system

Abstract

We provide the system. [Solution] Means for collecting publicly available document data, A method for extracting text data from collected document data, A method for preprocessing extracted text data to remove noise and normalize the text, A means for analyzing morphemes from preprocessed text data and extracting important topics, A means for generating a summary based on extracted topics, A means of saving the generated summary and providing it to the user, A system that includes this.

Description

Technical Field

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

Background Art

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

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In modern society, it is very difficult to quickly and accurately extract important information from publicly available document data, especially a large amount of text information such as meeting minutes and reports, and create a summary. Especially in the foreign affairs department, young people spend a huge amount of time creating summaries, but the problem is that productivity decreases during the process. The purpose of the present invention is to automate such laborious and time-consuming work and improve the business efficiency of enterprises and organizations by efficiently creating summaries.

Means for Solving the Problems

[0005] This invention provides a system for collecting, preprocessing, analyzing, and generating summaries of publicly available document data. The system first includes means for collecting publicly available document data by web crawling. Next, it includes means for extracting text data from the collected document data, and further means for preprocessing the text data to remove noise and normalize it. Subsequently, it includes means for analyzing the preprocessed text data using a morphological analysis tool and extracting important topics. Finally, it includes means for generating summaries using natural language generation technology based on the extracted topics, and means for saving the generated summaries and making them available to users. This system enables efficient summary creation, saving effort and time, and improving business productivity.

[0006] "Publicly available document data" refers to document information that is made publicly available through the internet or other media and is accessible to anyone.

[0007] "Text data" refers to string information extracted from document data, which has been formatted into a format that can be parsed and processed.

[0008] "Preprocessing" refers to the initial processing performed on text data, including preparatory work to facilitate analysis, such as noise reduction and normalization.

[0009] "Noise reduction" is the process of removing unnecessary information and special characters from text data, leaving only the necessary information.

[0010] "Normalization" is the process of unifying text data in different formats into a consistent format.

[0011] Morphological analysis is a technique used in natural language processing that divides text into individual words and analyzes the meaning and part of speech of each word.

[0012] A "topic" refers to the subject or theme within text data, and is the main content that the text deals with.

[0013] "Natural language generation technology" refers to the technology that automatically generates natural-sounding sentences based on data analyzed by a computer.

[0014] "Web crawling" is a technology that automatically visits web pages on the internet to collect information.

[0015] A "summary" is a short piece of text that summarizes the main points or content of a document.

[0016] "Saving" refers to the act of recording and retaining generated summaries and data in a database or storage device.

[0017] "Providing to users" refers to the act of transmitting saved summaries and data in a format that makes them viewable and usable by users who require them. [Brief explanation of the drawing]

[0018] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7]It is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] It is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] It shows an emotion map to which a plurality of emotions are mapped. [Figure 10] It shows an emotion map to which a plurality of emotions are mapped. [Figure 11] It is a sequence diagram showing the processing flow of the data processing system in Example 1. [Figure 12] It is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13] It is a sequence diagram showing the processing flow of the data processing system in Example 2 when an emotion engine is combined. [Figure 14] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 when an emotion engine is combined.

Modes for Carrying Out the Invention

[0019] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0020] First, the terms used in the following description will be explained.

[0021] In the following embodiments, the signed processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Furthermore, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), and APU (Accelerated Processing Unit).

[0022] In the following embodiments, signed RAM (Random Access Memory) is a memory that temporarily stores information and is used as work memory by the processor.

[0023] In the following embodiments, the signed storage is one or more non-volatile storage devices that store various programs and various parameters. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes.

[0024] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0025] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0026] [First Embodiment]

[0027] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0028] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0029] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0030] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0031] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0032] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0033] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0034] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0035] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0036] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0037] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0038] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0039] This invention is a system that collects publicly available document data, particularly meeting minutes and report data from the internet, preprocesses and analyzes it, and extracts and summarizes important points from this data to provide a summary. The following describes specific embodiments for implementing this invention.

[0040] Conceptual explanation of the program

[0041] overview

[0042] This system begins by acquiring publicly available document data, preprocessing and analyzing it to extract important information, and finally generating a summary to provide to the user. The following describes the specific actions of the server, terminal, and user at each step.

[0043] Data collection

[0044] The server collects meeting minutes data from a specified internet URL using web crawling technology. For example, if the target is the minutes of a research group published by the Ministry of Internal Affairs and Communications, the server accesses the relevant web page and downloads the HTML content.

[0045] Data extraction and preprocessing

[0046] The server extracts the meeting minutes text from the downloaded HTML content. During this process, it performs noise reduction to remove HTML tags, unnecessary special characters, line breaks, etc. Next, it performs normalization, such as converting full-width spaces to half-width spaces and unifying different expressions.

[0047] Text analysis and topic extraction

[0048] The server performs morphological analysis on the preprocessed text data. A morphological analysis tool (e.g., MeCab) is used here. As a result of the analysis, the text data is divided into words, and the part of speech of each word is identified. Based on the data obtained from this analysis, important topics are extracted using topic modeling (e.g., LDA or TF-IDF).

[0049] Summary Generation

[0050] The server uses natural language generation technology to create summaries based on the extracted topics. For example, it uses a template to concisely summarize the main points in the format "XX was discussed, and YY was decided." Grammatical checks are also performed during this process to generate consistent sentences.

[0051] Saving and providing summaries

[0052] The created summary is stored in the server's database. When a user requests a summary through their device, the server searches the stored summary and sends the relevant data to the user's device. The user can then view the summary on their device.

[0053] Specific example

[0054] For example, when the Ministry of Internal Affairs and Communications provides a summary of the latest research meeting minutes, the following process occurs when a user clicks the "View summary of the latest meeting minutes" button on their device.

[0055] 1. The user clicks a button on their device.

[0056] 2. The terminal sends a summary request to the server.

[0057] 3. The server generates a summary from the latest collected meeting minutes data and saves it to the database.

[0058] 4. The server searches the saved summary and sends it to the user's terminal.

[0059] 5. The terminal displays the summary received from the server on its screen, which the user can then view.

[0060] Through this series of processes, users can efficiently obtain important information, and the system of the present invention facilitates the understanding and sharing of meeting minutes.

[0061] The following describes the processing flow.

[0062] Step 1: Data Collection

[0063] The server accesses the specified URL (for example, a public webpage of a specific government agency).

[0064] The server uses web crawling technology to download the HTML content of web pages.

[0065] The server saves the downloaded data to temporary storage.

[0066] Step 2: Data Extraction

[0067] The server runs an HTML parser to extract the meeting minutes text from the saved HTML content.

[0068] The server extracts the text data while removing HTML tags.

[0069] Step 3: Data Preprocessing

[0070] As a preprocessing step, the server removes unnecessary special characters and multiple line breaks / spaces from the text data.

[0071] The server performs normalization, such as converting full-width spaces to half-width spaces within the text.

[0072] Step 4: Morphological Analysis

[0073] The server uses a morphological analysis tool (e.g., MeCab) to divide the preprocessed text data into words.

[0074] The server tags each word with its part of speech and records the results of morphological analysis.

[0075] Step 5: Topic Extraction

[0076] The server uses topic modeling techniques (e.g., LDA and TF-IDF) to extract important topics from morphologically analyzed data.

[0077] The server evaluates the frequency and importance of words related to each topic and lists the main topics.

[0078] Step 6: Summary Generation

[0079] The server generates a summary sentence based on the extracted topics, utilizing natural language generation technology.

[0080] The server constructs the text based on a specific template (e.g., "[Topic] was discussed, and [Result] was determined.").

[0081] The server performs a grammatical check on the generated summary sentence and makes corrections as needed.

[0082] Step 7: Save the summary

[0083] The server saves the generated summary to the database.

[0084] The server also records metadata related to the summary (e.g., publication date, meeting minutes title, attendees).

[0085] Step 8: Provide a summary

[0086] The user requests a summary through their device.

[0087] The server searches the database for the relevant summary and sends it to the user's terminal.

[0088] The terminal displays the received summary to the user.

[0089] As described above, a system is realized in which the server, terminal, and user cooperate at each step to efficiently generate and provide document summaries.

[0090] (Example 1)

[0091] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0092] The sheer volume of publicly available document data, particularly meeting minutes and reports found online, makes it difficult to efficiently extract key points and provide summaries. Furthermore, manually collecting, pre-processing, analyzing, and summarizing this data is time-consuming and labor-intensive. Therefore, there is a need for a system that automatically and efficiently collects and analyzes document data, extracts key points, and provides summaries.

[0093] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0094] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary using a generative AI model based on the extracted topics, means for verifying the generated summary using a grammar checking tool, and means for saving the generated summary and providing it to the user. This enables the user to efficiently obtain important information from a vast amount of document data.

[0095] "Publicly available document data" refers to document data that is generally accessible on the internet or in other public spheres.

[0096] "Web crawling" is a technology that automatically collects data from specified web pages and websites.

[0097] "HTML content" refers to data in HTML (Hypertext Markup Language) format that represents the structure and content of a web page.

[0098] "Preprocessing" refers to a series of processes that remove noise from data and normalize it before performing data analysis.

[0099] "Noise reduction" is the process of removing elements from text data that are unnecessary for analysis (e.g., HTML tags, special characters, unnecessary line breaks).

[0100] "Normalization" is a process that unifies different representations within text data to ensure data consistency.

[0101] Morphological analysis is a technique that divides text data into the smallest possible linguistic elements and analyzes the part of speech of each element.

[0102] A "morphological analysis tool" refers to software or algorithms used to perform morphological analysis (e.g., MeCab).

[0103] Topic modeling is a technique that automatically extracts important topics and themes from large amounts of text data (e.g., LDA, TF-IDF).

[0104] A "generative AI model" is an artificial intelligence model that uses machine learning to generate natural language (e.g., GPT-3(registered trademark)).

[0105] A "summary" is a concise summary that captures the key points of an entire document.

[0106] A "grammar check tool" is software or algorithms used to verify and correct the grammatical accuracy of generated text.

[0107] A "database" is a system that allows for the efficient storage, management, and retrieval of structured data.

[0108] A "terminal" refers to a device (e.g., a personal computer, smartphone, or tablet) that a user uses to access a server and view or manipulate data.

[0109] This invention is a system that collects publicly available document data, particularly meeting minutes and report data from the internet, preprocesses and analyzes it, and extracts and summarizes important points from this data. A detailed embodiment of this invention is described below.

[0110] Data collection

[0111] The server uses web crawling technology to collect meeting minutes data from specified internet URLs. Specifically, the server accesses URLs in a specified list sequentially, retrieves HTML content, and saves it to local storage. For example, it might access the URL of a publicly available research conference minutes page of a certain public institution and download the HTML file.

[0112] Data extraction and preprocessing

[0113] The server extracts the necessary meeting minutes text from the collected HTML content and performs preprocessing. It uses an HTML parser (e.g., Beautiful Soup) to remove HTML tags, unnecessary special characters, and line breaks. Furthermore, it performs normalization, such as converting full-width spaces to half-width spaces and unifying different expressions, as part of noise reduction. For example, the meeting minutes text " Chairperson: Thank you for joining us today. Extract "Chairman: Thank you for participating today." from the above.

[0114] Text analysis and topic extraction

[0115] The server performs morphological analysis on the preprocessed text data. For this analysis, a morphological analysis tool (e.g., MeCab) is used to divide the text into the smallest units of linguistic elements and identify the part of speech for each. For example, the sentence "Thank you for participating today." is divided into "today / noun," "wa / particle," "participation / noun," "arigato / verb," ​​"u / auxiliary verb," ​​and "gozaimasu / verb." Subsequently, topic modeling techniques (e.g., LDA or TF-IDF) are used to extract important topics from the text data. Frequently occurring keywords such as "research," "minutes," and "technology" are extracted.

[0116] Summary Generation

[0117] The server generates summaries using a generative AI model (e.g., GPT-3) based on important topics. For example, it might generate a summary such as, "The minutes mainly discussed advancements in new technologies, and future research directions were decided." The generated summaries are then checked for grammatical accuracy using a grammar checking tool and corrected as needed.

[0118] Saving and providing summaries

[0119] The generated summaries are stored in the server's database. The storage format includes fields such as "timestamp," "meeting minutes title," and "summary." When a user requests a summary through their terminal, the terminal sends a request to the server, the server searches the database for the corresponding summary, and sends it to the terminal. The terminal displays the summary received from the server on its screen, allowing the user to view it.

[0120] Specific example

[0121] For example, if a public institution is providing a summary of its latest meeting minutes, when a user clicks the "View Summary of Latest Meeting Minutes" button on their device, the following process takes place.

[0122] 1. The user clicks a button on their device.

[0123] 2. The terminal sends a summary request to the server.

[0124] 3. The server generates a summary from the latest meeting minutes data it has collected and saves it to the database.

[0125] 4. The server searches for the saved summary and sends it to the user's terminal.

[0126] 5. The terminal displays the summary received from the server on the screen, and the user views it.

[0127] As an example of a prompt, the process is carried out by inputting "Please provide a summary of the latest research conference minutes from a public institution" into the generating AI model.

[0128] This system allows users to efficiently obtain important information from vast amounts of document data.

[0129] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0130] Step 1:

[0131] The server collects publicly available document data. Specifically, the server receives a specified list of URLs as input and accesses each URL using web crawling technology. It downloads the HTML content of each URL and saves it to local storage. The output is the downloaded HTML file.

[0132] Step 2:

[0133] The server extracts text data from the collected HTML content. Specifically, it receives an HTML file as input and uses an HTML parser (e.g., Beautiful Soup) to extract the meeting minutes text. As a noise reduction process, HTML tags, unnecessary special characters, and line breaks are removed. The output is the meeting minutes text. For example, HTML content " Chairperson: Thank you for joining us today. Extract "Chairman: Thank you for participating today." from the above.

[0134] Step 3:

[0135] The server performs preprocessing on the extracted text data. Specifically, it receives the text data as input, converts full-width spaces to half-width spaces, and performs normalization to unify different expressions (e.g., AI, エーアイ). The output is normalized text data.

[0136] Step 4:

[0137] The server performs morphological analysis on the pre-processed text data. Specifically, it uses a morphological analysis tool (e.g., MeCab) to receive the text data as input, split it into words, and identify the part of speech of each word. The output is a list of the split words. For example, the sentence "Thank you for participating today." is split into "today / noun", "is / particle", "participation / noun", "thank you / verb", "u / auxiliary verb", and "gozaimasu / verb".

[0138] Step 5:

[0139] The server extracts important topics using topic modeling techniques. Specifically, it takes a list of words obtained as a result of morphological analysis as input and uses topic modeling techniques (e.g., LDA or TF-IDF) to extract important topics. The output is a list of the extracted topics.

[0140] Step 6:

[0141] The server generates a summary based on the extracted topics. Specifically, it takes a list of important topics as input and generates a summary using a generative AI model (e.g., GPT-3). The output is the generated summary. For example, it might generate a summary such as, "The meeting minutes mainly discussed the progress of new technologies, and future research directions were decided."

[0142] Step 7:

[0143] The server checks the generated summary using a grammar checking tool. Specifically, it receives the generated summary as input, uses the grammar checking tool to verify its grammatical accuracy, and makes corrections as needed. The output is the corrected summary.

[0144] Step 8:

[0145] The server saves the generated summary to the database. Specifically, it receives the modified summary as input and saves it to the database. The output is the summary saved in the database. The saving format is in the form of fields such as "timestamp," "meeting minutes title," and "summary."

[0146] Step 9:

[0147] The user requests the summary via their device. Specifically, the user clicks the "View summary of the latest meeting minutes" button on their device. This action causes the device to send a summary request to the server.

[0148] Step 10:

[0149] The server retrieves the relevant summary from the database and sends it to the user's terminal. Specifically, the server receives a summary request, retrieves the relevant summary from the database, and sends it to the terminal. The output is the sent summary.

[0150] Step 11:

[0151] The terminal displays the summary received from the server on its screen, which the user then views. Specifically, the terminal receives the summary as input and displays it on the screen. The output is a summary display that the user can view.

[0152] Therefore, users can efficiently obtain important information.

[0153] (Application Example 1)

[0154] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0155] While a vast amount of meeting minutes and reports are publicly available online, there is a lack of efficient means to collect this information, extract key points, and provide concise summaries. Traditional systems require significant time and effort for manual information gathering and summarization, and the sheer volume of information makes it highly likely that users will miss useful details. Furthermore, maintaining consistent summaries is difficult, highlighting the need for technology to appropriately extract essential information.

[0156] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0157] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary using a generative AI model based on the extracted topics, and means for storing the generated summary and providing it to a terminal. This makes it possible to efficiently extract important information from a vast amount of meeting minutes and reports published on the internet and to quickly provide high-quality summaries.

[0158] "Publicly available document data" refers to information data, including text, that is provided on the internet in a format that is publicly accessible.

[0159] "Text data" refers to information expressed as a string of characters, in a format that humans can read and understand.

[0160] "Preprocessing" refers to a series of processes performed on raw data, including noise reduction and text normalization.

[0161] "Noise" refers to unnecessary elements or strings in text data, information that is not needed for analysis.

[0162] "Normalization" is the process of transforming text data into a consistent format, unifying strings based on specific rules.

[0163] Morphological analysis is the process of dividing text data into morphemes, which are the smallest semantic elements that make up the data, and adding information such as part of speech.

[0164] A "topic" refers to the main theme or subject matter deemed important within the text data.

[0165] A "generative AI model" is an artificial intelligence that learns from large amounts of data and generates natural language text based on given conditions.

[0166] A "summary" is a short, concise summary of the main points and important aspects of a long text or document.

[0167] A "terminal" is an electronic device used by a user to receive and view information.

[0168] This invention relates to a system for efficiently collecting, preprocessing, and analyzing publicly available document data, extracting important information, and generating summaries. Specifically, this system operates in the following steps:

[0169] The server first collects document data from specified URLs on the internet. This is done using a web crawler. For example, it downloads the HTML content of a webpage and extracts specific meeting minutes data. Programming libraries used include requests and BeautifulSoup.

[0170] Next, the server preprocesses the collected document data. This includes denoising (e.g., removing unnecessary HTML tags and special characters) and text normalization (e.g., converting full-width spaces to half-width spaces). Python string manipulation functions are used for this process.

[0171] Subsequently, the server analyzes the pre-processed text data and performs morphological analysis. Specifically, it uses the morphological analysis tool janome.tokenizer. This analysis identifies the words and their parts of speech that make up the text data.

[0172] Furthermore, the server extracts topics from the analyzed data. Here, topics are extracted using TF-IDF (Inverse Document Frequency) and LDA (Latent Dirichlet Allocation). In this process, the libraries TfidfVectorizer and LatentDirichletAllocation are used.

[0173] Based on the extracted topics, the server generates a summary using a generative AI model. The generative AI model utilizes the OpenAI® API (e.g., GPT-3). An example of a prompt for this model is shown below.

[0174] Please generate a meeting minutes summary based on the following topics:

[0175] Topics: Digitalization, Policy, Budget, AI, Data Analysis

[0176] Summary:

[0177] Based on this prompt, the AI ​​model generates a natural and consistent summary.

[0178] The generated summary is stored in a database and provided to the user's device. Users can retrieve and view the meeting minutes summary by using an application on their device and entering a specific URL. This entire process allows for the quick and efficient acquisition of important information without requiring significant time and effort.

[0179] The hardware required to implement this system consists of an internet-connected server and user terminals. The software uses the Python programming language and its libraries, BeautifulSoup, janome, TfidfVectorizer, LatentDirichletAllocation, and the OpenAI API.

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

[0181] Step 1:

[0182] The server collects document data from an internet URL specified by the user. In this case, when the user enters a specific URL on their terminal, that information is sent to the server. The server uses the requests library to access the specified URL and download the HTML content of the webpage. The input to this process is the URL specified by the user, and the output is the downloaded HTML content.

[0183] Step 2:

[0184] The server extracts text data from the HTML content collected in Step 1. It uses the BeautifulSoup library to parse HTML tags and extract the text portion. This removes HTML tags and unnecessary special characters. The input is the collected HTML content, and the output is the extracted text data.

[0185] Step 3:

[0186] The server performs preprocessing on the extracted text data. Preprocessing includes noise reduction (removing unnecessary characters) and normalization (converting full-width spaces to half-width spaces, unifying different expressions, etc.). Python string manipulation functions are used at this stage. The input is the extracted text data, and the output is the preprocessed text data.

[0187] Step 4:

[0188] The server performs morphological analysis on the preprocessed text data. During this process, it uses janome.tokenizer to split the text data into individual words and identify the part of speech for each word. The input is the preprocessed text data, and the output is the split data with its part of speech information.

[0189] Step 5:

[0190] The server extracts important topics based on the data analyzed in step 4. Here, it uses TfidfVectorizer to calculate the importance of each topic and then models the main topics from the entire text using LatentDirichletAllocation. The input is data segmented word by word and its part-of-speech information, and the output is the extracted important topics.

[0191] Step 6:

[0192] The server generates a summary using a generative AI model based on the extracted topics. Here, the OpenAI API (e.g., GPT-3) is used to form prompt statements and send them to the AI ​​model. A concrete example of a generated prompt statement is as follows:

[0193] Please generate a meeting minutes summary based on the following topics:

[0194] Topics: Digitalization, Policy, Budget, AI, Data Analysis

[0195] Summary:

[0196] The input consists of the extracted key topics, and the output is the generated summary.

[0197] Step 7:

[0198] The server stores the generated summary in a database and provides it to the user's terminal upon request. When a user requests a summary from their terminal, the server retrieves the corresponding summary from the database and sends it to the user's terminal. The input is the user's request and the generated summary, and the output is the summary sent to the user's terminal.

[0199] Through the steps described above, a system is realized that allows users to efficiently and quickly obtain important summary information from document information on the internet.

[0200] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0201] This invention is a system that collects, preprocesses, and analyzes publicly available document data to extract important information and generate summaries, and also incorporates an emotion engine that recognizes the user's emotions. This configuration allows for dynamic adjustment of the optimal summary delivery method according to the user's emotional state.

[0202] Conceptual explanation of the program

[0203] overview

[0204] This system collects publicly available document data by web crawling, preprocesses it, performs morphological analysis and topic extraction, and then automatically generates summaries. In addition to this basic summary generation process, it uses a sentiment engine to analyze user sentiment and determine the optimal way to provide summaries. The specific actions of the server, terminal, and user at each step are described below.

[0205] Data collection and extraction

[0206] The server collects meeting minutes data from specified internet URLs using web crawling technology. For example, it downloads HTML content from the public web pages of specific government agencies. This content is temporarily stored on the server.

[0207] Next, the server extracts the meeting minutes text from the saved HTML content. Specifically, it uses an HTML parser to remove HTML tags and unnecessary special characters, and then extracts the necessary text data.

[0208] Data preprocessing and analysis

[0209] As a preprocessing step, the server removes unnecessary noise (special characters, multiple line breaks and spaces, etc.) from the extracted text data and performs normalization by converting full-width spaces to half-width spaces.

[0210] Next, morphological analysis is performed on the preprocessed text data. A morphological analysis tool (e.g., MeCab) is used to divide the text data into words, and each word is tagged with its part of speech. Based on this analyzed data, topic modeling techniques (e.g., LDA or TF-IDF) are used to extract important topics within the text. The extracted topics are listed and used in the subsequent summary generation process.

[0211] Summary generation and saving

[0212] The server uses natural language generation technology to create a summary based on the extracted topics. After generating a consistent sentence using a template (e.g., "[Topic] was discussed, and [Result] was decided."), it performs a grammatical check and saves the final summary to the database. At this time, metadata related to the summary (e.g., publication date, agenda title, attendees) is also recorded.

[0213] Emotion Engine Additions and Features

[0214] When a user requests a summary through their device, the emotion engine analyzes the user's input data (e.g., text, voice, facial expression data) to recognize their emotions. Based on the analyzed emotion data, the emotion engine determines the optimal way to provide the summary (e.g., text length, expression style, level of detail). This emotion-based adjustment ensures that the summary is provided in a way that is most easily understood and accepted by the user.

[0215] As a concrete example, consider a case where a user requests to "check the summary of the latest research conference minutes." The process in this case would be as follows:

[0216] 1. The user clicks the button to request a summary on their device.

[0217] 2. The terminal sends the user's request to the server.

[0218] 3. The server retrieves the latest summary from the database and has the emotion engine analyze the user's emotion data.

[0219] 4. The emotion engine adjusts how the summary is provided based on the user's emotion information (input data at the time of the request).

[0220] 5. The server sends the adjusted summary to the user's terminal.

[0221] 6. The terminal displays the summary received from the server to the user.

[0222] This system provides an optimal summary tailored to the user's emotional state, improving user understanding and satisfaction.

[0223] The following describes the processing flow.

[0224] Step 1: Data Collection

[0225] The server accesses a specified internet URL (e.g., a public webpage of a government agency) and downloads the HTML content of the webpage using web crawling technology.

[0226] The server saves the downloaded HTML content to temporary storage.

[0227] Step 2: Data Extraction

[0228] The server runs an HTML parser to extract the meeting minutes text from the saved HTML content.

[0229] The server removes HTML tags and extracts the necessary text data.

[0230] Step 3: Data Preprocessing

[0231] As a preprocessing step, the server performs noise reduction by removing unnecessary special characters and multiple line breaks / spaces from the text data.

[0232] The server performs a normalization process that converts full-width spaces to half-width spaces.

[0233] Step 4: Morphological Analysis

[0234] The server uses a morphological analysis tool (e.g., MeCab) to divide the preprocessed text data into words and tag each word with its part of speech.

[0235] The server lists the morphological analysis results.

[0236] Step 5: Topic Extraction

[0237] The server uses topic modeling techniques (e.g., LDA and TF-IDF) to extract important topics from morphologically analyzed data.

[0238] The server evaluates the frequency and importance of related words for each topic and lists the main topics.

[0239] Step 6: Summary Generation

[0240] The server uses natural language generation technology to generate summary sentences based on the extracted topics.

[0241] The server constructs consistent sentences based on a template (e.g., "[Topic] was discussed, and [Result] was determined.").

[0242] The server performs a grammatical check on the generated summary sentence and makes corrections as needed.

[0243] Step 7: Save the summary

[0244] The server saves the generated summary to the database.

[0245] The server also records metadata related to the summary (e.g., publication date, agenda title, attendees).

[0246] Step 8: Emotion Recognition

[0247] The user sends a summary request through their device.

[0248] The terminal collects user input data (e.g., text, voice, facial expression data) and sends it to the server.

[0249] The server uses an emotion engine to analyze the transmitted user data and recognize the user's emotions.

[0250] Step 9: Emotion-based summary adjustment

[0251] The server adjusts how the summary is provided (e.g., text length, format, level of detail) based on the analysis results of the emotion engine.

[0252] The server prepares the adjusted summary in the optimal format.

[0253] Step 10: Provide a summary

[0254] The server sends the adjusted summary to the user's terminal.

[0255] The terminal displays the received summary to the user.

[0256] Users view and utilize summaries optimized based on their emotions.

[0257] Through each of the above steps, the system efficiently generates and adjusts document data summaries and provides them in an appropriate format that suits the user's emotions.

[0258] (Example 2)

[0259] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0260] Conventional text summary generation systems have the problem of not being able to consider user sentiment when collecting publicly available document data and generating summaries, making it difficult to provide summaries that are optimal for the user. In addition, the method of providing summaries is fixed, which is a problem as it does not sufficiently improve user understanding and satisfaction.

[0261] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0262] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary based on the extracted topics, means for saving the generated summary and providing it to the user, and means for analyzing the user's sentiment information and adjusting the method of providing the summary based on the analyzed sentiment information. This makes it possible to provide an optimal summary that responds to the user's sentiment, thereby improving the user's understanding and satisfaction.

[0263] "Publicly available document data" refers to document information that is provided in a format accessible to anyone on the internet or in other public places.

[0264] "Means of collection" refers to technologies and methods that have the function of automatically collecting publicly available document data from the internet using specific algorithms or tools.

[0265] "Text data" refers to string information extracted from collected document data, and is the basic unit that constitutes the content of a document.

[0266] "Preprocessing" refers to the process of removing unnecessary information (special characters, spaces, line breaks, etc.) from text data and preparing the data for analysis.

[0267] "Normalization" refers to the operation of standardizing irregular elements in text data (such as converting full-width spaces to half-width spaces) during preprocessing.

[0268] A "morpheme" is the smallest semantic unit (word or phrase) that makes up text data, and is an element extracted through morphological analysis.

[0269] "Morphological analysis" refers to the process of dividing text data into morphemes, tagging each morpheme with its part of speech, and then analyzing it.

[0270] A "topic" refers to an important subject or subject matter within document data, extracted through morphological analysis or topic modeling.

[0271] "Means of extracting topics" refers to technologies and methods that use text analysis techniques to identify important subjects within a document and list them.

[0272] A "summary" refers to a short document that concisely summarizes the main points of the entire document based on the extracted topics.

[0273] "Means for generating summaries" refers to technologies and methods that have the function of creating a consistent summary sentence using natural language generation technology based on extracted topics.

[0274] "Means of providing to the user" refers to technologies and methods that have the functionality to send the generated summary to the user's terminal and display it.

[0275] "Emotional information" refers to emotional states and reactions analyzed from user input data (text, voice, facial expressions, etc.).

[0276] "Means of analyzing emotional information" refers to technologies and methods that analyze user input data and identify the emotions contained within it.

[0277] "Means of adjusting the delivery method" refers to technologies and methods that have the function of optimizing the content, format, and expression of the summary based on the analyzed emotional information.

[0278] The present invention is a system that collects, preprocesses, and analyzes publicly available document data to extract important information and generate summaries, and also incorporates an emotion engine that recognizes user emotions. Specific embodiments of the present invention are described below.

[0279] Data collection and extraction

[0280] The server collects the published document data from the specified Internet URLs using web crawling technology. For example, it downloads the HTML content from the web page of a specific public institution and temporarily saves it. The server then uses an HTML parser (such as Beautiful Soup) to remove unnecessary tags and special characters from the downloaded HTML content and extract the required text data.

[0281] Data preprocessing

[0282] The server performs normalization processing on the extracted text data. Specifically, it performs operations such as removing special characters and unnecessary whitespace, and converting full-width spaces to half-width spaces. This arranges the text data in a form suitable for analysis.

[0283] Text analysis and topic extraction

[0284] The server uses a morphological analysis tool (such as MeCab) to split the normalized text data into words and tag each word with its part of speech. Then, the server uses topic modeling techniques such as LDA (Latent Dirichlet Allocation) and TF-IDF (Term Frequency-Inverse Document Frequency) to extract important topics within the text.

[0285] Summary generation and saving

[0286] The server utilizes natural language generation technology based on the extracted topics to generate a coherent summary. It creates summary text using a template (such as "[Topic] was discussed and [Result] was determined.") After performing a grammar check on the generated summary, it is saved in a database along with relevant metadata (publication date, topic title, attendees, etc.).

[0287] Sentiment engine and summary provision

[0288] When a user requests a summary using their device, the device sends the request to the server. The server retrieves the latest summary from the database and has the emotion engine analyze the user's input data (text, voice, facial expression data). Based on the analysis results, the emotion engine adjusts how the summary is provided (e.g., text length, expression style, level of detail). The server sends the adjusted summary to the user's device, which then displays it to the user.

[0289] Specific example

[0290] For example, if a user requests to "check the summary of the latest research conference minutes," the processing flow is as follows:

[0291] 1. The user clicks the "Request Summary" button on their device.

[0292] 2. The device sends a request to the server.

[0293] 3. The server retrieves the latest summary from the database and has the emotion engine analyze the user's emotion data.

[0294] 4. The emotion engine adjusts how the summary is provided based on the user's emotional information.

[0295] 5. The server sends the adjusted summary to the user's terminal.

[0296] 6. The device displays Sally, and the user confirms it.

[0297] Example of a prompt

[0298] The following are examples of prompts to input into a generative AI model:

[0299] I would like to check the summary of the latest research meeting minutes. Please analyze the crawled meeting minutes data and generate a summary based on important topics. Also, please adjust the way of providing the summary based on the user's sentiment data (text, voice, facial expression data).

[0300] As described above, the present invention can improve the user's understanding and satisfaction by providing an optimal summary considering the user's sentiment.

[0301] The flow of the specific process in Example 2 will be described with reference to FIG. 13.

[0302] Step 1:

[0303] Data collection

[0304] The server uses web crawling technology to collect document data published from the specified URL. As input, the URL of a specific website is provided to the crawler, and as output, the HTML content is temporarily saved on the server.

[0305] As a specific operation, the crawler accesses the URL, downloads the HTML data in real time, and stores it in temporary storage.

[0306] Step 2:

[0307] HTML parsing and text extraction

[0308] The server extracts the necessary text data from the saved HTML content. As input, the saved HTML data is given, and as output, text data is generated.

[0309] As a specific operation, the server uses an HTML parser (e.g., Beautiful Soup) to remove HTML tags and unnecessary special characters to extract text. The extracted text is stored in a new variable.

[0310] Step 3:

[0311] Normalization process

[0312] The server performs normalization on the extracted text data. The extracted text data is given as input, and normalized text data is generated as output.

[0313] Specifically, the server uses regular expressions to remove special characters and unnecessary whitespace, and converts full-width spaces to half-width spaces. This prepares the text data for parsing.

[0314] Step 4:

[0315] Morphological analysis

[0316] The server performs morphological analysis on normalized text data. Normalized text data is provided as input, and the morphological analysis results are generated as output.

[0317] Specifically, the server uses a morphological analysis tool (for example, MeCab) to split the text data into words and tag each word with its part of speech.

[0318] Step 5:

[0319] Topic extraction

[0320] The server extracts topics based on the morphological analysis results. The input is the morphological analysis results, and the output is a list of the extracted topics.

[0321] In terms of specific operations, the server executes topic modeling algorithms such as LDA (Latent Dirichlet Allocation) and TF-IDF (Term Frequency-Inverse Document Frequency) to extract important topics.

[0322] Step 6:

[0323] Summary Generation

[0324] The server generates a summary based on the extracted topics. The extracted topics are given as input, and the generated summary is provided as output.

[0325] In practice, the server uses a natural language generation algorithm and a template (for example, "[Topic] was discussed, and [Result] was determined.") to create a consistent summary text.

[0326] Step 7:

[0327] Save summary

[0328] The server saves the generated summary. The generated summary and associated metadata (e.g., publication date, agenda title, attendees) are given as input, and the output is saved to the database.

[0329] Specifically, the server records summary text and metadata in the database.

[0330] Step 8:

[0331] Summary Request Acceptance

[0332] The user requests a summary through their device. The user's request is sent to the device as input, and the device then sends a request to the server.

[0333] Specifically, the user clicks the "Request Summary" button on their device. The device then sends this request to the server.

[0334] Step 9:

[0335] Emotion analysis

[0336] The server analyzes the received request and user sentiment data. User sentiment data (text, voice, facial expressions) is given as input, and sentiment analysis results are generated as output.

[0337] Specifically, the server uses an emotion engine to analyze emotional information from the user's input data and retrieves the results.

[0338] Step 10:

[0339] Adjustment of summary delivery method

[0340] The server adjusts the way the summary is provided based on the sentiment analysis results. The sentiment analysis results are given as input, and the adjusted summary is generated as output.

[0341] Specifically, the server adjusts the format and length of the summary text based on the sentiment analysis results.

[0342] Step 11:

[0343] Send Summary

[0344] The server sends the adjusted summary to the user terminal. The adjusted summary is sent from the server as input and received by the user terminal as output.

[0345] Specifically, the server uses a protocol to send the adjusted summary to the user's terminal.

[0346] Step 12:

[0347] Summary display

[0348] The terminal displays a summary to the user. The adjusted summary is received by the terminal as input and displayed to the user as output.

[0349] Specifically, the terminal displays the received summary on the user interface, allowing the user to review it.

[0350] (Application Example 2)

[0351] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0352] The vast amount of electronic payment data available today can be difficult for users to understand, and information provided without considering emotions can negatively impact the user experience. For example, simply providing a payment history is insufficient for a user who feels anxious about high spending after shopping. While there is a need for appropriate information tailored to the user's emotions, no system currently exists to achieve this. Therefore, a system is needed that recognizes the user's emotions and provides an optimal summary of their payment history based on those emotions.

[0353] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0354] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary based on the extracted topics, means for storing the generated summary and providing it to the user, means for recognizing the user's emotions, and means for adjusting the method of providing the summary based on the recognized emotions. This makes it possible to provide an optimal summary that responds to the user's emotions, thereby improving the user experience.

[0355] "Publicly available document data" refers to document data that is provided in a form accessible to anyone on the internet or other public environments.

[0356] "Means of collection" refers to methods and devices for obtaining document data from specified URLs or sources using technologies such as web crawling and HTTP requests.

[0357] "Means of extraction" refers to methods or devices used to extract necessary text information from collected data, such as HTML parsers or text analysis tools.

[0358] "Preprocessing" refers to the process of removing noise from data, normalizing it, and converting its format, essentially organizing and correcting the data before analysis.

[0359] "Noise" refers to unnecessary data or information that should be excluded from the analysis, such as special characters, multiple line breaks, and extra whitespace.

[0360] "Normalization" refers to the process of unifying the format and notation of data, such as converting full-width spaces to half-width spaces to conform to a standard format.

[0361] A "morpheme" is a fundamental unit that makes up text, such as a word or phrase—the smallest element that possesses meaning.

[0362] "Morphological analysis" refers to the process of dividing text data into words and phrases and tagging each with its part of speech.

[0363] "Important topics" refer to themes and topics that frequently appear in the text data, and are particularly useful or interesting to the user.

[0364] A "summary" is a concise compilation of key points extracted from lengthy texts or complex data, presented in a way that users can quickly understand.

[0365] "Means for generating summaries" refers to methods or devices that use topic modeling or natural language generation techniques to create consistent, short sentences from extracted information.

[0366] "Means of providing to the user" refers to methods and devices for presenting the generated summary in a form accessible to the user, such as smartphone apps and web interfaces.

[0367] "Means for recognizing user emotions" refers to methods and devices for detecting a user's emotional state using text analysis, speech recognition, facial expression analysis, etc.

[0368] "Means for adjusting the method of providing summaries based on recognized emotions" refers to methods or devices that receive user emotion information as analysis results and dynamically change the length, level of detail, and format of the summary provided.

[0369] The system for realizing this invention includes the following means: A server collects publicly available document data and uses an HTML parser and HTTP request technology to extract text data from the collected document data. Then, it preprocesses the extracted text data, removing noise and using MeCab as a tool to normalize the text. Topic modeling techniques such as LDA and TF-IDF are used to analyze morphemes from the preprocessed text data and extract important topics. When generating a summary based on the extracted topics, natural language generation technology is used, and the generated summary is stored in a database and includes an interface for providing it to the user.

[0370] Furthermore, to recognize user emotions, the system incorporates an emotion engine that uses text analysis, speech recognition, and facial expression analysis technologies. This emotion engine analyzes user input data (e.g., text, speech, and facial expression data) to recognize emotions and adjusts how the summary is provided based on the recognized emotions. Specifically, if a user sends a prompt such as, "I want a summary of my recent spending. I'm worried about the content, so I'd also like some saving advice," the emotion engine analyzes the user's emotions (anxiety) and appropriately adjusts the length and level of detail of the summary.

[0371] The hardware used will include a server, the user's smartphone or PC, and, if necessary, smart glasses or a head-mounted display. The specific software used will be Python programs, the requests library, BeautifulSoup, MeCab, the gensim library, and TextBlob. This will improve the user experience by providing users with summaries that appropriately reflect their emotional state when they want to learn about their spending and transaction history.

[0372] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0373] Step 1:

[0374] The user requests a summary of their electronic payment history using a device (smartphone or PC).

[0375] Input: The user enters the prompt message: "I would like a summary of my recent spending. I am also concerned about the content, so I would like some advice on saving money."

[0376] Output: Sends a user request to the server.

[0377] Step 2:

[0378] The terminal forwards the received request to the server.

[0379] Input: User prompt text.

[0380] Output: Sends a prompt message to the server.

[0381] Step 3:

[0382] The server collects electronic payment data from a specified internet URL.

[0383] Input: The specified URL.

[0384] Output: Collected electronic payment data.

[0385] The server uses the requests library to download HTML content from the specified URL and BeautifulSoup to extract text data from the HTML tags.

[0386] Step 4:

[0387] The server performs preprocessing on the extracted text data.

[0388] Input: Extracted text data.

[0389] Output: Noise-removed and normalized text data.

[0390] Specifically, it removes special characters, multiple line breaks, and unnecessary spaces, and converts full-width spaces to half-width spaces.

[0391] Step 5:

[0392] The server performs morphological analysis on the pre-processed text data and extracts important topics.

[0393] Input: Preprocessed text data.

[0394] Output: Key topics extracted.

[0395] We use MeCab to split the text into words and then perform topic modeling using LDA or TF-IDF.

[0396] Step 6:

[0397] The server generates a summary based on the extracted topics.

[0398] Input: Key topics extracted.

[0399] Output: The generated summary.

[0400] Using natural language generation technology, a coherent text containing extracted topics is created.

[0401] Step 7:

[0402] The server saves the generated summary to the database and prepares it for delivery.

[0403] Input: Generated summary.

[0404] Output: Summary saved in the database.

[0405] Grammar checks are also performed, and the final summary is formatted and saved.

[0406] Step 8:

[0407] The emotion engine analyzes the user's emotions.

[0408] Input: User prompt (e.g., "I'd like a summary of my recent expenses. I'm also concerned about the details, so I'd like some advice on saving money.").

[0409] Output: User's emotional state (e.g., anxiety).

[0410] Use TextBlob to analyze the user's input text and obtain a sentiment score.

[0411] Step 9:

[0412] The server generates a refined summary using an emotion engine.

[0413] Input: User's emotional state, generated summary.

[0414] Output: Adjusted summary.

[0415] The length, level of detail, and presentation of the summary are adjusted according to the user's emotional state.

[0416] Step 10:

[0417] The server sends the adjusted summary to the terminal.

[0418] Input: Adjusted summary.

[0419] Output: Sends a summary to the user's terminal.

[0420] The terminal displays a summary received from the server to the user.

[0421] Through the above processing steps, a user-friendly and emotionally resonant summary of the electronic payment history is provided.

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

[0423] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

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

[0425] [Second Embodiment]

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

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

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

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

[0430] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

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

[0432] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0433] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

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

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

[0436] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

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

[0438] This invention is a system that collects publicly available document data, particularly meeting minutes and report data from the internet, preprocesses and analyzes it, and extracts and summarizes important points from this data to provide a summary. The following describes specific embodiments for implementing this invention.

[0439] Conceptual explanation of the program

[0440] overview

[0441] This system begins by acquiring publicly available document data, preprocessing and analyzing it to extract important information, and finally generating a summary to provide to the user. The following describes the specific actions of the server, terminal, and user at each step.

[0442] Data collection

[0443] The server collects meeting minutes data from a specified internet URL using web crawling technology. For example, if the target is the minutes of a research group published by the Ministry of Internal Affairs and Communications, the server accesses the relevant web page and downloads the HTML content.

[0444] Data extraction and preprocessing

[0445] The server extracts the meeting minutes text from the downloaded HTML content. During this process, it performs noise reduction to remove HTML tags, unnecessary special characters, line breaks, etc. Next, it performs normalization, such as converting full-width spaces to half-width spaces and unifying different expressions.

[0446] Text analysis and topic extraction

[0447] The server performs morphological analysis on the preprocessed text data. A morphological analysis tool (e.g., MeCab) is used here. As a result of the analysis, the text data is divided into words, and the part of speech of each word is identified. Based on the data obtained from this analysis, important topics are extracted using topic modeling (e.g., LDA or TF-IDF).

[0448] Summary Generation

[0449] The server uses natural language generation technology to create summaries based on the extracted topics. For example, it uses a template to concisely summarize the main points in the format "XX was discussed, and YY was decided." Grammatical checks are also performed during this process to generate consistent sentences.

[0450] Saving and providing summaries

[0451] The created summary is stored in the server's database. When a user requests a summary through their device, the server searches the stored summary and sends the relevant data to the user's device. The user can then view the summary on their device.

[0452] Specific example

[0453] For example, when the Ministry of Internal Affairs and Communications provides a summary of the latest research meeting minutes, the following process occurs when a user clicks the "View summary of the latest meeting minutes" button on their device.

[0454] 1. The user clicks a button on their device.

[0455] 2. The terminal sends a summary request to the server.

[0456] 3. The server generates a summary from the latest collected meeting minutes data and saves it to the database.

[0457] 4. The server searches the saved summary and sends it to the user's terminal.

[0458] 5. The terminal displays the summary received from the server on its screen, which the user can then view.

[0459] Through this series of processes, users can efficiently obtain important information, and the system of the present invention facilitates the understanding and sharing of meeting minutes.

[0460] The following describes the processing flow.

[0461] Step 1: Data Collection

[0462] The server accesses the specified URL (for example, a public webpage of a specific government agency).

[0463] The server uses web crawling technology to download the HTML content of web pages.

[0464] The server saves the downloaded data to temporary storage.

[0465] Step 2: Data Extraction

[0466] The server runs an HTML parser to extract the meeting minutes text from the saved HTML content.

[0467] The server extracts the text data while removing HTML tags.

[0468] Step 3: Data Preprocessing

[0469] As a preprocessing step, the server removes unnecessary special characters and multiple line breaks / spaces from the text data.

[0470] The server performs normalization, such as converting full-width spaces to half-width spaces within the text.

[0471] Step 4: Morphological Analysis

[0472] The server uses a morphological analysis tool (e.g., MeCab) to divide the preprocessed text data into words.

[0473] The server tags each word with its part of speech and records the results of morphological analysis.

[0474] Step 5: Topic Extraction

[0475] The server uses topic modeling techniques (e.g., LDA and TF-IDF) to extract important topics from morphologically analyzed data.

[0476] The server evaluates the frequency and importance of words related to each topic and lists the main topics.

[0477] Step 6: Summary Generation

[0478] The server generates a summary sentence based on the extracted topics, utilizing natural language generation technology.

[0479] The server constructs the text based on a specific template (e.g., "[Topic] was discussed, and [Result] was determined.").

[0480] The server performs a grammatical check on the generated summary sentence and makes corrections as needed.

[0481] Step 7: Save the summary

[0482] The server saves the generated summary to the database.

[0483] The server also records metadata related to the summary (e.g., publication date, meeting minutes title, attendees).

[0484] Step 8: Provide a summary

[0485] The user requests a summary through their device.

[0486] The server searches the database for the relevant summary and sends it to the user's terminal.

[0487] The terminal displays the received summary to the user.

[0488] As described above, a system is realized in which the server, terminal, and user cooperate at each step to efficiently generate and provide document summaries.

[0489] (Example 1)

[0490] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0491] The sheer volume of publicly available document data, particularly meeting minutes and reports found online, makes it difficult to efficiently extract key points and provide summaries. Furthermore, manually collecting, pre-processing, analyzing, and summarizing this data is time-consuming and labor-intensive. Therefore, there is a need for a system that automatically and efficiently collects and analyzes document data, extracts key points, and provides summaries.

[0492] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0493] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary using a generative AI model based on the extracted topics, means for verifying the generated summary using a grammar checking tool, and means for saving the generated summary and providing it to the user. This enables the user to efficiently obtain important information from a vast amount of document data.

[0494] "Publicly available document data" refers to document data that is generally accessible on the internet or in other public spheres.

[0495] "Web crawling" is a technology that automatically collects data from specified web pages and websites.

[0496] "HTML content" refers to data in HTML (Hypertext Markup Language) format that represents the structure and content of a web page.

[0497] "Preprocessing" refers to a series of processes that remove noise from data and normalize it before performing data analysis.

[0498] "Noise reduction" is the process of removing elements from text data that are unnecessary for analysis (e.g., HTML tags, special characters, unnecessary line breaks).

[0499] "Normalization" is a process that unifies different representations within text data to ensure data consistency.

[0500] Morphological analysis is a technique that divides text data into the smallest possible linguistic elements and analyzes the part of speech of each element.

[0501] A "morphological analysis tool" refers to software or algorithms used to perform morphological analysis (e.g., MeCab).

[0502] Topic modeling is a technique that automatically extracts important topics and themes from large amounts of text data (e.g., LDA, TF-IDF).

[0503] A "generative AI model" is an artificial intelligence model that uses machine learning to generate natural language (e.g., GPT-3).

[0504] A "summary" is a concise summary that captures the key points of an entire document.

[0505] A "grammar check tool" is software or algorithms used to verify and correct the grammatical accuracy of generated text.

[0506] A "database" is a system that allows for the efficient storage, management, and retrieval of structured data.

[0507] A "terminal" refers to a device (e.g., a personal computer, smartphone, or tablet) that a user uses to access a server and view or manipulate data.

[0508] This invention is a system that collects publicly available document data, particularly meeting minutes and report data from the internet, preprocesses and analyzes it, and extracts and summarizes important points from this data. A detailed embodiment of this invention is described below.

[0509] Data collection

[0510] The server uses web crawling technology to collect meeting minutes data from specified internet URLs. Specifically, the server accesses URLs in a specified list sequentially, retrieves HTML content, and saves it to local storage. For example, it might access the URL of a publicly available research conference minutes page of a certain public institution and download the HTML file.

[0511] Data extraction and preprocessing

[0512] The server extracts the necessary meeting minutes text from the collected HTML content and performs preprocessing. It uses an HTML parser (e.g., Beautiful Soup) to remove HTML tags, unnecessary special characters, and line breaks. Furthermore, it performs normalization, such as converting full-width spaces to half-width spaces and unifying different expressions, as part of noise reduction. For example, the meeting minutes text " Chairperson: Thank you for joining us today. Extract "Chairman: Thank you for participating today." from the above.

[0513] Text analysis and topic extraction

[0514] The server performs morphological analysis on the preprocessed text data. For this analysis, a morphological analysis tool (e.g., MeCab) is used to divide the text into the smallest units of linguistic elements and identify the part of speech for each. For example, the sentence "Thank you for participating today." is divided into "today / noun," "wa / particle," "participation / noun," "arigato / verb," ​​"u / auxiliary verb," ​​and "gozaimasu / verb." Subsequently, topic modeling techniques (e.g., LDA or TF-IDF) are used to extract important topics from the text data. Frequently occurring keywords such as "research," "minutes," and "technology" are extracted.

[0515] Summary Generation

[0516] The server generates summaries using a generative AI model (e.g., GPT-3) based on important topics. For example, it might generate a summary such as, "The minutes mainly discussed advancements in new technologies, and future research directions were decided." The generated summaries are then checked for grammatical accuracy using a grammar checking tool and corrected as needed.

[0517] Saving and providing summaries

[0518] The generated summaries are stored in the server's database. The storage format includes fields such as "timestamp," "meeting minutes title," and "summary." When a user requests a summary through their terminal, the terminal sends a request to the server, the server searches the database for the corresponding summary, and sends it to the terminal. The terminal displays the summary received from the server on its screen, allowing the user to view it.

[0519] Specific example

[0520] For example, if a public institution is providing a summary of its latest meeting minutes, when a user clicks the "View Summary of Latest Meeting Minutes" button on their device, the following process takes place.

[0521] 1. The user clicks a button on their device.

[0522] 2. The terminal sends a summary request to the server.

[0523] 3. The server generates a summary from the latest meeting minutes data it has collected and saves it to the database.

[0524] 4. The server searches for the saved summary and sends it to the user's terminal.

[0525] 5. The terminal displays the summary received from the server on the screen, and the user views it.

[0526] As an example of a prompt, the process is carried out by inputting "Please provide a summary of the latest research conference minutes from a public institution" into the generating AI model.

[0527] This system allows users to efficiently obtain important information from vast amounts of document data.

[0528] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0529] Step 1:

[0530] The server collects publicly available document data. Specifically, the server receives a specified list of URLs as input and accesses each URL using web crawling technology. It downloads the HTML content of each URL and saves it to local storage. The output is the downloaded HTML file.

[0531] Step 2:

[0532] The server extracts text data from the collected HTML content. Specifically, it receives an HTML file as input and uses an HTML parser (e.g., Beautiful Soup) to extract the meeting minutes text. As a noise reduction process, HTML tags, unnecessary special characters, and line breaks are removed. The output is the meeting minutes text. For example, HTML content " Chairperson: Thank you for joining us today. Extract "Chairman: Thank you for participating today." from the above.

[0533] Step 3:

[0534] The server performs preprocessing on the extracted text data. Specifically, it receives the text data as input, converts full-width spaces to half-width spaces, and performs normalization to unify different expressions (e.g., AI, エーアイ). The output is normalized text data.

[0535] Step 4:

[0536] The server performs morphological analysis on the pre-processed text data. Specifically, it uses a morphological analysis tool (e.g., MeCab) to receive the text data as input, split it into words, and identify the part of speech of each word. The output is a list of the split words. For example, the sentence "Thank you for participating today." is split into "today / noun", "is / particle", "participation / noun", "thank you / verb", "u / auxiliary verb", and "gozaimasu / verb".

[0537] Step 5:

[0538] The server extracts important topics using topic modeling techniques. Specifically, it takes a list of words obtained as a result of morphological analysis as input and uses topic modeling techniques (e.g., LDA or TF-IDF) to extract important topics. The output is a list of the extracted topics.

[0539] Step 6:

[0540] The server generates a summary based on the extracted topics. Specifically, it takes a list of important topics as input and generates a summary using a generative AI model (e.g., GPT-3). The output is the generated summary. For example, it might generate a summary such as, "The meeting minutes mainly discussed the progress of new technologies, and future research directions were decided."

[0541] Step 7:

[0542] The server checks the generated summary using a grammar checking tool. Specifically, it receives the generated summary as input, uses the grammar checking tool to verify its grammatical accuracy, and makes corrections as needed. The output is the corrected summary.

[0543] Step 8:

[0544] The server saves the generated summary to the database. Specifically, it receives the modified summary as input and saves it to the database. The output is the summary saved in the database. The saving format is in the form of fields such as "timestamp," "meeting minutes title," and "summary."

[0545] Step 9:

[0546] The user requests the summary via their device. Specifically, the user clicks the "View summary of the latest meeting minutes" button on their device. This action causes the device to send a summary request to the server.

[0547] Step 10:

[0548] The server retrieves the relevant summary from the database and sends it to the user's terminal. Specifically, the server receives a summary request, retrieves the relevant summary from the database, and sends it to the terminal. The output is the sent summary.

[0549] Step 11:

[0550] The terminal displays the summary received from the server on its screen, which the user then views. Specifically, the terminal receives the summary as input and displays it on the screen. The output is a summary display that the user can view.

[0551] Therefore, users can efficiently obtain important information.

[0552] (Application Example 1)

[0553] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0554] While a vast amount of meeting minutes and reports are publicly available online, there is a lack of efficient means to collect this information, extract key points, and provide concise summaries. Traditional systems require significant time and effort for manual information gathering and summarization, and the sheer volume of information makes it highly likely that users will miss useful details. Furthermore, maintaining consistent summaries is difficult, highlighting the need for technology to appropriately extract essential information.

[0555] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0556] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary using a generative AI model based on the extracted topics, and means for storing the generated summary and providing it to a terminal. This makes it possible to efficiently extract important information from a vast amount of meeting minutes and reports published on the internet and to quickly provide high-quality summaries.

[0557] "Publicly available document data" refers to information data, including text, that is provided on the internet in a format that is publicly accessible.

[0558] "Text data" refers to information expressed as a string of characters, in a format that humans can read and understand.

[0559] "Preprocessing" refers to a series of processes performed on raw data, including noise reduction and text normalization.

[0560] "Noise" refers to unnecessary elements or strings in text data, information that is not needed for analysis.

[0561] "Normalization" is the process of transforming text data into a consistent format, unifying strings based on specific rules.

[0562] Morphological analysis is the process of dividing text data into morphemes, which are the smallest semantic elements that make up the data, and adding information such as part of speech.

[0563] A "topic" refers to the main theme or subject matter deemed important within the text data.

[0564] A "generative AI model" is an artificial intelligence that learns from large amounts of data and generates natural language text based on given conditions.

[0565] A "summary" is a short, concise summary of the main points and important aspects of a long text or document.

[0566] A "terminal" is an electronic device used by a user to receive and view information.

[0567] This invention relates to a system for efficiently collecting, preprocessing, and analyzing publicly available document data, extracting important information, and generating summaries. Specifically, this system operates in the following steps:

[0568] The server first collects document data from specified URLs on the internet. This is done using a web crawler. For example, it downloads the HTML content of a webpage and extracts specific meeting minutes data. Programming libraries used include requests and BeautifulSoup.

[0569] Next, the server preprocesses the collected document data. This includes denoising (e.g., removing unnecessary HTML tags and special characters) and text normalization (e.g., converting full-width spaces to half-width spaces). Python string manipulation functions are used for this process.

[0570] Subsequently, the server analyzes the pre-processed text data and performs morphological analysis. Specifically, it uses the morphological analysis tool janome.tokenizer. This analysis identifies the words and their parts of speech that make up the text data.

[0571] Furthermore, the server extracts topics from the analyzed data. Here, topics are extracted using TF-IDF (Inverse Document Frequency) and LDA (Latent Dirichlet Allocation). In this process, the libraries TfidfVectorizer and LatentDirichletAllocation are used.

[0572] Based on the extracted topics, the server generates a summary using a generative AI model. This generative AI model utilizes the OpenAI API (e.g., GPT-3). An example of a prompt for this model is shown below.

[0573] Please generate a meeting minutes summary based on the following topics:

[0574] Topics: Digitalization, Policy, Budget, AI, Data Analysis

[0575] Summary:

[0576] Based on this prompt, the AI ​​model generates a natural and consistent summary.

[0577] The generated summary is stored in a database and provided to the user's device. Users can retrieve and view the meeting minutes summary by using an application on their device and entering a specific URL. This entire process allows for the quick and efficient acquisition of important information without requiring significant time and effort.

[0578] The hardware required to implement this system consists of an internet-connected server and user terminals. The software uses the Python programming language and its libraries, BeautifulSoup, janome, TfidfVectorizer, LatentDirichletAllocation, and the OpenAI API.

[0579] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0580] Step 1:

[0581] The server collects document data from an internet URL specified by the user. In this case, when the user enters a specific URL on their terminal, that information is sent to the server. The server uses the requests library to access the specified URL and download the HTML content of the webpage. The input to this process is the URL specified by the user, and the output is the downloaded HTML content.

[0582] Step 2:

[0583] The server extracts text data from the HTML content collected in Step 1. It uses the BeautifulSoup library to parse HTML tags and extract the text portion. This removes HTML tags and unnecessary special characters. The input is the collected HTML content, and the output is the extracted text data.

[0584] Step 3:

[0585] The server performs preprocessing on the extracted text data. Preprocessing includes noise reduction (removing unnecessary characters) and normalization (converting full-width spaces to half-width spaces, unifying different expressions, etc.). Python string manipulation functions are used at this stage. The input is the extracted text data, and the output is the preprocessed text data.

[0586] Step 4:

[0587] The server performs morphological analysis on the preprocessed text data. During this process, it uses janome.tokenizer to split the text data into individual words and identify the part of speech for each word. The input is the preprocessed text data, and the output is the split data with its part of speech information.

[0588] Step 5:

[0589] The server extracts important topics based on the data analyzed in step 4. Here, it uses TfidfVectorizer to calculate the importance of each topic and then models the main topics from the entire text using LatentDirichletAllocation. The input is data segmented word by word and its part-of-speech information, and the output is the extracted important topics.

[0590] Step 6:

[0591] The server generates a summary using a generative AI model based on the extracted topics. Here, the OpenAI API (e.g., GPT-3) is used to form prompt statements and send them to the AI ​​model. A concrete example of a generated prompt statement is as follows:

[0592] Please generate a meeting minutes summary based on the following topics:

[0593] Topics: Digitalization, Policy, Budget, AI, Data Analysis

[0594] Summary:

[0595] The input consists of the extracted key topics, and the output is the generated summary.

[0596] Step 7:

[0597] The server stores the generated summary in a database and provides it to the user's terminal upon request. When a user requests a summary from their terminal, the server retrieves the corresponding summary from the database and sends it to the user's terminal. The input is the user's request and the generated summary, and the output is the summary sent to the user's terminal.

[0598] Through the steps described above, a system is realized that allows users to efficiently and quickly obtain important summary information from document information on the internet.

[0599] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0600] This invention is a system that collects, preprocesses, and analyzes publicly available document data to extract important information and generate summaries, and also incorporates an emotion engine that recognizes the user's emotions. This configuration allows for dynamic adjustment of the optimal summary delivery method according to the user's emotional state.

[0601] Conceptual explanation of the program

[0602] overview

[0603] This system collects publicly available document data by web crawling, preprocesses it, performs morphological analysis and topic extraction, and then automatically generates summaries. In addition to this basic summary generation process, it uses a sentiment engine to analyze user sentiment and determine the optimal way to provide summaries. The specific actions of the server, terminal, and user at each step are described below.

[0604] Data collection and extraction

[0605] The server collects meeting minutes data from specified internet URLs using web crawling technology. For example, it downloads HTML content from the public web pages of specific government agencies. This content is temporarily stored on the server.

[0606] Next, the server extracts the meeting minutes text from the saved HTML content. Specifically, it uses an HTML parser to remove HTML tags and unnecessary special characters, and then extracts the necessary text data.

[0607] Data preprocessing and analysis

[0608] As a preprocessing step, the server removes unnecessary noise (special characters, multiple line breaks and spaces, etc.) from the extracted text data and performs normalization by converting full-width spaces to half-width spaces.

[0609] Next, morphological analysis is performed on the preprocessed text data. A morphological analysis tool (e.g., MeCab) is used to divide the text data into words, and each word is tagged with its part of speech. Based on this analyzed data, topic modeling techniques (e.g., LDA or TF-IDF) are used to extract important topics within the text. The extracted topics are listed and used in the subsequent summary generation process.

[0610] Summary generation and saving

[0611] The server uses natural language generation technology to create a summary based on the extracted topics. After generating a consistent sentence using a template (e.g., "[Topic] was discussed, and [Result] was decided."), it performs a grammatical check and saves the final summary to the database. At this time, metadata related to the summary (e.g., publication date, agenda title, attendees) is also recorded.

[0612] Emotion Engine Additions and Features

[0613] When a user requests a summary through their device, the emotion engine analyzes the user's input data (e.g., text, voice, facial expression data) to recognize their emotions. Based on the analyzed emotion data, the emotion engine determines the optimal way to provide the summary (e.g., text length, expression style, level of detail). This emotion-based adjustment ensures that the summary is provided in a way that is most easily understood and accepted by the user.

[0614] As a concrete example, consider a case where a user requests to "check the summary of the latest research conference minutes." The process in this case would be as follows:

[0615] 1. The user clicks the button to request a summary on their device.

[0616] 2. The terminal sends the user's request to the server.

[0617] 3. The server retrieves the latest summary from the database and has the emotion engine analyze the user's emotion data.

[0618] 4. The emotion engine adjusts how the summary is provided based on the user's emotion information (input data at the time of the request).

[0619] 5. The server sends the adjusted summary to the user's terminal.

[0620] 6. The terminal displays the summary received from the server to the user.

[0621] This system provides an optimal summary tailored to the user's emotional state, improving user understanding and satisfaction.

[0622] The following describes the processing flow.

[0623] Step 1: Data Collection

[0624] The server accesses a specified internet URL (e.g., a public webpage of a government agency) and downloads the HTML content of the webpage using web crawling technology.

[0625] The server saves the downloaded HTML content to temporary storage.

[0626] Step 2: Data Extraction

[0627] The server runs an HTML parser to extract the meeting minutes text from the saved HTML content.

[0628] The server removes HTML tags and extracts the necessary text data.

[0629] Step 3: Data Preprocessing

[0630] As a preprocessing step, the server performs noise reduction by removing unnecessary special characters and multiple line breaks / spaces from the text data.

[0631] The server performs a normalization process that converts full-width spaces to half-width spaces.

[0632] Step 4: Morphological Analysis

[0633] The server uses a morphological analysis tool (e.g., MeCab) to divide the preprocessed text data into words and tag each word with its part of speech.

[0634] The server lists the morphological analysis results.

[0635] Step 5: Topic Extraction

[0636] The server uses topic modeling techniques (e.g., LDA and TF-IDF) to extract important topics from morphologically analyzed data.

[0637] The server evaluates the frequency and importance of related words for each topic and lists the main topics.

[0638] Step 6: Summary Generation

[0639] The server uses natural language generation technology to generate summary sentences based on the extracted topics.

[0640] The server constructs consistent sentences based on a template (e.g., "[Topic] was discussed, and [Result] was determined.").

[0641] The server performs a grammatical check on the generated summary sentence and makes corrections as needed.

[0642] Step 7: Save the summary

[0643] The server saves the generated summary to the database.

[0644] The server also records metadata related to the summary (e.g., publication date, agenda title, attendees).

[0645] Step 8: Emotion Recognition

[0646] The user sends a summary request through their device.

[0647] The terminal collects user input data (e.g., text, voice, facial expression data) and sends it to the server.

[0648] The server uses an emotion engine to analyze the transmitted user data and recognize the user's emotions.

[0649] Step 9: Emotion-based summary adjustment

[0650] The server adjusts how the summary is provided (e.g., text length, format, level of detail) based on the analysis results of the emotion engine.

[0651] The server prepares the adjusted summary in the optimal format.

[0652] Step 10: Provide a summary

[0653] The server sends the adjusted summary to the user's terminal.

[0654] The terminal displays the received summary to the user.

[0655] Users view and utilize summaries optimized based on their emotions.

[0656] Through each of the above steps, the system efficiently generates and adjusts document data summaries and provides them in an appropriate format that suits the user's emotions.

[0657] (Example 2)

[0658] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0659] Conventional text summary generation systems have the problem of not being able to consider user sentiment when collecting publicly available document data and generating summaries, making it difficult to provide summaries that are optimal for the user. In addition, the method of providing summaries is fixed, which is a problem as it does not sufficiently improve user understanding and satisfaction.

[0660] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0661] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary based on the extracted topics, means for saving the generated summary and providing it to the user, and means for analyzing the user's sentiment information and adjusting the method of providing the summary based on the analyzed sentiment information. This makes it possible to provide an optimal summary that responds to the user's sentiment, thereby improving the user's understanding and satisfaction.

[0662] "Publicly available document data" refers to document information that is provided in a format accessible to anyone on the internet or in other public places.

[0663] "Means of collection" refers to technologies and methods that have the function of automatically collecting publicly available document data from the internet using specific algorithms or tools.

[0664] "Text data" refers to string information extracted from collected document data, and is the basic unit that constitutes the content of a document.

[0665] "Preprocessing" refers to the process of removing unnecessary information (special characters, spaces, line breaks, etc.) from text data and preparing the data for analysis.

[0666] "Normalization" refers to the operation of standardizing irregular elements in text data (such as converting full-width spaces to half-width spaces) during preprocessing.

[0667] A "morpheme" is the smallest semantic unit (word or phrase) that makes up text data, and is an element extracted through morphological analysis.

[0668] "Morphological analysis" refers to the process of dividing text data into morphemes, tagging each morpheme with its part of speech, and then analyzing it.

[0669] A "topic" refers to an important subject or subject matter within document data, extracted through morphological analysis or topic modeling.

[0670] "Means of extracting topics" refers to technologies and methods that use text analysis techniques to identify important subjects within a document and list them.

[0671] A "summary" refers to a short document that concisely summarizes the main points of the entire document based on the extracted topics.

[0672] "Means for generating summaries" refers to technologies and methods that have the function of creating a consistent summary sentence using natural language generation technology based on extracted topics.

[0673] "Means of providing to the user" refers to technologies and methods that have the functionality to send the generated summary to the user's terminal and display it.

[0674] "Emotional information" refers to emotional states and reactions analyzed from user input data (text, voice, facial expressions, etc.).

[0675] "Means of analyzing emotional information" refers to technologies and methods that analyze user input data and identify the emotions contained within it.

[0676] "Means of adjusting the delivery method" refers to technologies and methods that have the function of optimizing the content, format, and expression of the summary based on the analyzed emotional information.

[0677] The present invention is a system that collects, preprocesses, and analyzes publicly available document data to extract important information and generate summaries, and also incorporates an emotion engine that recognizes user emotions. Specific embodiments of the present invention are described below.

[0678] Data collection and extraction

[0679] The server uses web crawling technology to collect publicly available document data from specified internet URLs. For example, it downloads HTML content from a specific public institution's webpage and stores it temporarily. The server then uses an HTML parser (such as Beautiful Soup) to remove unnecessary tags and special characters from the downloaded HTML content and extract the necessary text data.

[0680] Data preprocessing

[0681] The server performs normalization on the extracted text data. Specifically, it removes special characters and unnecessary whitespace, and converts full-width spaces to half-width spaces. This prepares the text data for analysis.

[0682] Text analysis and topic extraction

[0683] The server uses a morphological analysis tool (e.g., MeCab) to segment the normalized text data into words and tags each word with its part of speech. The server then uses topic modeling techniques such as LDA (Latent Dirichlet Allocation) and TF-IDF (Term Frequency-Inverse Document Frequency) to extract important topics from the text.

[0684] Summary generation and saving

[0685] The server utilizes natural language generation technology based on the extracted topics to generate consistent summaries. It uses templates (e.g., "[Topic] was discussed, and [Result] was decided.") to create the summary text. The generated summaries undergo grammatical checks and are then stored in a database along with relevant metadata (publication date, agenda title, attendees, etc.).

[0686] Emotion engine and summary provided

[0687] When a user requests a summary using their device, the device sends the request to the server. The server retrieves the latest summary from the database and has the emotion engine analyze the user's input data (text, voice, facial expression data). Based on the analysis results, the emotion engine adjusts how the summary is provided (e.g., text length, expression style, level of detail). The server sends the adjusted summary to the user's device, which then displays it to the user.

[0688] Specific example

[0689] For example, if a user requests to "check the summary of the latest research conference minutes," the processing flow is as follows:

[0690] 1. The user clicks the "Request Summary" button on their device.

[0691] 2. The device sends a request to the server.

[0692] 3. The server retrieves the latest summary from the database and has the emotion engine analyze the user's emotion data.

[0693] 4. The emotion engine adjusts how the summary is provided based on the user's emotional information.

[0694] 5. The server sends the adjusted summary to the user's terminal.

[0695] 6. The device displays Sally, and the user confirms it.

[0696] Example of a prompt

[0697] The following are examples of prompts to input into a generative AI model:

[0698] I would like to review summaries of the latest research conference proceedings. Please analyze the crawled conference data and generate summaries based on key topics. Additionally, please adjust the summary presentation method based on user sentiment data (text, voice, and facial expression data).

[0699] As described above, the present invention can improve user understanding and satisfaction by providing an optimal summary that takes user emotions into consideration.

[0700] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0701] Step 1:

[0702] Data collection

[0703] The server uses web crawling technology to collect publicly available document data from specified URLs. As input, the URL of a specific website is provided to the crawler, and as output, the HTML content is temporarily stored on the server.

[0704] Specifically, the crawler accesses a URL, downloads HTML data in real time, and stores it in temporary storage.

[0705] Step 2:

[0706] HTML parsing and text extraction

[0707] The server extracts the necessary text data from the stored HTML content. The stored HTML data is given as input, and the text data is generated as output.

[0708] Specifically, the server uses an HTML parser (for example, Beautiful Soup) to remove HTML tags and unnecessary special characters and extract the text. The extracted text is then stored in a new variable.

[0709] Step 3:

[0710] Normalization process

[0711] The server performs normalization on the extracted text data. The extracted text data is given as input, and normalized text data is generated as output.

[0712] Specifically, the server uses regular expressions to remove special characters and unnecessary whitespace, and converts full-width spaces to half-width spaces. This prepares the text data for parsing.

[0713] Step 4:

[0714] Morphological analysis

[0715] The server performs morphological analysis on normalized text data. Normalized text data is provided as input, and the morphological analysis results are generated as output.

[0716] Specifically, the server uses a morphological analysis tool (for example, MeCab) to split the text data into words and tag each word with its part of speech.

[0717] Step 5:

[0718] Topic extraction

[0719] The server extracts topics based on the morphological analysis results. The input is the morphological analysis results, and the output is a list of the extracted topics.

[0720] In terms of specific operations, the server executes topic modeling algorithms such as LDA (Latent Dirichlet Allocation) and TF-IDF (Term Frequency-Inverse Document Frequency) to extract important topics.

[0721] Step 6:

[0722] Summary Generation

[0723] The server generates a summary based on the extracted topics. The extracted topics are given as input, and the generated summary is provided as output.

[0724] In practice, the server uses a natural language generation algorithm and a template (for example, "[Topic] was discussed, and [Result] was determined.") to create a consistent summary text.

[0725] Step 7:

[0726] Save summary

[0727] The server saves the generated summary. The generated summary and associated metadata (e.g., publication date, agenda title, attendees) are given as input, and the output is saved to the database.

[0728] Specifically, the server records summary text and metadata in the database.

[0729] Step 8:

[0730] Summary Request Acceptance

[0731] The user requests a summary through their device. The user's request is sent to the device as input, and the device then sends a request to the server.

[0732] Specifically, the user clicks the "Request Summary" button on their device. The device then sends this request to the server.

[0733] Step 9:

[0734] Emotion analysis

[0735] The server analyzes the received request and user sentiment data. User sentiment data (text, voice, facial expressions) is given as input, and sentiment analysis results are generated as output.

[0736] Specifically, the server uses an emotion engine to analyze emotional information from the user's input data and retrieves the results.

[0737] Step 10:

[0738] Adjustment of summary delivery method

[0739] The server adjusts the way the summary is provided based on the sentiment analysis results. The sentiment analysis results are given as input, and the adjusted summary is generated as output.

[0740] Specifically, the server adjusts the format and length of the summary text based on the sentiment analysis results.

[0741] Step 11:

[0742] Send Summary

[0743] The server sends the adjusted summary to the user terminal. The adjusted summary is sent from the server as input and received by the user terminal as output.

[0744] Specifically, the server uses a protocol to send the adjusted summary to the user's terminal.

[0745] Step 12:

[0746] Summary display

[0747] The terminal displays a summary to the user. The adjusted summary is received by the terminal as input and displayed to the user as output.

[0748] Specifically, the terminal displays the received summary on the user interface, allowing the user to review it.

[0749] (Application Example 2)

[0750] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0751] The vast amount of electronic payment data available today can be difficult for users to understand, and information provided without considering emotions can negatively impact the user experience. For example, simply providing a payment history is insufficient for a user who feels anxious about high spending after shopping. While there is a need for appropriate information tailored to the user's emotions, no system currently exists to achieve this. Therefore, a system is needed that recognizes the user's emotions and provides an optimal summary of their payment history based on those emotions.

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

[0753] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary based on the extracted topics, means for storing the generated summary and providing it to the user, means for recognizing the user's emotions, and means for adjusting the method of providing the summary based on the recognized emotions. This makes it possible to provide an optimal summary that responds to the user's emotions, thereby improving the user experience.

[0754] "Publicly available document data" refers to document data that is provided in a form accessible to anyone on the internet or other public environments.

[0755] "Means of collection" refers to methods and devices for obtaining document data from specified URLs or sources using technologies such as web crawling and HTTP requests.

[0756] "Means of extraction" refers to methods or devices used to extract necessary text information from collected data, such as HTML parsers or text analysis tools.

[0757] "Preprocessing" refers to the process of removing noise from data, normalizing it, and converting its format, essentially organizing and correcting the data before analysis.

[0758] "Noise" refers to unnecessary data or information that should be excluded from the analysis, such as special characters, multiple line breaks, and extra whitespace.

[0759] "Normalization" refers to the process of unifying the format and notation of data, such as converting full-width spaces to half-width spaces to conform to a standard format.

[0760] A "morpheme" is a fundamental unit that makes up text, such as a word or phrase—the smallest element that possesses meaning.

[0761] "Morphological analysis" refers to the process of dividing text data into words and phrases and tagging each with its part of speech.

[0762] "Important topics" refer to themes and topics that frequently appear in the text data, and are particularly useful or interesting to the user.

[0763] A "summary" is a concise compilation of key points extracted from lengthy texts or complex data, presented in a way that users can quickly understand.

[0764] "Means for generating summaries" refers to methods or devices that use topic modeling or natural language generation techniques to create consistent, short sentences from extracted information.

[0765] "Means of providing to the user" refers to methods and devices for presenting the generated summary in a form accessible to the user, such as smartphone apps and web interfaces.

[0766] "Means for recognizing user emotions" refers to methods and devices for detecting a user's emotional state using text analysis, speech recognition, facial expression analysis, etc.

[0767] "Means for adjusting the method of providing summaries based on recognized emotions" refers to methods or devices that receive user emotion information as analysis results and dynamically change the length, level of detail, and format of the summary provided.

[0768] The system for realizing this invention includes the following means: A server collects publicly available document data and uses an HTML parser and HTTP request technology to extract text data from the collected document data. Then, it preprocesses the extracted text data, removing noise and using MeCab as a tool to normalize the text. Topic modeling techniques such as LDA and TF-IDF are used to analyze morphemes from the preprocessed text data and extract important topics. When generating a summary based on the extracted topics, natural language generation technology is used, and the generated summary is stored in a database and includes an interface for providing it to the user.

[0769] Furthermore, to recognize user emotions, the system incorporates an emotion engine that uses text analysis, speech recognition, and facial expression analysis technologies. This emotion engine analyzes user input data (e.g., text, speech, and facial expression data) to recognize emotions and adjusts how the summary is provided based on the recognized emotions. Specifically, if a user sends a prompt such as, "I want a summary of my recent spending. I'm worried about the content, so I'd also like some saving advice," the emotion engine analyzes the user's emotions (anxiety) and appropriately adjusts the length and level of detail of the summary.

[0770] The hardware used will include a server, the user's smartphone or PC, and, if necessary, smart glasses or a head-mounted display. The specific software used will be Python programs, the requests library, BeautifulSoup, MeCab, the gensim library, and TextBlob. This will improve the user experience by providing users with summaries that appropriately reflect their emotional state when they want to learn about their spending and transaction history.

[0771] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0772] Step 1:

[0773] The user requests a summary of their electronic payment history using a device (smartphone or PC).

[0774] Input: The user enters the prompt message: "I would like a summary of my recent spending. I am also concerned about the content, so I would like some advice on saving money."

[0775] Output: Sends a user request to the server.

[0776] Step 2:

[0777] The terminal forwards the received request to the server.

[0778] Input: User prompt text.

[0779] Output: Sends a prompt message to the server.

[0780] Step 3:

[0781] The server collects electronic payment data from a specified internet URL.

[0782] Input: The specified URL.

[0783] Output: Collected electronic payment data.

[0784] The server uses the requests library to download HTML content from the specified URL and BeautifulSoup to extract text data from the HTML tags.

[0785] Step 4:

[0786] The server performs preprocessing on the extracted text data.

[0787] Input: Extracted text data.

[0788] Output: Noise-removed and normalized text data.

[0789] Specifically, it removes special characters, multiple line breaks, and unnecessary spaces, and converts full-width spaces to half-width spaces.

[0790] Step 5:

[0791] The server performs morphological analysis on the pre-processed text data and extracts important topics.

[0792] Input: Preprocessed text data.

[0793] Output: Key topics extracted.

[0794] We use MeCab to split the text into words and then perform topic modeling using LDA or TF-IDF.

[0795] Step 6:

[0796] The server generates a summary based on the extracted topics.

[0797] Input: Key topics extracted.

[0798] Output: The generated summary.

[0799] Using natural language generation technology, a coherent text containing extracted topics is created.

[0800] Step 7:

[0801] The server saves the generated summary to the database and prepares it for delivery.

[0802] Input: Generated summary.

[0803] Output: Summary saved in the database.

[0804] Grammar checks are also performed, and the final summary is formatted and saved.

[0805] Step 8:

[0806] The emotion engine analyzes the user's emotions.

[0807] Input: User prompt (e.g., "I'd like a summary of my recent expenses. I'm also concerned about the details, so I'd like some advice on saving money.").

[0808] Output: User's emotional state (e.g., anxiety).

[0809] Use TextBlob to analyze the user's input text and obtain a sentiment score.

[0810] Step 9:

[0811] The server generates a refined summary using an emotion engine.

[0812] Input: User's emotional state, generated summary.

[0813] Output: Adjusted summary.

[0814] The length, level of detail, and presentation of the summary are adjusted according to the user's emotional state.

[0815] Step 10:

[0816] The server sends the adjusted summary to the terminal.

[0817] Input: Adjusted summary.

[0818] Output: Sends a summary to the user's terminal.

[0819] The terminal displays a summary received from the server to the user.

[0820] Through the above processing steps, a user-friendly and emotionally resonant summary of the electronic payment history is provided.

[0821] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0822] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0823] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0824] [Third Embodiment]

[0825] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0826] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0827] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0828] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0829] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0830] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0831] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0832] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0833] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0834] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0835] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0836] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0837] This invention is a system that collects publicly available document data, particularly meeting minutes and report data from the internet, preprocesses and analyzes it, and extracts and summarizes important points from this data to provide a summary. The following describes specific embodiments for implementing this invention.

[0838] Conceptual explanation of the program

[0839] overview

[0840] This system begins by acquiring publicly available document data, preprocessing and analyzing it to extract important information, and finally generating a summary to provide to the user. The following describes the specific actions of the server, terminal, and user at each step.

[0841] Data collection

[0842] The server collects meeting minutes data from a specified internet URL using web crawling technology. For example, if the target is the minutes of a research group published by the Ministry of Internal Affairs and Communications, the server accesses the relevant web page and downloads the HTML content.

[0843] Data extraction and preprocessing

[0844] The server extracts the meeting minutes text from the downloaded HTML content. During this process, it performs noise reduction to remove HTML tags, unnecessary special characters, line breaks, etc. Next, it performs normalization, such as converting full-width spaces to half-width spaces and unifying different expressions.

[0845] Text analysis and topic extraction

[0846] The server performs morphological analysis on the preprocessed text data. A morphological analysis tool (e.g., MeCab) is used here. As a result of the analysis, the text data is divided into words, and the part of speech of each word is identified. Based on the data obtained from this analysis, important topics are extracted using topic modeling (e.g., LDA or TF-IDF).

[0847] Summary Generation

[0848] The server uses natural language generation technology to create summaries based on the extracted topics. For example, it uses a template to concisely summarize the main points in the format "XX was discussed, and YY was decided." Grammatical checks are also performed during this process to generate consistent sentences.

[0849] Saving and providing summaries

[0850] The created summary is stored in the server's database. When a user requests a summary through their device, the server searches the stored summary and sends the relevant data to the user's device. The user can then view the summary on their device.

[0851] Specific example

[0852] For example, when the Ministry of Internal Affairs and Communications provides a summary of the latest research meeting minutes, the following process occurs when a user clicks the "View summary of the latest meeting minutes" button on their device.

[0853] 1. The user clicks a button on their device.

[0854] 2. The terminal sends a summary request to the server.

[0855] 3. The server generates a summary from the latest collected meeting minutes data and saves it to the database.

[0856] 4. The server searches the saved summary and sends it to the user's terminal.

[0857] 5. The terminal displays the summary received from the server on its screen, which the user can then view.

[0858] Through this series of processes, users can efficiently obtain important information, and the system of the present invention facilitates the understanding and sharing of meeting minutes.

[0859] The following describes the processing flow.

[0860] Step 1: Data Collection

[0861] The server accesses the specified URL (for example, a public webpage of a specific government agency).

[0862] The server uses web crawling technology to download the HTML content of web pages.

[0863] The server saves the downloaded data to temporary storage.

[0864] Step 2: Data Extraction

[0865] The server runs an HTML parser to extract the meeting minutes text from the saved HTML content.

[0866] The server extracts the text data while removing HTML tags.

[0867] Step 3: Data Preprocessing

[0868] As a preprocessing step, the server removes unnecessary special characters and multiple line breaks / spaces from the text data.

[0869] The server performs normalization, such as converting full-width spaces to half-width spaces within the text.

[0870] Step 4: Morphological Analysis

[0871] The server uses a morphological analysis tool (e.g., MeCab) to divide the preprocessed text data into words.

[0872] The server tags each word with its part of speech and records the results of morphological analysis.

[0873] Step 5: Topic Extraction

[0874] The server uses topic modeling techniques (e.g., LDA and TF-IDF) to extract important topics from morphologically analyzed data.

[0875] The server evaluates the frequency and importance of words related to each topic and lists the main topics.

[0876] Step 6: Summary Generation

[0877] The server generates a summary sentence based on the extracted topics, utilizing natural language generation technology.

[0878] The server constructs the text based on a specific template (e.g., "[Topic] was discussed, and [Result] was determined.").

[0879] The server performs a grammatical check on the generated summary sentence and makes corrections as needed.

[0880] Step 7: Save the summary

[0881] The server saves the generated summary to the database.

[0882] The server also records metadata related to the summary (e.g., publication date, meeting minutes title, attendees).

[0883] Step 8: Provide a summary

[0884] The user requests a summary through their device.

[0885] The server searches the database for the relevant summary and sends it to the user's terminal.

[0886] The terminal displays the received summary to the user.

[0887] As described above, a system is realized in which the server, terminal, and user cooperate at each step to efficiently generate and provide document summaries.

[0888] (Example 1)

[0889] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0890] The sheer volume of publicly available document data, particularly meeting minutes and reports found online, makes it difficult to efficiently extract key points and provide summaries. Furthermore, manually collecting, pre-processing, analyzing, and summarizing this data is time-consuming and labor-intensive. Therefore, there is a need for a system that automatically and efficiently collects and analyzes document data, extracts key points, and provides summaries.

[0891] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0892] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary using a generative AI model based on the extracted topics, means for verifying the generated summary using a grammar checking tool, and means for saving the generated summary and providing it to the user. This enables the user to efficiently obtain important information from a vast amount of document data.

[0893] "Publicly available document data" refers to document data that is generally accessible on the internet or in other public spheres.

[0894] "Web crawling" is a technology that automatically collects data from specified web pages and websites.

[0895] "HTML content" refers to data in HTML (Hypertext Markup Language) format that represents the structure and content of a web page.

[0896] "Preprocessing" refers to a series of processes that remove noise from data and normalize it before performing data analysis.

[0897] "Noise reduction" is the process of removing elements from text data that are unnecessary for analysis (e.g., HTML tags, special characters, unnecessary line breaks).

[0898] "Normalization" is a process that unifies different representations within text data to ensure data consistency.

[0899] Morphological analysis is a technique that divides text data into the smallest possible linguistic elements and analyzes the part of speech of each element.

[0900] A "morphological analysis tool" refers to software or algorithms used to perform morphological analysis (e.g., MeCab).

[0901] Topic modeling is a technique that automatically extracts important topics and themes from large amounts of text data (e.g., LDA, TF-IDF).

[0902] A "generative AI model" is an artificial intelligence model that uses machine learning to generate natural language (e.g., GPT-3).

[0903] A "summary" is a concise summary that captures the key points of an entire document.

[0904] A "grammar check tool" is software or algorithms used to verify and correct the grammatical accuracy of generated text.

[0905] A "database" is a system that allows for the efficient storage, management, and retrieval of structured data.

[0906] A "terminal" refers to a device (e.g., a personal computer, smartphone, or tablet) that a user uses to access a server and view or manipulate data.

[0907] This invention is a system that collects publicly available document data, particularly meeting minutes and report data from the internet, preprocesses and analyzes it, and extracts and summarizes important points from this data. A detailed embodiment of this invention is described below.

[0908] Data collection

[0909] The server uses web crawling technology to collect meeting minutes data from specified internet URLs. Specifically, the server accesses URLs in a specified list sequentially, retrieves HTML content, and saves it to local storage. For example, it might access the URL of a publicly available research conference minutes page of a certain public institution and download the HTML file.

[0910] Data extraction and preprocessing

[0911] The server extracts the necessary meeting minutes text from the collected HTML content and performs preprocessing. It uses an HTML parser (e.g., Beautiful Soup) to remove HTML tags, unnecessary special characters, and line breaks. Furthermore, it performs normalization, such as converting full-width spaces to half-width spaces and unifying different expressions, as part of noise reduction. For example, the meeting minutes text " Chairperson: Thank you for joining us today. Extract "Chairman: Thank you for participating today." from the above.

[0912] Text analysis and topic extraction

[0913] The server performs morphological analysis on the preprocessed text data. For this analysis, a morphological analysis tool (e.g., MeCab) is used to divide the text into the smallest units of linguistic elements and identify the part of speech for each. For example, the sentence "Thank you for participating today." is divided into "today / noun," "wa / particle," "participation / noun," "arigato / verb," ​​"u / auxiliary verb," ​​and "gozaimasu / verb." Subsequently, topic modeling techniques (e.g., LDA or TF-IDF) are used to extract important topics from the text data. Frequently occurring keywords such as "research," "minutes," and "technology" are extracted.

[0914] Summary Generation

[0915] The server generates summaries using a generative AI model (e.g., GPT-3) based on important topics. For example, it might generate a summary such as, "The minutes mainly discussed advancements in new technologies, and future research directions were decided." The generated summaries are then checked for grammatical accuracy using a grammar checking tool and corrected as needed.

[0916] Saving and providing summaries

[0917] The generated summaries are stored in the server's database. The storage format includes fields such as "timestamp," "meeting minutes title," and "summary." When a user requests a summary through their terminal, the terminal sends a request to the server, the server searches the database for the corresponding summary, and sends it to the terminal. The terminal displays the summary received from the server on its screen, allowing the user to view it.

[0918] Specific example

[0919] For example, if a public institution is providing a summary of its latest meeting minutes, when a user clicks the "View Summary of Latest Meeting Minutes" button on their device, the following process takes place.

[0920] 1. The user clicks a button on their device.

[0921] 2. The terminal sends a summary request to the server.

[0922] 3. The server generates a summary from the latest meeting minutes data it has collected and saves it to the database.

[0923] 4. The server searches for the saved summary and sends it to the user's terminal.

[0924] 5. The terminal displays the summary received from the server on the screen, and the user views it.

[0925] As an example of a prompt, the process is carried out by inputting "Please provide a summary of the latest research conference minutes from a public institution" into the generating AI model.

[0926] This system allows users to efficiently obtain important information from vast amounts of document data.

[0927] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0928] Step 1:

[0929] The server collects publicly available document data. Specifically, the server receives a specified list of URLs as input and accesses each URL using web crawling technology. It downloads the HTML content of each URL and saves it to local storage. The output is the downloaded HTML file.

[0930] Step 2:

[0931] The server extracts text data from the collected HTML content. Specifically, it receives an HTML file as input and uses an HTML parser (e.g., Beautiful Soup) to extract the meeting minutes text. As a noise reduction process, HTML tags, unnecessary special characters, and line breaks are removed. The output is the meeting minutes text. For example, HTML content " Chairperson: Thank you for joining us today. Extract "Chairman: Thank you for participating today." from the above.

[0932] Step 3:

[0933] The server performs preprocessing on the extracted text data. Specifically, it receives the text data as input, converts full-width spaces to half-width spaces, and performs normalization to unify different expressions (e.g., AI, エーアイ). The output is normalized text data.

[0934] Step 4:

[0935] The server performs morphological analysis on the pre-processed text data. Specifically, it uses a morphological analysis tool (e.g., MeCab) to receive the text data as input, split it into words, and identify the part of speech of each word. The output is a list of the split words. For example, the sentence "Thank you for participating today." is split into "today / noun", "is / particle", "participation / noun", "thank you / verb", "u / auxiliary verb", and "gozaimasu / verb".

[0936] Step 5:

[0937] The server extracts important topics using topic modeling techniques. Specifically, it takes a list of words obtained as a result of morphological analysis as input and uses topic modeling techniques (e.g., LDA or TF-IDF) to extract important topics. The output is a list of the extracted topics.

[0938] Step 6:

[0939] The server generates a summary based on the extracted topics. Specifically, it takes a list of important topics as input and generates a summary using a generative AI model (e.g., GPT-3). The output is the generated summary. For example, it might generate a summary such as, "The meeting minutes mainly discussed the progress of new technologies, and future research directions were decided."

[0940] Step 7:

[0941] The server checks the generated summary using a grammar checking tool. Specifically, it receives the generated summary as input, uses the grammar checking tool to verify its grammatical accuracy, and makes corrections as needed. The output is the corrected summary.

[0942] Step 8:

[0943] The server saves the generated summary to the database. Specifically, it receives the modified summary as input and saves it to the database. The output is the summary saved in the database. The saving format is in the form of fields such as "timestamp," "meeting minutes title," and "summary."

[0944] Step 9:

[0945] The user requests the summary via their device. Specifically, the user clicks the "View summary of the latest meeting minutes" button on their device. This action causes the device to send a summary request to the server.

[0946] Step 10:

[0947] The server retrieves the relevant summary from the database and sends it to the user's terminal. Specifically, the server receives a summary request, retrieves the relevant summary from the database, and sends it to the terminal. The output is the sent summary.

[0948] Step 11:

[0949] The terminal displays the summary received from the server on its screen, which the user then views. Specifically, the terminal receives the summary as input and displays it on the screen. The output is a summary display that the user can view.

[0950] Therefore, users can efficiently obtain important information.

[0951] (Application Example 1)

[0952] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0953] While a vast amount of meeting minutes and reports are publicly available online, there is a lack of efficient means to collect this information, extract key points, and provide concise summaries. Traditional systems require significant time and effort for manual information gathering and summarization, and the sheer volume of information makes it highly likely that users will miss useful details. Furthermore, maintaining consistent summaries is difficult, highlighting the need for technology to appropriately extract essential information.

[0954] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0955] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary using a generative AI model based on the extracted topics, and means for storing the generated summary and providing it to a terminal. This makes it possible to efficiently extract important information from a vast amount of meeting minutes and reports published on the internet and to quickly provide high-quality summaries.

[0956] "Publicly available document data" refers to information data, including text, that is provided on the internet in a format that is publicly accessible.

[0957] "Text data" refers to information expressed as a string of characters, in a format that humans can read and understand.

[0958] "Preprocessing" refers to a series of processes performed on raw data, including noise reduction and text normalization.

[0959] "Noise" refers to unnecessary elements or strings in text data, information that is not needed for analysis.

[0960] "Normalization" is the process of transforming text data into a consistent format, unifying strings based on specific rules.

[0961] Morphological analysis is the process of dividing text data into morphemes, which are the smallest semantic elements that make up the data, and adding information such as part of speech.

[0962] A "topic" refers to the main theme or subject matter deemed important within the text data.

[0963] A "generative AI model" is an artificial intelligence that learns from large amounts of data and generates natural language text based on given conditions.

[0964] A "summary" is a short, concise summary of the main points and important aspects of a long text or document.

[0965] A "terminal" is an electronic device used by a user to receive and view information.

[0966] This invention relates to a system for efficiently collecting, preprocessing, and analyzing publicly available document data, extracting important information, and generating summaries. Specifically, this system operates in the following steps:

[0967] The server first collects document data from specified URLs on the internet. This is done using a web crawler. For example, it downloads the HTML content of a webpage and extracts specific meeting minutes data. Programming libraries used include requests and BeautifulSoup.

[0968] Next, the server preprocesses the collected document data. This includes denoising (e.g., removing unnecessary HTML tags and special characters) and text normalization (e.g., converting full-width spaces to half-width spaces). Python string manipulation functions are used for this process.

[0969] Subsequently, the server analyzes the pre-processed text data and performs morphological analysis. Specifically, it uses the morphological analysis tool janome.tokenizer. This analysis identifies the words and their parts of speech that make up the text data.

[0970] Furthermore, the server extracts topics from the analyzed data. Here, topics are extracted using TF-IDF (Inverse Document Frequency) and LDA (Latent Dirichlet Allocation). In this process, the libraries TfidfVectorizer and LatentDirichletAllocation are used.

[0971] Based on the extracted topics, the server generates a summary using a generative AI model. This generative AI model utilizes the OpenAI API (e.g., GPT-3). An example of a prompt for this model is shown below.

[0972] Please generate a meeting minutes summary based on the following topics:

[0973] Topics: Digitalization, Policy, Budget, AI, Data Analysis

[0974] Summary:

[0975] Based on this prompt, the AI ​​model generates a natural and consistent summary.

[0976] The generated summary is stored in a database and provided to the user's device. Users can retrieve and view the meeting minutes summary by using an application on their device and entering a specific URL. This entire process allows for the quick and efficient acquisition of important information without requiring significant time and effort.

[0977] The hardware required to implement this system consists of an internet-connected server and user terminals. The software uses the Python programming language and its libraries, BeautifulSoup, janome, TfidfVectorizer, LatentDirichletAllocation, and the OpenAI API.

[0978] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0979] Step 1:

[0980] The server collects document data from an internet URL specified by the user. In this case, when the user enters a specific URL on their terminal, that information is sent to the server. The server uses the requests library to access the specified URL and download the HTML content of the webpage. The input to this process is the URL specified by the user, and the output is the downloaded HTML content.

[0981] Step 2:

[0982] The server extracts text data from the HTML content collected in Step 1. It uses the BeautifulSoup library to parse HTML tags and extract the text portion. This removes HTML tags and unnecessary special characters. The input is the collected HTML content, and the output is the extracted text data.

[0983] Step 3:

[0984] The server performs preprocessing on the extracted text data. Preprocessing includes noise reduction (removing unnecessary characters) and normalization (converting full-width spaces to half-width spaces, unifying different expressions, etc.). Python string manipulation functions are used at this stage. The input is the extracted text data, and the output is the preprocessed text data.

[0985] Step 4:

[0986] The server performs morphological analysis on the preprocessed text data. During this process, it uses janome.tokenizer to split the text data into individual words and identify the part of speech for each word. The input is the preprocessed text data, and the output is the split data with its part of speech information.

[0987] Step 5:

[0988] The server extracts important topics based on the data analyzed in step 4. Here, it uses TfidfVectorizer to calculate the importance of each topic and then models the main topics from the entire text using LatentDirichletAllocation. The input is data segmented word by word and its part-of-speech information, and the output is the extracted important topics.

[0989] Step 6:

[0990] The server generates a summary using a generative AI model based on the extracted topics. Here, the OpenAI API (e.g., GPT-3) is used to form prompt statements and send them to the AI ​​model. A concrete example of a generated prompt statement is as follows:

[0991] Please generate a meeting minutes summary based on the following topics:

[0992] Topics: Digitalization, Policy, Budget, AI, Data Analysis

[0993] Summary:

[0994] The input consists of the extracted key topics, and the output is the generated summary.

[0995] Step 7:

[0996] The server stores the generated summary in a database and provides it to the user's terminal upon request. When a user requests a summary from their terminal, the server retrieves the corresponding summary from the database and sends it to the user's terminal. The input is the user's request and the generated summary, and the output is the summary sent to the user's terminal.

[0997] Through the steps described above, a system is realized that allows users to efficiently and quickly obtain important summary information from document information on the internet.

[0998] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0999] This invention is a system that collects, preprocesses, and analyzes publicly available document data to extract important information and generate summaries, and also incorporates an emotion engine that recognizes the user's emotions. This configuration allows for dynamic adjustment of the optimal summary delivery method according to the user's emotional state.

[1000] Conceptual explanation of the program

[1001] overview

[1002] This system collects publicly available document data by web crawling, preprocesses it, performs morphological analysis and topic extraction, and then automatically generates summaries. In addition to this basic summary generation process, it uses a sentiment engine to analyze user sentiment and determine the optimal way to provide summaries. The specific actions of the server, terminal, and user at each step are described below.

[1003] Data collection and extraction

[1004] The server collects meeting minutes data from specified internet URLs using web crawling technology. For example, it downloads HTML content from the public web pages of specific government agencies. This content is temporarily stored on the server.

[1005] Next, the server extracts the meeting minutes text from the saved HTML content. Specifically, it uses an HTML parser to remove HTML tags and unnecessary special characters, and then extracts the necessary text data.

[1006] Data preprocessing and analysis

[1007] As a preprocessing step, the server removes unnecessary noise (special characters, multiple line breaks and spaces, etc.) from the extracted text data and performs normalization by converting full-width spaces to half-width spaces.

[1008] Next, morphological analysis is performed on the preprocessed text data. A morphological analysis tool (e.g., MeCab) is used to divide the text data into words, and each word is tagged with its part of speech. Based on this analyzed data, topic modeling techniques (e.g., LDA or TF-IDF) are used to extract important topics within the text. The extracted topics are listed and used in the subsequent summary generation process.

[1009] Summary generation and saving

[1010] The server uses natural language generation technology to create a summary based on the extracted topics. After generating a consistent sentence using a template (e.g., "[Topic] was discussed, and [Result] was decided."), it performs a grammatical check and saves the final summary to the database. At this time, metadata related to the summary (e.g., publication date, agenda title, attendees) is also recorded.

[1011] Emotion Engine Additions and Features

[1012] When a user requests a summary through their device, the emotion engine analyzes the user's input data (e.g., text, voice, facial expression data) to recognize their emotions. Based on the analyzed emotion data, the emotion engine determines the optimal way to provide the summary (e.g., text length, expression style, level of detail). This emotion-based adjustment ensures that the summary is provided in a way that is most easily understood and accepted by the user.

[1013] As a concrete example, consider a case where a user requests to "check the summary of the latest research conference minutes." The process in this case would be as follows:

[1014] 1. The user clicks the button to request a summary on their device.

[1015] 2. The terminal sends the user's request to the server.

[1016] 3. The server retrieves the latest summary from the database and has the emotion engine analyze the user's emotion data.

[1017] 4. The emotion engine adjusts how the summary is provided based on the user's emotion information (input data at the time of the request).

[1018] 5. The server sends the adjusted summary to the user's terminal.

[1019] 6. The terminal displays the summary received from the server to the user.

[1020] This system provides an optimal summary tailored to the user's emotional state, improving user understanding and satisfaction.

[1021] The following describes the processing flow.

[1022] Step 1: Data Collection

[1023] The server accesses a specified internet URL (e.g., a public webpage of a government agency) and downloads the HTML content of the webpage using web crawling technology.

[1024] The server saves the downloaded HTML content to temporary storage.

[1025] Step 2: Data Extraction

[1026] The server runs an HTML parser to extract the meeting minutes text from the saved HTML content.

[1027] The server removes HTML tags and extracts the necessary text data.

[1028] Step 3: Data Preprocessing

[1029] As a preprocessing step, the server performs noise reduction by removing unnecessary special characters and multiple line breaks / spaces from the text data.

[1030] The server performs a normalization process that converts full-width spaces to half-width spaces.

[1031] Step 4: Morphological Analysis

[1032] The server uses a morphological analysis tool (e.g., MeCab) to divide the preprocessed text data into words and tag each word with its part of speech.

[1033] The server lists the morphological analysis results.

[1034] Step 5: Topic Extraction

[1035] The server uses topic modeling techniques (e.g., LDA and TF-IDF) to extract important topics from morphologically analyzed data.

[1036] The server evaluates the frequency and importance of related words for each topic and lists the main topics.

[1037] Step 6: Summary Generation

[1038] The server uses natural language generation technology to generate summary sentences based on the extracted topics.

[1039] The server constructs consistent sentences based on a template (e.g., "[Topic] was discussed, and [Result] was determined.").

[1040] The server performs a grammatical check on the generated summary sentence and makes corrections as needed.

[1041] Step 7: Save the summary

[1042] The server saves the generated summary to the database.

[1043] The server also records metadata related to the summary (e.g., publication date, agenda title, attendees).

[1044] Step 8: Emotion Recognition

[1045] The user sends a summary request through their device.

[1046] The terminal collects user input data (e.g., text, voice, facial expression data) and sends it to the server.

[1047] The server uses an emotion engine to analyze the transmitted user data and recognize the user's emotions.

[1048] Step 9: Emotion-based summary adjustment

[1049] The server adjusts how the summary is provided (e.g., text length, format, level of detail) based on the analysis results of the emotion engine.

[1050] The server prepares the adjusted summary in the optimal format.

[1051] Step 10: Provide a summary

[1052] The server sends the adjusted summary to the user's terminal.

[1053] The terminal displays the received summary to the user.

[1054] Users view and utilize summaries optimized based on their emotions.

[1055] Through each of the above steps, the system efficiently generates and adjusts document data summaries and provides them in an appropriate format that suits the user's emotions.

[1056] (Example 2)

[1057] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1058] Conventional text summary generation systems have the problem of not being able to consider user sentiment when collecting publicly available document data and generating summaries, making it difficult to provide summaries that are optimal for the user. In addition, the method of providing summaries is fixed, which is a problem as it does not sufficiently improve user understanding and satisfaction.

[1059] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1060] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary based on the extracted topics, means for saving the generated summary and providing it to the user, and means for analyzing the user's sentiment information and adjusting the method of providing the summary based on the analyzed sentiment information. This makes it possible to provide an optimal summary that responds to the user's sentiment, thereby improving the user's understanding and satisfaction.

[1061] "Publicly available document data" refers to document information that is provided in a format accessible to anyone on the internet or in other public places.

[1062] "Means of collection" refers to technologies and methods that have the function of automatically collecting publicly available document data from the internet using specific algorithms or tools.

[1063] "Text data" refers to string information extracted from collected document data, and is the basic unit that constitutes the content of a document.

[1064] "Preprocessing" refers to the process of removing unnecessary information (special characters, spaces, line breaks, etc.) from text data and preparing the data for analysis.

[1065] "Normalization" refers to the operation of standardizing irregular elements in text data (such as converting full-width spaces to half-width spaces) during preprocessing.

[1066] A "morpheme" is the smallest semantic unit (word or phrase) that makes up text data, and is an element extracted through morphological analysis.

[1067] "Morphological analysis" refers to the process of dividing text data into morphemes, tagging each morpheme with its part of speech, and then analyzing it.

[1068] A "topic" refers to an important subject or subject matter within document data, extracted through morphological analysis or topic modeling.

[1069] "Means of extracting topics" refers to technologies and methods that use text analysis techniques to identify important subjects within a document and list them.

[1070] A "summary" refers to a short document that concisely summarizes the main points of the entire document based on the extracted topics.

[1071] "Means for generating summaries" refers to technologies and methods that have the function of creating a consistent summary sentence using natural language generation technology based on extracted topics.

[1072] "Means of providing to the user" refers to technologies and methods that have the functionality to send the generated summary to the user's terminal and display it.

[1073] "Emotional information" refers to emotional states and reactions analyzed from user input data (text, voice, facial expressions, etc.).

[1074] "Means of analyzing emotional information" refers to technologies and methods that analyze user input data and identify the emotions contained within it.

[1075] "Means of adjusting the delivery method" refers to technologies and methods that have the function of optimizing the content, format, and expression of the summary based on the analyzed emotional information.

[1076] The present invention is a system that collects, preprocesses, and analyzes publicly available document data to extract important information and generate summaries, and also incorporates an emotion engine that recognizes user emotions. Specific embodiments of the present invention are described below.

[1077] Data collection and extraction

[1078] The server uses web crawling technology to collect publicly available document data from specified internet URLs. For example, it downloads HTML content from a specific public institution's webpage and stores it temporarily. The server then uses an HTML parser (such as Beautiful Soup) to remove unnecessary tags and special characters from the downloaded HTML content and extract the necessary text data.

[1079] Data preprocessing

[1080] The server performs normalization on the extracted text data. Specifically, it removes special characters and unnecessary whitespace, and converts full-width spaces to half-width spaces. This prepares the text data for analysis.

[1081] Text analysis and topic extraction

[1082] The server uses a morphological analysis tool (e.g., MeCab) to segment the normalized text data into words and tags each word with its part of speech. The server then uses topic modeling techniques such as LDA (Latent Dirichlet Allocation) and TF-IDF (Term Frequency-Inverse Document Frequency) to extract important topics from the text.

[1083] Summary generation and saving

[1084] The server utilizes natural language generation technology based on the extracted topics to generate consistent summaries. It uses templates (e.g., "[Topic] was discussed, and [Result] was decided.") to create the summary text. The generated summaries undergo grammatical checks and are then stored in a database along with relevant metadata (publication date, agenda title, attendees, etc.).

[1085] Emotion engine and summary provided

[1086] When a user requests a summary using their device, the device sends the request to the server. The server retrieves the latest summary from the database and has the emotion engine analyze the user's input data (text, voice, facial expression data). Based on the analysis results, the emotion engine adjusts how the summary is provided (e.g., text length, expression style, level of detail). The server sends the adjusted summary to the user's device, which then displays it to the user.

[1087] Specific example

[1088] For example, if a user requests to "check the summary of the latest research conference minutes," the processing flow is as follows:

[1089] 1. The user clicks the "Request Summary" button on their device.

[1090] 2. The device sends a request to the server.

[1091] 3. The server retrieves the latest summary from the database and has the emotion engine analyze the user's emotion data.

[1092] 4. The emotion engine adjusts how the summary is provided based on the user's emotional information.

[1093] 5. The server sends the adjusted summary to the user's terminal.

[1094] 6. The device displays Sally, and the user confirms it.

[1095] Example of a prompt

[1096] The following are examples of prompts to input into a generative AI model:

[1097] I would like to review summaries of the latest research conference proceedings. Please analyze the crawled conference data and generate summaries based on key topics. Additionally, please adjust the summary presentation method based on user sentiment data (text, voice, and facial expression data).

[1098] As described above, the present invention can improve user understanding and satisfaction by providing an optimal summary that takes user emotions into consideration.

[1099] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1100] Step 1:

[1101] Data collection

[1102] The server uses web crawling technology to collect publicly available document data from specified URLs. As input, the URL of a specific website is provided to the crawler, and as output, the HTML content is temporarily stored on the server.

[1103] Specifically, the crawler accesses a URL, downloads HTML data in real time, and stores it in temporary storage.

[1104] Step 2:

[1105] HTML parsing and text extraction

[1106] The server extracts the necessary text data from the stored HTML content. The stored HTML data is given as input, and the text data is generated as output.

[1107] Specifically, the server uses an HTML parser (for example, Beautiful Soup) to remove HTML tags and unnecessary special characters and extract the text. The extracted text is then stored in a new variable.

[1108] Step 3:

[1109] Normalization process

[1110] The server performs normalization on the extracted text data. The extracted text data is given as input, and normalized text data is generated as output.

[1111] Specifically, the server uses regular expressions to remove special characters and unnecessary whitespace, and converts full-width spaces to half-width spaces. This prepares the text data for parsing.

[1112] Step 4:

[1113] Morphological analysis

[1114] The server performs morphological analysis on normalized text data. Normalized text data is provided as input, and the morphological analysis results are generated as output.

[1115] Specifically, the server uses a morphological analysis tool (for example, MeCab) to split the text data into words and tag each word with its part of speech.

[1116] Step 5:

[1117] Topic extraction

[1118] The server extracts topics based on the morphological analysis results. The input is the morphological analysis results, and the output is a list of the extracted topics.

[1119] In terms of specific operations, the server executes topic modeling algorithms such as LDA (Latent Dirichlet Allocation) and TF-IDF (Term Frequency-Inverse Document Frequency) to extract important topics.

[1120] Step 6:

[1121] Summary Generation

[1122] The server generates a summary based on the extracted topics. The extracted topics are given as input, and the generated summary is provided as output.

[1123] In practice, the server uses a natural language generation algorithm and a template (for example, "[Topic] was discussed, and [Result] was determined.") to create a consistent summary text.

[1124] Step 7:

[1125] Save summary

[1126] The server saves the generated summary. The generated summary and associated metadata (e.g., publication date, agenda title, attendees) are given as input, and the output is saved to the database.

[1127] Specifically, the server records summary text and metadata in the database.

[1128] Step 8:

[1129] Summary Request Acceptance

[1130] The user requests a summary through their device. The user's request is sent to the device as input, and the device then sends a request to the server.

[1131] Specifically, the user clicks the "Request Summary" button on their device. The device then sends this request to the server.

[1132] Step 9:

[1133] Emotion analysis

[1134] The server analyzes the received request and user sentiment data. User sentiment data (text, voice, facial expressions) is given as input, and sentiment analysis results are generated as output.

[1135] Specifically, the server uses an emotion engine to analyze emotional information from the user's input data and retrieves the results.

[1136] Step 10:

[1137] Adjustment of summary delivery method

[1138] The server adjusts the way the summary is provided based on the sentiment analysis results. The sentiment analysis results are given as input, and the adjusted summary is generated as output.

[1139] Specifically, the server adjusts the format and length of the summary text based on the sentiment analysis results.

[1140] Step 11:

[1141] Send Summary

[1142] The server sends the adjusted summary to the user terminal. The adjusted summary is sent from the server as input and received by the user terminal as output.

[1143] Specifically, the server uses a protocol to send the adjusted summary to the user's terminal.

[1144] Step 12:

[1145] Summary display

[1146] The terminal displays a summary to the user. The adjusted summary is received by the terminal as input and displayed to the user as output.

[1147] Specifically, the terminal displays the received summary on the user interface, allowing the user to review it.

[1148] (Application Example 2)

[1149] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1150] The vast amount of electronic payment data available today can be difficult for users to understand, and information provided without considering emotions can negatively impact the user experience. For example, simply providing a payment history is insufficient for a user who feels anxious about high spending after shopping. While there is a need for appropriate information tailored to the user's emotions, no system currently exists to achieve this. Therefore, a system is needed that recognizes the user's emotions and provides an optimal summary of their payment history based on those emotions.

[1151] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1152] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary based on the extracted topics, means for storing the generated summary and providing it to the user, means for recognizing the user's emotions, and means for adjusting the method of providing the summary based on the recognized emotions. This makes it possible to provide an optimal summary that responds to the user's emotions, thereby improving the user experience.

[1153] "Publicly available document data" refers to document data that is provided in a form accessible to anyone on the internet or other public environments.

[1154] "Means of collection" refers to methods and devices for obtaining document data from specified URLs or sources using technologies such as web crawling and HTTP requests.

[1155] "Means of extraction" refers to methods or devices used to extract necessary text information from collected data, such as HTML parsers or text analysis tools.

[1156] "Preprocessing" refers to the process of removing noise from data, normalizing it, and converting its format, essentially organizing and correcting the data before analysis.

[1157] "Noise" refers to unnecessary data or information that should be excluded from the analysis, such as special characters, multiple line breaks, and extra whitespace.

[1158] "Normalization" refers to the process of unifying the format and notation of data, such as converting full-width spaces to half-width spaces to conform to a standard format.

[1159] A "morpheme" is a fundamental unit that makes up text, such as a word or phrase—the smallest element that possesses meaning.

[1160] "Morphological analysis" refers to the process of dividing text data into words and phrases and tagging each with its part of speech.

[1161] "Important topics" refer to themes and topics that frequently appear in the text data, and are particularly useful or interesting to the user.

[1162] A "summary" is a concise compilation of key points extracted from lengthy texts or complex data, presented in a way that users can quickly understand.

[1163] "Means for generating summaries" refers to methods or devices that use topic modeling or natural language generation techniques to create consistent, short sentences from extracted information.

[1164] "Means of providing to the user" refers to methods and devices for presenting the generated summary in a form accessible to the user, such as smartphone apps and web interfaces.

[1165] "Means for recognizing user emotions" refers to methods and devices for detecting a user's emotional state using text analysis, speech recognition, facial expression analysis, etc.

[1166] "Means for adjusting the method of providing summaries based on recognized emotions" refers to methods or devices that receive user emotion information as analysis results and dynamically change the length, level of detail, and format of the summary provided.

[1167] The system for realizing this invention includes the following means: A server collects publicly available document data and uses an HTML parser and HTTP request technology to extract text data from the collected document data. Then, it preprocesses the extracted text data, removing noise and using MeCab as a tool to normalize the text. Topic modeling techniques such as LDA and TF-IDF are used to analyze morphemes from the preprocessed text data and extract important topics. When generating a summary based on the extracted topics, natural language generation technology is used, and the generated summary is stored in a database and includes an interface for providing it to the user.

[1168] Furthermore, to recognize user emotions, the system incorporates an emotion engine that uses text analysis, speech recognition, and facial expression analysis technologies. This emotion engine analyzes user input data (e.g., text, speech, and facial expression data) to recognize emotions and adjusts how the summary is provided based on the recognized emotions. Specifically, if a user sends a prompt such as, "I want a summary of my recent spending. I'm worried about the content, so I'd also like some saving advice," the emotion engine analyzes the user's emotions (anxiety) and appropriately adjusts the length and level of detail of the summary.

[1169] The hardware used will include a server, the user's smartphone or PC, and, if necessary, smart glasses or a head-mounted display. The specific software used will be Python programs, the requests library, BeautifulSoup, MeCab, the gensim library, and TextBlob. This will improve the user experience by providing users with summaries that appropriately reflect their emotional state when they want to learn about their spending and transaction history.

[1170] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1171] Step 1:

[1172] The user requests a summary of their electronic payment history using a device (smartphone or PC).

[1173] Input: The user enters the prompt message: "I would like a summary of my recent spending. I am also concerned about the content, so I would like some advice on saving money."

[1174] Output: Sends a user request to the server.

[1175] Step 2:

[1176] The terminal forwards the received request to the server.

[1177] Input: User prompt text.

[1178] Output: Sends a prompt message to the server.

[1179] Step 3:

[1180] The server collects electronic payment data from a specified internet URL.

[1181] Input: The specified URL.

[1182] Output: Collected electronic payment data.

[1183] The server uses the requests library to download HTML content from the specified URL and BeautifulSoup to extract text data from the HTML tags.

[1184] Step 4:

[1185] The server performs preprocessing on the extracted text data.

[1186] Input: Extracted text data.

[1187] Output: Noise-removed and normalized text data.

[1188] Specifically, it removes special characters, multiple line breaks, and unnecessary spaces, and converts full-width spaces to half-width spaces.

[1189] Step 5:

[1190] The server performs morphological analysis on the pre-processed text data and extracts important topics.

[1191] Input: Preprocessed text data.

[1192] Output: Key topics extracted.

[1193] We use MeCab to split the text into words and then perform topic modeling using LDA or TF-IDF.

[1194] Step 6:

[1195] The server generates a summary based on the extracted topics.

[1196] Input: Key topics extracted.

[1197] Output: The generated summary.

[1198] Using natural language generation technology, a coherent text containing extracted topics is created.

[1199] Step 7:

[1200] The server saves the generated summary to the database and prepares it for delivery.

[1201] Input: Generated summary.

[1202] Output: Summary saved in the database.

[1203] Grammar checks are also performed, and the final summary is formatted and saved.

[1204] Step 8:

[1205] The emotion engine analyzes the user's emotions.

[1206] Input: User prompt (e.g., "I'd like a summary of my recent expenses. I'm also concerned about the details, so I'd like some advice on saving money.").

[1207] Output: User's emotional state (e.g., anxiety).

[1208] Use TextBlob to analyze the user's input text and obtain a sentiment score.

[1209] Step 9:

[1210] The server generates a refined summary using an emotion engine.

[1211] Input: User's emotional state, generated summary.

[1212] Output: Adjusted summary.

[1213] The length, level of detail, and presentation of the summary are adjusted according to the user's emotional state.

[1214] Step 10:

[1215] The server sends the adjusted summary to the terminal.

[1216] Input: Adjusted summary.

[1217] Output: Sends a summary to the user's terminal.

[1218] The terminal displays a summary received from the server to the user.

[1219] Through the above processing steps, a user-friendly and emotionally resonant summary of the electronic payment history is provided.

[1220] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1221] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1222] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[1223] [Fourth Embodiment]

[1224] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1225] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1226] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1227] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[1228] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1229] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1230] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1231] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[1232] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1233] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1234] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1235] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1236] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1237] This invention is a system that collects publicly available document data, particularly meeting minutes and report data from the internet, preprocesses and analyzes it, and extracts and summarizes important points from this data to provide a summary. The following describes specific embodiments for implementing this invention.

[1238] Conceptual explanation of the program

[1239] overview

[1240] This system begins by acquiring publicly available document data, preprocessing and analyzing it to extract important information, and finally generating a summary to provide to the user. The following describes the specific actions of the server, terminal, and user at each step.

[1241] Data collection

[1242] The server collects meeting minutes data from a specified internet URL using web crawling technology. For example, if the target is the minutes of a research group published by the Ministry of Internal Affairs and Communications, the server accesses the relevant web page and downloads the HTML content.

[1243] Data extraction and preprocessing

[1244] The server extracts the meeting minutes text from the downloaded HTML content. During this process, it performs noise reduction to remove HTML tags, unnecessary special characters, line breaks, etc. Next, it performs normalization, such as converting full-width spaces to half-width spaces and unifying different expressions.

[1245] Text analysis and topic extraction

[1246] The server performs morphological analysis on the preprocessed text data. A morphological analysis tool (e.g., MeCab) is used here. As a result of the analysis, the text data is divided into words, and the part of speech of each word is identified. Based on the data obtained from this analysis, important topics are extracted using topic modeling (e.g., LDA or TF-IDF).

[1247] Summary Generation

[1248] The server uses natural language generation technology to create summaries based on the extracted topics. For example, it uses a template to concisely summarize the main points in the format "XX was discussed, and YY was decided." Grammatical checks are also performed during this process to generate consistent sentences.

[1249] Saving and providing summaries

[1250] The created summary is stored in the server's database. When a user requests a summary through their device, the server searches the stored summary and sends the relevant data to the user's device. The user can then view the summary on their device.

[1251] Specific example

[1252] For example, when the Ministry of Internal Affairs and Communications provides a summary of the latest research meeting minutes, the following process occurs when a user clicks the "View summary of the latest meeting minutes" button on their device.

[1253] 1. The user clicks a button on their device.

[1254] 2. The terminal sends a summary request to the server.

[1255] 3. The server generates a summary from the latest collected meeting minutes data and saves it to the database.

[1256] 4. The server searches the saved summary and sends it to the user's terminal.

[1257] 5. The terminal displays the summary received from the server on its screen, which the user can then view.

[1258] Through this series of processes, users can efficiently obtain important information, and the system of the present invention facilitates the understanding and sharing of meeting minutes.

[1259] The following describes the processing flow.

[1260] Step 1: Data Collection

[1261] The server accesses the specified URL (for example, a public webpage of a specific government agency).

[1262] The server uses web crawling technology to download the HTML content of web pages.

[1263] The server saves the downloaded data to temporary storage.

[1264] Step 2: Data Extraction

[1265] The server runs an HTML parser to extract the meeting minutes text from the saved HTML content.

[1266] The server extracts the text data while removing HTML tags.

[1267] Step 3: Data Preprocessing

[1268] As a preprocessing step, the server removes unnecessary special characters and multiple line breaks / spaces from the text data.

[1269] The server performs normalization, such as converting full-width spaces to half-width spaces within the text.

[1270] Step 4: Morphological Analysis

[1271] The server uses a morphological analysis tool (e.g., MeCab) to divide the preprocessed text data into words.

[1272] The server tags each word with its part of speech and records the results of morphological analysis.

[1273] Step 5: Topic Extraction

[1274] The server uses topic modeling techniques (e.g., LDA and TF-IDF) to extract important topics from morphologically analyzed data.

[1275] The server evaluates the frequency and importance of words related to each topic and lists the main topics.

[1276] Step 6: Summary Generation

[1277] The server generates a summary sentence based on the extracted topics, utilizing natural language generation technology.

[1278] The server constructs the text based on a specific template (e.g., "[Topic] was discussed, and [Result] was determined.").

[1279] The server performs a grammatical check on the generated summary sentence and makes corrections as needed.

[1280] Step 7: Save the summary

[1281] The server saves the generated summary to the database.

[1282] The server also records metadata related to the summary (e.g., publication date, meeting minutes title, attendees).

[1283] Step 8: Provide a summary

[1284] The user requests a summary through their device.

[1285] The server searches the database for the relevant summary and sends it to the user's terminal.

[1286] The terminal displays the received summary to the user.

[1287] As described above, a system is realized in which the server, terminal, and user cooperate at each step to efficiently generate and provide document summaries.

[1288] (Example 1)

[1289] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1290] The sheer volume of publicly available document data, particularly meeting minutes and reports found online, makes it difficult to efficiently extract key points and provide summaries. Furthermore, manually collecting, pre-processing, analyzing, and summarizing this data is time-consuming and labor-intensive. Therefore, there is a need for a system that automatically and efficiently collects and analyzes document data, extracts key points, and provides summaries.

[1291] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1292] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary using a generative AI model based on the extracted topics, means for verifying the generated summary using a grammar checking tool, and means for saving the generated summary and providing it to the user. This enables the user to efficiently obtain important information from a vast amount of document data.

[1293] "Publicly available document data" refers to document data that is generally accessible on the internet or in other public spheres.

[1294] "Web crawling" is a technology that automatically collects data from specified web pages and websites.

[1295] "HTML content" refers to data in HTML (Hypertext Markup Language) format that represents the structure and content of a web page.

[1296] "Preprocessing" refers to a series of processes that remove noise from data and normalize it before performing data analysis.

[1297] "Noise reduction" is the process of removing elements from text data that are unnecessary for analysis (e.g., HTML tags, special characters, unnecessary line breaks).

[1298] "Normalization" is a process that unifies different representations within text data to ensure data consistency.

[1299] Morphological analysis is a technique that divides text data into the smallest possible linguistic elements and analyzes the part of speech of each element.

[1300] A "morphological analysis tool" refers to software or algorithms used to perform morphological analysis (e.g., MeCab).

[1301] Topic modeling is a technique that automatically extracts important topics and themes from large amounts of text data (e.g., LDA, TF-IDF).

[1302] A "generative AI model" is an artificial intelligence model that uses machine learning to generate natural language (e.g., GPT-3).

[1303] A "summary" is a concise summary that captures the key points of an entire document.

[1304] A "grammar check tool" is software or algorithms used to verify and correct the grammatical accuracy of generated text.

[1305] A "database" is a system that allows for the efficient storage, management, and retrieval of structured data.

[1306] A "terminal" refers to a device (e.g., a personal computer, smartphone, or tablet) that a user uses to access a server and view or manipulate data.

[1307] This invention is a system that collects publicly available document data, particularly meeting minutes and report data from the internet, preprocesses and analyzes it, and extracts and summarizes important points from this data. A detailed embodiment of this invention is described below.

[1308] Data collection

[1309] The server uses web crawling technology to collect meeting minutes data from specified internet URLs. Specifically, the server accesses URLs in a specified list sequentially, retrieves HTML content, and saves it to local storage. For example, it might access the URL of a publicly available research conference minutes page of a certain public institution and download the HTML file.

[1310] Data extraction and preprocessing

[1311] The server extracts the necessary meeting minutes text from the collected HTML content and performs preprocessing. It uses an HTML parser (e.g., Beautiful Soup) to remove HTML tags, unnecessary special characters, and line breaks. Furthermore, it performs normalization, such as converting full-width spaces to half-width spaces and unifying different expressions, as part of noise reduction. For example, the meeting minutes text " Chairperson: Thank you for joining us today. Extract "Chairman: Thank you for participating today." from the above.

[1312] Text analysis and topic extraction

[1313] The server performs morphological analysis on the preprocessed text data. For this analysis, a morphological analysis tool (e.g., MeCab) is used to divide the text into the smallest units of linguistic elements and identify the part of speech for each. For example, the sentence "Thank you for participating today." is divided into "today / noun," "wa / particle," "participation / noun," "arigato / verb," ​​"u / auxiliary verb," ​​and "gozaimasu / verb." Subsequently, topic modeling techniques (e.g., LDA or TF-IDF) are used to extract important topics from the text data. Frequently occurring keywords such as "research," "minutes," and "technology" are extracted.

[1314] Summary Generation

[1315] The server generates summaries using a generative AI model (e.g., GPT-3) based on important topics. For example, it might generate a summary such as, "The minutes mainly discussed advancements in new technologies, and future research directions were decided." The generated summaries are then checked for grammatical accuracy using a grammar checking tool and corrected as needed.

[1316] Saving and providing summaries

[1317] The generated summaries are stored in the server's database. The storage format includes fields such as "timestamp," "meeting minutes title," and "summary." When a user requests a summary through their terminal, the terminal sends a request to the server, the server searches the database for the corresponding summary, and sends it to the terminal. The terminal displays the summary received from the server on its screen, allowing the user to view it.

[1318] Specific example

[1319] For example, if a public institution is providing a summary of its latest meeting minutes, when a user clicks the "View Summary of Latest Meeting Minutes" button on their device, the following process takes place.

[1320] 1. The user clicks a button on their device.

[1321] 2. The terminal sends a summary request to the server.

[1322] 3. The server generates a summary from the latest meeting minutes data it has collected and saves it to the database.

[1323] 4. The server searches for the saved summary and sends it to the user's terminal.

[1324] 5. The terminal displays the summary received from the server on the screen, and the user views it.

[1325] As an example of a prompt, the process is carried out by inputting "Please provide a summary of the latest research conference minutes from a public institution" into the generating AI model.

[1326] This system allows users to efficiently obtain important information from vast amounts of document data.

[1327] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1328] Step 1:

[1329] The server collects publicly available document data. Specifically, the server receives a specified list of URLs as input and accesses each URL using web crawling technology. It downloads the HTML content of each URL and saves it to local storage. The output is the downloaded HTML file.

[1330] Step 2:

[1331] The server extracts text data from the collected HTML content. Specifically, it receives an HTML file as input and uses an HTML parser (e.g., Beautiful Soup) to extract the meeting minutes text. As a noise reduction process, HTML tags, unnecessary special characters, and line breaks are removed. The output is the meeting minutes text. For example, HTML content " Chairperson: Thank you for joining us today. Extract "Chairman: Thank you for participating today." from the above.

[1332] Step 3:

[1333] The server performs preprocessing on the extracted text data. Specifically, it receives the text data as input, converts full-width spaces to half-width spaces, and performs normalization to unify different expressions (e.g., AI, エーアイ). The output is normalized text data.

[1334] Step 4:

[1335] The server performs morphological analysis on the pre-processed text data. Specifically, it uses a morphological analysis tool (e.g., MeCab) to receive the text data as input, split it into words, and identify the part of speech of each word. The output is a list of the split words. For example, the sentence "Thank you for participating today." is split into "today / noun", "is / particle", "participation / noun", "thank you / verb", "u / auxiliary verb", and "gozaimasu / verb".

[1336] Step 5:

[1337] The server extracts important topics using topic modeling techniques. Specifically, it takes a list of words obtained as a result of morphological analysis as input and uses topic modeling techniques (e.g., LDA or TF-IDF) to extract important topics. The output is a list of the extracted topics.

[1338] Step 6:

[1339] The server generates a summary based on the extracted topics. Specifically, it takes a list of important topics as input and generates a summary using a generative AI model (e.g., GPT-3). The output is the generated summary. For example, it might generate a summary such as, "The meeting minutes mainly discussed the progress of new technologies, and future research directions were decided."

[1340] Step 7:

[1341] The server checks the generated summary using a grammar checking tool. Specifically, it receives the generated summary as input, uses the grammar checking tool to verify its grammatical accuracy, and makes corrections as needed. The output is the corrected summary.

[1342] Step 8:

[1343] The server saves the generated summary to the database. Specifically, it receives the modified summary as input and saves it to the database. The output is the summary saved in the database. The saving format is in the form of fields such as "timestamp," "meeting minutes title," and "summary."

[1344] Step 9:

[1345] The user requests the summary via their device. Specifically, the user clicks the "View summary of the latest meeting minutes" button on their device. This action causes the device to send a summary request to the server.

[1346] Step 10:

[1347] The server retrieves the relevant summary from the database and sends it to the user's terminal. Specifically, the server receives a summary request, retrieves the relevant summary from the database, and sends it to the terminal. The output is the sent summary.

[1348] Step 11:

[1349] The terminal displays the summary received from the server on its screen, which the user then views. Specifically, the terminal receives the summary as input and displays it on the screen. The output is a summary display that the user can view.

[1350] Therefore, users can efficiently obtain important information.

[1351] (Application Example 1)

[1352] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1353] While a vast amount of meeting minutes and reports are publicly available online, there is a lack of efficient means to collect this information, extract key points, and provide concise summaries. Traditional systems require significant time and effort for manual information gathering and summarization, and the sheer volume of information makes it highly likely that users will miss useful details. Furthermore, maintaining consistent summaries is difficult, highlighting the need for technology to appropriately extract essential information.

[1354] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1355] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary using a generative AI model based on the extracted topics, and means for storing the generated summary and providing it to a terminal. This makes it possible to efficiently extract important information from a vast amount of meeting minutes and reports published on the internet and to quickly provide high-quality summaries.

[1356] "Publicly available document data" refers to information data, including text, that is provided on the internet in a format that is publicly accessible.

[1357] "Text data" refers to information expressed as a string of characters, in a format that humans can read and understand.

[1358] "Preprocessing" refers to a series of processes performed on raw data, including noise reduction and text normalization.

[1359] "Noise" refers to unnecessary elements or strings in text data, information that is not needed for analysis.

[1360] "Normalization" is the process of transforming text data into a consistent format, unifying strings based on specific rules.

[1361] Morphological analysis is the process of dividing text data into morphemes, which are the smallest semantic elements that make up the data, and adding information such as part of speech.

[1362] A "topic" refers to the main theme or subject matter deemed important within the text data.

[1363] A "generative AI model" is an artificial intelligence that learns from large amounts of data and generates natural language text based on given conditions.

[1364] A "summary" is a short, concise summary of the main points and important aspects of a long text or document.

[1365] A "terminal" is an electronic device used by a user to receive and view information.

[1366] This invention relates to a system for efficiently collecting, preprocessing, and analyzing publicly available document data, extracting important information, and generating summaries. Specifically, this system operates in the following steps:

[1367] The server first collects document data from specified URLs on the internet. This is done using a web crawler. For example, it downloads the HTML content of a webpage and extracts specific meeting minutes data. Programming libraries used include requests and BeautifulSoup.

[1368] Next, the server preprocesses the collected document data. This includes denoising (e.g., removing unnecessary HTML tags and special characters) and text normalization (e.g., converting full-width spaces to half-width spaces). Python string manipulation functions are used for this process.

[1369] Subsequently, the server analyzes the pre-processed text data and performs morphological analysis. Specifically, it uses the morphological analysis tool janome.tokenizer. This analysis identifies the words and their parts of speech that make up the text data.

[1370] Furthermore, the server extracts topics from the analyzed data. Here, topics are extracted using TF-IDF (Inverse Document Frequency) and LDA (Latent Dirichlet Allocation). In this process, the libraries TfidfVectorizer and LatentDirichletAllocation are used.

[1371] Based on the extracted topics, the server generates a summary using a generative AI model. This generative AI model utilizes the OpenAI API (e.g., GPT-3). An example of a prompt for this model is shown below.

[1372] Please generate a meeting minutes summary based on the following topics:

[1373] Topics: Digitalization, Policy, Budget, AI, Data Analysis

[1374] Summary:

[1375] Based on this prompt, the AI ​​model generates a natural and consistent summary.

[1376] The generated summary is stored in a database and provided to the user's device. Users can retrieve and view the meeting minutes summary by using an application on their device and entering a specific URL. This entire process allows for the quick and efficient acquisition of important information without requiring significant time and effort.

[1377] The hardware required to implement this system consists of an internet-connected server and user terminals. The software uses the Python programming language and its libraries, BeautifulSoup, janome, TfidfVectorizer, LatentDirichletAllocation, and the OpenAI API.

[1378] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1379] Step 1:

[1380] The server collects document data from an internet URL specified by the user. In this case, when the user enters a specific URL on their terminal, that information is sent to the server. The server uses the requests library to access the specified URL and download the HTML content of the webpage. The input to this process is the URL specified by the user, and the output is the downloaded HTML content.

[1381] Step 2:

[1382] The server extracts text data from the HTML content collected in Step 1. It uses the BeautifulSoup library to parse HTML tags and extract the text portion. This removes HTML tags and unnecessary special characters. The input is the collected HTML content, and the output is the extracted text data.

[1383] Step 3:

[1384] The server performs preprocessing on the extracted text data. Preprocessing includes noise reduction (removing unnecessary characters) and normalization (converting full-width spaces to half-width spaces, unifying different expressions, etc.). Python string manipulation functions are used at this stage. The input is the extracted text data, and the output is the preprocessed text data.

[1385] Step 4:

[1386] The server performs morphological analysis on the preprocessed text data. During this process, it uses janome.tokenizer to split the text data into individual words and identify the part of speech for each word. The input is the preprocessed text data, and the output is the split data with its part of speech information.

[1387] Step 5:

[1388] The server extracts important topics based on the data analyzed in step 4. Here, it uses TfidfVectorizer to calculate the importance of each topic and then models the main topics from the entire text using LatentDirichletAllocation. The input is data segmented word by word and its part-of-speech information, and the output is the extracted important topics.

[1389] Step 6:

[1390] The server generates a summary using a generative AI model based on the extracted topics. Here, the OpenAI API (e.g., GPT-3) is used to form prompt statements and send them to the AI ​​model. A concrete example of a generated prompt statement is as follows:

[1391] Please generate a meeting minutes summary based on the following topics:

[1392] Topics: Digitalization, Policy, Budget, AI, Data Analysis

[1393] Summary:

[1394] The input consists of the extracted key topics, and the output is the generated summary.

[1395] Step 7:

[1396] The server stores the generated summary in a database and provides it to the user's terminal upon request. When a user requests a summary from their terminal, the server retrieves the corresponding summary from the database and sends it to the user's terminal. The input is the user's request and the generated summary, and the output is the summary sent to the user's terminal.

[1397] Through the steps described above, a system is realized that allows users to efficiently and quickly obtain important summary information from document information on the internet.

[1398] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1399] This invention is a system that collects, preprocesses, and analyzes publicly available document data to extract important information and generate summaries, and also incorporates an emotion engine that recognizes the user's emotions. This configuration allows for dynamic adjustment of the optimal summary delivery method according to the user's emotional state.

[1400] Conceptual explanation of the program

[1401] overview

[1402] This system collects publicly available document data by web crawling, preprocesses it, performs morphological analysis and topic extraction, and then automatically generates summaries. In addition to this basic summary generation process, it uses a sentiment engine to analyze user sentiment and determine the optimal way to provide summaries. The specific actions of the server, terminal, and user at each step are described below.

[1403] Data collection and extraction

[1404] The server collects meeting minutes data from specified internet URLs using web crawling technology. For example, it downloads HTML content from the public web pages of specific government agencies. This content is temporarily stored on the server.

[1405] Next, the server extracts the meeting minutes text from the saved HTML content. Specifically, it uses an HTML parser to remove HTML tags and unnecessary special characters, and then extracts the necessary text data.

[1406] Data preprocessing and analysis

[1407] As a preprocessing step, the server removes unnecessary noise (special characters, multiple line breaks and spaces, etc.) from the extracted text data and performs normalization by converting full-width spaces to half-width spaces.

[1408] Next, morphological analysis is performed on the preprocessed text data. A morphological analysis tool (e.g., MeCab) is used to divide the text data into words, and each word is tagged with its part of speech. Based on this analyzed data, topic modeling techniques (e.g., LDA or TF-IDF) are used to extract important topics within the text. The extracted topics are listed and used in the subsequent summary generation process.

[1409] Summary generation and saving

[1410] The server uses natural language generation technology to create a summary based on the extracted topics. After generating a consistent sentence using a template (e.g., "[Topic] was discussed, and [Result] was decided."), it performs a grammatical check and saves the final summary to the database. At this time, metadata related to the summary (e.g., publication date, agenda title, attendees) is also recorded.

[1411] Emotion Engine Additions and Features

[1412] When a user requests a summary through their device, the emotion engine analyzes the user's input data (e.g., text, voice, facial expression data) to recognize their emotions. Based on the analyzed emotion data, the emotion engine determines the optimal way to provide the summary (e.g., text length, expression style, level of detail). This emotion-based adjustment ensures that the summary is provided in a way that is most easily understood and accepted by the user.

[1413] As a concrete example, consider a case where a user requests to "check the summary of the latest research conference minutes." The process in this case would be as follows:

[1414] 1. The user clicks the button to request a summary on their device.

[1415] 2. The terminal sends the user's request to the server.

[1416] 3. The server retrieves the latest summary from the database and has the emotion engine analyze the user's emotion data.

[1417] 4. The emotion engine adjusts how the summary is provided based on the user's emotion information (input data at the time of the request).

[1418] 5. The server sends the adjusted summary to the user's terminal.

[1419] 6. The terminal displays the summary received from the server to the user.

[1420] This system provides an optimal summary tailored to the user's emotional state, improving user understanding and satisfaction.

[1421] The following describes the processing flow.

[1422] Step 1: Data Collection

[1423] The server accesses a specified internet URL (e.g., a public webpage of a government agency) and downloads the HTML content of the webpage using web crawling technology.

[1424] The server saves the downloaded HTML content to temporary storage.

[1425] Step 2: Data Extraction

[1426] The server runs an HTML parser to extract the meeting minutes text from the saved HTML content.

[1427] The server removes HTML tags and extracts the necessary text data.

[1428] Step 3: Data Preprocessing

[1429] As a preprocessing step, the server performs noise reduction by removing unnecessary special characters and multiple line breaks / spaces from the text data.

[1430] The server performs a normalization process that converts full-width spaces to half-width spaces.

[1431] Step 4: Morphological Analysis

[1432] The server uses a morphological analysis tool (e.g., MeCab) to divide the preprocessed text data into words and tag each word with its part of speech.

[1433] The server lists the morphological analysis results.

[1434] Step 5: Topic Extraction

[1435] The server uses topic modeling techniques (e.g., LDA and TF-IDF) to extract important topics from morphologically analyzed data.

[1436] The server evaluates the frequency and importance of related words for each topic and lists the main topics.

[1437] Step 6: Summary Generation

[1438] The server uses natural language generation technology to generate summary sentences based on the extracted topics.

[1439] The server constructs consistent sentences based on a template (e.g., "[Topic] was discussed, and [Result] was determined.").

[1440] The server performs a grammatical check on the generated summary sentence and makes corrections as needed.

[1441] Step 7: Save the summary

[1442] The server saves the generated summary to the database.

[1443] The server also records metadata related to the summary (e.g., publication date, agenda title, attendees).

[1444] Step 8: Emotion Recognition

[1445] The user sends a summary request through their device.

[1446] The terminal collects user input data (e.g., text, voice, facial expression data) and sends it to the server.

[1447] The server uses an emotion engine to analyze the transmitted user data and recognize the user's emotions.

[1448] Step 9: Emotion-based summary adjustment

[1449] The server adjusts how the summary is provided (e.g., text length, format, level of detail) based on the analysis results of the emotion engine.

[1450] The server prepares the adjusted summary in the optimal format.

[1451] Step 10: Provide a summary

[1452] The server sends the adjusted summary to the user's terminal.

[1453] The terminal displays the received summary to the user.

[1454] Users view and utilize summaries optimized based on their emotions.

[1455] Through each of the above steps, the system efficiently generates and adjusts document data summaries and provides them in an appropriate format that suits the user's emotions.

[1456] (Example 2)

[1457] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1458] Conventional text summary generation systems have the problem of not being able to consider user sentiment when collecting publicly available document data and generating summaries, making it difficult to provide summaries that are optimal for the user. In addition, the method of providing summaries is fixed, which is a problem as it does not sufficiently improve user understanding and satisfaction.

[1459] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1460] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary based on the extracted topics, means for saving the generated summary and providing it to the user, and means for analyzing the user's sentiment information and adjusting the method of providing the summary based on the analyzed sentiment information. This makes it possible to provide an optimal summary that responds to the user's sentiment, thereby improving the user's understanding and satisfaction.

[1461] "Publicly available document data" refers to document information that is provided in a format accessible to anyone on the internet or in other public places.

[1462] "Means of collection" refers to technologies and methods that have the function of automatically collecting publicly available document data from the internet using specific algorithms or tools.

[1463] "Text data" refers to string information extracted from collected document data, and is the basic unit that constitutes the content of a document.

[1464] "Preprocessing" refers to the process of removing unnecessary information (special characters, spaces, line breaks, etc.) from text data and preparing the data for analysis.

[1465] "Normalization" refers to the operation of standardizing irregular elements in text data (such as converting full-width spaces to half-width spaces) during preprocessing.

[1466] A "morpheme" is the smallest semantic unit (word or phrase) that makes up text data, and is an element extracted through morphological analysis.

[1467] "Morphological analysis" refers to the process of dividing text data into morphemes, tagging each morpheme with its part of speech, and then analyzing it.

[1468] A "topic" refers to an important subject or subject matter within document data, extracted through morphological analysis or topic modeling.

[1469] "Means of extracting topics" refers to technologies and methods that use text analysis techniques to identify important subjects within a document and list them.

[1470] A "summary" refers to a short document that concisely summarizes the main points of the entire document based on the extracted topics.

[1471] "Means for generating summaries" refers to technologies and methods that have the function of creating a consistent summary sentence using natural language generation technology based on extracted topics.

[1472] "Means of providing to the user" refers to technologies and methods that have the functionality to send the generated summary to the user's terminal and display it.

[1473] "Emotional information" refers to emotional states and reactions analyzed from user input data (text, voice, facial expressions, etc.).

[1474] "Means of analyzing emotional information" refers to technologies and methods that analyze user input data and identify the emotions contained within it.

[1475] "Means of adjusting the delivery method" refers to technologies and methods that have the function of optimizing the content, format, and expression of the summary based on the analyzed emotional information.

[1476] The present invention is a system that collects, preprocesses, and analyzes publicly available document data to extract important information and generate summaries, and also incorporates an emotion engine that recognizes user emotions. Specific embodiments of the present invention are described below.

[1477] Data collection and extraction

[1478] The server uses web crawling technology to collect publicly available document data from specified internet URLs. For example, it downloads HTML content from a specific public institution's webpage and stores it temporarily. The server then uses an HTML parser (such as Beautiful Soup) to remove unnecessary tags and special characters from the downloaded HTML content and extract the necessary text data.

[1479] Data preprocessing

[1480] The server performs normalization on the extracted text data. Specifically, it removes special characters and unnecessary whitespace, and converts full-width spaces to half-width spaces. This prepares the text data for analysis.

[1481] Text analysis and topic extraction

[1482] The server uses a morphological analysis tool (e.g., MeCab) to segment the normalized text data into words and tags each word with its part of speech. The server then uses topic modeling techniques such as LDA (Latent Dirichlet Allocation) and TF-IDF (Term Frequency-Inverse Document Frequency) to extract important topics from the text.

[1483] Summary generation and saving

[1484] The server utilizes natural language generation technology based on the extracted topics to generate consistent summaries. It uses templates (e.g., "[Topic] was discussed, and [Result] was decided.") to create the summary text. The generated summaries undergo grammatical checks and are then stored in a database along with relevant metadata (publication date, agenda title, attendees, etc.).

[1485] Emotion engine and summary provided

[1486] When a user requests a summary using their device, the device sends the request to the server. The server retrieves the latest summary from the database and has the emotion engine analyze the user's input data (text, voice, facial expression data). Based on the analysis results, the emotion engine adjusts how the summary is provided (e.g., text length, expression style, level of detail). The server sends the adjusted summary to the user's device, which then displays it to the user.

[1487] Specific example

[1488] For example, if a user requests to "check the summary of the latest research conference minutes," the processing flow is as follows:

[1489] 1. The user clicks the "Request Summary" button on their device.

[1490] 2. The device sends a request to the server.

[1491] 3. The server retrieves the latest summary from the database and has the emotion engine analyze the user's emotion data.

[1492] 4. The emotion engine adjusts how the summary is provided based on the user's emotional information.

[1493] 5. The server sends the adjusted summary to the user's terminal.

[1494] 6. The device displays Sally, and the user confirms it.

[1495] Example of a prompt

[1496] The following are examples of prompts to input into a generative AI model:

[1497] I would like to review summaries of the latest research conference proceedings. Please analyze the crawled conference data and generate summaries based on key topics. Additionally, please adjust the summary presentation method based on user sentiment data (text, voice, and facial expression data).

[1498] As described above, the present invention can improve user understanding and satisfaction by providing an optimal summary that takes user emotions into consideration.

[1499] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1500] Step 1:

[1501] Data collection

[1502] The server uses web crawling technology to collect publicly available document data from specified URLs. As input, the URL of a specific website is provided to the crawler, and as output, the HTML content is temporarily stored on the server.

[1503] Specifically, the crawler accesses a URL, downloads HTML data in real time, and stores it in temporary storage.

[1504] Step 2:

[1505] HTML parsing and text extraction

[1506] The server extracts the necessary text data from the stored HTML content. The stored HTML data is given as input, and the text data is generated as output.

[1507] Specifically, the server uses an HTML parser (for example, Beautiful Soup) to remove HTML tags and unnecessary special characters and extract the text. The extracted text is then stored in a new variable.

[1508] Step 3:

[1509] Normalization process

[1510] The server performs normalization on the extracted text data. The extracted text data is given as input, and normalized text data is generated as output.

[1511] Specifically, the server uses regular expressions to remove special characters and unnecessary whitespace, and converts full-width spaces to half-width spaces. This prepares the text data for parsing.

[1512] Step 4:

[1513] Morphological analysis

[1514] The server performs morphological analysis on normalized text data. Normalized text data is provided as input, and the morphological analysis results are generated as output.

[1515] Specifically, the server uses a morphological analysis tool (for example, MeCab) to split the text data into words and tag each word with its part of speech.

[1516] Step 5:

[1517] Topic extraction

[1518] The server extracts topics based on the morphological analysis results. The input is the morphological analysis results, and the output is a list of the extracted topics.

[1519] In terms of specific operations, the server executes topic modeling algorithms such as LDA (Latent Dirichlet Allocation) and TF-IDF (Term Frequency-Inverse Document Frequency) to extract important topics.

[1520] Step 6:

[1521] Summary Generation

[1522] The server generates a summary based on the extracted topics. The extracted topics are given as input, and the generated summary is provided as output.

[1523] In practice, the server uses a natural language generation algorithm and a template (for example, "[Topic] was discussed, and [Result] was determined.") to create a consistent summary text.

[1524] Step 7:

[1525] Save summary

[1526] The server saves the generated summary. The generated summary and associated metadata (e.g., publication date, agenda title, attendees) are given as input, and the output is saved to the database.

[1527] Specifically, the server records summary text and metadata in the database.

[1528] Step 8:

[1529] Summary Request Acceptance

[1530] The user requests a summary through their device. The user's request is sent to the device as input, and the device then sends a request to the server.

[1531] Specifically, the user clicks the "Request Summary" button on their device. The device then sends this request to the server.

[1532] Step 9:

[1533] Emotion analysis

[1534] The server analyzes the received request and user sentiment data. User sentiment data (text, voice, facial expressions) is given as input, and sentiment analysis results are generated as output.

[1535] Specifically, the server uses an emotion engine to analyze emotional information from the user's input data and retrieves the results.

[1536] Step 10:

[1537] Adjustment of summary delivery method

[1538] The server adjusts the way the summary is provided based on the sentiment analysis results. The sentiment analysis results are given as input, and the adjusted summary is generated as output.

[1539] Specifically, the server adjusts the format and length of the summary text based on the sentiment analysis results.

[1540] Step 11:

[1541] Send Summary

[1542] The server sends the adjusted summary to the user terminal. The adjusted summary is sent from the server as input and received by the user terminal as output.

[1543] Specifically, the server uses a protocol to send the adjusted summary to the user's terminal.

[1544] Step 12:

[1545] Summary display

[1546] The terminal displays a summary to the user. The adjusted summary is received by the terminal as input and displayed to the user as output.

[1547] Specifically, the terminal displays the received summary on the user interface, allowing the user to review it.

[1548] (Application Example 2)

[1549] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1550] The vast amount of electronic payment data available today can be difficult for users to understand, and information provided without considering emotions can negatively impact the user experience. For example, simply providing a payment history is insufficient for a user who feels anxious about high spending after shopping. While there is a need for appropriate information tailored to the user's emotions, no system currently exists to achieve this. Therefore, a system is needed that recognizes the user's emotions and provides an optimal summary of their payment history based on those emotions.

[1551] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1552] In this invention, the server includes means for collecting publicly available document data, means for extracting text data from the collected document data, means for preprocessing the extracted text data to remove noise and normalize the text, means for analyzing morphemes from the preprocessed text data and extracting important topics, means for generating a summary based on the extracted topics, means for storing the generated summary and providing it to the user, means for recognizing the user's emotions, and means for adjusting the method of providing the summary based on the recognized emotions. This makes it possible to provide an optimal summary that responds to the user's emotions, thereby improving the user experience.

[1553] "Publicly available document data" refers to document data that is provided in a form accessible to anyone on the internet or other public environments.

[1554] "Means of collection" refers to methods and devices for obtaining document data from specified URLs or sources using technologies such as web crawling and HTTP requests.

[1555] "Means of extraction" refers to methods or devices used to extract necessary text information from collected data, such as HTML parsers or text analysis tools.

[1556] "Preprocessing" refers to the process of removing noise from data, normalizing it, and converting its format, essentially organizing and correcting the data before analysis.

[1557] "Noise" refers to unnecessary data or information that should be excluded from the analysis, such as special characters, multiple line breaks, and extra whitespace.

[1558] "Normalization" refers to the process of unifying the format and notation of data, such as converting full-width spaces to half-width spaces to conform to a standard format.

[1559] A "morpheme" is a fundamental unit that makes up text, such as a word or phrase—the smallest element that possesses meaning.

[1560] "Morphological analysis" refers to the process of dividing text data into words and phrases and tagging each with its part of speech.

[1561] "Important topics" refer to themes and topics that frequently appear in the text data, and are particularly useful or interesting to the user.

[1562] A "summary" is a concise compilation of key points extracted from lengthy texts or complex data, presented in a way that users can quickly understand.

[1563] "Means for generating summaries" refers to methods or devices that use topic modeling or natural language generation techniques to create consistent, short sentences from extracted information.

[1564] "Means of providing to the user" refers to methods and devices for presenting the generated summary in a form accessible to the user, such as smartphone apps and web interfaces.

[1565] "Means for recognizing user emotions" refers to methods and devices for detecting a user's emotional state using text analysis, speech recognition, facial expression analysis, etc.

[1566] "Means for adjusting the method of providing summaries based on recognized emotions" refers to methods or devices that receive user emotion information as analysis results and dynamically change the length, level of detail, and format of the summary provided.

[1567] The system for realizing this invention includes the following means: A server collects publicly available document data and uses an HTML parser and HTTP request technology to extract text data from the collected document data. Then, it preprocesses the extracted text data, removing noise and using MeCab as a tool to normalize the text. Topic modeling techniques such as LDA and TF-IDF are used to analyze morphemes from the preprocessed text data and extract important topics. When generating a summary based on the extracted topics, natural language generation technology is used, and the generated summary is stored in a database and includes an interface for providing it to the user.

[1568] Furthermore, to recognize user emotions, the system incorporates an emotion engine that uses text analysis, speech recognition, and facial expression analysis technologies. This emotion engine analyzes user input data (e.g., text, speech, and facial expression data) to recognize emotions and adjusts how the summary is provided based on the recognized emotions. Specifically, if a user sends a prompt such as, "I want a summary of my recent spending. I'm worried about the content, so I'd also like some saving advice," the emotion engine analyzes the user's emotions (anxiety) and appropriately adjusts the length and level of detail of the summary.

[1569] The hardware used will include a server, the user's smartphone or PC, and, if necessary, smart glasses or a head-mounted display. The specific software used will be Python programs, the requests library, BeautifulSoup, MeCab, the gensim library, and TextBlob. This will improve the user experience by providing users with summaries that appropriately reflect their emotional state when they want to learn about their spending and transaction history.

[1570] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1571] Step 1:

[1572] The user requests a summary of their electronic payment history using a device (smartphone or PC).

[1573] Input: The user enters the prompt message: "I would like a summary of my recent spending. I am also concerned about the content, so I would like some advice on saving money."

[1574] Output: Sends a user request to the server.

[1575] Step 2:

[1576] The terminal forwards the received request to the server.

[1577] Input: User prompt text.

[1578] Output: Sends a prompt message to the server.

[1579] Step 3:

[1580] The server collects electronic payment data from a specified internet URL.

[1581] Input: The specified URL.

[1582] Output: Collected electronic payment data.

[1583] The server uses the requests library to download HTML content from the specified URL and BeautifulSoup to extract text data from the HTML tags.

[1584] Step 4:

[1585] The server performs preprocessing on the extracted text data.

[1586] Input: Extracted text data.

[1587] Output: Noise-removed and normalized text data.

[1588] Specifically, it removes special characters, multiple line breaks, and unnecessary spaces, and converts full-width spaces to half-width spaces.

[1589] Step 5:

[1590] The server performs morphological analysis on the pre-processed text data and extracts important topics.

[1591] Input: Preprocessed text data.

[1592] Output: Key topics extracted.

[1593] We use MeCab to split the text into words and then perform topic modeling using LDA or TF-IDF.

[1594] Step 6:

[1595] The server generates a summary based on the extracted topics.

[1596] Input: Key topics extracted.

[1597] Output: The generated summary.

[1598] Using natural language generation technology, a coherent text containing extracted topics is created.

[1599] Step 7:

[1600] The server saves the generated summary to the database and prepares it for delivery.

[1601] Input: Generated summary.

[1602] Output: Summary saved in the database.

[1603] Grammar checks are also performed, and the final summary is formatted and saved.

[1604] Step 8:

[1605] The emotion engine analyzes the user's emotions.

[1606] Input: User prompt (e.g., "I'd like a summary of my recent expenses. I'm also concerned about the details, so I'd like some advice on saving money.").

[1607] Output: User's emotional state (e.g., anxiety).

[1608] Use TextBlob to analyze the user's input text and obtain a sentiment score.

[1609] Step 9:

[1610] The server generates a refined summary using an emotion engine.

[1611] Input: User's emotional state, generated summary.

[1612] Output: Adjusted summary.

[1613] The length, level of detail, and presentation of the summary are adjusted according to the user's emotional state.

[1614] Step 10:

[1615] The server sends the adjusted summary to the terminal.

[1616] Input: Adjusted summary.

[1617] Output: Sends a summary to the user's terminal.

[1618] The terminal displays a summary received from the server to the user.

[1619] Through the above processing steps, a user-friendly and emotionally resonant summary of the electronic payment history is provided.

[1620] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1621] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1622] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[1623] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1624] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[1625] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[1626] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[1627] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[1628] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[1629] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[1630] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[1631] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[1632] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[1633] 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.

[1634] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[1635] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[1636] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[1637] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[1638] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[1639] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[1640] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[1641] The following is further disclosed regarding the embodiments described above.

[1642] (Claim 1)

[1643] Means for collecting publicly available document data,

[1644] A method for extracting text data from collected document data,

[1645] A method for preprocessing extracted text data to remove noise and normalize the text,

[1646] A means for analyzing morphemes from preprocessed text data and extracting important topics,

[1647] A means for generating a summary based on extracted topics,

[1648] A means of saving the generated summary and providing it to the user,

[1649] A system that includes this.

[1650] (Claim 2)

[1651] The system according to claim 1, which collects publicly available document data by crawling from the web.

[1652] (Claim 3)

[1653] The system according to claim 1, which analyzes text data using a morphological analysis tool.

[1654] "Example 1"

[1655] (Claim 1)

[1656] Means for collecting publicly available document data,

[1657] A method for extracting text data from collected document data,

[1658] A method for preprocessing extracted text data to remove noise and normalize the text,

[1659] A means for analyzing morphemes from preprocessed text data and extracting important topics,

[1660] A means of generating a summary using a generative AI model based on extracted topics,

[1661] A method for checking the generated summary using a grammar checking tool,

[1662] A means of saving the generated summary and providing it to the user,

[1663] A system that includes this.

[1664] (Claim 2)

[1665] The system according to claim 1, comprising means for crawling and collecting publicly available document data from the web.

[1666] (Claim 3)

[1667] The system according to claim 1, comprising means for analyzing text data using a morphological analysis tool.

[1668] "Application Example 1"

[1669] (Claim 1)

[1670] Means for collecting publicly available document data,

[1671] A method for extracting text data from collected document data,

[1672] A method for preprocessing extracted text data to remove noise and normalize the text,

[1673] A means for analyzing morphemes from preprocessed text data and extracting important topics,

[1674] A means of generating a summary using a generative AI model based on extracted topics,

[1675] A means of saving the generated summary and providing it to the terminal,

[1676] A system that includes this.

[1677] (Claim 2)

[1678] The system according to claim 1, which collects publicly available document data by crawling it from the internet.

[1679] (Claim 3)

[1680] The system according to claim 1, which analyzes text data using a morphological analysis tool.

[1681] "Example 2 of combining an emotion engine"

[1682] (Claim 1)

[1683] Means for collecting publicly available document data,

[1684] A method for extracting text data from collected document data,

[1685] A method for preprocessing extracted text data to remove noise and normalize the text,

[1686] A means for analyzing morphemes from preprocessed text data and extracting important topics,

[1687] A means for generating a summary based on extracted topics,

[1688] A means of saving the generated summary and providing it to the user,

[1689] A means for analyzing user sentiment information and adjusting the method of providing summaries based on the analyzed sentiment information,

[1690] A system that includes this.

[1691] (Claim 2)

[1692] The system according to claim 1, which collects publicly available document data by crawling from the web.

[1693] (Claim 3)

[1694] The system according to claim 1, which analyzes text data using a morphological analysis tool.

[1695] "Application example 2 when combining with an emotional engine"

[1696] (Claim 1)

[1697] Means for collecting publicly available document data,

[1698] A method for extracting text data from collected document data,

[1699] A method for preprocessing extracted text data to remove noise and normalize the text,

[1700] A means for analyzing morphemes from preprocessed text data and extracting important topics,

[1701] A means for generating a summary based on extracted topics,

[1702] A means of saving the generated summary and providing it to the user,

[1703] Means of recognizing user emotions,

[1704] Means for adjusting the way summaries are provided based on perceived emotions,

[1705] A system that includes this.

[1706] (Claim 2)

[1707] The system according to claim 1, which collects publicly available document data by crawling from the web.

[1708] (Claim 3)

[1709] The system according to claim 1, which analyzes text data using a morphological analysis tool. [Explanation of symbols]

[1710] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. Means for collecting publicly available document data, A method for extracting text data from collected document data, A method for preprocessing extracted text data to remove noise and normalize the text, A means for analyzing morphemes from preprocessed text data and extracting important topics, A means for generating a summary based on extracted topics, A means of saving the generated summary and providing it to the user, A system that includes this.

2. The system according to claim 1, which collects publicly available document data by crawling from the web.

3. The system according to claim 1, which analyzes text data using a morphological analysis tool.

Images