meeting and the information obtained.

[0578] Input: What was said during the meeting and what information was captured

[0579] Action: Generate minutes

[0580] Output: Generated meeting minutes

[0581] Step 11:

[0582] The server links the generated minutes to the user's account and sends a notification.

[0583] Input: Generated minutes

[0584] Action: Send meeting minutes link and notification

[0585] Output: User notification

[0586] Step 12:

[0587] Users receive notifications and can review and edit the minutes.

[0588] Input: Notified link

[0589] Action: Check and edit minutes

[0590] Output: Edited transcript

[0591] In this way, the system of the present invention can recognize and analyze statements made during a meeting in real time, quickly provide relevant information, and automatically generate and notify minutes after the meeting has ended.

[0592] (Application example 1)

[0593] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0594] Modern conference systems provide functions such as speech recognition, information search, and meeting minutes generation, but they lack the ability to provide product information in physical stores or real-time reference to inventory information. Therefore, there is a need for systems that not only provide effective information during meetings but also improve everyday shopping experiences. Furthermore, by creating an environment where customers can obtain product information in real time, it is necessary to improve shopping efficiency and satisfaction.

[0595] 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.

[0596] In this invention, the server includes a speech recognition means for recognizing utterances made during a conference in real time, a conversion means for converting the utterances recognized by the speech recognition means into text data, an analysis means for analyzing the text data converted by the conversion means, a display means for providing detailed product information and inventory information in a physical store, and a documentation means for automatically documenting the conference contents. This not only enables efficient information provision during a conference, but also improves the customer experience in a physical store and makes it possible to provide product information and inventory information in real time.

[0597] "Speech recognition means" refers to a device or software that recognizes speech during a meeting in real time.

[0598] The "conversion means" is a device or software that has the function of converting the speech content recognized by the speech recognition means into text data.

[0599] The "analysis means" is a device or software that has the function of analyzing the text data converted by the conversion means and extracting related keywords and phrases.

[0600] The "search means" is a device or software that has the function of searching a database for related information based on the keywords extracted by the analysis means.

[0601] The "display means" is a device or software for displaying the related information obtained by the search means to the user.

[0602] "Documentation means" refers to a device or software that has the function of automatically documenting the contents of a meeting.

[0603] A "large-scale language model" is an advanced natural language processing model used to analyze text data, and is trained on large amounts of text data.

[0604] "Display means for providing detailed product information and inventory information in a physical store" refers to a device or software for displaying detailed product information and inventory information in real time in a physical store based on customer requests.

[0605] The system of the present invention recognizes comments made during a meeting in real time, provides relevant information, and improves meeting efficiency. Furthermore, by incorporating a function to provide real-time product information and inventory information in physical stores, the customer shopping experience can be improved. Specific embodiments are described below.

[0606] System configuration

[0607] The system consists of the following components:

[0608] Voice recognition means

[0609] Recognizes what is said in a meeting in real time. This feature uses a microphone to capture what the user is saying and speech recognition software to process the audio data.

[0610] Conversion Method

[0611] The speech recognized by the speech recognition means is converted into text data. In this process, a speech recognition API (e.g., Google Cloud Speech-to-Text API) is used.

[0612] Analysis means

[0613] The text data converted by the conversion method is analyzed to extract related keywords and phrases. Here, a large-scale language model (a generative AI model such as GPT-2) is used.

[0614] Search methods

[0615] Based on the keywords extracted by the analysis means, related information is searched from a database, which may include meeting materials, sales data, product inventory information, etc.

[0616] Display means

[0617] Related information obtained by the search tool is displayed to the user using a smartphone, tablet, smart glasses, or other device.

[0618] Documentation Methods

[0619] Automatically document meeting content. This function automatically generates meeting minutes based on speech recognition and text data.

[0620] How to operate the system

[0621] The specific operation method is to follow the steps below.

[0622] 1. Speech recognition during meetings

[0623] When the user says, "What other colors do you have for this item?", the voice recognition means captures this.

[0624] 2. Audio data conversion

[0625] The voice data is converted into text data via a voice recognition API.

[0626] 3. Text Data Analysis

[0627] The converted text data is analyzed by an analysis means, and important keywords (in this case, "product" and "color") are extracted.

[0628] 4. Information Search

[0629] Based on the extracted keywords, the search means searches the database for related product information and inventory information.

[0630] 5. Information display

[0631] The searched information is provided to the user by a display means. For example, a message such as "This product is available in red, blue, and green" is displayed on the screen of a smartphone.

[0632] Examples of concrete examples and prompts

[0633] As a specific example, the response when a user asks a question about a product will be shown.

[0634] Specific examples

[0635] User: "What other colors does this product come in?"

[0636] App: "This item comes in red, blue, and green."

[0637] Prompt Sentence Examples

[0638] "Generate a response when the user requests a different color."

[0639] In this way, the system of the present invention allows users to quickly and accurately obtain necessary information not only during meetings but also in physical stores, which not only improves the efficiency of meetings but also the customer's shopping experience.

[0640] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0641] Step 1:

[0642] Capture speech

[0643] The user says, "What other colors do you have for this product?" The device (such as a smartphone or smart glasses) captures this speech through a microphone. The input is speech data, which is collected for conversion into text data in a later step.

[0644] Step 2:

[0645] Voice Recognition

[0646] The captured voice data is sent from the device to a voice recognition API (for example, Google Cloud Speech-to-Text API). This API converts the voice data into text data and returns the conversion result to the device. The input is voice data and the output is text data.

[0647] Step 3:

[0648] Sending text data

[0649] The converted text data is sent from the terminal to the server. The server receives this text data and proceeds to the next analysis process. The input is text data, and the output is data transfer to the server.

[0650] Step 4:

[0651] Text data analysis

[0652] The server inputs the received text data into an analysis means (a generative AI model, such as GPT-2). The analysis means analyzes the text data and extracts important keywords and phrases (in this case, "product" and "color"). The input is the text data, and the output is the extracted keywords.

[0653] Step 5:

[0654] Information Search

[0655] The server uses a search engine to search for related product information and inventory information from a database based on the extracted keywords. The database contains past conversation history, product information, inventory information, etc. The input is the keywords, and the output is related information.

[0656] Step 6:

[0657] Information ranking

[0658] The server ranks the retrieved information based on relevance and priority. The most appropriate information is selected and sent to the device in the next step. The input is the relevant information, and the output is the ranked information.

[0659] Step 7:

[0660] Providing information

[0661] The server sends the ranking information to the terminal. The terminal receives it and displays it to the user. The user sees a message on the screen of their smartphone saying, "This product is available in red, blue, and green." The input is the ranking information, and the output is the display content.

[0662] Step 8:

[0663] Documenting meeting content

[0664] The server automatically generates meeting minutes based on comments made during the meeting and related information. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The input is comments and related information, and the output is the minutes and notification.

[0665] 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.

[0666] The system of the present invention recognizes and analyzes comments made during a meeting in real time, provides relevant information, and recognizes the user's emotions, presents information and creates minutes accordingly, thereby improving the efficiency and effectiveness of the meeting. Specific embodiments of the present invention are described below.

[0667] Building a database

[0668] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[0669] Speech Recognition and Analysis

[0670] The device captures the audio during the meeting in real time. The captured audio data is sent to a speech recognition API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM). The LLM extracts important keywords and phrases from the text data.

[0671] Real-time search and information provision

[0672] The server searches the database for relevant information based on the keywords extracted by the analysis means. The retrieved information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is sent to the terminal and displayed to the user.

[0673] Automatic documentation of meeting content

[0674] The server automatically documents what is said during the meeting and the information obtained as a result. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The user receives a notification and can review and edit the minutes.

[0675] Introducing emotion recognition

[0676] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to the emotion engine, which analyzes and recognizes the user's emotions (e.g., excitement, confusion, interest, etc.).

[0677] Coordination of information provision

[0678] Based on the emotional data recognized by the emotion engine, the server automatically adjusts how it presents relevant information: for example, if the user is confused, it will provide additional explanations or detailed materials, while if the user expresses interest, it will provide relevant, in-depth information.

[0679] Emotion tagging

[0680] The recognized emotional information is also reflected in the minutes of the meeting. The server assigns emotional tags to each part of the minutes, allowing for a more detailed understanding of the progress of the meeting.

[0681] Specific examples

[0682] Interaction during the meeting

[0683] 1. User: "I'm not sure if this data is correct."

[0684] 2. The device captures this utterance and sends it to the speech recognition API.

[0685] 3. The speech recognition API converts the speech into text "It is unclear whether this data is correct." and sends it to the server.

[0686] 4. The server analyzes the text using a large-scale language model and extracts important keywords: "data," "correct," and "unknown."

[0687] 5. The server searches the database based on these keywords and retrieves the relevant verification procedure materials.

[0688] 6. The server ranks the retrieved materials and sends the most relevant information to the terminal.

[0689] 7. The terminal displays the relevant information to the user.

[0690] 8. At the same time, the device sends data to the emotion engine to analyze emotions from the user's facial expressions and voice.

[0691] 9. The emotion engine recognizes that the user is "confused" and issues an alert.

[0692] 10. The server sends the terminal again materials containing more detailed explanations based on the emotion recognition.

[0693] Post-conference report materials

[0694] 1. The server automatically generates meeting minutes based on the voice data captured during the meeting and the voice recognition results.

[0695] 2. The generated minutes are tagged with the recognized emotion tags along with the content of each part of the speech.

[0696] 3. The minutes are linked to the user's account by the server.

[0697] 4. The server sends a notification to the user after the meeting ends, including a link to the minutes.

[0698] 5. The user confirms the notification and opens the automatically generated minutes.

[0699] 6. If necessary, the user can edit the minutes and use them for the next meeting or report.

[0700] In this way, the system of the present invention, which combines an emotion engine, can further improve the efficiency and effectiveness of meetings by providing real-time information during the meeting and generating detailed minutes after the meeting, as well as recognizing the user's emotions and presenting information accordingly.

[0701] The processing flow will be explained below.

[0702] Step 1:

[0703] The server receives user-provided materials, past conversation history, and email data, imports them into a database, checks that the data is properly formatted, and reports errors if there are any anomalies.

[0704] Step 2:

[0705] The server analyzes the content of the imported data and automatically generates related keywords and tags to add them as metadata. Based on this metadata, it builds an index to improve search performance.

[0706] Step 3:

[0707] The device continues to capture audio during the meeting in real time, and the captured audio data is sent to the speech recognition API at predetermined intervals.

[0708] Step 4:

[0709] The speech recognition API converts the received speech data into text data and sends the converted text data to the server.

[0710] Step 5:

[0711] The server then inputs the received text data into a large-scale language model (LLM) to analyze the conversation, extracting important keywords and phrases from the text.

[0712] Step 6:

[0713] The device captures the user's voice, facial expressions, and physical movements in real time and transmits this data to the emotion engine.

[0714] Step 7:

[0715] The emotion engine analyzes the received data and recognizes and classifies the user's emotions (e.g., excitement, confusion, interest, etc.).

[0716] Step 8:

[0717] The server searches the database for relevant information based on the emotion data recognized by the emotion engine and the keywords extracted by the LLM, and then retrieves the search results and ranks them based on relevance and priority.

[0718] Step 9:

[0719] The server selects the most relevant information from the ranked search results and automatically adjusts the way the information is presented based on the user's emotions.

[0720] Step 10:

[0721] The server sends tailored information to the terminal, for example, sending additional explanations or detailed materials if the user is confused.

[0722] Step 11:

[0723] The device visually displays the received relevant information and materials to the user, and can display the information in a pop-up display or a new window on the user's screen.

[0724] Step 12:

[0725] The server automatically generates minutes of the meeting based on the voice data captured during the meeting and the corresponding voice recognition results.

[0726] Step 13:

[0727] The server assigns emotion tags to the generated minutes and automatically associates them with the user's account after the meeting ends.

[0728] Step 14:

[0729] After the meeting, the server sends a notification to the user with a link to the minutes, allowing the user to view and edit the minutes.

[0730] Step 15:

[0731] The user opens the notification, reviews the automatically generated minutes, and, if necessary, completes the minutes by checking for errors and adding additional comments or corrections.

[0732] In this way, by performing specific operations for each processing step, the system of the present invention can efficiently provide information during a meeting and generate minutes after the meeting. In addition, by recognizing the user's emotions and adjusting the information presentation method accordingly, the quality of the meeting can be further improved.

[0733] Example 2

[0734] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0735] Conventional meeting support systems could recognize comments in real time, present relevant information, and create minutes, but they had limitations in providing information that took user emotions into account and automatically documenting meeting content. In addition, there were cases where users were unable to provide appropriate support because they did not understand the emotions behind their comments. Furthermore, because the emotions of participants were not reflected in post-meeting minutes, it was difficult to accurately grasp the progress and effectiveness of the meeting.

[0736] 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.

[0737] In this invention, the server includes speech recognition means for recognizing utterances made during a conference in real time, conversion means for converting the utterances recognized by the speech recognition means into text data, analysis means for analyzing the text data converted by the conversion means, search means for searching a database for related information based on keywords extracted by the analysis means, display means for displaying the related information acquired by the search means, emotion recognition means for recognizing emotion data of users during the conference, means for adjusting the method of presenting information based on the emotion data acquired by the emotion recognition means, documentation means for automatically documenting the conference content, and means for reflecting the emotion information in documents automatically generated by the documentation means. This makes it possible to grasp the progress of the conference and the emotions of the participants in real time, and to provide more effective support during and after the conference.

[0738] "Speech recognition means" refers to a device or software for recognizing speech during a meeting in real time.

[0739] The "conversion means" refers to a device or software that has the function of converting the speech content recognized by the speech recognition means into text data.

[0740] The "analysis means" refers to a device or software for analyzing the text data converted by the conversion means and extracting important keywords and phrases.

[0741] The "search means" refers to a device or software for searching a database for related information based on the keywords extracted by the analysis means.

[0742] The "display means" refers to a device or software that visually presents the related information obtained by the search means to the user.

[0743] "Emotion recognition means" refers to a device or software for analyzing a user's voice, facial expression, or physical movement during a meeting to recognize emotional data.

[0744] "Emotional data" refers to data that expresses a user's emotional state (e.g., excitement, confusion, interest, etc.).

[0745] "Documentation tool" refers to a device or software that automatically documents what is said and related information during a meeting.

[0746] "Emotion information" refers to information indicating the emotional state of the user acquired by emotion recognition means.

[0747] "Large-scale language model" refers to a highly trained machine learning model used for natural language processing analysis.

[0748] The conference support system according to the present invention recognizes comments made during a conference in real time, provides related information, creates minutes, and recognizes the user's emotions and presents information based on those emotions.

[0749] Building a database

[0750] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[0751] For example, when a user uploads relevant materials before a meeting, the server updates the database based on those materials and automatically assigns necessary keywords and tags.

[0752] Speech Recognition and Analysis

[0753] The device captures audio during the meeting in real time. The captured audio data is sent to a speech recognition API such as Google Cloud Speech-to-Text API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM) such as OpenAI's GPT-4. The LLM extracts important keywords and phrases from the text data.

[0754] For example, if a user says, "I'm not sure if this data is correct," the device captures it and sends it to a speech recognition API, which generates text data. This text data is then sent to a server and analyzed using a large-scale language model.

[0755] Real-time search and information provision

[0756] The server searches the database based on the keywords extracted by the analysis means. The search results are ranked based on relevance and priority, and the most relevant information is selected. The selected information is sent to the terminal and displayed to the user.

[0757] As a specific example, the server searches for relevant verification procedure documents based on keywords such as "data," "correct," and "unknown," and sends the most relevant information to the user's terminal.

[0758] Automatic documentation of meeting content

[0759] All statements made during the meeting and related information automatically acquired are documented by the server. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. Users can view and edit the minutes through the notification.

[0760] Specifically, after the meeting ends, the server automatically generates minutes and creates a link to the user's account. The contents of the meeting and related information are all organized in text format.

[0761] Introducing emotion recognition

[0762] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to a pre-installed emotion recognition engine, which analyzes and recognizes the user's emotions (e.g., excitement, confusion, interest, etc.).

[0763] For example, the device's camera and microphone capture the user's reactions and send the real-time data to the emotion engine, which then analyzes the user's emotions.

[0764] Coordination of information provision

[0765] The server adjusts the content and method of information presented based on the emotional data recognized by the emotion engine. For example, if the emotion engine determines that the user is confused, the server will present additional explanations or detailed materials. If the user shows interest, the server will provide related in-depth information.

[0766] As a specific example, if the emotion engine detects that the user is confused, the server retransmits detailed information to the terminal and displays it to the user.

[0767] Emotion tagging

[0768] The server reflects the emotional information recognized during the meeting in the minutes. By assigning appropriate emotional tags to each part of the minutes, the progress of the meeting can be understood in more detail.

[0769] For example, the server automatically generates emotion tags for each comment and inserts them into the minutes, recording not only the content of the meeting but also changes in participants' emotions.

[0770] By combining these functions, a system can provide real-time information during a meeting, generate detailed minutes after the meeting, and present information according to the user's emotions. Such a system will further improve the efficiency and effectiveness of meetings.

[0771] Prompt Sentence Examples

[0772] For example, you can check the system's operation by using the following prompt sentence:

[0773] It is unclear whether this data is correct.

[0774] The server analyzes the text data to search for relevant information, and also recognizes the user's emotions to adjust the way information is presented.

[0775] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0776] Step 1: Importing data and building the database

[0777] The server imports documents uploaded by users, past conversation history, and email data, and builds a database based on these. It analyzes the content of each document in the imported data and assigns relevant keywords and tags. The input is file data in various formats uploaded by users, and the output is an indexed database. Specifically, the server scans the documents and automatically extracts relevant keywords using NLP (natural language processing) technology, updating the database accordingly.

[0778] Step 2: Capture and convert audio

[0779] The device captures speech during the meeting in real time. The captured voice data is sent to a speech recognition API such as the Google Cloud Speech-to-Text API and converted into text data. The input is the real-time captured voice data, and the output is the recognized text data. Specifically, the microphone captures the voice and sends it to the speech recognition API, which then converts the voice into text and sends it back to the server.

[0780] Step 3: Analyzing the text data

[0781] The server receives the converted text data and analyzes it using an internal large-scale language model (for example, OpenAI's generative AI model). The analysis means extracts important keywords and phrases from the text data. The input is the text data received from the speech recognition API, and the output is the extracted keywords and phrases. Specifically, the large-scale language model analyzes the text and automatically extracts specific keywords and phrases from it.

[0782] Step 4: Find related information

[0783] The server searches for relevant information from a database based on the extracted keywords. The search results are ranked based on relevance and priority. The input is the extracted keywords, and the output is the ranked search results. Specifically, the server executes a database query to retrieve relevant materials and data, and ranks them based on priority.

[0784] Step 5: Viewing search results

[0785] The device displays the search results sent from the server to the user, allowing the user to see the most relevant information in real time. The input is the ranked search results sent from the server, and the output is the specific information displayed on the user's screen. In specific operation, the screen display portion of the device displays the relevant information through a user interface.

[0786] Step 6: Capturing and analyzing user emotions

[0787] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to an emotion recognition engine. The emotion recognition engine analyzes the acquired emotion data and recognizes the user's emotions. The input is the captured real-time user data, and the output is analyzed emotion data. Specifically, the camera and microphone capture the user's voice and facial movements and send this data to the emotion engine.

[0788] Step 7: Tailor your information based on emotions

[0789] The server adjusts the content and format of the information it presents based on the emotional data analyzed by the emotion recognition engine. For example, if it recognizes that the user is confused, the server will provide additional explanations or detailed materials. The input is the emotional data received from the emotion engine, and the output is the adjusted presentation of information. Specifically, the server dynamically changes the content and format of the information it presents based on the emotional data and sends it to the device.

[0790] Step 8: Automatically document meeting content

[0791] The server documents what was said during the meeting and any related information that was automatically acquired. After the meeting ends, this document is linked to the user's account as minutes, and a notification is sent. The input is the analyzed text data and related information, and the output is the automatically generated minutes. Specifically, the server collects, organizes, and documents the data from the meeting, saves the results in the user's account, and sends a notification.

[0792] Step 9: Sentiment tagging of meeting notes

[0793] The server also reflects the emotional information recognized during the meeting in the minutes. This allows each part of the minutes to be assigned an emotional tag. The input is the emotional data and the generated minutes, and the output is a detailed minutes with the emotional tags. Specifically, the server automatically generates an emotional tag corresponding to each utterance and adds it to the corresponding part of the minutes.

[0794] (Application example 2)

[0795] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0796] While conventional conferencing systems can recognize comments made during meetings in real time and provide relevant information, they provide information without considering the emotions of workers working on-site, making it difficult to provide efficient work support. There is also a need for a system that can appropriately recognize workers' emotions and stress levels and adjust and provide information accordingly.

[0797] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a voice recognition means for recognizing utterances made during a meeting in real time; a conversion means for converting the utterances recognized by the voice recognition means into text data; an analysis means for analyzing the text data converted by the conversion means; a search means for searching a database for related information based on keywords extracted by the analysis means; a display means for displaying the related information acquired by the search means; a documentation means for automatically documenting the meeting contents; an expression recognition means for capturing the facial expressions of workers at the workplace and recognizing their emotions in real time; and an emotion analysis means for adjusting the presentation method of related information based on the emotion data obtained by the expression recognition means. This enables efficient provision of information and work support in accordance with the emotions of workers.

[0798] The "voice recognition means" is a means for recognizing speech during a meeting in real time and converting the speech data into text.

[0799] The "conversion means" is a means for converting the speech content recognized by the speech recognition means into text data.

[0800] The "analysis means" is a means for analyzing the text data converted by the conversion means and extracting important keywords and phrases.

[0801] The "search means" is a means for searching the database for related information based on the keywords extracted by the analysis means.

[0802] The "display means" is a means for visually displaying the related information obtained by the search means to the user.

[0803] The "documentation means" is a means for automatically recording the contents of a meeting in a document format and notifying the user of the minutes generated after the meeting has ended.

[0804] The "facial expression recognition means" is a means for capturing the facial expressions of workers at the workplace and recognizing their emotions in real time.

[0805] The "emotion analysis means" is a means for adjusting the presentation method of related information based on the emotion data obtained by the facial expression recognition means.

[0806] The system of the present invention recognizes and analyzes worker conversations and emotions in real time during work meetings and daily tasks in a factory, and provides relevant work procedures, manuals, and past history information. Furthermore, by analyzing worker emotions and providing appropriate work instructions and supplementary explanations accordingly, it is possible to improve work efficiency and safety. This system includes the following hardware and software configuration:

[0807] First, the Google Cloud Speech-to-Text API is used as a speech recognition means to recognize statements made during meetings in real time. This converts the voice data into text data. The converted text data is sent to a server and analyzed using a large-scale language model (for example, OpenAI GPT). The analysis means extracts important keywords and phrases, and a search means operates to search a database for related information based on those keywords. The searched information is ranked based on relevance and priority, and the most appropriate information is sent from the server to the terminal and visually presented to the user via the terminal's display means. The display means can be a standard monitor, smartphone, or tablet.

[0808] Furthermore, a camera is used as an expression recognition means to capture the facial expressions of workers on the workplace. The captured facial expression data is analyzed using OpenCV to recognize the worker's emotions (e.g., confusion, excitement, interest, etc.) in real time. This emotion data is processed by the emotion analysis means, and the way related information is presented is automatically adjusted. For example, if the worker is confused, additional explanations or detailed materials are presented. On the other hand, if the worker shows interest, relevant in-depth information is provided.

[0809] The meeting content is automatically documented by the server, and meeting minutes are generated based on the captured audio data and its analysis. The generated minutes are linked to the user's account, and a notification is sent after the meeting ends.

[0810] Specific examples

[0811] Example 1: Providing information in response to a worker's confusion

[0812] 1. Worker: "I don't know how to set up this machine."

[0813] 2. The server captures this utterance and sends it to the speech recognition API.

[0814] 3. The speech recognition API converts the speech into text "I don't know how to set up this machine" and sends it to the server.

[0815] 4. The server analyzes the text using a large-scale language model and extracts important keywords such as "machine," "settings," and "don't know."

[0816] 5. The server searches the database based on these keywords and retrieves the relevant configuration instructions.

[0817] 6. At the same time, the facial expression recognition means captures the worker's facial expression and recognizes that he is confused.

[0818] 7. The server will provide further detailed configuration instructions and additional instructions.

[0819] Prompt Sentence Examples

[0820] User input: "I don't know how to configure this machine."

[0821] Prompt for generative AI model: "Please provide relevant instructions or manuals based on the following sentence: 'I don't know how to set up this machine.'"

[0822] In this way, the system according to the present invention recognizes the emotions of workers in real time and provides efficient and appropriate information and support accordingly.

[0823] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0824] Step 1:

[0825] A user speaks while working or in a meeting. This speech is captured in real time by the microphone of the terminal. The input is the user's voice, and the output is the captured voice data.

[0826] Step 2:

[0827] The device sends the captured voice data to the Google Cloud Speech-to-Text API, where the data is converted into text data. The output from the API is text data in string format.

[0828] Step 3:

[0829] The server receives the text data. This text data is then input into a large-scale language model such as OpenAI GPT to extract important keywords and phrases. This involves data calculations such as TEKST data analysis and keyword extraction, and the output is a list of keywords.

[0830] Step 4:

[0831] The server then searches a database based on the extracted keyword list to retrieve related information. The input is the keyword list, and the database search outputs related materials, procedure manuals, etc.

[0832] Step 5:

[0833] The server transmits the acquired related information to the terminal, where the data transfer operation is performed and the information is provided to the user by the display means. The information visually presented using the terminal's display is the output.

[0834] Step 6:

[0835] At the same time, the terminal uses a camera to capture the worker's facial expression, whose input is the worker's facial image data and whose output is the captured facial data.

[0836] Step 7:

[0837] The device uses OpenCV to recognize facial expressions and analyzes the facial expressions of the worker from the captured facial data. The input is facial data, and emotion analysis is performed as data calculation. The analysis results are output as emotional data such as confusion, interest, and excitement.

[0838] Step 8:

[0839] The server receives the emotion data and adjusts the presentation of related information based on it. For example, if the worker is confused, it will send a document with additional detailed explanations to the terminal again. The input data is emotion data, and the presentation information is adjusted as a data calculation. The output is the readjusted related information.

[0840] Step 9:

[0841] The server automatically documents the meeting contents. It generates meeting minutes using the captured audio data and the analysis results based on it. The input data is audio data and text data, and the data is processed into documentation. The output is the generated minutes.

[0842] Step 10:

[0843] The generated minutes are linked to the user's account by the server and a notification is sent after the meeting ends. The output is a notification with a link to the minutes.

[0844] Through the above steps, the system according to the present invention recognizes the emotions of workers in real time and provides efficient and appropriate information and support accordingly.

[0845] 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.

[0846] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0847] 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.

[0848] [Third embodiment]

[0849] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0850] 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.

[0851] 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).

[0852] 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.

[0853] 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.

[0854] 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).

[0855] 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.

[0856] 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.

[0857] 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.

[0858] 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.

[0859] 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.

[0860] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0861] The system according to the present invention is designed to improve the efficiency of meetings by recognizing and analyzing comments made during meetings in real time and providing relevant information. Specific embodiments of the present invention are described below.

[0862] Building a database

[0863] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[0864] Speech Recognition and Analysis

[0865] The device captures the audio during the meeting in real time. The captured audio data is sent to a speech recognition API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM). The LLM extracts important keywords and phrases from the text data.

[0866] Real-time search and information provision

[0867] The server searches the database for relevant information based on the keywords extracted by the analysis means. The retrieved information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is sent to the terminal and displayed to the user.

[0868] Automatic documentation of meeting content

[0869] The server automatically documents what is said during the meeting and the information obtained as a result. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The user receives a notification and can review and edit the minutes.

[0870] Specific examples

[0871] Interaction during the meeting

[0872] 1. User: "Oh, what's last month's sales data?"

[0873] 2. The device captures this utterance and sends it to the speech recognition API.

[0874] 3. The speech recognition API converts the speech into text ("What was last month's sales data?") and sends it to the server.

[0875] 4. The server inputs the text into LLM and extracts related keywords such as "last month" and "sales data."

[0876] 5. The server searches the database based on these keywords to retrieve relevant sales data.

[0877] 6. The server sends the acquired sales data to the terminal, which displays it to the user.

[0878] 7. User: "That's the data, thank you," and continues the meeting.

[0879] Post-conference report materials

[0880] 1. The server generates minutes of the meeting based on the speech and speech-recognized text during the meeting.

[0881] 2. The generated minutes are linked to the user's account by the server.

[0882] 3. The server sends the user a notification after the meeting ends, including a link to the minutes.

[0883] 4. The user confirms the notification and opens the generated minutes.

[0884] 5. If necessary, the user can edit the minutes and use them for the next meeting or report.

[0885] In this way, the present invention not only quickly provides necessary materials and information during a meeting, but also automates the creation of minutes after the meeting, thereby significantly improving work efficiency.

[0886] The processing flow will be explained below.

[0887] Step 1:

[0888] The server receives user-provided materials, past conversation history, and email data, imports them into a database, checks that the data is properly formatted, and reports errors if there are any anomalies.

[0889] Step 2:

[0890] The server analyzes the content of the imported data and automatically generates related keywords and tags to add them as metadata. Based on this metadata, it builds an index to improve search performance.

[0891] Step 3:

[0892] The device continues to capture audio during the meeting in real time, and the captured audio data is sent to the speech recognition API at predetermined intervals.

[0893] Step 4:

[0894] The speech recognition API converts the received speech data into text data and sends the converted text data to the server.

[0895] Step 5:

[0896] The server then inputs the received text data into a large-scale language model (LLM) to analyze the conversation, extracting important keywords and phrases from the text.

[0897] Step 6:

[0898] The server searches for relevant information from an indexed database based on the extracted keywords or phrases, retrieves the search results, and ranks them based on relevance and priority.

[0899] Step 7:

[0900] The server selects the most relevant information from the ranked search results and sends it to the device, based on criteria such as keyword match and importance of the material.

[0901] Step 8:

[0902] The device visually displays the received relevant information and materials to the user, and can display the information in a pop-up display or a new window on the user's screen.

[0903] Step 9:

[0904] The server automatically generates meeting minutes based on the voice data captured during the meeting and the corresponding voice recognition results, and the generated minutes are automatically associated with the user's account after the meeting ends.

[0905] Step 10:

[0906] The server notifies the user of the generated minutes and sends a link so that the user can view and edit the minutes.

[0907] Step 11:

[0908] The user opens the notification, reviews the automatically generated minutes, and, if necessary, completes the minutes by checking for errors and adding additional comments or corrections.

[0909] In this way, by performing specific operations for each processing step, the system according to the present invention can efficiently provide information during a meeting and generate minutes after the meeting.

[0910] Example 1

[0911] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0912] Conventional conferencing systems lack the ability to recognize and analyze what is said during a meeting in real time and provide relevant information, resulting in reduced meeting efficiency. Meeting minutes also need to be created manually after the meeting, which also requires time and effort. Furthermore, there are insufficient search and ranking methods to ensure the accuracy and relevance of the information provided in real time. To solve these problems, a more efficient and accurate meeting support system is needed.

[0913] 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.

[0914] In this invention, the server includes a speech recognition unit that recognizes speech during a conference in real time, a conversion unit that converts speech content recognized by the speech recognition unit into text data, an analysis unit that analyzes the text data converted by the conversion unit, a search unit that searches a database for related information based on keywords extracted by the analysis unit, a display unit that displays the related information acquired by the search unit, a documentation unit that automatically documents the conference content, a link unit that links the document generated by the documentation unit to a user account, a notification unit that sends a notification to the user after the conference ends, a speech capture unit that captures speech using a microphone, a transmission unit that transmits the captured speech data to a speech recognition API, and a ranking unit that ranks the results of the database search. This allows speech during a conference to be recognized in real time and related information to be quickly provided, thereby improving the efficiency of the conference and automatically generating and notifying minutes after the conference ends.

[0915] "Speech recognition means" refers to a device or program that recognizes speech during a meeting in real time and acquires the speech data.

[0916] The "conversion means" refers to a device or program that converts the speech content recognized by the speech recognition means into text data.

[0917] The "analysis means" refers to a device or program that analyzes the text data converted by the conversion means and extracts important keywords and phrases.

[0918] The "search means" refers to a device or program that searches a database for related information based on the keywords extracted by the analysis means.

[0919] The "display means" refers to a device or program that visually displays the related information obtained by the search means to the user.

[0920] "Documentation means" refers to a device or program that automatically documents the contents of a meeting and generates meeting minutes.

[0921] "Linking means" refers to a device or program that associates the document generated by the documentation means with a user account.

[0922] "Notification means" refers to a device or program that sends a notification to the user after the conference has ended.

[0923] "Audio capture means" refers to a device or program that uses a microphone to capture audio during a meeting.

[0924] "Transmission means" refers to the device or program that transmits the captured voice data to the voice recognition API.

[0925] "Ranking means" refers to a device or program that ranks the results of a database search based on relevance or priority.

[0926] A "large-scale language model" is a natural language processing model generated based on massive amounts of data, and has the ability to extract important keywords and phrases from text data.

[0927] The system according to the present invention is designed to improve the efficiency of meetings by recognizing and analyzing comments made during meetings in real time and providing relevant information. Specific embodiments of the present invention will be described below.

[0928] Building a database

[0929] The server imports users' documents, past conversation history, and email data. This data is stored in a database such as MongoDB. The server then analyzes the imported data using Python's natural language processing (NLP) library (e.g., NLTK or spaCy) and assigns relevant keywords and tags based on the content of each document. This enables relevant information to be provided quickly and accurately when searching.

[0930] Speech Recognition and Analysis

[0931] The device uses a microphone to capture audio during the meeting in real time. The captured audio data is sent to the Google Speech-to-Text API and converted into text data. This text data is then sent to a server. The server then analyzes the text data using a large-scale language model (e.g., OpenAI's GPT-3) to extract important keywords and phrases.

[0932] Real-time search and information provision

[0933] The server searches the database based on the extracted keywords to retrieve relevant information. This information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is then sent to the terminal and displayed to the user.

[0934] Automatic documentation of meeting content

[0935] The server automatically documents what is said and the information obtained during the meeting. This process uses a Python document generation tool (e.g., ReportLab). After the meeting ends, the generated minutes are linked to the user's account, and a notification is sent via email or a chat tool (e.g., Slack). The user receives a notification and can review and edit the minutes.

[0936] Specific examples

[0937] For example, if a user says "I want to know last month's sales data" during a meeting, the device captures this utterance and sends the audio data to the Google Speech-to-Text API. The speech recognition API converts the speech into text "I want to know last month's sales data" and sends it to the server. The server inputs this text into GPT-3 and extracts the keywords "last month" and "sales data." The server then searches the database based on these keywords to retrieve relevant sales data. The retrieved data is sent to the device and displayed to the user. The user can then say, "That's the data, thank you," and continue the meeting.

[0938] In this invention, examples of prompt sentences include:

[0939] "I want to know last month's sales data."

[0940] "Tell me about the latest market trends"

[0941] "What's the next step for this project?"

[0942] In this way, the present invention not only quickly provides necessary materials and information during a meeting, but also automates the creation of minutes after the meeting, thereby significantly improving work efficiency.

[0943] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0944] Step 1:

[0945] The terminal uses a microphone to capture audio during the meeting in real time.

[0946] Input: Speech during the meeting (audio data)

[0947] Action: Capture audio data with a microphone

[0948] Output: Captured audio data

[0949] Step 2:

[0950] The device sends the captured audio data to the Google Speech-to-Text API.

[0951] Input: Captured audio data

[0952] What it does: Sends audio data to the Google Speech-to-Text API

[0953] Output: Audio data is sent to the API

[0954] Step 3:

[0955] The speech recognition API converts the speech data into text data and sends the text data to the server.

[0956] Input: Transmitted audio data

[0957] Operation: Converts voice data into text data

[0958] Output: Converted text data

[0959] Step 4:

[0960] The server inputs text data sent from the Google Speech-to-Text API into GPT-3.

[0961] Input: Converted text data

[0962] Action: Input text data into GPT-3

[0963] Output: Data analyzed by GPT-3

[0964] Step 5:

[0965] The server extracts important keywords and phrases from the data analyzed by GPT-3.

[0966] Input: Parsed data

[0967] What it does: Extracts important keywords and phrases

[0968] Output: Extracted keywords and phrases

[0969] Step 6:

[0970] The server searches the MongoDB database based on the extracted keywords.

[0971] Input: Extracted keywords or phrases

[0972] How it works: Generating and executing database search queries

[0973] Output: Related search results

[0974] Step 7:

[0975] The server ranks the search results based on relevance and priority to select the most suitable results.

[0976] Input: Search results

[0977] Behavior: Ranking based on relevance and priority

[0978] Output: Best results

[0979] Step 8:

[0980] The server transmits the selected information to the terminal.

[0981] Input: Best results

[0982] Action: Send information to device

[0983] Output: Sending information to a terminal

[0984] Step 9:

[0985] The terminal displays the received information to the user.

[0986] Input: Submitted information

[0987] Action: Display information

[0988] Output: Display data that the user can see

[0989] Step 10:

[0990] The server generates meeting minutes using Python's ReportLab based on comments made during the meeting and the information obtained.

[0991] Input: What was said during the meeting and what information was captured

[0992] Action: Generate minutes

[0993] Output: Generated meeting minutes

[0994] Step 11:

[0995] The server links the generated minutes to the user's account and sends a notification.

[0996] Input: Generated minutes

[0997] Action: Send meeting minutes link and notification

[0998] Output: User notification

[0999] Step 12:

[1000] Users receive notifications and can review and edit the minutes.

[1001] Input: Notified link

[1002] Action: Check and edit minutes

[1003] Output: Edited transcript

[1004] In this way, the system of the present invention can recognize and analyze statements made during a meeting in real time, quickly provide relevant information, and automatically generate and notify minutes after the meeting has ended.

[1005] (Application example 1)

[1006] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1007] Modern conference systems provide functions such as speech recognition, information search, and meeting minutes generation, but they lack the ability to provide product information in physical stores or real-time reference to inventory information. Therefore, there is a need for systems that not only provide effective information during meetings but also improve everyday shopping experiences. Furthermore, by creating an environment where customers can obtain product information in real time, it is necessary to improve shopping efficiency and satisfaction.

[1008] 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.

[1009] In this invention, the server includes a speech recognition means for recognizing utterances made during a conference in real time, a conversion means for converting the utterances recognized by the speech recognition means into text data, an analysis means for analyzing the text data converted by the conversion means, a display means for providing detailed product information and inventory information in a physical store, and a documentation means for automatically documenting the conference contents. This not only enables efficient information provision during a conference, but also improves the customer experience in a physical store and makes it possible to provide product information and inventory information in real time.

[1010] "Speech recognition means" refers to a device or software that recognizes speech during a meeting in real time.

[1011] The "conversion means" is a device or software that has the function of converting the speech content recognized by the speech recognition means into text data.

[1012] The "analysis means" is a device or software that has the function of analyzing the text data converted by the conversion means and extracting related keywords and phrases.

[1013] The "search means" is a device or software that has the function of searching a database for related information based on the keywords extracted by the analysis means.

[1014] The "display means" is a device or software for displaying the related information obtained by the search means to the user.

[1015] "Documentation means" refers to a device or software that has the function of automatically documenting the contents of a meeting.

[1016] A "large-scale language model" is an advanced natural language processing model used to analyze text data, and is trained on large amounts of text data.

[1017] "Display means for providing detailed product information and inventory information in a physical store" refers to a device or software for displaying detailed product information and inventory information in real time in a physical store based on customer requests.

[1018] The system of the present invention recognizes comments made during a meeting in real time, provides relevant information, and improves meeting efficiency. Furthermore, by incorporating a function to provide real-time product information and inventory information in physical stores, the customer shopping experience can be improved. Specific embodiments are described below.

[1019] System configuration

[1020] The system consists of the following components:

[1021] Voice recognition means

[1022] Recognizes what is said in a meeting in real time. This feature uses a microphone to capture what the user is saying and speech recognition software to process the audio data.

[1023] Conversion Method

[1024] The speech recognized by the speech recognition means is converted into text data. In this process, a speech recognition API (e.g., Google Cloud Speech-to-Text API) is used.

[1025] Analysis means

[1026] The text data converted by the conversion method is analyzed to extract related keywords and phrases. Here, a large-scale language model (a generative AI model such as GPT-2) is used.

[1027] Search methods

[1028] Based on the keywords extracted by the analysis means, related information is searched from a database, which may include meeting materials, sales data, product inventory information, etc.

[1029] Display means

[1030] Related information obtained by the search tool is displayed to the user using a smartphone, tablet, smart glasses, or other device.

[1031] Documentation Methods

[1032] Automatically document meeting content. This function automatically generates meeting minutes based on speech recognition and text data.

[1033] How to operate the system

[1034] The specific operation method is to follow the steps below.

[1035] 1. Speech recognition during meetings

[1036] When the user says, "What other colors do you have for this item?", the voice recognition means captures this.

[1037] 2. Audio data conversion

[1038] The voice data is converted into text data via a voice recognition API.

[1039] 3. Text Data Analysis

[1040] The converted text data is analyzed by an analysis means, and important keywords (in this case, "product" and "color") are extracted.

[1041] 4. Information Search

[1042] Based on the extracted keywords, the search means searches the database for related product information and inventory information.

[1043] 5. Information display

[1044] The searched information is provided to the user by a display means. For example, a message such as "This product is available in red, blue, and green" is displayed on the screen of a smartphone.

[1045] Examples of concrete examples and prompts

[1046] As a specific example, the response when a user asks a question about a product will be shown.

[1047] Specific examples

[1048] User: "What other colors does this product come in?"

[1049] App: "This item comes in red, blue, and green."

[1050] Prompt Sentence Examples

[1051] "Generate a response when the user requests a different color."

[1052] In this way, the system of the present invention allows users to quickly and accurately obtain necessary information not only during meetings but also in physical stores, which not only improves the efficiency of meetings but also the customer's shopping experience.

[1053] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1054] Step 1:

[1055] Capture speech

[1056] The user says, "What other colors do you have for this product?" The device (such as a smartphone or smart glasses) captures this speech through a microphone. The input is speech data, which is collected for conversion into text data in a later step.

[1057] Step 2:

[1058] Voice Recognition

[1059] The captured voice data is sent from the device to a voice recognition API (for example, Google Cloud Speech-to-Text API). This API converts the voice data into text data and returns the conversion result to the device. The input is voice data and the output is text data.

[1060] Step 3:

[1061] Sending text data

[1062] The converted text data is sent from the terminal to the server. The server receives this text data and proceeds to the next analysis process. The input is text data, and the output is data transfer to the server.

[1063] Step 4:

[1064] Text data analysis

[1065] The server inputs the received text data into an analysis means (a generative AI model, such as GPT-2). The analysis means analyzes the text data and extracts important keywords and phrases (in this case, "product" and "color"). The input is the text data, and the output is the extracted keywords.

[1066] Step 5:

[1067] Information Search

[1068] The server uses a search engine to search for related product information and inventory information from a database based on the extracted keywords. The database contains past conversation history, product information, inventory information, etc. The input is the keywords, and the output is related information.

[1069] Step 6:

[1070] Information ranking

[1071] The server ranks the retrieved information based on relevance and priority. The most appropriate information is selected and sent to the device in the next step. The input is the relevant information, and the output is the ranked information.

[1072] Step 7:

[1073] Providing information

[1074] The server sends the ranking information to the terminal. The terminal receives it and displays it to the user. The user sees a message on the screen of their smartphone saying, "This product is available in red, blue, and green." The input is the ranking information, and the output is the display content.

[1075] Step 8:

[1076] Documenting meeting content

[1077] The server automatically generates meeting minutes based on comments made during the meeting and related information. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The input is comments and related information, and the output is the minutes and notification.

[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] The system of the present invention recognizes and analyzes comments made during a meeting in real time, provides relevant information, and recognizes the user's emotions, presents information and creates minutes accordingly, thereby improving the efficiency and effectiveness of the meeting. Specific embodiments of the present invention are described below.

[1080] Building a database

[1081] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[1082] Speech Recognition and Analysis

[1083] The device captures the audio during the meeting in real time. The captured audio data is sent to a speech recognition API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM). The LLM extracts important keywords and phrases from the text data.

[1084] Real-time search and information provision

[1085] The server searches the database for relevant information based on the keywords extracted by the analysis means. The retrieved information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is sent to the terminal and displayed to the user.

[1086] Automatic documentation of meeting content

[1087] The server automatically documents what is said during the meeting and the information obtained as a result. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The user receives a notification and can review and edit the minutes.

[1088] Introducing emotion recognition

[1089] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to the emotion engine, which analyzes and recognizes the user's emotions (e.g., excitement, confusion, interest, etc.).

[1090] Coordination of information provision

[1091] Based on the emotional data recognized by the emotion engine, the server automatically adjusts how it presents relevant information: for example, if the user is confused, it will provide additional explanations or detailed materials, while if the user expresses interest, it will provide relevant, in-depth information.

[1092] Emotion tagging

[1093] The recognized emotional information is also reflected in the minutes of the meeting. The server assigns emotional tags to each part of the minutes, allowing for a more detailed understanding of the progress of the meeting.

[1094] Specific examples

[1095] Interaction during the meeting

[1096] 1. User: "I'm not sure if this data is correct."

[1097] 2. The device captures this utterance and sends it to the speech recognition API.

[1098] 3. The speech recognition API converts the speech into text "It is unclear whether this data is correct." and sends it to the server.

[1099] 4. The server analyzes the text using a large-scale language model and extracts important keywords: "data," "correct," and "unknown."

[1100] 5. The server searches the database based on these keywords and retrieves the relevant verification procedure materials.

[1101] 6. The server ranks the retrieved materials and sends the most relevant information to the terminal.

[1102] 7. The terminal displays the relevant information to the user.

[1103] 8. At the same time, the device sends data to the emotion engine to analyze emotions from the user's facial expressions and voice.

[1104] 9. The emotion engine recognizes that the user is "confused" and issues an alert.

[1105] 10. The server sends the terminal again materials containing more detailed explanations based on the emotion recognition.

[1106] Post-conference report materials

[1107] 1. The server automatically generates meeting minutes based on the voice data captured during the meeting and the voice recognition results.

[1108] 2. The generated minutes are tagged with the recognized emotion tags along with the content of each part of the speech.

[1109] 3. The minutes are linked to the user's account by the server.

[1110] 4. The server sends a notification to the user after the meeting ends, including a link to the minutes.

[1111] 5. The user confirms the notification and opens the automatically generated minutes.

[1112] 6. If necessary, the user can edit the minutes and use them for the next meeting or report.

[1113] In this way, the system of the present invention, which combines an emotion engine, can further improve the efficiency and effectiveness of meetings by providing real-time information during the meeting and generating detailed minutes after the meeting, as well as recognizing the user's emotions and presenting information accordingly.

[1114] The processing flow will be explained below.

[1115] Step 1:

[1116] The server receives user-provided materials, past conversation history, and email data, imports them into a database, checks that the data is properly formatted, and reports errors if there are any anomalies.

[1117] Step 2:

[1118] The server analyzes the content of the imported data and automatically generates related keywords and tags to add them as metadata. Based on this metadata, it builds an index to improve search performance.

[1119] Step 3:

[1120] The device continues to capture audio during the meeting in real time, and the captured audio data is sent to the speech recognition API at predetermined intervals.

[1121] Step 4:

[1122] The speech recognition API converts the received speech data into text data and sends the converted text data to the server.

[1123] Step 5:

[1124] The server then inputs the received text data into a large-scale language model (LLM) to analyze the conversation, extracting important keywords and phrases from the text.

[1125] Step 6:

[1126] The device captures the user's voice, facial expressions, and physical movements in real time and transmits this data to the emotion engine.

[1127] Step 7:

[1128] The emotion engine analyzes the received data and recognizes and classifies the user's emotions (e.g., excitement, confusion, interest, etc.).

[1129] Step 8:

[1130] The server searches the database for relevant information based on the emotion data recognized by the emotion engine and the keywords extracted by the LLM, and then retrieves the search results and ranks them based on relevance and priority.

[1131] Step 9:

[1132] The server selects the most relevant information from the ranked search results and automatically adjusts the way the information is presented based on the user's emotions.

[1133] Step 10:

[1134] The server sends tailored information to the terminal, for example, sending additional explanations or detailed materials if the user is confused.

[1135] Step 11:

[1136] The device visually displays the received relevant information and materials to the user, and can display the information in a pop-up display or a new window on the user's screen.

[1137] Step 12:

[1138] The server automatically generates minutes of the meeting based on the voice data captured during the meeting and the corresponding voice recognition results.

[1139] Step 13:

[1140] The server assigns emotion tags to the generated minutes and automatically associates them with the user's account after the meeting ends.

[1141] Step 14:

[1142] After the meeting, the server sends a notification to the user with a link to the minutes, allowing the user to view and edit the minutes.

[1143] Step 15:

[1144] The user opens the notification, reviews the automatically generated minutes, and, if necessary, completes the minutes by checking for errors and adding additional comments or corrections.

[1145] In this way, by performing specific operations for each processing step, the system of the present invention can efficiently provide information during a meeting and generate minutes after the meeting. In addition, by recognizing the user's emotions and adjusting the information presentation method accordingly, the quality of the meeting can be further improved.

[1146] Example 2

[1147] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1148] Conventional meeting support systems could recognize comments in real time, present relevant information, and create minutes, but they had limitations in providing information that took user emotions into account and automatically documenting meeting content. In addition, there were cases where users were unable to provide appropriate support because they did not understand the emotions behind their comments. Furthermore, because the emotions of participants were not reflected in post-meeting minutes, it was difficult to accurately grasp the progress and effectiveness of the meeting.

[1149] 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.

[1150] In this invention, the server includes speech recognition means for recognizing utterances made during a conference in real time, conversion means for converting the utterances recognized by the speech recognition means into text data, analysis means for analyzing the text data converted by the conversion means, search means for searching a database for related information based on keywords extracted by the analysis means, display means for displaying the related information acquired by the search means, emotion recognition means for recognizing emotion data of users during the conference, means for adjusting the method of presenting information based on the emotion data acquired by the emotion recognition means, documentation means for automatically documenting the conference content, and means for reflecting the emotion information in documents automatically generated by the documentation means. This makes it possible to grasp the progress of the conference and the emotions of the participants in real time, and to provide more effective support during and after the conference.

[1151] "Speech recognition means" refers to a device or software for recognizing speech during a meeting in real time.

[1152] The "conversion means" refers to a device or software that has the function of converting the speech content recognized by the speech recognition means into text data.

[1153] The "analysis means" refers to a device or software for analyzing the text data converted by the conversion means and extracting important keywords and phrases.

[1154] The "search means" refers to a device or software for searching a database for related information based on the keywords extracted by the analysis means.

[1155] The "display means" refers to a device or software that visually presents the related information obtained by the search means to the user.

[1156] "Emotion recognition means" refers to a device or software for analyzing a user's voice, facial expression, or physical movement during a meeting to recognize emotional data.

[1157] "Emotional data" refers to data that expresses a user's emotional state (e.g., excitement, confusion, interest, etc.).

[1158] "Documentation tool" refers to a device or software that automatically documents what is said and related information during a meeting.

[1159] "Emotion information" refers to information indicating the emotional state of the user acquired by emotion recognition means.

[1160] "Large-scale language model" refers to a highly trained machine learning model used for natural language processing analysis.

[1161] The conference support system according to the present invention recognizes comments made during a conference in real time, provides related information, creates minutes, and recognizes the user's emotions and presents information based on those emotions.

[1162] Building a database

[1163] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[1164] For example, when a user uploads relevant materials before a meeting, the server updates the database based on those materials and automatically assigns necessary keywords and tags.

[1165] Speech Recognition and Analysis

[1166] The device captures audio during the meeting in real time. The captured audio data is sent to a speech recognition API such as Google Cloud Speech-to-Text API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM) such as OpenAI's GPT-4. The LLM extracts important keywords and phrases from the text data.

[1167] For example, if a user says, "I'm not sure if this data is correct," the device captures it and sends it to a speech recognition API, which generates text data. This text data is then sent to a server and analyzed using a large-scale language model.

[1168] Real-time search and information provision

[1169] The server searches the database based on the keywords extracted by the analysis means. The search results are ranked based on relevance and priority, and the most relevant information is selected. The selected information is sent to the terminal and displayed to the user.

[1170] As a specific example, the server searches for relevant verification procedure documents based on keywords such as "data," "correct," and "unknown," and sends the most relevant information to the user's terminal.

[1171] Automatic documentation of meeting content

[1172] All statements made during the meeting and related information automatically acquired are documented by the server. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. Users can view and edit the minutes through the notification.

[1173] Specifically, after the meeting ends, the server automatically generates minutes and creates a link to the user's account. The contents of the meeting and related information are all organized in text format.

[1174] Introducing emotion recognition

[1175] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to a pre-installed emotion recognition engine, which analyzes and recognizes the user's emotions (e.g., excitement, confusion, interest, etc.).

[1176] For example, the device's camera and microphone capture the user's reactions and send the real-time data to the emotion engine, which then analyzes the user's emotions.

[1177] Coordination of information provision

[1178] The server adjusts the content and method of information presented based on the emotional data recognized by the emotion engine. For example, if the emotion engine determines that the user is confused, the server will present additional explanations or detailed materials. If the user shows interest, the server will provide related in-depth information.

[1179] As a specific example, if the emotion engine detects that the user is confused, the server retransmits detailed information to the terminal and displays it to the user.

[1180] Emotion tagging

[1181] The server reflects the emotional information recognized during the meeting in the minutes. By assigning appropriate emotional tags to each part of the minutes, the progress of the meeting can be understood in more detail.

[1182] For example, the server automatically generates emotion tags for each comment and inserts them into the minutes, recording not only the content of the meeting but also changes in participants' emotions.

[1183] By combining these functions, a system can provide real-time information during a meeting, generate detailed minutes after the meeting, and present information according to the user's emotions. Such a system will further improve the efficiency and effectiveness of meetings.

[1184] Prompt Sentence Examples

[1185] For example, you can check the system's operation by using the following prompt sentence:

[1186] It is unclear whether this data is correct.

[1187] The server analyzes the text data to search for relevant information, and also recognizes the user's emotions to adjust the way information is presented.

[1188] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1189] Step 1: Importing data and building the database

[1190] The server imports documents uploaded by users, past conversation history, and email data, and builds a database based on these. It analyzes the content of each document in the imported data and assigns relevant keywords and tags. The input is file data in various formats uploaded by users, and the output is an indexed database. Specifically, the server scans the documents and automatically extracts relevant keywords using NLP (natural language processing) technology, updating the database accordingly.

[1191] Step 2: Capture and convert audio

[1192] The device captures speech during the meeting in real time. The captured voice data is sent to a speech recognition API such as the Google Cloud Speech-to-Text API and converted into text data. The input is the real-time captured voice data, and the output is the recognized text data. Specifically, the microphone captures the voice and sends it to the speech recognition API, which then converts the voice into text and sends it back to the server.

[1193] Step 3: Analyzing the text data

[1194] The server receives the converted text data and analyzes it using an internal large-scale language model (for example, OpenAI's generative AI model). The analysis means extracts important keywords and phrases from the text data. The input is the text data received from the speech recognition API, and the output is the extracted keywords and phrases. Specifically, the large-scale language model analyzes the text and automatically extracts specific keywords and phrases from it.

[1195] Step 4: Find related information

[1196] The server searches for relevant information from a database based on the extracted keywords. The search results are ranked based on relevance and priority. The input is the extracted keywords, and the output is the ranked search results. Specifically, the server executes a database query to retrieve relevant materials and data, and ranks them based on priority.

[1197] Step 5: Viewing search results

[1198] The device displays the search results sent from the server to the user, allowing the user to see the most relevant information in real time. The input is the ranked search results sent from the server, and the output is the specific information displayed on the user's screen. In specific operation, the screen display portion of the device displays the relevant information through a user interface.

[1199] Step 6: Capturing and analyzing user emotions

[1200] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to an emotion recognition engine. The emotion recognition engine analyzes the acquired emotion data and recognizes the user's emotions. The input is the captured real-time user data, and the output is analyzed emotion data. Specifically, the camera and microphone capture the user's voice and facial movements and send this data to the emotion engine.

[1201] Step 7: Tailor your information based on emotions

[1202] The server adjusts the content and format of the information it presents based on the emotional data analyzed by the emotion recognition engine. For example, if it recognizes that the user is confused, the server will provide additional explanations or detailed materials. The input is the emotional data received from the emotion engine, and the output is the adjusted presentation of information. Specifically, the server dynamically changes the content and format of the information it presents based on the emotional data and sends it to the device.

[1203] Step 8: Automatically document meeting content

[1204] The server documents what was said during the meeting and any related information that was automatically acquired. After the meeting ends, this document is linked to the user's account as minutes, and a notification is sent. The input is the analyzed text data and related information, and the output is the automatically generated minutes. Specifically, the server collects, organizes, and documents the data from the meeting, saves the results in the user's account, and sends a notification.

[1205] Step 9: Sentiment tagging of meeting notes

[1206] The server also reflects the emotional information recognized during the meeting in the minutes. This allows each part of the minutes to be assigned an emotional tag. The input is the emotional data and the generated minutes, and the output is a detailed minutes with the emotional tags. Specifically, the server automatically generates an emotional tag corresponding to each utterance and adds it to the corresponding part of the minutes.

[1207] (Application example 2)

[1208] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1209] While conventional conferencing systems can recognize comments made during meetings in real time and provide relevant information, they provide information without considering the emotions of workers working on-site, making it difficult to provide efficient work support. There is also a need for a system that can appropriately recognize workers' emotions and stress levels and adjust and provide information accordingly.

[1210] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a voice recognition means for recognizing utterances made during a meeting in real time; a conversion means for converting the utterances recognized by the voice recognition means into text data; an analysis means for analyzing the text data converted by the conversion means; a search means for searching a database for related information based on keywords extracted by the analysis means; a display means for displaying the related information acquired by the search means; a documentation means for automatically documenting the meeting contents; an expression recognition means for capturing the facial expressions of workers at the workplace and recognizing their emotions in real time; and an emotion analysis means for adjusting the presentation method of related information based on the emotion data obtained by the expression recognition means. This enables efficient provision of information and work support in accordance with the emotions of workers.

[1211] The "voice recognition means" is a means for recognizing speech during a meeting in real time and converting the speech data into text.

[1212] The "conversion means" is a means for converting the speech content recognized by the speech recognition means into text data.

[1213] The "analysis means" is a means for analyzing the text data converted by the conversion means and extracting important keywords and phrases.

[1214] The "search means" is a means for searching the database for related information based on the keywords extracted by the analysis means.

[1215] The "display means" is a means for visually displaying the related information obtained by the search means to the user.

[1216] The "documentation means" is a means for automatically recording the contents of a meeting in a document format and notifying the user of the minutes generated after the meeting has ended.

[1217] The "facial expression recognition means" is a means for capturing the facial expressions of workers at the workplace and recognizing their emotions in real time.

[1218] The "emotion analysis means" is a means for adjusting the presentation method of related information based on the emotion data obtained by the facial expression recognition means.

[1219] The system of the present invention recognizes and analyzes worker conversations and emotions in real time during work meetings and daily tasks in a factory, and provides relevant work procedures, manuals, and past history information. Furthermore, by analyzing worker emotions and providing appropriate work instructions and supplementary explanations accordingly, it is possible to improve work efficiency and safety. This system includes the following hardware and software configuration:

[1220] First, the Google Cloud Speech-to-Text API is used as a speech recognition means to recognize statements made during meetings in real time. This converts the voice data into text data. The converted text data is sent to a server and analyzed using a large-scale language model (for example, OpenAI GPT). The analysis means extracts important keywords and phrases, and a search means operates to search a database for related information based on those keywords. The searched information is ranked based on relevance and priority, and the most appropriate information is sent from the server to the terminal and visually presented to the user via the terminal's display means. The display means can be a standard monitor, smartphone, or tablet.

[1221] Furthermore, a camera is used as an expression recognition means to capture the facial expressions of workers on the workplace. The captured facial expression data is analyzed using OpenCV to recognize the worker's emotions (e.g., confusion, excitement, interest, etc.) in real time. This emotion data is processed by the emotion analysis means, and the way related information is presented is automatically adjusted. For example, if the worker is confused, additional explanations or detailed materials are presented. On the other hand, if the worker shows interest, relevant in-depth information is provided.

[1222] The meeting content is automatically documented by the server, and meeting minutes are generated based on the captured audio data and its analysis. The generated minutes are linked to the user's account, and a notification is sent after the meeting ends.

[1223] Specific examples

[1224] Example 1: Providing information in response to a worker's confusion

[1225] 1. Worker: "I don't know how to set up this machine."

[1226] 2. The server captures this utterance and sends it to the speech recognition API.

[1227] 3. The speech recognition API converts the speech into text "I don't know how to set up this machine" and sends it to the server.

[1228] 4. The server analyzes the text using a large-scale language model and extracts important keywords such as "machine," "settings," and "don't know."

[1229] 5. The server searches the database based on these keywords and retrieves the relevant configuration instructions.

[1230] 6. At the same time, the facial expression recognition means captures the worker's facial expression and recognizes that he is confused.

[1231] 7. The server will provide further detailed configuration instructions and additional instructions.

[1232] Prompt Sentence Examples

[1233] User input: "I don't know how to configure this machine."

[1234] Prompt for generative AI model: "Please provide relevant instructions or manuals based on the following sentence: 'I don't know how to set up this machine.'"

[1235] In this way, the system according to the present invention recognizes the emotions of workers in real time and provides efficient and appropriate information and support accordingly.

[1236] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1237] Step 1:

[1238] A user speaks while working or in a meeting. This speech is captured in real time by the microphone of the terminal. The input is the user's voice, and the output is the captured voice data.

[1239] Step 2:

[1240] The device sends the captured voice data to the Google Cloud Speech-to-Text API, where the data is converted into text data. The output from the API is text data in string format.

[1241] Step 3:

[1242] The server receives the text data. This text data is then input into a large-scale language model such as OpenAI GPT to extract important keywords and phrases. This involves data calculations such as TEKST data analysis and keyword extraction, and the output is a list of keywords.

[1243] Step 4:

[1244] The server then searches a database based on the extracted keyword list to retrieve related information. The input is the keyword list, and the database search outputs related materials, procedure manuals, etc.

[1245] Step 5:

[1246] The server transmits the acquired related information to the terminal, where the data transfer operation is performed and the information is provided to the user by the display means. The information visually presented using the terminal's display is the output.

[1247] Step 6:

[1248] At the same time, the terminal uses a camera to capture the worker's facial expression, whose input is the worker's facial image data and whose output is the captured facial data.

[1249] Step 7:

[1250] The device uses OpenCV to recognize facial expressions and analyzes the facial expressions of the worker from the captured facial data. The input is facial data, and emotion analysis is performed as data calculation. The analysis results are output as emotional data such as confusion, interest, and excitement.

[1251] Step 8:

[1252] The server receives the emotion data and adjusts the presentation of related information based on it. For example, if the worker is confused, it will send a document with additional detailed explanations to the terminal again. The input data is emotion data, and the presentation information is adjusted as a data calculation. The output is the readjusted related information.

[1253] Step 9:

[1254] The server automatically documents the meeting contents. It generates meeting minutes using the captured audio data and the analysis results based on it. The input data is audio data and text data, and the data is processed into documentation. The output is the generated minutes.

[1255] Step 10:

[1256] The generated minutes are linked to the user's account by the server and a notification is sent after the meeting ends. The output is a notification with a link to the minutes.

[1257] Through the above steps, the system according to the present invention recognizes the emotions of workers in real time and provides efficient and appropriate information and support accordingly.

[1258] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1259] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1260] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1261] [Fourth embodiment]

[1262] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1263] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1264] 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).

[1265] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1266] 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.

[1267] 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).

[1268] 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.

[1269] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1270] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1271] 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.

[1272] 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.

[1273] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1274] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1275] The system according to the present invention is designed to improve the efficiency of meetings by recognizing and analyzing comments made during meetings in real time and providing relevant information. Specific embodiments of the present invention are described below.

[1276] Building a database

[1277] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[1278] Speech Recognition and Analysis

[1279] The device captures the audio during the meeting in real time. The captured audio data is sent to a speech recognition API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM). The LLM extracts important keywords and phrases from the text data.

[1280] Real-time search and information provision

[1281] The server searches the database for relevant information based on the keywords extracted by the analysis means. The retrieved information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is sent to the terminal and displayed to the user.

[1282] Automatic documentation of meeting content

[1283] The server automatically documents what is said during the meeting and the information obtained as a result. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The user receives a notification and can review and edit the minutes.

[1284] Specific examples

[1285] Interaction during the meeting

[1286] 1. User: "Oh, what's last month's sales data?"

[1287] 2. The device captures this utterance and sends it to the speech recognition API.

[1288] 3. The speech recognition API converts the speech into text ("What was last month's sales data?") and sends it to the server.

[1289] 4. The server inputs the text into LLM and extracts related keywords such as "last month" and "sales data."

[1290] 5. The server searches the database based on these keywords to retrieve relevant sales data.

[1291] 6. The server sends the acquired sales data to the terminal, which displays it to the user.

[1292] 7. User: "That's the data, thank you," and continues the meeting.

[1293] Post-conference report materials

[1294] 1. The server generates minutes of the meeting based on the speech and speech-recognized text during the meeting.

[1295] 2. The generated minutes are linked to the user's account by the server.

[1296] 3. The server sends the user a notification after the meeting ends, including a link to the minutes.

[1297] 4. The user confirms the notification and opens the generated minutes.

[1298] 5. If necessary, the user can edit the minutes and use them for the next meeting or report.

[1299] In this way, the present invention not only quickly provides necessary materials and information during a meeting, but also automates the creation of minutes after the meeting, thereby significantly improving work efficiency.

[1300] The processing flow will be explained below.

[1301] Step 1:

[1302] The server receives user-provided materials, past conversation history, and email data, imports them into a database, checks that the data is properly formatted, and reports errors if there are any anomalies.

[1303] Step 2:

[1304] The server analyzes the content of the imported data and automatically generates related keywords and tags to add them as metadata. Based on this metadata, it builds an index to improve search performance.

[1305] Step 3:

[1306] The device continues to capture audio during the meeting in real time, and the captured audio data is sent to the speech recognition API at predetermined intervals.

[1307] Step 4:

[1308] The speech recognition API converts the received speech data into text data and sends the converted text data to the server.

[1309] Step 5:

[1310] The server then inputs the received text data into a large-scale language model (LLM) to analyze the conversation, extracting important keywords and phrases from the text.

[1311] Step 6:

[1312] The server searches for relevant information from an indexed database based on the extracted keywords or phrases, retrieves the search results, and ranks them based on relevance and priority.

[1313] Step 7:

[1314] The server selects the most relevant information from the ranked search results and sends it to the device, based on criteria such as keyword match and importance of the material.

[1315] Step 8:

[1316] The device visually displays the received relevant information and materials to the user, and can display the information in a pop-up display or a new window on the user's screen.

[1317] Step 9:

[1318] The server automatically generates meeting minutes based on the voice data captured during the meeting and the corresponding voice recognition results, and the generated minutes are automatically associated with the user's account after the meeting ends.

[1319] Step 10:

[1320] The server notifies the user of the generated minutes and sends a link so that the user can view and edit the minutes.

[1321] Step 11:

[1322] The user opens the notification, reviews the automatically generated minutes, and, if necessary, completes the minutes by checking for errors and adding additional comments or corrections.

[1323] In this way, by performing specific operations for each processing step, the system according to the present invention can efficiently provide information during a meeting and generate minutes after the meeting.

[1324] Example 1

[1325] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1326] Conventional conferencing systems lack the ability to recognize and analyze what is said during a meeting in real time and provide relevant information, resulting in reduced meeting efficiency. Meeting minutes also need to be created manually after the meeting, which also requires time and effort. Furthermore, there are insufficient search and ranking methods to ensure the accuracy and relevance of the information provided in real time. To solve these problems, a more efficient and accurate meeting support system is needed.

[1327] 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.

[1328] In this invention, the server includes a speech recognition unit that recognizes speech during a conference in real time, a conversion unit that converts speech content recognized by the speech recognition unit into text data, an analysis unit that analyzes the text data converted by the conversion unit, a search unit that searches a database for related information based on keywords extracted by the analysis unit, a display unit that displays the related information acquired by the search unit, a documentation unit that automatically documents the conference content, a link unit that links the document generated by the documentation unit to a user account, a notification unit that sends a notification to the user after the conference ends, a speech capture unit that captures speech using a microphone, a transmission unit that transmits the captured speech data to a speech recognition API, and a ranking unit that ranks the results of the database search. This allows speech during a conference to be recognized in real time and related information to be quickly provided, thereby improving the efficiency of the conference and automatically generating and notifying minutes after the conference ends.

[1329] "Speech recognition means" refers to a device or program that recognizes speech during a meeting in real time and acquires the speech data.

[1330] The "conversion means" refers to a device or program that converts the speech content recognized by the speech recognition means into text data.

[1331] The "analysis means" refers to a device or program that analyzes the text data converted by the conversion means and extracts important keywords and phrases.

[1332] The "search means" refers to a device or program that searches a database for related information based on the keywords extracted by the analysis means.

[1333] The "display means" refers to a device or program that visually displays the related information obtained by the search means to the user.

[1334] "Documentation means" refers to a device or program that automatically documents the contents of a meeting and generates meeting minutes.

[1335] "Linking means" refers to a device or program that associates the document generated by the documentation means with a user account.

[1336] "Notification means" refers to a device or program that sends a notification to the user after the conference has ended.

[1337] "Audio capture means" refers to a device or program that uses a microphone to capture audio during a meeting.

[1338] "Transmission means" refers to the device or program that transmits the captured voice data to the voice recognition API.

[1339] "Ranking means" refers to a device or program that ranks the results of a database search based on relevance or priority.

[1340] A "large-scale language model" is a natural language processing model generated based on massive amounts of data, and has the ability to extract important keywords and phrases from text data.

[1341] The system according to the present invention is designed to improve the efficiency of meetings by recognizing and analyzing comments made during meetings in real time and providing relevant information. Specific embodiments of the present invention will be described below.

[1342] Building a database

[1343] The server imports users' documents, past conversation history, and email data. This data is stored in a database such as MongoDB. The server then analyzes the imported data using Python's natural language processing (NLP) library (e.g., NLTK or spaCy) and assigns relevant keywords and tags based on the content of each document. This enables relevant information to be provided quickly and accurately when searching.

[1344] Speech Recognition and Analysis

[1345] The device uses a microphone to capture audio during the meeting in real time. The captured audio data is sent to the Google Speech-to-Text API and converted into text data. This text data is then sent to a server. The server then analyzes the text data using a large-scale language model (e.g., OpenAI's GPT-3) to extract important keywords and phrases.

[1346] Real-time search and information provision

[1347] The server searches the database based on the extracted keywords to retrieve relevant information. This information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is then sent to the terminal and displayed to the user.

[1348] Automatic documentation of meeting content

[1349] The server automatically documents what is said and the information obtained during the meeting. This process uses a Python document generation tool (e.g., ReportLab). After the meeting ends, the generated minutes are linked to the user's account, and a notification is sent via email or a chat tool (e.g., Slack). The user receives a notification and can review and edit the minutes.

[1350] Specific examples

[1351] For example, if a user says "I want to know last month's sales data" during a meeting, the device captures this utterance and sends the audio data to the Google Speech-to-Text API. The speech recognition API converts the speech into text "I want to know last month's sales data" and sends it to the server. The server inputs this text into GPT-3 and extracts the keywords "last month" and "sales data." The server then searches the database based on these keywords to retrieve relevant sales data. The retrieved data is sent to the device and displayed to the user. The user can then say, "That's the data, thank you," and continue the meeting.

[1352] In this invention, examples of prompt sentences include:

[1353] "I want to know last month's sales data."

[1354] "Tell me about the latest market trends"

[1355] "What's the next step for this project?"

[1356] In this way, the present invention not only quickly provides necessary materials and information during a meeting, but also automates the creation of minutes after the meeting, thereby significantly improving work efficiency.

[1357] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1358] Step 1:

[1359] The terminal uses a microphone to capture audio during the meeting in real time.

[1360] Input: Speech during the meeting (audio data)

[1361] Action: Capture audio data with a microphone

[1362] Output: Captured audio data

[1363] Step 2:

[1364] The device sends the captured audio data to the Google Speech-to-Text API.

[1365] Input: Captured audio data

[1366] What it does: Sends audio data to the Google Speech-to-Text API

[1367] Output: Audio data is sent to the API

[1368] Step 3:

[1369] The speech recognition API converts the speech data into text data and sends the text data to the server.

[1370] Input: Transmitted audio data

[1371] Operation: Converts voice data into text data

[1372] Output: Converted text data

[1373] Step 4:

[1374] The server inputs text data sent from the Google Speech-to-Text API into GPT-3.

[1375] Input: Converted text data

[1376] Action: Input text data into GPT-3

[1377] Output: Data analyzed by GPT-3

[1378] Step 5:

[1379] The server extracts important keywords and phrases from the data analyzed by GPT-3.

[1380] Input: Parsed data

[1381] What it does: Extracts important keywords and phrases

[1382] Output: Extracted keywords and phrases

[1383] Step 6:

[1384] The server searches the MongoDB database based on the extracted keywords.

[1385] Input: Extracted keywords or phrases

[1386] How it works: Generating and executing database search queries

[1387] Output: Related search results

[1388] Step 7:

[1389] The server ranks the search results based on relevance and priority to select the most suitable results.

[1390] Input: Search results

[1391] Behavior: Ranking based on relevance and priority

[1392] Output: Best results

[1393] Step 8:

[1394] The server transmits the selected information to the terminal.

[1395] Input: Best results

[1396] Action: Send information to device

[1397] Output: Sending information to a terminal

[1398] Step 9:

[1399] The terminal displays the received information to the user.

[1400] Input: Submitted information

[1401] Action: Display information

[1402] Output: Display data that the user can see

[1403] Step 10:

[1404] The server generates meeting minutes using Python's ReportLab based on comments made during the meeting and the information obtained.

[1405] Input: What was said during the meeting and what information was captured

[1406] Action: Generate minutes

[1407] Output: Generated meeting minutes

[1408] Step 11:

[1409] The server links the generated minutes to the user's account and sends a notification.

[1410] Input: Generated minutes

[1411] Action: Send meeting minutes link and notification

[1412] Output: User notification

[1413] Step 12:

[1414] Users receive notifications and can review and edit the minutes.

[1415] Input: Notified link

[1416] Action: Check and edit minutes

[1417] Output: Edited transcript

[1418] In this way, the system of the present invention can recognize and analyze statements made during a meeting in real time, quickly provide relevant information, and automatically generate and notify minutes after the meeting has ended.

[1419] (Application example 1)

[1420] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1421] Modern conference systems provide functions such as speech recognition, information search, and meeting minutes generation, but they lack the ability to provide product information in physical stores or real-time reference to inventory information. Therefore, there is a need for systems that not only provide effective information during meetings but also improve everyday shopping experiences. Furthermore, by creating an environment where customers can obtain product information in real time, it is necessary to improve shopping efficiency and satisfaction.

[1422] 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.

[1423] In this invention, the server includes a speech recognition means for recognizing utterances made during a conference in real time, a conversion means for converting the utterances recognized by the speech recognition means into text data, an analysis means for analyzing the text data converted by the conversion means, a display means for providing detailed product information and inventory information in a physical store, and a documentation means for automatically documenting the conference contents. This not only enables efficient information provision during a conference, but also improves the customer experience in a physical store and makes it possible to provide product information and inventory information in real time.

[1424] "Speech recognition means" refers to a device or software that recognizes speech during a meeting in real time.

[1425] The "conversion means" is a device or software that has the function of converting the speech content recognized by the speech recognition means into text data.

[1426] The "analysis means" is a device or software that has the function of analyzing the text data converted by the conversion means and extracting related keywords and phrases.

[1427] The "search means" is a device or software that has the function of searching a database for related information based on the keywords extracted by the analysis means.

[1428] The "display means" is a device or software for displaying the related information obtained by the search means to the user.

[1429] "Documentation means" refers to a device or software that has the function of automatically documenting the contents of a meeting.

[1430] A "large-scale language model" is an advanced natural language processing model used to analyze text data, and is trained on large amounts of text data.

[1431] "Display means for providing detailed product information and inventory information in a physical store" refers to a device or software for displaying detailed product information and inventory information in real time in a physical store based on customer requests.

[1432] The system of the present invention recognizes comments made during a meeting in real time, provides relevant information, and improves meeting efficiency. Furthermore, by incorporating a function to provide real-time product information and inventory information in physical stores, the customer shopping experience can be improved. Specific embodiments are described below.

[1433] System configuration

[1434] The system consists of the following components:

[1435] Voice recognition means

[1436] Recognizes what is said in a meeting in real time. This feature uses a microphone to capture what the user is saying and speech recognition software to process the audio data.

[1437] Conversion Method

[1438] The speech recognized by the speech recognition means is converted into text data. In this process, a speech recognition API (e.g., Google Cloud Speech-to-Text API) is used.

[1439] Analysis means

[1440] The text data converted by the conversion method is analyzed to extract related keywords and phrases. Here, a large-scale language model (a generative AI model such as GPT-2) is used.

[1441] Search methods

[1442] Based on the keywords extracted by the analysis means, related information is searched from a database, which may include meeting materials, sales data, product inventory information, etc.

[1443] Display means

[1444] Related information obtained by the search tool is displayed to the user using a smartphone, tablet, smart glasses, or other device.

[1445] Documentation Methods

[1446] Automatically document meeting content. This function automatically generates meeting minutes based on speech recognition and text data.

[1447] How to operate the system

[1448] The specific operation method is to follow the steps below.

[1449] 1. Speech recognition during meetings

[1450] When the user says, "What other colors do you have for this item?", the voice recognition means captures this.

[1451] 2. Audio data conversion

[1452] The voice data is converted into text data via a voice recognition API.

[1453] 3. Text Data Analysis

[1454] The converted text data is analyzed by an analysis means, and important keywords (in this case, "product" and "color") are extracted.

[1455] 4. Information Search

[1456] Based on the extracted keywords, the search means searches the database for related product information and inventory information.

[1457] 5. Information display

[1458] The searched information is provided to the user by a display means. For example, a message such as "This product is available in red, blue, and green" is displayed on the screen of a smartphone.

[1459] Examples of concrete examples and prompts

[1460] As a specific example, the response when a user asks a question about a product will be shown.

[1461] Specific examples

[1462] User: "What other colors does this product come in?"

[1463] App: "This item comes in red, blue, and green."

[1464] Prompt Sentence Examples

[1465] "Generate a response when the user requests a different color."

[1466] In this way, the system of the present invention allows users to quickly and accurately obtain necessary information not only during meetings but also in physical stores, which not only improves the efficiency of meetings but also the customer's shopping experience.

[1467] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1468] Step 1:

[1469] Capture speech

[1470] The user says, "What other colors do you have for this product?" The device (such as a smartphone or smart glasses) captures this speech through a microphone. The input is speech data, which is collected for conversion into text data in a later step.

[1471] Step 2:

[1472] Voice Recognition

[1473] The captured voice data is sent from the device to a voice recognition API (for example, Google Cloud Speech-to-Text API). This API converts the voice data into text data and returns the conversion result to the device. The input is voice data and the output is text data.

[1474] Step 3:

[1475] Sending text data

[1476] The converted text data is sent from the terminal to the server. The server receives this text data and proceeds to the next analysis process. The input is text data, and the output is data transfer to the server.

[1477] Step 4:

[1478] Text data analysis

[1479] The server inputs the received text data into an analysis means (a generative AI model, such as GPT-2). The analysis means analyzes the text data and extracts important keywords and phrases (in this case, "product" and "color"). The input is the text data, and the output is the extracted keywords.

[1480] Step 5:

[1481] Information Search

[1482] The server uses a search engine to search for related product information and inventory information from a database based on the extracted keywords. The database contains past conversation history, product information, inventory information, etc. The input is the keywords, and the output is related information.

[1483] Step 6:

[1484] Information ranking

[1485] The server ranks the retrieved information based on relevance and priority. The most appropriate information is selected and sent to the device in the next step. The input is the relevant information, and the output is the ranked information.

[1486] Step 7:

[1487] Providing information

[1488] The server sends the ranking information to the terminal. The terminal receives it and displays it to the user. The user sees a message on the screen of their smartphone saying, "This product is available in red, blue, and green." The input is the ranking information, and the output is the display content.

[1489] Step 8:

[1490] Documenting meeting content

[1491] The server automatically generates meeting minutes based on comments made during the meeting and related information. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The input is comments and related information, and the output is the minutes and notification.

[1492] 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.

[1493] The system of the present invention recognizes and analyzes comments made during a meeting in real time, provides relevant information, and recognizes the user's emotions, presents information and creates minutes accordingly, thereby improving the efficiency and effectiveness of the meeting. Specific embodiments of the present invention are described below.

[1494] Building a database

[1495] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[1496] Speech Recognition and Analysis

[1497] The device captures the audio during the meeting in real time. The captured audio data is sent to a speech recognition API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM). The LLM extracts important keywords and phrases from the text data.

[1498] Real-time search and information provision

[1499] The server searches the database for relevant information based on the keywords extracted by the analysis means. The retrieved information is ranked based on relevance and priority, and the most appropriate information is selected. The selected information is sent to the terminal and displayed to the user.

[1500] Automatic documentation of meeting content

[1501] The server automatically documents what is said during the meeting and the information obtained as a result. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. The user receives a notification and can review and edit the minutes.

[1502] Introducing emotion recognition

[1503] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to the emotion engine, which analyzes and recognizes the user's emotions (e.g., excitement, confusion, interest, etc.).

[1504] Coordination of information provision

[1505] Based on the emotional data recognized by the emotion engine, the server automatically adjusts how it presents relevant information: for example, if the user is confused, it will provide additional explanations or detailed materials, while if the user expresses interest, it will provide relevant, in-depth information.

[1506] Emotion tagging

[1507] The recognized emotional information is also reflected in the minutes of the meeting. The server assigns emotional tags to each part of the minutes, allowing for a more detailed understanding of the progress of the meeting.

[1508] Specific examples

[1509] Interaction during the meeting

[1510] 1. User: "I'm not sure if this data is correct."

[1511] 2. The device captures this utterance and sends it to the speech recognition API.

[1512] 3. The speech recognition API converts the speech into text "It is unclear whether this data is correct." and sends it to the server.

[1513] 4. The server analyzes the text using a large-scale language model and extracts important keywords: "data," "correct," and "unknown."

[1514] 5. The server searches the database based on these keywords and retrieves the relevant verification procedure materials.

[1515] 6. The server ranks the retrieved materials and sends the most relevant information to the terminal.

[1516] 7. The terminal displays the relevant information to the user.

[1517] 8. At the same time, the device sends data to the emotion engine to analyze emotions from the user's facial expressions and voice.

[1518] 9. The emotion engine recognizes that the user is "confused" and issues an alert.

[1519] 10. The server sends the terminal again materials containing more detailed explanations based on the emotion recognition.

[1520] Post-conference report materials

[1521] 1. The server automatically generates meeting minutes based on the voice data captured during the meeting and the voice recognition results.

[1522] 2. The generated minutes are tagged with the recognized emotion tags along with the content of each part of the speech.

[1523] 3. The minutes are linked to the user's account by the server.

[1524] 4. The server sends a notification to the user after the meeting ends, including a link to the minutes.

[1525] 5. The user confirms the notification and opens the automatically generated minutes.

[1526] 6. If necessary, the user can edit the minutes and use them for the next meeting or report.

[1527] In this way, the system of the present invention, which combines an emotion engine, can further improve the efficiency and effectiveness of meetings by providing real-time information during the meeting and generating detailed minutes after the meeting, as well as recognizing the user's emotions and presenting information accordingly.

[1528] The processing flow will be explained below.

[1529] Step 1:

[1530] The server receives user-provided materials, past conversation history, and email data, imports them into a database, checks that the data is properly formatted, and reports errors if there are any anomalies.

[1531] Step 2:

[1532] The server analyzes the content of the imported data and automatically generates related keywords and tags to add them as metadata. Based on this metadata, it builds an index to improve search performance.

[1533] Step 3:

[1534] The device continues to capture audio during the meeting in real time, and the captured audio data is sent to the speech recognition API at predetermined intervals.

[1535] Step 4:

[1536] The speech recognition API converts the received speech data into text data and sends the converted text data to the server.

[1537] Step 5:

[1538] The server then inputs the received text data into a large-scale language model (LLM) to analyze the conversation, extracting important keywords and phrases from the text.

[1539] Step 6:

[1540] The device captures the user's voice, facial expressions, and physical movements in real time and transmits this data to the emotion engine.

[1541] Step 7:

[1542] The emotion engine analyzes the received data and recognizes and classifies the user's emotions (e.g., excitement, confusion, interest, etc.).

[1543] Step 8:

[1544] The server searches the database for relevant information based on the emotion data recognized by the emotion engine and the keywords extracted by the LLM, and then retrieves the search results and ranks them based on relevance and priority.

[1545] Step 9:

[1546] The server selects the most relevant information from the ranked search results and automatically adjusts the way the information is presented based on the user's emotions.

[1547] Step 10:

[1548] The server sends tailored information to the terminal, for example, sending additional explanations or detailed materials if the user is confused.

[1549] Step 11:

[1550] The device visually displays the received relevant information and materials to the user, and can display the information in a pop-up display or a new window on the user's screen.

[1551] Step 12:

[1552] The server automatically generates minutes of the meeting based on the voice data captured during the meeting and the corresponding voice recognition results.

[1553] Step 13:

[1554] The server assigns emotion tags to the generated minutes and automatically associates them with the user's account after the meeting ends.

[1555] Step 14:

[1556] After the meeting, the server sends a notification to the user with a link to the minutes, allowing the user to view and edit the minutes.

[1557] Step 15:

[1558] The user opens the notification, reviews the automatically generated minutes, and, if necessary, completes the minutes by checking for errors and adding additional comments or corrections.

[1559] In this way, by performing specific operations for each processing step, the system of the present invention can efficiently provide information during a meeting and generate minutes after the meeting. In addition, by recognizing the user's emotions and adjusting the information presentation method accordingly, the quality of the meeting can be further improved.

[1560] Example 2

[1561] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1562] Conventional meeting support systems could recognize comments in real time, present relevant information, and create minutes, but they had limitations in providing information that took user emotions into account and automatically documenting meeting content. In addition, there were cases where users were unable to provide appropriate support because they did not understand the emotions behind their comments. Furthermore, because the emotions of participants were not reflected in post-meeting minutes, it was difficult to accurately grasp the progress and effectiveness of the meeting.

[1563] 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.

[1564] In this invention, the server includes speech recognition means for recognizing utterances made during a conference in real time, conversion means for converting the utterances recognized by the speech recognition means into text data, analysis means for analyzing the text data converted by the conversion means, search means for searching a database for related information based on keywords extracted by the analysis means, display means for displaying the related information acquired by the search means, emotion recognition means for recognizing emotion data of users during the conference, means for adjusting the method of presenting information based on the emotion data acquired by the emotion recognition means, documentation means for automatically documenting the conference content, and means for reflecting the emotion information in documents automatically generated by the documentation means. This makes it possible to grasp the progress of the conference and the emotions of the participants in real time, and to provide more effective support during and after the conference.

[1565] "Speech recognition means" refers to a device or software for recognizing speech during a meeting in real time.

[1566] The "conversion means" refers to a device or software that has the function of converting the speech content recognized by the speech recognition means into text data.

[1567] The "analysis means" refers to a device or software for analyzing the text data converted by the conversion means and extracting important keywords and phrases.

[1568] The "search means" refers to a device or software for searching a database for related information based on the keywords extracted by the analysis means.

[1569] The "display means" refers to a device or software that visually presents the related information obtained by the search means to the user.

[1570] "Emotion recognition means" refers to a device or software for analyzing a user's voice, facial expression, or physical movement during a meeting to recognize emotional data.

[1571] "Emotional data" refers to data that expresses a user's emotional state (e.g., excitement, confusion, interest, etc.).

[1572] "Documentation tool" refers to a device or software that automatically documents what is said and related information during a meeting.

[1573] "Emotion information" refers to information indicating the emotional state of the user acquired by emotion recognition means.

[1574] "Large-scale language model" refers to a highly trained machine learning model used for natural language processing analysis.

[1575] The conference support system according to the present invention recognizes comments made during a conference in real time, provides related information, creates minutes, and recognizes the user's emotions and presents information based on those emotions.

[1576] Building a database

[1577] The server creates a database containing the user's documents, past conversation history, and email data. It analyzes the content of each imported document and assigns relevant keywords and tags to the data, enabling it to quickly and accurately provide relevant information when searching.

[1578] For example, when a user uploads relevant materials before a meeting, the server updates the database based on those materials and automatically assigns necessary keywords and tags.

[1579] Speech Recognition and Analysis

[1580] The device captures audio during the meeting in real time. The captured audio data is sent to a speech recognition API such as Google Cloud Speech-to-Text API and converted into text data. This text data is then sent to a server where it is analyzed by a large-scale language model (LLM) such as OpenAI's GPT-4. The LLM extracts important keywords and phrases from the text data.

[1581] For example, if a user says, "I'm not sure if this data is correct," the device captures it and sends it to a speech recognition API, which generates text data. This text data is then sent to a server and analyzed using a large-scale language model.

[1582] Real-time search and information provision

[1583] The server searches the database based on the keywords extracted by the analysis means. The search results are ranked based on relevance and priority, and the most relevant information is selected. The selected information is sent to the terminal and displayed to the user.

[1584] As a specific example, the server searches for relevant verification procedure documents based on keywords such as "data," "correct," and "unknown," and sends the most relevant information to the user's terminal.

[1585] Automatic documentation of meeting content

[1586] All statements made during the meeting and related information automatically acquired are documented by the server. After the meeting ends, the generated minutes are linked to the user's account and a notification is sent. Users can view and edit the minutes through the notification.

[1587] Specifically, after the meeting ends, the server automatically generates minutes and creates a link to the user's account. The contents of the meeting and related information are all organized in text format.

[1588] Introducing emotion recognition

[1589] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to a pre-installed emotion recognition engine, which analyzes and recognizes the user's emotions (e.g., excitement, confusion, interest, etc.).

[1590] For example, the device's camera and microphone capture the user's reactions and send the real-time data to the emotion engine, which then analyzes the user's emotions.

[1591] Coordination of information provision

[1592] The server adjusts the content and method of information presented based on the emotional data recognized by the emotion engine. For example, if the emotion engine determines that the user is confused, the server will present additional explanations or detailed materials. If the user shows interest, the server will provide related in-depth information.

[1593] As a specific example, if the emotion engine detects that the user is confused, the server retransmits detailed information to the terminal and displays it to the user.

[1594] Emotion tagging

[1595] The server reflects the emotional information recognized during the meeting in the minutes. By assigning appropriate emotional tags to each part of the minutes, the progress of the meeting can be understood in more detail.

[1596] For example, the server automatically generates emotion tags for each comment and inserts them into the minutes, recording not only the content of the meeting but also changes in participants' emotions.

[1597] By combining these functions, a system can provide real-time information during a meeting, generate detailed minutes after the meeting, and present information according to the user's emotions. Such a system will further improve the efficiency and effectiveness of meetings.

[1598] Prompt Sentence Examples

[1599] For example, you can check the system's operation by using the following prompt sentence:

[1600] It is unclear whether this data is correct.

[1601] The server analyzes the text data to search for relevant information, and also recognizes the user's emotions to adjust the way information is presented.

[1602] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1603] Step 1: Importing data and building the database

[1604] The server imports documents uploaded by users, past conversation history, and email data, and builds a database based on these. It analyzes the content of each document in the imported data and assigns relevant keywords and tags. The input is file data in various formats uploaded by users, and the output is an indexed database. Specifically, the server scans the documents and automatically extracts relevant keywords using NLP (natural language processing) technology, updating the database accordingly.

[1605] Step 2: Capture and convert audio

[1606] The device captures speech during the meeting in real time. The captured voice data is sent to a speech recognition API such as the Google Cloud Speech-to-Text API and converted into text data. The input is the real-time captured voice data, and the output is the recognized text data. Specifically, the microphone captures the voice and sends it to the speech recognition API, which then converts the voice into text and sends it back to the server.

[1607] Step 3: Analyzing the text data

[1608] The server receives the converted text data and analyzes it using an internal large-scale language model (for example, OpenAI's generative AI model). The analysis means extracts important keywords and phrases from the text data. The input is the text data received from the speech recognition API, and the output is the extracted keywords and phrases. Specifically, the large-scale language model analyzes the text and automatically extracts specific keywords and phrases from it.

[1609] Step 4: Find related information

[1610] The server searches for relevant information from a database based on the extracted keywords. The search results are ranked based on relevance and priority. The input is the extracted keywords, and the output is the ranked search results. Specifically, the server executes a database query to retrieve relevant materials and data, and ranks them based on priority.

[1611] Step 5: Viewing search results

[1612] The device displays the search results sent from the server to the user, allowing the user to see the most relevant information in real time. The input is the ranked search results sent from the server, and the output is the specific information displayed on the user's screen. In specific operation, the screen display portion of the device displays the relevant information through a user interface.

[1613] Step 6: Capturing and analyzing user emotions

[1614] The device captures the user's voice, facial expressions, and physical movements in real time and sends this data to an emotion recognition engine. The emotion recognition engine analyzes the acquired emotion data and recognizes the user's emotions. The input is the captured real-time user data, and the output is analyzed emotion data. Specifically, the camera and microphone capture the user's voice and facial movements and send this data to the emotion engine.

[1615] Step 7: Tailor your information based on emotions

[1616] The server adjusts the content and format of the information it presents based on the emotional data analyzed by the emotion recognition engine. For example, if it recognizes that the user is confused, the server will provide additional explanations or detailed materials. The input is the emotional data received from the emotion engine, and the output is the adjusted presentation of information. Specifically, the server dynamically changes the content and format of the information it presents based on the emotional data and sends it to the device.

[1617] Step 8: Automatically document meeting content

[1618] The server documents what was said during the meeting and any related information that was automatically acquired. After the meeting ends, this document is linked to the user's account as minutes, and a notification is sent. The input is the analyzed text data and related information, and the output is the automatically generated minutes. Specifically, the server collects, organizes, and documents the data from the meeting, saves the results in the user's account, and sends a notification.

[1619] Step 9: Sentiment tagging of meeting notes

[1620] The server also reflects the emotional information recognized during the meeting in the minutes. This allows each part of the minutes to be assigned an emotional tag. The input is the emotional data and the generated minutes, and the output is a detailed minutes with the emotional tags. Specifically, the server automatically generates an emotional tag corresponding to each utterance and adds it to the corresponding part of the minutes.

[1621] (Application example 2)

[1622] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1623] While conventional conferencing systems can recognize comments made during meetings in real time and provide relevant information, they provide information without considering the emotions of workers working on-site, making it difficult to provide efficient work support. There is also a need for a system that can appropriately recognize workers' emotions and stress levels and adjust and provide information accordingly.

[1624] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: a voice recognition means for recognizing utterances made during a meeting in real time; a conversion means for converting the utterances recognized by the voice recognition means into text data; an analysis means for analyzing the text data converted by the conversion means; a search means for searching a database for related information based on keywords extracted by the analysis means; a display means for displaying the related information acquired by the search means; a documentation means for automatically documenting the meeting contents; an expression recognition means for capturing the facial expressions of workers at the workplace and recognizing their emotions in real time; and an emotion analysis means for adjusting the presentation method of related information based on the emotion data obtained by the expression recognition means. This enables efficient provision of information and work support in accordance with the emotions of workers.

[1625] The "voice recognition means" is a means for recognizing speech during a meeting in real time and converting the speech data into text.

[1626] The "conversion means" is a means for converting the speech content recognized by the speech recognition means into text data.

[1627] The "analysis means" is a means for analyzing the text data converted by the conversion means and extracting important keywords and phrases.

[1628] The "search means" is a means for searching the database for related information based on the keywords extracted by the analysis means.

[1629] The "display means" is a means for visually displaying the related information obtained by the search means to the user.

[1630] The "documentation means" is a means for automatically recording the contents of a meeting in a document format and notifying the user of the minutes generated after the meeting has ended.

[1631] The "facial expression recognition means" is a means for capturing the facial expressions of workers at the workplace and recognizing their emotions in real time.

[1632] The "emotion analysis means" is a means for adjusting the presentation method of related information based on the emotion data obtained by the facial expression recognition means.

[1633] The system of the present invention recognizes and analyzes worker conversations and emotions in real time during work meetings and daily tasks in a factory, and provides relevant work procedures, manuals, and past history information. Furthermore, by analyzing worker emotions and providing appropriate work instructions and supplementary explanations accordingly, it is possible to improve work efficiency and safety. This system includes the following hardware and software configuration:

[1634] First, the Google Cloud Speech-to-Text API is used as a speech recognition means to recognize statements made during meetings in real time. This converts the voice data into text data. The converted text data is sent to a server and analyzed using a large-scale language model (for example, OpenAI GPT). The analysis means extracts important keywords and phrases, and a search means operates to search a database for related information based on those keywords. The searched information is ranked based on relevance and priority, and the most appropriate information is sent from the server to the terminal and visually presented to the user via the terminal's display means. The display means can be a standard monitor, smartphone, or tablet.

[1635] Furthermore, a camera is used as an expression recognition means to capture the facial expressions of workers on the workplace. The captured facial expression data is analyzed using OpenCV to recognize the worker's emotions (e.g., confusion, excitement, interest, etc.) in real time. This emotion data is processed by the emotion analysis means, and the way related information is presented is automatically adjusted. For example, if the worker is confused, additional explanations or detailed materials are presented. On the other hand, if the worker shows interest, relevant in-depth information is provided.

[1636] The meeting content is automatically documented by the server, and meeting minutes are generated based on the captured audio data and its analysis. The generated minutes are linked to the user's account, and a notification is sent after the meeting ends.

[1637] Specific examples

[1638] Example 1: Providing information in response to a worker's confusion

[1639] 1. Worker: "I don't know how to set up this machine."

[1640] 2. The server captures this utterance and sends it to the speech recognition API.

[1641] 3. The speech recognition API converts the speech into text "I don't know how to set up this machine" and sends it to the server.

[1642] 4. The server analyzes the text using a large-scale language model and extracts important keywords such as "machine," "settings," and "don't know."

[1643] 5. The server searches the database based on these keywords and retrieves the relevant configuration instructions.

[1644] 6. At the same time, the facial expression recognition means captures the worker's facial expression and recognizes that he is confused.

[1645] 7. The server will provide further detailed configuration instructions and additional instructions.

[1646] Prompt Sentence Examples

[1647] User input: "I don't know how to configure this machine."

[1648] Prompt for generative AI model: "Please provide relevant instructions or manuals based on the following sentence: 'I don't know how to set up this machine.'"

[1649] In this way, the system according to the present invention recognizes the emotions of workers in real time and provides efficient and appropriate information and support accordingly.

[1650] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1651] Step 1:

[1652] A user speaks while working or in a meeting. This speech is captured in real time by the microphone of the terminal. The input is the user's voice, and the output is the captured voice data.

[1653] Step 2:

[1654] The device sends the captured voice data to the Google Cloud Speech-to-Text API, where the data is converted into text data. The output from the API is text data in string format.

[1655] Step 3:

[1656] The server receives the text data. This text data is then input into a large-scale language model such as OpenAI GPT to extract important keywords and phrases. This involves data calculations such as TEKST data analysis and keyword extraction, and the output is a list of keywords.

[1657] Step 4:

[1658] The server then searches a database based on the extracted keyword list to retrieve related information. The input is the keyword list, and the database search outputs related materials, procedure manuals, etc.

[1659] Step 5:

[1660] The server transmits the acquired related information to the terminal, where the data transfer operation is performed and the information is provided to the user by the display means. The information visually presented using the terminal's display is the output.

[1661] Step 6:

[1662] At the same time, the terminal uses a camera to capture the worker's facial expression, whose input is the worker's facial image data and whose output is the captured facial data.

[1663] Step 7:

[1664] The device uses OpenCV to recognize facial expressions and analyzes the facial expressions of the worker from the captured facial data. The input is facial data, and emotion analysis is performed as data calculation. The analysis results are output as emotional data such as confusion, interest, and excitement.

[1665] Step 8:

[1666] The server receives the emotion data and adjusts the presentation of related information based on it. For example, if the worker is confused, it will send a document with additional detailed explanations to the terminal again. The input data is emotion data, and the presentation information is adjusted as a data calculation. The output is the readjusted related information.

[1667] Step 9:

[1668] The server automatically documents the meeting contents. It generates meeting minutes using the captured audio data and the analysis results based on it. The input data is audio data and text data, and the data is processed into documentation. The output is the generated minutes.

[1669] Step 10:

[1670] The generated minutes are linked to the user's account by the server and a notification is sent after the meeting ends. The output is a notification with a link to the minutes.

[1671] Through the above steps, the system according to the present invention recognizes the emotions of workers in real time and provides efficient and appropriate information and support accordingly.

[1672] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1673] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1674] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1675] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1676] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1677] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1678] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1679] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1680] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1681] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1682] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1683] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1684] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1685] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1686] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1687] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1688] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1689] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1690] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1691] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1692] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1693] The following is further disclosed regarding the above embodiment.

[1694] (Claim 1)

[1695] A speech recognition means for recognizing speeches made during a meeting in real time;

[1696] A conversion means for converting the speech content recognized by the speech recognition means into text data;

[1697] analysis means for analyzing the text data converted by the conversion means;

[1698] a search means for searching a database for related information based on the keywords extracted by the analysis means;

[1699] a display means for displaying the related information acquired by the search means;

[1700] a documentation means for automatically documenting the contents of the meeting;

[1701] A system including:

[1702] (Claim 2)

[1703] 2. The system of claim 1, wherein the analyzing means extracts keywords using a large-scale language model.

[1704] (Claim 3)

[1705] 2. The system according to claim 1, wherein the documentation means includes means for notifying the user of the automatically generated minutes after the meeting has ended.

[1706] "Example 1"

[1707] (Claim 1)

[1708] A speech recognition means for recognizing speeches made during a meeting in real time;

[1709] A conversion means for converting the speech content recognized by the speech recognition means into text data;

[1710] analysis means for analyzing the text data converted by the conversion means;

[1711] a search means for searching a database for related information based on the keywords extracted by the analysis means;

[1712] a display means for displaying the related information acquired by the search means;

[1713] a documentation means for automatically documenting the contents of the meeting;

[1714] linking means for linking the documentation generated by the documentation means to a user account;

[1715] a notification means for sending a notification to the user after the conference has ended;

[1716] an audio capture means for capturing audio using a microphone;

[1717] a transmitting means for transmitting the captured voice data to the voice recognition API;

[1718] A ranking method for ranking the results of a database search

[1719] A system including:

[1720] (Claim 2)

[1721] 2. The system of claim 1, wherein the analyzing means extracts keywords using a large-scale language model.

[1722] (Claim 3)

[1723] 2. The system according to claim 1, wherein the documenting means includes means for generating minutes based on statements made during the meeting and information obtained.

[1724] "Application Example 1"

[1725] (Claim 1)

[1726] A speech recognition means for recognizing speeches made during a meeting in real time;

[1727] A conversion means for converting the speech content recognized by the speech recognition means into text data;

[1728] analysis means for analyzing the text data converted by the conversion means;

[1729] a search means for searching a database for related information based on the keywords extracted by the analysis means;

[1730] a display means for displaying the related information acquired by the search means;

[1731] A display means for providing detailed product information and inventory information in a physical store;

[1732] a documentation means for automatically documenting the contents of the meeting;

[1733] A system including:

[1734] (Claim 2)

[1735] 2. The system of claim 1, wherein the analyzing means extracts keywords using a large-scale language model.

[1736] (Claim 3)

[1737] 2. The system according to claim 1, wherein the documentation means includes means for notifying the user of the automatically generated minutes after the meeting has ended.

[1738] (Claim 4)

[1739] The system according to claim 1, wherein product information and inventory information are provided in real time using keywords generated based on comments.

[1740] "Example 2: Combining Emotion Engines"

[1741] (Claim 1)

[1742] A speech recognition means for recognizing speeches made during a meeting in real time;

[1743] A conversion means for converting the speech content recognized by the speech recognition means into text data;

[1744] analysis means for analyzing the text data converted by the conversion means;

[1745] a search means for searching a database for related information based on the keywords extracted by the analysis means;

[1746] a display means for displaying the related information acquired by the search means;

[1747] emotion recognition means for recognizing emotion data of a user during a conference;

[1748] means for adjusting a method of presenting information based on emotion data acquired by the emotion recognition means;

[1749] a documentation means for automatically documenting the contents of the meeting;

[1750] a means for reflecting emotional information in a document automatically generated by the documentation means;

[1751] A system including:

[1752] (Claim 2)

[1753] 2. The system of claim 1, wherein the analysis means extracts keywords using a generative AI model.

[1754] (Claim 3)

[1755] 2. The system according to claim 1, wherein the documentation means includes means for notifying the user of the automatically generated minutes after the meeting has ended.

[1756] "Application example 2 when combining emotion engines"

[1757] (Claim 1)

[1758] A speech recognition means for recognizing speeches made during a meeting in real time;

[1759] A conversion means for converting the speech content recognized by the speech recognition means into text data;

[1760] analysis means for analyzing the text data converted by the conversion means;

[1761] a search means for searching a database for related information based on the keywords extracted by the analysis means;

[1762] a display means for displaying the related information acquired by the search means;

[1763] a documentation means for automatically documenting the contents of the meeting;

[1764] An expression recognition means for capturing facial expressions of workers at a workplace and recognizing their emotions in real time;

[1765] emotion analysis means for adjusting a presentation method of related information based on emotion data obtained by the facial expression recognition means;

[1766] A system including:

[1767] (Claim 2)

[1768] 2. The system of claim 1, wherein the analyzing means extracts keywords using a large-scale language model.

[1769] (Claim 3)

[1770] 2. The system according to claim 1, wherein the documentation means includes means for notifying the user of the automatically generated minutes after the meeting has ended. [Explanation of symbols]

[1771] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A speech recognition means for recognizing speeches made during a meeting in real time; A conversion means for converting the speech content recognized by the speech recognition means into text data; analysis means for analyzing the text data converted by the conversion means; a search means for searching a database for related information based on the keywords extracted by the analysis means; a display means for displaying the related information acquired by the search means; a documentation means for automatically documenting the contents of the meeting; A system including:

2. 2. The system of claim 1, wherein the analyzing means extracts keywords using a large-scale language model.

3. 2. The system according to claim 1, wherein the documenting means includes means for notifying the user of the automatically generated minutes after the meeting has ended.

Images