System

Abstract

An object of a system according to an embodiment is to efficiently examine comments collected from a plurality of platforms on the Internet and identify a user who leaves a dangerous comment.SOLUTION: A system according to an embodiment includes a comment collection unit, a comment review unit, a comment classification unit, and a user extraction unit. The comment collection unit collects comments from a plurality of platforms on the Internet. The comment reviewing unit reviews the comments collected by the comment collecting unit. The comment classifying unit classifies the comments reviewed by the comment reviewing unit. The user extraction unit extracts users who have left dangerous comments from among the comments classified by the comment classification unit.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] Conventional technologies have had the problem of making it difficult to efficiently scrutinize comments collected from multiple platforms on the Internet and identify users who leave dangerous comments.

[0005] The system according to the embodiment aims to efficiently scrutinize comments collected from multiple platforms on the Internet and identify users who leave dangerous comments. [Means for solving the problem]

[0006] The system according to the embodiment includes a comment collection unit, a comment scrutiny unit, a comment classification unit, and a user extraction unit. The comment collection unit collects comments from multiple platforms on the Internet. The comment scrutiny unit scrutinizes the comments collected by the comment collection unit. The comment classification unit classifies the comments scrutinized by the comment scrutiny unit. The user extraction unit extracts users who leave dangerous comments from among the comments classified by the comment classification unit. [Effects of the Invention]

[0007] The system according to the embodiment can efficiently scrutinize comments collected from multiple platforms on the Internet and identify users who leave dangerous comments. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. DETAILED DESCRIPTION OF THE INVENTION

[0009] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0010] First, the terms used in the following description will be explained.

[0011] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, the processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), or a TPU (Tensor Processing Unit).

[0012] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0013] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0014] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), and Bluetooth (registered trademark).

[0015] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0016] [First embodiment] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0017] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0018] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0019] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0020] The reception device 38 includes a touch panel 38A and a microphone 38B, and receives user input. The touch panel 38A detects contact with a pointer (for example, a pen or a finger) to receive user input by the touch of the pointer. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 (see FIG. 2) acquires the data indicating the user input.

[0021] Output device 40 includes a display 40A and a speaker 40B, and presents data to a user by outputting the data in a form of expression that the user can perceive (e.g., audio and / or text). Display 40A displays visible information such as text and images in accordance with instructions from processor 46. Speaker 40B outputs audio in accordance with instructions from processor 46. Camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0022] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0023] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0024] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0025] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate a user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotion, including estimation and prediction of the user's emotion, but is not limited to these examples. Furthermore, the estimation and prediction of emotion also includes, for example, emotion analysis.

[0026] In the smart device 14, the specific processing is performed by the processor 46. The storage 50 stores a specific processing program 60. The specific processing program 60 is used together with the specific processing program 56 by the data processing system 10. The processor 46 reads the specific processing program 60 from the storage 50 and executes the read specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as the control unit 46A in accordance with the specific processing program 60 executed on the RAM 48. Note that the smart device 14 has a data generation model and an emotion identification model similar to the data generation model 58 and the emotion identification model 59, and can also perform processing similar to that of the specific processing unit 290 using these models.

[0027] Note that a device other than the data processing device 12 may have the data generation model 58. For example, a server device (e.g., a generation server) may have the data generation model 58. In this case, the data processing device 12 obtains a processing result (prediction result, etc.) using the data generation model 58 by communicating with the server device having the data generation model 58. Furthermore, the data processing device 12 may be a server device, or may be a terminal device owned by a user (e.g., a mobile phone, a robot, a home appliance, etc.). Next, an example of processing by the data processing system 10 according to the first embodiment will be described.

[0028] (Example 1) An AI system according to an embodiment of the present invention is a system that scrutinizes anti-comments against companies and individuals, picks out slanderous and violent comments, and checks for and determines whether they contain dangerous content. This allows the AI ​​system to efficiently scrutinize anti-comments against companies and individuals and pick out slanderous and violent comments. It can also identify users who leave dangerous comments and take appropriate measures.

[0029] The AI ​​system according to the embodiment includes a comment collection unit, a comment scrutiny unit, a comment classification unit, and a user extraction unit. The comment collection unit collects comments from multiple platforms on the Internet. For example, it automatically collects comments from social media, blogs, bulletin boards, and the like. The comment collection unit also uses a generative AI to estimate the poster's emotions in real time when collecting comments, and prioritizes the collection of comments that express particularly negative emotions. For example, it analyzes the poster's emotions from their writing style and vocabulary to collect comments that express strong feelings of anger or sadness. The comment collection unit also expands the platforms it collects comments from, and collects comments from the dark web and anonymous bulletin boards. For example, it automatically collects comments related to specific keywords or topics and stores them in a database. The comment collection unit also analyzes the metadata of the collected comments (e.g., posting time, posting location, posting device) to identify trends in anti-comments during specific time periods or locations. For example, it identifies time periods or areas where anti-comments are most prevalent based on the posting time and posting location. The comment collection unit also extracts text from audio comments and video comments and collects them as anti-comments. For example, speech recognition technology is used to convert voice comments into text and collect comments containing negative content. The comment collection unit also automatically translates comments in different languages ​​to collect anti-comments from a global perspective. For example, comments in English, French, Chinese, etc. are automatically translated to collect comments containing negative content. The comment collection unit also uses emotion estimation functionality to analyze the emotional tone of the collected comments and prioritize the collection of comments that show particularly strong emotional reactions. For example, comments containing strong emotions such as anger or sadness are collected. The comment review unit then reviews the collected comments. For example, generative AI is used to understand the context of the comments and conduct a context-based review rather than simple keyword matching. For example, the context before and after the comment is analyzed to identify comments containing defamatory or violent content. The comment review unit also refers to a database of similar past comments when reviewing comments to improve accuracy. For example, new comments are reviewed based on past defamatory or violent comments.Furthermore, the comment screening unit uses emotion estimation to evaluate the emotional impact of the screened comments and classify comments that have a particularly strong emotional impact. For example, it classifies comments that express strong emotions such as anger or sadness. The comment screening unit also analyzes text within images and videos to classify them as anti-comments. For example, it extracts text from images and classifies comments that contain defamatory or violent content. Furthermore, the comment screening unit considers expressions unique to different cultures and regions and screens and classifies based on cultural background. For example, it analyzes slang and expressions used in specific cultures and regions to classify comments that contain negative content. The comment classification unit classifies the screened comments. For example, it uses generative AI to analyze the emotional tone of the comments and prioritizes classifying comments that show particularly strong emotional reactions. For example, it classifies comments that express strong emotions such as anger or sadness. The user extraction unit extracts users who have left dangerous comments from the classified comments. For example, generative AI is used to analyze the intent behind the selected comments and evaluate their actual risk, rather than mere slander. For example, the context and wording of the comment are analyzed to assess the likelihood of actual action being taken. The user extraction unit also analyzes the past behavioral history of the poster of the selected comment to assess the likelihood of recidivism. For example, the likelihood of recidivism is assessed based on a history of past postings of slanderous or violent comments. Furthermore, the user extraction unit uses emotion estimation to evaluate the emotional impact of the selected comments and prioritizes comments that have a particularly strong emotional impact. For example, comments that express strong emotions such as anger or sadness are selected. The user extraction unit also identifies the groups or communities behind the selected comments and detects organized slander. For example, if a specific group or community is engaged in a coordinated slanderous activity, this can be identified as organized activity. Furthermore, the user extraction unit cross-references the selected comments across different platforms to identify consistent and dangerous comments. For example, if the same comment is posted on multiple platforms, the unit identifies that comment.As a result, the AI ​​system according to the embodiment can efficiently scrutinize anti-comments against companies or individuals and pick out slanderous or violent comments. It can also identify users who leave dangerous comments and take appropriate measures. For example, when proposing countermeasures, the output unit takes into account the social network of the dangerous user and takes measures against other related users. To take the social network into account, the generation AI analyzes the relationships between users. For example, it analyzes the behavior of users who belong to the same group or community and takes measures against other related users. To take the social network into account when proposing countermeasures, the generation AI uses social network analysis technology. For example, it analyzes the connections and relationships between users and takes measures against other related users.

[0030] The comment collection unit can expand the platforms it collects to also collect comments from the dark web and anonymous message boards. The comment collection unit, for example, uses generation AI to collect comments from the dark web and anonymous message boards. For example, it automatically collects comments related to specific keywords or topics and stores them in a database. To expand the platforms it collects, the generation AI also collects comments from multiple sources. For example, it collects comments from social media, blogs, message boards, the dark web, etc., and analyzes them comprehensively. Furthermore, to efficiently collect comments from the dark web and anonymous message boards, the generation AI uses specific crawling techniques. For example, it automatically collects comments related to specific topics or keywords and stores them in a database. This expands the platforms it collects, thereby collecting a wider range of comments and improving the accuracy of detecting dangerous comments.

[0031] The comment collection unit can analyze the metadata of the collected comments to identify trends in anti-comments at specific times and locations. The comment collection unit, for example, analyzes the metadata of the collected comments to identify trends in anti-comments at specific times and locations. For example, it identifies times and areas where anti-comments are most prevalent based on the posting time and location. It also uses generative AI to analyze the metadata of the collected comments to identify trends in anti-comments on specific devices and platforms. For example, it analyzes comments from specific smartphones and social media apps. Furthermore, it can identify trends in anti-comments related to specific events or occurrences through metadata analysis. For example, it can analyze the tendency for anti-comments to increase when specific news or events occur. This allows effective countermeasures to be taken by identifying trends in anti-comments at specific times and locations.

[0032] The comment collection unit can extract the text from audio comments or video comments and collect it as anti-comments. The comment collection unit, for example, uses a generation AI to extract text from audio comments and collect it as anti-comments. For example, it uses voice recognition technology to convert audio comments into text and collect comments containing negative content. To extract text from video comments, the generation AI combines voice recognition technology and video analysis technology. For example, it converts audio in a video into text and collects comments containing negative content. Furthermore, when extracting text from audio comments or video comments, the generation AI detects specific keywords or phrases. For example, it prioritizes the collection of comments containing slander or violence. This allows text to be extracted from audio comments and video comments, thereby collecting a wider variety of comments and improving the accuracy of detecting dangerous comments.

[0033] The comment inspection unit understands the context of the comment and can perform context-based inspection rather than simple keyword matching. The comment inspection unit, for example, uses generation AI to understand the context of the comment and perform context-based inspection rather than simple keyword matching. For example, it analyzes the context before and after the comment to identify comments that contain defamatory or violent content. In addition, to perform context analysis, the generation AI uses natural language processing technology. For example, it understands the meaning and intent of the comment and inspects comments that contain negative content. Furthermore, the generation AI learns from past comment data to understand the context of the comment. For example, it analyzes the context based on similar comments and inspects comments that contain defamatory or violent content. This reduces false positives and improves accuracy by performing context-based inspection.

[0034] The comment scrutiny unit can improve accuracy by referring to a database of similar past comments when scrutinizing comments. The comment scrutiny unit, for example, uses a generation AI to refer to a database of similar past comments when scrutinizing comments to improve accuracy. For example, new comments are scrutinized based on past defamatory or violent comments. In addition, in order to refer to the database of similar comments, the generation AI learns from past comment data. For example, it analyzes patterns and characteristics of past comments and scrutinizes new comments. Furthermore, the generation AI improves accuracy by referring to the database of similar past comments. For example, it scrutinizes new comments based on the trends and characteristics of past comments. In this way, accuracy is improved by referring to the database of similar past comments.

[0035] The comment scrutiny unit can also analyze text within images and videos and classify them as anti-comments. For example, the comment scrutiny unit uses generation AI to analyze text within images and videos and classify them as anti-comments. For example, it extracts text within images and classifies comments containing defamatory or violent content. To analyze text within videos, the generation AI combines voice recognition technology and video analysis technology. For example, it converts audio within videos into text and classifies comments containing negative content. Furthermore, when analyzing text within images and videos, the generation AI detects specific keywords and phrases. For example, it prioritizes the classification of comments containing defamatory or violent content. By analyzing text within images and videos, this allows for scrutiny of a wider variety of comments and improves the accuracy of detecting dangerous comments.

[0036] The comment inspection unit can take into account expressions specific to different cultures and regions and perform inspection and classification based on cultural background. The comment inspection unit, for example, uses generative AI to perform inspection and classification based on cultural background, taking into account expressions specific to different cultures and regions. For example, it analyzes slang and expressions used in a particular culture or region and classifies comments containing negative content. In addition, to perform inspection based on cultural background, the generative AI learns data from different cultures and regions. For example, it analyzes words and expressions used in a particular culture or region and classifies comments containing negative content. Furthermore, the generative AI uses natural language processing technology to take into account expressions specific to different cultures and regions. For example, it detects words and expressions used in a particular culture or region and classifies comments containing negative content. This improves the accuracy of inspection and classification by taking into account expressions specific to different cultures and regions.

[0037] The user extraction unit can analyze the intent of dangerous comments and assess the actual danger, rather than mere slander. The user extraction unit, for example, uses generation AI to analyze the intent behind the picked comments and assess the actual danger, rather than mere slander. For example, it analyzes the context and wording of the comment to evaluate the likelihood that it will actually be acted upon. In addition, to analyze the intent behind the comments, the generation AI uses natural language processing technology. For example, it understands the meaning and intent of the comment and assesses the actual danger. Furthermore, the generation AI learns from past comment data to analyze the intent behind the picked comments. For example, it analyzes the intent based on similar comments and assesses the actual danger. This allows more effective measures to be taken by assessing the actual danger.

[0038] The user extraction unit can analyze the past behavioral history of the poster of the picked comment and evaluate the possibility of recidivism. The user extraction unit, for example, uses a generation AI to analyze the past behavioral history of the poster of the picked comment and evaluate the possibility of recidivism. For example, the possibility of recidivism is evaluated based on a history of past postings of defamatory or violent comments. In addition, to analyze the poster's past behavioral history, the generation AI learns past comment data. For example, it analyzes the patterns and characteristics of past comments and evaluates the possibility of recidivism. Furthermore, the generation AI uses a specific algorithm to analyze the past behavioral history of the poster of the picked comment. For example, it evaluates the possibility of recidivism based on the frequency and content of past comments. This allows more effective measures to be taken by evaluating the possibility of recidivism.

[0039] The user extraction unit can cross-reference the picked comments across different platforms to identify consistent, dangerous comments. The user extraction unit, for example, uses a generation AI to cross-reference the picked comments across different platforms to identify consistent, dangerous comments. For example, if a comment with the same content is posted on multiple platforms, it identifies that comment. To perform the cross-referencing, the generation AI collects comments from multiple platforms and performs a comparative analysis. For example, it analyzes comments collected from social media, blogs, bulletin boards, etc. to identify consistent, dangerous comments. Furthermore, the generation AI uses a specific algorithm to cross-reference the picked comments across different platforms. For example, it identifies consistent, dangerous comments based on the content of the comment and information about the poster. This allows for more effective countermeasures to be taken by identifying consistent, dangerous comments.

[0040] The user extraction unit can analyze the past posting history of risky users and evaluate the likelihood of recidivism. The user extraction unit, for example, uses generation AI to analyze the past posting history of risky users and evaluate the likelihood of recidivism. For example, the likelihood of recidivism is evaluated based on a history of past postings of defamatory or violent comments. In addition, to analyze the posting history, the generation AI learns past comment data. For example, it analyzes the patterns and characteristics of past comments to evaluate the likelihood of recidivism. Furthermore, the generation AI uses a specific algorithm to analyze the past posting history of risky users. For example, it evaluates the likelihood of recidivism based on the frequency and content of past comments. This allows more effective measures to be taken by evaluating the likelihood of recidivism.

[0041] The user extraction unit can analyze the social network of a risky user and identify other related risky users. The user extraction unit, for example, uses a generation AI to analyze the social network of a risky user and identify other related risky users. For example, it analyzes users belonging to a specific group or community and identifies related risky users. To analyze the social network, the generation AI also analyzes the relationships between users. For example, it analyzes comments made by users belonging to the same group or community and identifies related risky users. Furthermore, the generation AI uses social network analysis technology to analyze the social network of a risky user. For example, it analyzes connections and relationships between users and identifies related risky users. This allows more effective measures to be taken by identifying other related risky users.

[0042] The user extraction unit can cross-reference user behavior across different platforms to identify consistent, risky users. For example, the user extraction unit uses a generation AI to cross-reference user behavior across different platforms to identify consistent, risky users. For example, if the same user posts abusive or violent comments on multiple platforms, the unit can identify that user. To perform this cross-referencing, the generation AI collects user behavior data from multiple platforms and performs comparative analysis. For example, it analyzes data collected from social media, blogs, bulletin boards, etc. to identify consistent, risky users. Furthermore, the generation AI uses a specific algorithm to cross-reference user behavior across different platforms. For example, it can identify consistent, risky users based on the content of user posts and behavioral patterns. This allows for more effective countermeasures to be implemented by identifying consistent, risky users.

[0043] The user extraction unit can automatically translate the posts of risky users into different languages ​​and extract risky users from a global perspective. The user extraction unit, for example, uses generation AI to automatically translate the posts of risky users into different languages ​​and extract risky users from a global perspective. For example, it automatically translates posts in English, French, Chinese, etc. to extract users who contain negative content. It also uses machine translation technology to translate posts in different languages ​​in real time and extract risky users. For example, it automatically translates posts collected from social media and bulletin boards to extract users who contain negative content. Furthermore, when automatically translating posts in different languages, the generation AI detects specific keywords and phrases. For example, it prioritizes the extraction of users who contain slanderous or violent content. As a result, by automatically translating posts in different languages, it is possible to extract risky users from a global perspective and improve detection accuracy.

[0044] The user extraction unit can analyze past countermeasure data and propose the most effective countermeasure. The user extraction unit, for example, uses a generation AI to analyze past countermeasure data and propose the most effective countermeasure. For example, it proposes suspending an account or sending a warning message based on countermeasures that have been successful in the past. Furthermore, to analyze past countermeasure data, the generation AI learns from past cases. For example, it analyzes the effectiveness and results of past countermeasures and proposes the most effective countermeasure. Furthermore, the generation AI uses a specific algorithm to analyze past countermeasure data. For example, it proposes the most effective countermeasure based on the success rate and effectiveness of past countermeasures. In this way, it is possible to propose the most effective countermeasure by analyzing past countermeasure data.

[0045] The user extraction unit can cross-reference the effectiveness of countermeasures across different platforms and propose consistent countermeasures. The user extraction unit, for example, uses a generation AI to cross-reference the effectiveness of countermeasures across different platforms and propose consistent countermeasures. For example, it compares and analyzes the effectiveness of countermeasures on social media, blogs, bulletin boards, etc., and proposes the most effective countermeasures. In addition, to perform the cross-referencing, the generation AI collects countermeasure data from multiple platforms and performs a comparative analysis. For example, it analyzes the effectiveness of countermeasures on different platforms and proposes consistent countermeasures. Furthermore, the generation AI uses a specific algorithm to cross-reference the effectiveness of countermeasures across different platforms. For example, it proposes consistent countermeasures based on the effectiveness and results of the countermeasures. This allows for more effective countermeasures to be taken by proposing consistent countermeasures.

[0046] The user extraction unit can consider expressions specific to different cultures and regions when proposing countermeasures and propose countermeasures based on cultural backgrounds. The user extraction unit, for example, uses a generation AI to consider expressions specific to different cultures and regions when proposing countermeasures and propose countermeasures based on cultural backgrounds. For example, it considers words and expressions used in a specific culture or region and proposes countermeasures. In addition, to propose countermeasures based on cultural backgrounds, the generation AI learns data from different cultures and regions. For example, it analyzes words and expressions used in a specific culture or region and proposes countermeasures. Furthermore, the generation AI uses natural language processing technology to consider expressions specific to different cultures and regions when proposing countermeasures. For example, it detects words and expressions used in a specific culture or region and proposes countermeasures. This allows more effective countermeasures to be taken by proposing countermeasures based on cultural backgrounds.

[0047] The system according to the embodiment is not limited to the above-described example, and various modifications are possible, for example, as follows.

[0048] The comment collection unit can also analyze the metadata of collected comments to identify trends in anti-comments at specific times and locations. For example, it can identify times and areas where anti-comments are most prevalent based on the time and location of posting. Generative AI can also be used to analyze the metadata of collected comments to identify trends in anti-comments on specific devices and platforms. For example, it can analyze comments from specific smartphones or social media apps. Furthermore, metadata analysis can also identify trends in anti-comments related to specific events or occurrences. For example, it can analyze the tendency for anti-comments to increase when specific news or events occur. This can reveal trends in anti-comments at specific times and locations, allowing for effective countermeasures to be implemented.

[0049] The comment collection unit can also extract text from audio and video comments and collect it as anti-comments. For example, it can use voice recognition technology to convert audio comments into text and collect comments containing negative content. To extract text from video comments, the generation AI can also combine voice recognition technology with video analysis technology. For example, it can convert audio in videos into text and collect comments containing negative content. Furthermore, when extracting text from audio and video comments, the generation AI can also detect specific keywords and phrases. For example, it can prioritize the collection of comments containing defamatory or violent content. This allows for the extraction of text from audio and video comments, thereby collecting a wider variety of comments and improving the accuracy of detecting dangerous comments.

[0050] The comment collection unit can automatically translate comments in different languages ​​and collect anti-comments from a global perspective. For example, it can automatically translate comments in English, French, Chinese, and other languages ​​to collect comments containing negative content. The generation AI can also detect specific keywords and phrases when automatically translating comments in different languages. For example, it can prioritize the collection of comments containing defamatory or violent content. Furthermore, the generation AI can use natural language processing technology to automatically translate comments in different languages. For example, it can analyze the grammar and expressions of specific languages ​​to collect comments containing negative content. This allows automatic translation of comments in different languages ​​to collect anti-comments from a global perspective and improve detection accuracy.

[0051] The comment screening unit can also analyze text within images and videos and classify them as anti-comments. For example, it can extract text within images and classify comments containing defamatory or violent content. To analyze text within videos, the generative AI can also combine speech recognition technology and video analysis technology. For example, it can convert the audio within a video into text and classify comments containing negative content. Furthermore, when analyzing text within images and videos, the generative AI can detect specific keywords and phrases. For example, it can prioritize the classification of comments containing defamatory or violent content. By analyzing text within images and videos, this allows for screening a wider variety of comments and improves the accuracy of detecting dangerous comments.

[0052] The comment screening unit can take into account expressions specific to different cultures and regions and perform screening and classification based on cultural background. For example, it can analyze slang and expressions used in a particular culture or region and classify comments containing negative content. In addition, to perform screening based on cultural background, the generative AI can also learn data from different cultures and regions. For example, it can analyze words and expressions used in a particular culture or region and classify comments containing negative content. Furthermore, the generative AI can use natural language processing technology to take into account expressions specific to different cultures and regions. For example, it can detect words and expressions used in a particular culture or region and classify comments containing negative content. This improves the accuracy of screening and classification by taking into account expressions specific to different cultures and regions.

[0053] The user extraction unit can analyze the intent of dangerous comments and assess their actual risk, rather than simply slander. For example, it can analyze the context and wording of the comment to assess the likelihood of actual action being taken. The generation AI can also use natural language processing technology to analyze the intent behind the comments. For example, it can understand the meaning and intent of the comment and assess its actual risk. Furthermore, the generation AI can learn from past comment data to analyze the intent behind the selected comment. For example, it can analyze the intent based on similar comments and assess the actual risk. This allows for more effective countermeasures to be taken by assessing the actual risk.

[0054] The user extraction unit can analyze the past behavioral history of the poster of the selected comment and evaluate the likelihood of recidivism. For example, the likelihood of recidivism can be evaluated based on a history of posting defamatory or violent comments in the past. In addition, to analyze the poster's past behavioral history, the generation AI can also learn from past comment data. For example, it can analyze the patterns and characteristics of past comments to evaluate the likelihood of recidivism. Furthermore, the generation AI can use specific algorithms to analyze the past behavioral history of the poster of the selected comment. For example, it can evaluate the likelihood of recidivism based on the frequency and content of past comments. This allows more effective measures to be taken by evaluating the likelihood of recidivism.

[0055] The user extraction unit can cross-reference the picked comments across different platforms to identify consistently dangerous comments. For example, if a comment with the same content is posted on multiple platforms, it can identify that comment. To perform this cross-referencing, the generation AI can also collect comments from multiple platforms and perform comparative analysis. For example, it can analyze comments collected from social media, blogs, message boards, etc. to identify consistently dangerous comments. Furthermore, the generation AI can use specific algorithms to cross-reference the picked comments across different platforms. For example, it can identify consistently dangerous comments based on the content of the comment and information about the poster. This allows for more effective countermeasures to be taken by identifying consistently dangerous comments.

[0056] The processing flow of the first embodiment will be briefly explained below.

[0057] Step 1: The comment collection unit collects comments from multiple online platforms. For example, it automatically collects comments from social media, blogs, bulletin boards, etc. The comment collection unit also uses generative AI to estimate the poster's emotions in real time as the comments are collected, and prioritizes the collection of comments with particularly negative sentiment. The comment collection unit then expands the platforms it collects to include comments from the dark web and anonymous bulletin boards. It analyzes the metadata of the collected comments (posting time, posting location, posting device, etc.) to identify trends in anti-comments at specific times and locations. It also extracts text from audio and video comments and collects them as anti-comments. It automatically translates comments in different languages ​​to collect anti-comments from a global perspective. It uses an emotion estimation function to analyze the emotional tone of the collected comments and prioritize the collection of comments that evoke particularly strong emotional reactions. Step 2: The comment screening unit screens the collected comments. Using generative AI, it understands the context of the comment and performs a context-based screening rather than simply keyword matching. It references a database of similar past comments to improve accuracy. It uses an emotion estimation function to evaluate the emotional impact of the screened comments and classifies comments that have a particularly strong emotional impact. It also analyzes text within images and videos and classifies them as anti-comments. It screens and classifies based on cultural background, taking into account expressions unique to different cultures and regions. Step 3: The comment classification unit classifies the scrutinized comments. Using generative AI, it analyzes the emotional tone of the comments and prioritizes comments that evoke particularly strong emotional responses. Step 4: The user extraction unit extracts users who leave dangerous comments from the classified comments. Generative AI is used to analyze the intention behind the picked comments and evaluate their actual risk. The past behavioral history of the picked comment poster is analyzed to evaluate the likelihood of recidivism. Sentiment estimation is used to evaluate the emotional impact of the picked comments and prioritize comments with particularly strong emotional impact. The groups and communities behind the picked comments are identified to detect organized slander. Picked comments are cross-referenced across different platforms to identify consistent dangerous comments.

[0058] (Example 2) An AI system according to an embodiment of the present invention is a system that scrutinizes anti-comments against companies and individuals, picks out slanderous and violent comments, and checks for and determines whether they contain dangerous content. This allows the AI ​​system to efficiently scrutinize anti-comments against companies and individuals and pick out slanderous and violent comments. It can also identify users who leave dangerous comments and take appropriate measures.

[0059] The AI ​​system according to the embodiment includes a comment collection unit, a comment scrutiny unit, a comment classification unit, and a user extraction unit. The comment collection unit collects comments from multiple platforms on the Internet. For example, it automatically collects comments from social media, blogs, bulletin boards, and the like. The comment collection unit also uses a generative AI to estimate the poster's emotions in real time when collecting comments, and prioritizes the collection of comments that express particularly negative emotions. For example, it analyzes the poster's emotions from their writing style and vocabulary to collect comments that express strong feelings of anger or sadness. The comment collection unit also expands the platforms it collects comments from, and collects comments from the dark web and anonymous bulletin boards. For example, it automatically collects comments related to specific keywords or topics and stores them in a database. The comment collection unit also analyzes the metadata of the collected comments (e.g., posting time, posting location, posting device) to identify trends in anti-comments during specific time periods or locations. For example, it identifies time periods or areas where anti-comments are most prevalent based on the posting time and posting location. The comment collection unit also extracts text from audio comments and video comments and collects them as anti-comments. For example, speech recognition technology is used to convert voice comments into text and collect comments containing negative content. The comment collection unit also automatically translates comments in different languages ​​to collect anti-comments from a global perspective. For example, comments in English, French, Chinese, etc. are automatically translated to collect comments containing negative content. The comment collection unit also uses emotion estimation functionality to analyze the emotional tone of the collected comments and prioritize the collection of comments that show particularly strong emotional reactions. For example, comments containing strong emotions such as anger or sadness are collected. The comment review unit then reviews the collected comments. For example, generative AI is used to understand the context of the comments and conduct a context-based review rather than simple keyword matching. For example, the context before and after the comment is analyzed to identify comments containing defamatory or violent content. The comment review unit also refers to a database of similar past comments when reviewing comments to improve accuracy. For example, new comments are reviewed based on past defamatory or violent comments.Furthermore, the comment screening unit uses emotion estimation to evaluate the emotional impact of the screened comments and classify comments that have a particularly strong emotional impact. For example, it classifies comments that express strong emotions such as anger or sadness. The comment screening unit also analyzes text within images and videos to classify them as anti-comments. For example, it extracts text from images and classifies comments that contain defamatory or violent content. Furthermore, the comment screening unit considers expressions unique to different cultures and regions and screens and classifies based on cultural background. For example, it analyzes slang and expressions used in specific cultures and regions to classify comments that contain negative content. The comment classification unit classifies the screened comments. For example, it uses generative AI to analyze the emotional tone of the comments and prioritizes classifying comments that show particularly strong emotional reactions. For example, it classifies comments that express strong emotions such as anger or sadness. The user extraction unit extracts users who have left dangerous comments from the classified comments. For example, generative AI is used to analyze the intent behind the selected comments and evaluate their actual risk, rather than mere slander. For example, the context and wording of the comment are analyzed to assess the likelihood of actual action being taken. The user extraction unit also analyzes the past behavioral history of the poster of the selected comment to assess the likelihood of recidivism. For example, the likelihood of recidivism is assessed based on a history of past postings of slanderous or violent comments. Furthermore, the user extraction unit uses emotion estimation to evaluate the emotional impact of the selected comments and prioritizes comments that have a particularly strong emotional impact. For example, comments that express strong emotions such as anger or sadness are selected. The user extraction unit also identifies the groups or communities behind the selected comments and detects organized slander. For example, if a specific group or community is engaged in a coordinated slanderous activity, this can be identified as organized activity. Furthermore, the user extraction unit cross-references the selected comments across different platforms to identify consistent and dangerous comments. For example, if the same comment is posted on multiple platforms, the unit identifies that comment.As a result, the AI ​​system according to the embodiment can efficiently scrutinize anti-comments against companies or individuals and pick out slanderous or violent comments. It can also identify users who leave dangerous comments and take appropriate measures. For example, when proposing countermeasures, the output unit takes into account the social network of the dangerous user and takes measures against other related users. To take the social network into account, the generation AI analyzes the relationships between users. For example, it analyzes the behavior of users who belong to the same group or community and takes measures against other related users. To take the social network into account when proposing countermeasures, the generation AI uses social network analysis technology. For example, it analyzes the connections and relationships between users and takes measures against other related users.

[0060] The comment collection unit can estimate the poster's emotions in real time when collecting comments, and prioritize collecting comments with particularly negative emotions. The comment collection unit, for example, uses a generation AI to estimate the poster's emotions in real time when collecting comments, and prioritize collecting comments with particularly negative emotions. For example, it analyzes emotions from the poster's writing style and vocabulary and collects comments with strong emotions such as anger or sadness. In addition, to estimate the poster's emotions in real time, the generation AI analyzes the content of the comments using natural language processing technology. For example, it detects specific keywords and phrases and prioritizes collecting comments with negative emotions. Furthermore, the generation AI analyzes the context and tone of the comments to estimate the poster's emotions in real time. For example, it detects words and expressions that indicate negative emotions and prioritizes collecting those comments. This prioritizes collecting comments with negative emotions, improving the accuracy of detecting dangerous comments.

[0061] The comment collection unit can expand the platforms it collects to also collect comments from the dark web and anonymous message boards. The comment collection unit, for example, uses generation AI to collect comments from the dark web and anonymous message boards. For example, it automatically collects comments related to specific keywords or topics and stores them in a database. To expand the platforms it collects, the generation AI also collects comments from multiple sources. For example, it collects comments from social media, blogs, message boards, the dark web, etc., and analyzes them comprehensively. Furthermore, to efficiently collect comments from the dark web and anonymous message boards, the generation AI uses specific crawling techniques. For example, it automatically collects comments related to specific topics or keywords and stores them in a database. This expands the platforms it collects, thereby collecting a wider range of comments and improving the accuracy of detecting dangerous comments.

[0062] The comment collection unit can analyze the metadata of the collected comments to identify trends in anti-comments at specific times and locations. The comment collection unit, for example, analyzes the metadata of the collected comments to identify trends in anti-comments at specific times and locations. For example, it identifies times and areas where anti-comments are most prevalent based on the posting time and location. It also uses generative AI to analyze the metadata of the collected comments to identify trends in anti-comments on specific devices and platforms. For example, it analyzes comments from specific smartphones and social media apps. Furthermore, it can identify trends in anti-comments related to specific events or occurrences through metadata analysis. For example, it can analyze the tendency for anti-comments to increase when specific news or events occur. This allows effective countermeasures to be taken by identifying trends in anti-comments at specific times and locations.

[0063] The comment collection unit can extract the text from audio comments or video comments and collect it as anti-comments. The comment collection unit, for example, uses a generation AI to extract text from audio comments and collect it as anti-comments. For example, it uses voice recognition technology to convert audio comments into text and collect comments containing negative content. To extract text from video comments, the generation AI combines voice recognition technology and video analysis technology. For example, it converts audio in a video into text and collects comments containing negative content. Furthermore, when extracting text from audio comments or video comments, the generation AI detects specific keywords or phrases. For example, it prioritizes the collection of comments containing slander or violence. This allows text to be extracted from audio comments and video comments, thereby collecting a wider variety of comments and improving the accuracy of detecting dangerous comments.

[0064] The comment collection unit can use the emotion estimation function to analyze the emotional tone of the collected comments and prioritize collecting comments that indicate particularly strong emotional reactions. The comment collection unit can, for example, use the generation AI to analyze the emotional tone of the collected comments and prioritize collecting comments that indicate particularly strong emotional reactions. For example, comments with strong emotions of anger or sadness are collected. The emotion estimation function can also be used to calculate an emotion score for the collected comments and prioritize collecting comments that indicate particularly strong emotional reactions. For example, comments with a high emotion score are collected. Furthermore, the generation AI uses natural language processing technology to analyze the emotional tone of the collected comments. For example, it detects specific keywords and phrases and collects comments that indicate particularly strong emotional reactions. This prioritizes collecting comments that indicate particularly strong emotional reactions, thereby improving the accuracy of detecting dangerous comments.

[0065] The comment inspection unit understands the context of the comment and can perform context-based inspection rather than simple keyword matching. The comment inspection unit, for example, uses generation AI to understand the context of the comment and perform context-based inspection rather than simple keyword matching. For example, it analyzes the context before and after the comment to identify comments that contain defamatory or violent content. In addition, to perform context analysis, the generation AI uses natural language processing technology. For example, it understands the meaning and intent of the comment and inspects comments that contain negative content. Furthermore, the generation AI learns from past comment data to understand the context of the comment. For example, it analyzes the context based on similar comments and inspects comments that contain defamatory or violent content. This reduces false positives and improves accuracy by performing context-based inspection.

[0066] The comment scrutiny unit can improve accuracy by referring to a database of similar past comments when scrutinizing comments. The comment scrutiny unit, for example, uses a generation AI to refer to a database of similar past comments when scrutinizing comments to improve accuracy. For example, new comments are scrutinized based on past defamatory or violent comments. In addition, in order to refer to the database of similar comments, the generation AI learns from past comment data. For example, it analyzes patterns and characteristics of past comments and scrutinizes new comments. Furthermore, the generation AI improves accuracy by referring to the database of similar past comments. For example, it scrutinizes new comments based on the trends and characteristics of past comments. In this way, accuracy is improved by referring to the database of similar past comments.

[0067] The comment inspection unit can use an emotion estimation function to evaluate the emotional impact of the inspected comments and classify comments that have a particularly strong emotional impact. The comment inspection unit, for example, uses a generation AI to evaluate the emotional impact of the inspected comments using the emotion estimation function and classify comments that have a particularly strong emotional impact. For example, it classifies comments that have a strong emotion such as anger or sadness. The emotion estimation function is also used to calculate an emotion score for the inspected comments and classify comments that have a particularly strong emotional impact. For example, it classifies comments with a high emotion score. Furthermore, the generation AI uses natural language processing technology to evaluate the emotional impact of the inspected comments. For example, it detects specific keywords and phrases and classifies comments that have a strong emotional impact. This classifies comments that have a strong emotional impact, thereby improving the accuracy of detecting dangerous comments.

[0068] The comment scrutiny unit can also analyze text within images and videos and classify them as anti-comments. For example, the comment scrutiny unit uses generation AI to analyze text within images and videos and classify them as anti-comments. For example, it extracts text within images and classifies comments containing defamatory or violent content. To analyze text within videos, the generation AI combines voice recognition technology and video analysis technology. For example, it converts audio within videos into text and classifies comments containing negative content. Furthermore, when analyzing text within images and videos, the generation AI detects specific keywords and phrases. For example, it prioritizes the classification of comments containing defamatory or violent content. By analyzing text within images and videos, this allows for scrutiny of a wider variety of comments and improves the accuracy of detecting dangerous comments.

[0069] The comment inspection unit can take into account expressions specific to different cultures and regions and perform inspection and classification based on cultural background. The comment inspection unit, for example, uses generative AI to perform inspection and classification based on cultural background, taking into account expressions specific to different cultures and regions. For example, it analyzes slang and expressions used in a particular culture or region and classifies comments containing negative content. In addition, to perform inspection based on cultural background, the generative AI learns data from different cultures and regions. For example, it analyzes words and expressions used in a particular culture or region and classifies comments containing negative content. Furthermore, the generative AI uses natural language processing technology to take into account expressions specific to different cultures and regions. For example, it detects words and expressions used in a particular culture or region and classifies comments containing negative content. This improves the accuracy of inspection and classification by taking into account expressions specific to different cultures and regions.

[0070] The comment inspection unit can use the emotion estimation function to analyze the emotional tone of comments and prioritize classification of comments that indicate particularly strong emotional reactions. The comment inspection unit can, for example, use the generation AI to analyze the emotional tone of comments and prioritize classification of comments that indicate particularly strong emotional reactions. For example, it classifies comments with strong emotions such as anger or sadness. The emotion estimation function can also be used to calculate the emotion score of comments and prioritize classification of comments that indicate particularly strong emotional reactions. For example, it classifies comments with a high emotion score. Furthermore, the generation AI uses natural language processing technology to analyze the emotional tone of comments. For example, it detects specific keywords and phrases and classifies comments that indicate particularly strong emotional reactions. This prioritizes classification of comments that indicate particularly strong emotional reactions, thereby improving the accuracy of detecting dangerous comments.

[0071] The user extraction unit can analyze the intent of dangerous comments and assess the actual danger, rather than mere slander. The user extraction unit, for example, uses generation AI to analyze the intent behind the picked comments and assess the actual danger, rather than mere slander. For example, it analyzes the context and wording of the comment to evaluate the likelihood that it will actually be acted upon. In addition, to analyze the intent behind the comments, the generation AI uses natural language processing technology. For example, it understands the meaning and intent of the comment and assesses the actual danger. Furthermore, the generation AI learns from past comment data to analyze the intent behind the picked comments. For example, it analyzes the intent based on similar comments and assesses the actual danger. This allows more effective measures to be taken by assessing the actual danger.

[0072] The user extraction unit can analyze the past behavioral history of the poster of the picked comment and evaluate the possibility of recidivism. The user extraction unit, for example, uses a generation AI to analyze the past behavioral history of the poster of the picked comment and evaluate the possibility of recidivism. For example, the possibility of recidivism is evaluated based on a history of past postings of defamatory or violent comments. In addition, to analyze the poster's past behavioral history, the generation AI learns past comment data. For example, it analyzes the patterns and characteristics of past comments and evaluates the possibility of recidivism. Furthermore, the generation AI uses a specific algorithm to analyze the past behavioral history of the poster of the picked comment. For example, it evaluates the possibility of recidivism based on the frequency and content of past comments. This allows more effective measures to be taken by evaluating the possibility of recidivism.

[0073] The user extraction unit can use the emotion estimation function to evaluate the emotional impact of the picked comments and prioritize comments that have a particularly strong emotional impact. The user extraction unit can, for example, use a generation AI to evaluate the emotional impact of the picked comments and prioritize comments that have a particularly strong emotional impact. For example, comments that express strong emotions such as anger or sadness are picked up. The emotion estimation function can also be used to calculate an emotion score for the picked comments and prioritize comments that have a particularly strong emotional impact. For example, comments with a high emotion score are picked up. Furthermore, the generation AI uses natural language processing technology to evaluate the emotional impact of the picked comments. For example, specific keywords and phrases can be detected to pick up comments that have a particularly strong emotional impact. This prioritizes picking up comments that have a strong emotional impact, thereby improving the accuracy of detecting dangerous comments.

[0074] The user extraction unit can cross-reference the picked comments across different platforms to identify consistent, dangerous comments. The user extraction unit, for example, uses a generation AI to cross-reference the picked comments across different platforms to identify consistent, dangerous comments. For example, if a comment with the same content is posted on multiple platforms, it identifies that comment. To perform the cross-referencing, the generation AI collects comments from multiple platforms and performs a comparative analysis. For example, it analyzes comments collected from social media, blogs, bulletin boards, etc. to identify consistent, dangerous comments. Furthermore, the generation AI uses a specific algorithm to cross-reference the picked comments across different platforms. For example, it identifies consistent, dangerous comments based on the content of the comment and information about the poster. This allows for more effective countermeasures to be taken by identifying consistent, dangerous comments.

[0075] The user extraction unit can use the emotion estimation function to analyze the emotional tone of the picked comments and prioritize comments that evoke particularly strong emotional reactions. The user extraction unit can, for example, use a generation AI to analyze the emotional tone of the picked comments and prioritize comments that evoke particularly strong emotional reactions. For example, comments with strong emotions such as anger or sadness are picked up. The emotion estimation function can also be used to calculate an emotion score for the picked comments and prioritize comments that evoke particularly strong emotional reactions. For example, comments with a high emotion score are picked up. Furthermore, the generation AI uses natural language processing technology to analyze the emotional tone of the picked comments. For example, it can detect specific keywords and phrases and pick out comments that evoke particularly strong emotional reactions. This prioritizes picking out comments that evoke particularly strong emotional reactions, thereby improving the accuracy of detecting dangerous comments.

[0076] The user extraction unit can analyze the past posting history of risky users and evaluate the likelihood of recidivism. The user extraction unit, for example, uses generation AI to analyze the past posting history of risky users and evaluate the likelihood of recidivism. For example, the likelihood of recidivism is evaluated based on a history of past postings of defamatory or violent comments. In addition, to analyze the posting history, the generation AI learns past comment data. For example, it analyzes the patterns and characteristics of past comments to evaluate the likelihood of recidivism. Furthermore, the generation AI uses a specific algorithm to analyze the past posting history of risky users. For example, it evaluates the likelihood of recidivism based on the frequency and content of past comments. This allows more effective measures to be taken by evaluating the likelihood of recidivism.

[0077] The user extraction unit can analyze the social network of a risky user and identify other related risky users. The user extraction unit, for example, uses a generation AI to analyze the social network of a risky user and identify other related risky users. For example, it analyzes users belonging to a specific group or community and identifies related risky users. To analyze the social network, the generation AI also analyzes the relationships between users. For example, it analyzes comments made by users belonging to the same group or community and identifies related risky users. Furthermore, the generation AI uses social network analysis technology to analyze the social network of a risky user. For example, it analyzes connections and relationships between users and identifies related risky users. This allows more effective measures to be taken by identifying other related risky users.

[0078] The user extraction unit can use the emotion estimation function to analyze the emotional tone of the posts of risky users and preferentially extract users who show particularly strong emotional reactions. The user extraction unit can, for example, use a generation AI to analyze the emotional tone of the posts of risky users and preferentially extract users who show particularly strong emotional reactions. For example, it extracts users who show strong emotions of anger or sadness. The emotion estimation function can also be used to calculate an emotion score for the posts of risky users and preferentially extract users who show particularly strong emotional reactions. For example, it extracts users with a high emotion score. Furthermore, the generation AI uses natural language processing technology to analyze the emotional tone of the posts of risky users. For example, it detects specific keywords and phrases and extracts users who show particularly strong emotional reactions. This prioritizes the extraction of users who show particularly strong emotional reactions, thereby improving the accuracy of detecting risky users.

[0079] The user extraction unit can cross-reference user behavior across different platforms to identify consistent, risky users. For example, the user extraction unit uses a generation AI to cross-reference user behavior across different platforms to identify consistent, risky users. For example, if the same user posts abusive or violent comments on multiple platforms, the unit can identify that user. To perform this cross-referencing, the generation AI collects user behavior data from multiple platforms and performs comparative analysis. For example, it analyzes data collected from social media, blogs, bulletin boards, etc. to identify consistent, risky users. Furthermore, the generation AI uses a specific algorithm to cross-reference user behavior across different platforms. For example, it can identify consistent, risky users based on the content of user posts and behavioral patterns. This allows for more effective countermeasures to be implemented by identifying consistent, risky users.

[0080] The user extraction unit can automatically translate the posts of risky users into different languages ​​and extract risky users from a global perspective. The user extraction unit, for example, uses generation AI to automatically translate the posts of risky users into different languages ​​and extract risky users from a global perspective. For example, it automatically translates posts in English, French, Chinese, etc. to extract users who contain negative content. It also uses machine translation technology to translate posts in different languages ​​in real time and extract risky users. For example, it automatically translates posts collected from social media and bulletin boards to extract users who contain negative content. Furthermore, when automatically translating posts in different languages, the generation AI detects specific keywords and phrases. For example, it prioritizes the extraction of users who contain slanderous or violent content. As a result, by automatically translating posts in different languages, it is possible to extract risky users from a global perspective and improve detection accuracy.

[0081] The user extraction unit can use the emotion estimation function to analyze the emotional tone of the posts of risky users and preferentially extract users who show particularly strong emotional reactions. The user extraction unit can, for example, use a generation AI to analyze the emotional tone of the posts of risky users and preferentially extract users who show particularly strong emotional reactions. For example, it extracts users who show strong emotions of anger or sadness. The emotion estimation function can also be used to calculate an emotion score for the posts of risky users and preferentially extract users who show particularly strong emotional reactions. For example, it extracts users with a high emotion score. Furthermore, the generation AI uses natural language processing technology to analyze the emotional tone of the posts of risky users. For example, it detects specific keywords and phrases and extracts users who show particularly strong emotional reactions. This prioritizes the extraction of users who show particularly strong emotional reactions, thereby improving the accuracy of detecting risky users.

[0082] The user extraction unit can analyze past countermeasure data and propose the most effective countermeasure. The user extraction unit, for example, uses a generation AI to analyze past countermeasure data and propose the most effective countermeasure. For example, it proposes suspending an account or sending a warning message based on countermeasures that have been successful in the past. Furthermore, to analyze past countermeasure data, the generation AI learns from past cases. For example, it analyzes the effectiveness and results of past countermeasures and proposes the most effective countermeasure. Furthermore, the generation AI uses a specific algorithm to analyze past countermeasure data. For example, it proposes the most effective countermeasure based on the success rate and effectiveness of past countermeasures. In this way, it is possible to propose the most effective countermeasure by analyzing past countermeasure data.

[0083] The user extraction unit can use the emotion estimation function to predict the user's emotional reaction when proposing a countermeasure and propose the countermeasure that will elicit the most positive reaction. The user extraction unit can, for example, use a generation AI to predict the user's emotional reaction when proposing a countermeasure and propose the countermeasure that will elicit the most positive reaction. For example, it adjusts the content and tone of the warning message to predict the user's reaction. The emotion estimation function also calculates the user's emotional score when proposing a countermeasure and proposes the countermeasure that will elicit the most positive reaction. For example, it prioritizes the proposal of countermeasures with high emotion scores. Furthermore, the generation AI uses natural language processing technology to predict the user's emotional reaction when proposing a countermeasure. For example, it detects specific keywords and phrases and proposes the countermeasure that will elicit the most positive reaction. In this way, it is possible to predict the user's emotional reaction and propose the countermeasure that will elicit the most positive reaction.

[0084] The user extraction unit can cross-reference the effectiveness of countermeasures across different platforms and propose consistent countermeasures. The user extraction unit, for example, uses a generation AI to cross-reference the effectiveness of countermeasures across different platforms and propose consistent countermeasures. For example, it compares and analyzes the effectiveness of countermeasures on social media, blogs, bulletin boards, etc., and proposes the most effective countermeasures. In addition, to perform the cross-referencing, the generation AI collects countermeasure data from multiple platforms and performs a comparative analysis. For example, it analyzes the effectiveness of countermeasures on different platforms and proposes consistent countermeasures. Furthermore, the generation AI uses a specific algorithm to cross-reference the effectiveness of countermeasures across different platforms. For example, it proposes consistent countermeasures based on the effectiveness and results of the countermeasures. This allows for more effective countermeasures to be taken by proposing consistent countermeasures.

[0085] The user extraction unit can consider expressions specific to different cultures and regions when proposing countermeasures and propose countermeasures based on cultural backgrounds. The user extraction unit, for example, uses a generation AI to consider expressions specific to different cultures and regions when proposing countermeasures and propose countermeasures based on cultural backgrounds. For example, it considers words and expressions used in a specific culture or region and proposes countermeasures. In addition, to propose countermeasures based on cultural backgrounds, the generation AI learns data from different cultures and regions. For example, it analyzes words and expressions used in a specific culture or region and proposes countermeasures. Furthermore, the generation AI uses natural language processing technology to consider expressions specific to different cultures and regions when proposing countermeasures. For example, it detects words and expressions used in a specific culture or region and proposes countermeasures. This allows more effective countermeasures to be taken by proposing countermeasures based on cultural backgrounds.

[0086] The user extraction unit can use the emotion estimation function to predict the user's emotional reaction when proposing a countermeasure and propose the countermeasure that will elicit the most positive reaction. The user extraction unit can, for example, use a generation AI to predict the user's emotional reaction when proposing a countermeasure and propose the countermeasure that will elicit the most positive reaction. For example, it adjusts the content and tone of the warning message to predict the user's reaction. The emotion estimation function also calculates the user's emotional score when proposing a countermeasure and proposes the countermeasure that will elicit the most positive reaction. For example, it prioritizes the proposal of countermeasures with high emotion scores. Furthermore, the generation AI uses natural language processing technology to predict the user's emotional reaction when proposing a countermeasure. For example, it detects specific keywords and phrases and proposes the countermeasure that will elicit the most positive reaction. In this way, it is possible to predict the user's emotional reaction and propose the countermeasure that will elicit the most positive reaction.

[0087] The system according to the embodiment is not limited to the above-described example, and various modifications are possible, for example, as follows.

[0088] The comment collection unit can also analyze the metadata of collected comments to identify trends in anti-comments at specific times and locations. For example, it can identify times and areas where anti-comments are most prevalent based on the time and location of posting. Generative AI can also be used to analyze the metadata of collected comments to identify trends in anti-comments on specific devices and platforms. For example, it can analyze comments from specific smartphones or social media apps. Furthermore, metadata analysis can also identify trends in anti-comments related to specific events or occurrences. For example, it can analyze the tendency for anti-comments to increase when specific news or events occur. This can reveal trends in anti-comments at specific times and locations, allowing for effective countermeasures to be implemented.

[0089] The comment collection unit can also extract text from audio and video comments and collect it as anti-comments. For example, it can use voice recognition technology to convert audio comments into text and collect comments containing negative content. To extract text from video comments, the generation AI can also combine voice recognition technology with video analysis technology. For example, it can convert audio in videos into text and collect comments containing negative content. Furthermore, when extracting text from audio and video comments, the generation AI can also detect specific keywords and phrases. For example, it can prioritize the collection of comments containing defamatory or violent content. This allows for the extraction of text from audio and video comments, thereby collecting a wider variety of comments and improving the accuracy of detecting dangerous comments.

[0090] The comment collection unit can automatically translate comments in different languages ​​and collect anti-comments from a global perspective. For example, it can automatically translate comments in English, French, Chinese, and other languages ​​to collect comments containing negative content. The generation AI can also detect specific keywords and phrases when automatically translating comments in different languages. For example, it can prioritize the collection of comments containing defamatory or violent content. Furthermore, the generation AI can use natural language processing technology to automatically translate comments in different languages. For example, it can analyze the grammar and expressions of specific languages ​​to collect comments containing negative content. This allows automatic translation of comments in different languages ​​to collect anti-comments from a global perspective and improve detection accuracy.

[0091] The comment screening unit can also analyze text within images and videos and classify them as anti-comments. For example, it can extract text within images and classify comments containing defamatory or violent content. To analyze text within videos, the generative AI can also combine speech recognition technology and video analysis technology. For example, it can convert the audio within a video into text and classify comments containing negative content. Furthermore, when analyzing text within images and videos, the generative AI can detect specific keywords and phrases. For example, it can prioritize the classification of comments containing defamatory or violent content. By analyzing text within images and videos, this allows for screening a wider variety of comments and improves the accuracy of detecting dangerous comments.

[0092] The comment screening unit can take into account expressions specific to different cultures and regions and perform screening and classification based on cultural background. For example, it can analyze slang and expressions used in a particular culture or region and classify comments containing negative content. In addition, to perform screening based on cultural background, the generative AI can also learn data from different cultures and regions. For example, it can analyze words and expressions used in a particular culture or region and classify comments containing negative content. Furthermore, the generative AI can use natural language processing technology to take into account expressions specific to different cultures and regions. For example, it can detect words and expressions used in a particular culture or region and classify comments containing negative content. This improves the accuracy of screening and classification by taking into account expressions specific to different cultures and regions.

[0093] The comment inspection unit can use the emotion estimation function to analyze the emotional tone of comments and prioritize classification of comments that evoke particularly strong emotional responses. For example, it can classify comments that evoke strong emotions such as anger or sadness. The emotion estimation function can also be used to calculate the emotion score of comments and prioritize classification of comments that evoke particularly strong emotional responses. For example, it can classify comments with a high emotion score. Furthermore, the generation AI can use natural language processing technology to analyze the emotional tone of comments. For example, it can detect specific keywords and phrases and classify comments that evoke particularly strong emotional responses. This improves the accuracy of detecting dangerous comments by prioritizing classification of comments that evoke particularly strong emotional responses.

[0094] The user extraction unit can analyze the intent of dangerous comments and assess their actual risk, rather than simply slander. For example, it can analyze the context and wording of the comment to assess the likelihood of actual action being taken. The generation AI can also use natural language processing technology to analyze the intent behind the comments. For example, it can understand the meaning and intent of the comment and assess its actual risk. Furthermore, the generation AI can learn from past comment data to analyze the intent behind the selected comment. For example, it can analyze the intent based on similar comments and assess the actual risk. This allows for more effective countermeasures to be taken by assessing the actual risk.

[0095] The user extraction unit can analyze the past behavioral history of the poster of the selected comment and evaluate the likelihood of recidivism. For example, the likelihood of recidivism can be evaluated based on a history of posting defamatory or violent comments in the past. In addition, to analyze the poster's past behavioral history, the generation AI can also learn from past comment data. For example, it can analyze the patterns and characteristics of past comments to evaluate the likelihood of recidivism. Furthermore, the generation AI can use specific algorithms to analyze the past behavioral history of the poster of the selected comment. For example, it can evaluate the likelihood of recidivism based on the frequency and content of past comments. This allows more effective measures to be taken by evaluating the likelihood of recidivism.

[0096] The user extraction unit can use the emotion estimation function to evaluate the emotional impact of the picked comments and prioritize comments that have a particularly strong emotional impact. For example, comments that express strong emotions such as anger or sadness can be picked up. The emotion estimation function can also be used to calculate an emotion score for the picked comments and prioritize comments that have a particularly strong emotional impact. For example, comments with a high emotion score can be picked up. Furthermore, the generation AI can use natural language processing technology to evaluate the emotional impact of the picked comments. For example, specific keywords and phrases can be detected to pick up comments that have a strong emotional impact. This prioritizes picking up comments that have a strong emotional impact, thereby improving the accuracy of detecting dangerous comments.

[0097] The user extraction unit can cross-reference the picked comments across different platforms to identify consistently dangerous comments. For example, if a comment with the same content is posted on multiple platforms, it can identify that comment. To perform this cross-referencing, the generation AI can also collect comments from multiple platforms and perform comparative analysis. For example, it can analyze comments collected from social media, blogs, message boards, etc. to identify consistently dangerous comments. Furthermore, the generation AI can use specific algorithms to cross-reference the picked comments across different platforms. For example, it can identify consistently dangerous comments based on the content of the comment and information about the poster. This allows for more effective countermeasures to be taken by identifying consistently dangerous comments.

[0098] The processing flow of the second embodiment will be briefly explained below.

[0099] Step 1: The comment collection unit collects comments from multiple online platforms. For example, it automatically collects comments from social media, blogs, bulletin boards, etc. The comment collection unit also uses generative AI to estimate the poster's emotions in real time as the comments are collected, and prioritizes the collection of comments with particularly negative sentiment. The comment collection unit then expands the platforms it collects to include comments from the dark web and anonymous bulletin boards. It analyzes the metadata of the collected comments (posting time, posting location, posting device, etc.) to identify trends in anti-comments at specific times and locations. It also extracts text from audio and video comments and collects them as anti-comments. It automatically translates comments in different languages ​​to collect anti-comments from a global perspective. It uses an emotion estimation function to analyze the emotional tone of the collected comments and prioritize the collection of comments that evoke particularly strong emotional reactions. Step 2: The comment screening unit screens the collected comments. Using generative AI, it understands the context of the comment and performs a context-based screening rather than simply keyword matching. It references a database of similar past comments to improve accuracy. It uses an emotion estimation function to evaluate the emotional impact of the screened comments and classifies comments that have a particularly strong emotional impact. It also analyzes text within images and videos and classifies them as anti-comments. It screens and classifies based on cultural background, taking into account expressions unique to different cultures and regions. Step 3: The comment classification unit classifies the scrutinized comments. Using generative AI, it analyzes the emotional tone of the comments and prioritizes comments that evoke particularly strong emotional responses. Step 4: The user extraction unit extracts users who leave dangerous comments from the classified comments. Generative AI is used to analyze the intention behind the picked comments and evaluate their actual risk. The past behavioral history of the picked comment poster is analyzed to evaluate the likelihood of recidivism. Sentiment estimation is used to evaluate the emotional impact of the picked comments and prioritize comments with particularly strong emotional impact. The groups and communities behind the picked comments are identified to detect organized slander. Picked comments are cross-referenced across different platforms to identify consistent dangerous comments.

[0100] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0101] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> Examples of generative AIs include the data generation model 58, such as a neural network model (e.g., a neural network model), and a neural network model (e.g., a neural network model). The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating speech, text data indicating text, and image data indicating an image is also input to the data generation model 58. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The specification processing unit 290 performs the above-mentioned specification processing using the data generation model 58. The data generation model 58 may be a fine-tuned model so as to output an inference result from a prompt that does not include an instruction. In this case, the data generation model 58 can output an inference result from a prompt that does not include an instruction. The data processing device 12 and the like include multiple types of data generation models 58, and the data generation model 58 includes AIs other than the generative AI. The AI ​​other than the generative AI may be, for example, linear regression, logistic regression, decision tree, random forest, support vector machine (SVM), k-means clustering, convolutional neural network (CNN), recurrent neural network (RNN), generative adversarial network (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. The AI ​​may also be an AI agent. When the processes of each of the above-mentioned parts are performed by AI, the processes may be performed in part or entirely by AI, but are not limited to these examples. The processes performed by AI, including the generative AI, may be replaced with rule-based processes.

[0102] Furthermore, the processing by the data processing system 10 described above is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart device 14, but may also be executed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart device 14. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information necessary for processing from the smart device 14 or an external device, and the smart device 14 acquires or collects information necessary for processing from the data processing device 12 or an external device.

[0103] [Second embodiment] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0104] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0105] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN and / or a LAN.

[0106] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0107] The microphone 238 receives instructions and the like from the user by receiving voice uttered by the user. The microphone 238 captures the voice uttered by the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to instructions from the processor 46.

[0108] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the user's surroundings (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0109] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0110] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0111] 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.

[0112] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate a user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotion, including estimation and prediction of the user's emotion, but is not limited to these examples. Furthermore, the estimation and prediction of emotion also includes, for example, emotion analysis.

[0113] In the smart glasses 214, the specific processing is performed by the processor 46. A specific processing program 60 is stored in the storage 50. The processor 46 reads the specific processing program 60 from the storage 50 and executes the read specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as the control unit 46A in accordance with the specific processing program 60 executed on the RAM 48. The smart glasses 214 also have a data generation model and an emotion identification model similar to the data generation model 58 and the emotion identification model 59, and can perform processing similar to that of the specific processing unit 290 using these models.

[0114] Note that a device other than the data processing device 12 may have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 communicates with the server device having the data generation model 58 to obtain a processing result (such as a prediction result) using the data generation model 58. Furthermore, the data processing device 12 may be a server device, or may be a terminal device (for example, a mobile phone, a robot, a home appliance, etc.) owned by a user.

[0115] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0116] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 receives a prompt containing an instruction, as well as inference data such as voice data representing speech, text data representing text, and image data representing an image. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The identification processing unit 290 performs the above-mentioned identification processing using the data generation model 58. The data generation model 58 may be a fine-tuned model so as to output an inference result from a prompt that does not include an instruction. In this case, the data generation model 58 can output an inference result from a prompt that does not include an instruction. The data processing device 12 and the like include multiple types of data generation models 58, and the data generation model 58 includes AIs other than the generative AI. The AI ​​other than the generative AI may be, for example, linear regression, logistic regression, decision tree, random forest, support vector machine (SVM), k-means clustering, convolutional neural network (CNN), recurrent neural network (RNN), generative adversarial network (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. The AI ​​may also be an AI agent. When the processes of each of the above-mentioned parts are performed by AI, the processes may be performed in part or entirely by AI, but are not limited to these examples. The processes performed by AI, including the generative AI, may be replaced with rule-based processes.

[0117] The data processing system 210 according to the second embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 210 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the smart glasses 214, but may also be executed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the smart glasses 214. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the smart glasses 214 or an external device, etc., and the smart glasses 214 acquires or collects information required for processing from the data processing device 12 or an external device, etc.

[0118] [Third embodiment] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0119] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0120] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN and / or a LAN.

[0121] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0122] The microphone 238 receives instructions and the like from the user by receiving voice uttered by the user. The microphone 238 captures the voice uttered by the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to instructions from the processor 46.

[0123] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the user's surroundings (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0124] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0125] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0126] 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.

[0127] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate a user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotion, including estimation and prediction of the user's emotion, but is not limited to these examples. Furthermore, the estimation and prediction of emotion also includes, for example, emotion analysis.

[0128] In the headset type terminal 314, the specific processing is performed by the processor 46. A specific processing program 60 is stored in the storage 50. The processor 46 reads the specific processing program 60 from the storage 50 and executes the read specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as the control unit 46A in accordance with the specific processing program 60 executed on the RAM 48. Note that the headset type terminal 314 has a data generation model and an emotion identification model similar to the data generation model 58 and the emotion identification model 59, and can also perform processing similar to that of the specific processing unit 290 using these models.

[0129] Note that a device other than the data processing device 12 may have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 communicates with the server device having the data generation model 58 to obtain a processing result (such as a prediction result) using the data generation model 58. Furthermore, the data processing device 12 may be a server device, or may be a terminal device (for example, a mobile phone, a robot, a home appliance, etc.) owned by a user.

[0130] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0131] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 receives a prompt containing an instruction, as well as inference data such as voice data representing speech, text data representing text, and image data representing an image. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The identification processing unit 290 performs the above-mentioned identification processing using the data generation model 58. The data generation model 58 may be a fine-tuned model so as to output an inference result from a prompt that does not include an instruction. In this case, the data generation model 58 can output an inference result from a prompt that does not include an instruction. The data processing device 12 and the like include multiple types of data generation models 58, and the data generation model 58 includes AIs other than the generative AI. The AI ​​other than the generative AI may be, for example, linear regression, logistic regression, decision tree, random forest, support vector machine (SVM), k-means clustering, convolutional neural network (CNN), recurrent neural network (RNN), generative adversarial network (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. The AI ​​may also be an AI agent. When the processes of each of the above-mentioned parts are performed by AI, the processes may be performed in part or entirely by AI, but are not limited to these examples. The processes performed by AI, including the generative AI, may be replaced with rule-based processes.

[0132] The data processing system 310 according to the third embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 310 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the headset type terminal 314, but may also be executed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the headset type terminal 314. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the headset type terminal 314 or an external device, etc., and the headset type terminal 314 acquires or collects information required for processing from the data processing device 12 or an external device, etc.

[0133] [Fourth embodiment] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[0134] 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.

[0135] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN and / or a LAN.

[0136] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[0137] The microphone 238 receives instructions and the like from the user by receiving voice uttered by the user. The microphone 238 captures the voice uttered by the user, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to instructions from the processor 46.

[0138] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS image sensor or a CCD image sensor, and captures images of the user's surroundings (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0139] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0140] The control 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 emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[0141] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0142] 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.

[0143] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290. The identification processing unit 290 can estimate a user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion. The emotion estimation function (emotion identification function) using the emotion identification model 59 performs various estimations and predictions regarding the user's emotion, including estimation and prediction of the user's emotion, but is not limited to these examples. Furthermore, the estimation and prediction of emotion also includes, for example, emotion analysis.

[0144] In the robot 414, the specific processing is performed by the processor 46. A specific processing program 60 is stored in the storage 50. The processor 46 reads the specific processing program 60 from the storage 50 and executes the read specific processing program 60 on the RAM 48. The specific processing is realized by the processor 46 operating as the control unit 46A in accordance with the specific processing program 60 executed on the RAM 48. The robot 414 has a data generation model and an emotion identification model similar to the data generation model 58 and the emotion identification model 59, and can also perform processing similar to that of the specific processing unit 290 using these models.

[0145] Note that a device other than the data processing device 12 may have the data generation model 58. For example, a server device may have the data generation model 58. In this case, the data processing device 12 communicates with the server device having the data generation model 58 to obtain a processing result (such as a prediction result) using the data generation model 58. Furthermore, the data processing device 12 may be a server device, or may be a terminal device (for example, a mobile phone, a robot, a home appliance, etc.) owned by a user.

[0146] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0147] The data generation model 58 is a so-called generative AI. An example of the data generation model 58 is a generative AI such as ChatGPT. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 receives a prompt containing an instruction, as well as inference data such as voice data representing speech, text data representing text, and image data representing an image. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization. The identification processing unit 290 performs the above-mentioned identification processing using the data generation model 58. The data generation model 58 may be a fine-tuned model so as to output an inference result from a prompt that does not include an instruction. In this case, the data generation model 58 can output an inference result from a prompt that does not include an instruction. The data processing device 12 and the like include multiple types of data generation models 58, and the data generation model 58 includes AIs other than the generative AI. The AI ​​other than the generative AI may be, for example, linear regression, logistic regression, decision tree, random forest, support vector machine (SVM), k-means clustering, convolutional neural network (CNN), recurrent neural network (RNN), generative adversarial network (GAN), or naive Bayes, and can perform various processes, but is not limited to these examples. The AI ​​may also be an AI agent. When the processes of each of the above-mentioned parts are performed by AI, the processes may be performed in part or entirely by AI, but are not limited to these examples. The processes performed by AI, including the generative AI, may be replaced with rule-based processes.

[0148] The data processing system 410 according to the fourth embodiment performs the same processing as the data processing system 10 according to the first embodiment. The processing by the data processing system 410 is executed by the specific processing unit 290 of the data processing device 12 or the control unit 46A of the robot 414, but may also be executed by the specific processing unit 290 of the data processing device 12 and the control unit 46A of the robot 414. Furthermore, the specific processing unit 290 of the data processing device 12 acquires or collects information required for processing from the robot 414 or an external device, etc., and the robot 414 acquires or collects information required for processing from the data processing device 12 or an external device, etc.

[0149] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[0150] FIG. 9 illustrates an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and behaviors arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion encompasses both emotions and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[0151] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[0152] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[0153] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is expressed, and when they approach the ideal, a state of pleasure is expressed. Emotions can also be created for robots, cars, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is expressed, and when they approach the ideal, a state of pleasure is expressed. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on speech emotion recognition and brain physiological signal analysis systems for emotions, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the area called "reaction," where sensation is dominant. The right half of the emotion map lists emotions belonging to the area called "situation," where situational awareness is dominant.

[0154] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[0155] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[0156] In the above embodiment, an example was given in which a specific process is performed by one computer 22, but the technology disclosed herein is not limited to this, and distributed processing of the specific process may be performed by multiple computers including computer 22.

[0157] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[0158] 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.

[0159] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[0160] The hardware resource for executing a specific process can be any of the following types of processors: A processor, for example, is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. A processor also includes a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[0161] The hardware resource that executes the specific process may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific process may be a single processor.

[0162] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[0163] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[0164] In the above example, the first to fourth embodiments have been described separately, but some or all of these embodiments may be combined. The smart device 14, smart glasses 214, headset terminal 314, and robot 414 are merely examples, and they may be combined, or other devices may be used. In the above example, the first and second embodiments have been described separately, but they may be combined.

[0165] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[0166] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference. [Explanation of symbols]

[0167] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot

Claims

1. a comment collection unit that collects comments from multiple platforms on the Internet; a comment review unit that reviews the comments collected by the comment collection unit; a comment classification unit that classifies the comments reviewed by the comment review unit; a user extraction unit that extracts users who have left dangerous comments from among the comments classified by the comment classification unit. A system characterized by:

2. The comment collection unit When collecting comments, the system estimates the poster's sentiment in real time and prioritizes collecting comments with particularly negative sentiment.

2. The system of claim 1.

3. The comment collection unit Expand the scope of the platforms mentioned above to collect comments from the dark web and anonymous message boards.

2. The system of claim 1.

4. The comment collection unit Analyzing the metadata of collected comments reveals trends in anti-comments at specific times and locations.

2. The system of claim 1.

5. The comment collection unit Extract text from audio or video comments and collect it as anti-comments 2. The system of claim 1.

6. The comment collection unit Automatically translate comments in different languages ​​and collect anti-comments from a global perspective.

2. The system of claim 1.

Images