System

Abstract

A system is provided.SOLUTION: A system comprising: a terminal for receiving a malicious reply; a server for collecting past posts of a poster of the malicious reply; means for extracting frequent words from the past posts; means for analyzing the extracted frequent words; means for extracting images from the past posts; means for analyzing the extracted images; means for predicting a persona of the poster based on the analysis; and means for providing the predicted persona information to the terminal.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] Due to the high level of anonymity on social media, when users receive malicious replies (hereafter referred to as "kuso-replies"), they often feel excessive anxiety because they do not know the identity of the attacker.The objective of this invention is to alleviate this anxiety by providing a system that analyzes frequently used words and images from the attacker's past posts and predicts and presents a standard persona of the attacker. [Means for solving the problem]

[0005] In this invention, when a user receives a 'kuso-reply', the device first sends the content and the poster's ID to the system. The server then collects the poster's past posts using a social media API. From the collected posts, frequently occurring words are extracted using natural language processing (NLP) technology, and images are also extracted and analyzed. Based on the analysis results, a machine learning model is used to predict the poster's age, gender, occupation, hobbies, and preferences, and persona information is generated. Finally, the generated persona information is provided to the device and presented to the user. This allows the user to understand the profile of the attacker and reduce anxiety.

[0006] A "terminal" is a device that allows a user to use SNS and has the function of receiving shitty replies.

[0007] A "server" is a computer system that works in conjunction with terminals via a network to collect, analyze, and provide data.

[0008] "Malicious replies" refer to posts on social media that are intended to attack or slander others.

[0009] "Poster" refers to the owner of the account that posted the malicious reply on social media.

[0010] "Past posts" refers to all posts that the poster has made in the past on social media.

[0011] "SNS API" means an application programming interface provided by a social media platform that enables data collection.

[0012] "Frequent words" refer to words that appear particularly frequently in past posting data.

[0013] "Image" refers to visual content uploaded by a poster to a social networking site.

[0014] "Natural language processing (NLP) technology" is a technology for analyzing text data and extracting meaning and patterns.

[0015] A "machine learning model" is an algorithm that learns the characteristics of data and makes predictions and classifications.

[0016] A "persona" is a character image inferred from a combination of the poster's age, gender, occupation, hobbies, and preferences.

[0017] "Providing" means the act of sending analyzed and predicted data to a terminal and presenting it in a form that can be seen by the user. [Brief explanation of the drawings]

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

[0019] 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.

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

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

[0022] 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.

[0023] 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.

[0024] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0025] 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."

[0026] [First embodiment]

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

[0028] 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.

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

[0030] 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.

[0031] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0032] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0033] 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.

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

[0035] 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.

[0036] 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.

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

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

[0039] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the profile of the attacker and displays it to the user, thereby reducing anxiety. This system uses the user's device, server, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[0040] System Configuration

[0041] 1. User Device

[0042] It is a device that allows users to use SNS and can receive Kuso-Replies. It also has the function of sending the content of Kuso-Replies and the poster ID to the server.

[0043] 2. Server

[0044] The server uses SNS APIs to collect data on attackers' past posts based on the Kuso-Reply information received from the user's device. The collected data is then subjected to text and image analysis to extract frequently occurring words and image features. Furthermore, a machine learning model is used to predict the attacker's persona, and the results are sent to the user's device.

[0045] 3. Social Media API

[0046] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[0047] Program processing

[0048] 1. Receiving shitty replies

[0049] A user receives a shitty reply on social media and checks its content and the poster's ID.

[0050] The device sends the content of the shit reply and the poster ID to the server.

[0051] 2. Attacker Identification and Data Collection

[0052] The server receives the ID of the user who posted the kusorip and uses the SNS API to collect the attacker's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0053] 3. Extracting and analyzing text data

[0054] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques.

[0055] 4. Image Data Extraction and Analysis

[0056] The server extracts image data from the collected submissions, extracts image features using image analysis technology (e.g., computer vision and deep learning), and assigns tags such as characters and symbols to the images.

[0057] 5. Data integration for persona prediction

[0058] The server combines the results of text analysis and image analysis to create a comprehensive dataset, which is then fed into a machine learning model to predict the attacker's age, gender, occupation, hobbies, and preferences.

[0059] 6. Persona generation and provision

[0060] The server generates a standard persona of an attacker based on the prediction results of the machine learning model. The generated persona information is structured in a format that is easy for users to understand.

[0061] The device provides the user with the persona information received from the server and displays information such as, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0062] Specific examples

[0063] Let's say a user receives a malicious reply saying, "Your opinion is off the mark." The device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API and analyzes the text data and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict the persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." The prediction result is then sent to the user's device, and the attacker's standard persona information is presented to the user.

[0064] The above is a specific embodiment for carrying out the present invention, which allows the user to understand the personality of the sender of the malicious reply and reduce anxiety.

[0065] The processing flow will be explained below.

[0066] Step 1:

[0067] A user receives a malicious reply on a social networking site, which appears as a reply to a specific post.

[0068] Step 2:

[0069] The device obtains the content of the Kuso-Reply and the poster's ID, and sends this to the system's server, including the text content of the Kuso-Reply and the poster's account ID.

[0070] Step 3:

[0071] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0072] Step 4:

[0073] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and creates a clean dataset.

[0074] Step 5:

[0075] The server uses natural language processing (NLP) technology to analyze the extracted text data. This analysis identifies frequently occurring words and key phrases. Specifically, it uses morphological analysis and TF-IDF (Term Frequency-Inverse Document Frequency) techniques.

[0076] Step 6:

[0077] The server extracts the image portion from the collected post data, which includes all image files included in each post.

[0078] Step 7:

[0079] The server uses image analysis technology (e.g., computer vision and deep learning) to analyze the features of the extracted images. Specifically, it performs object recognition and assigns tags such as characters and symbols.

[0080] Step 8:

[0081] The server combines the results of the text analysis and the image analysis, creating a comprehensive dataset.

[0082] Step 9:

[0083] The server then inputs the combined data set into a machine learning model to predict the attacker's persona, including attributes such as age, gender, occupation, and hobbies.

[0084] Step 10:

[0085] The server generates a standard persona for an attacker based on the predictions obtained from the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[0086] Step 11:

[0087] The server transmits the generated persona information to the user's terminal.

[0088] Step 12:

[0089] The device presents the received persona information to the user. This is displayed on the screen in a format that is easy for the user to confirm. For example, it might say, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0090] By going through these steps, users can understand the profile of the attacker and reduce their anxiety.

[0091] Example 1

[0092] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0093] Users who receive malicious replies (kuso-replies) on social media often feel anxious and stressed. However, typical countermeasures include ignoring the replies or blocking the user, which does not fundamentally alleviate the anxiety. Furthermore, there are limited ways to specifically identify the attacker, leaving users plagued by fear of the so-called "invisible enemy." The present invention aims to address these issues by predicting the attacker's profile and presenting it to users, thereby providing them with specific information and reducing their anxiety.

[0094] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0095] In this invention, the server includes a device for receiving malicious replies, a device for collecting past posts by the poster of the malicious replies, means for extracting text from the collected past posts and performing preprocessing, means for analyzing frequently occurring words and key phrases from the preprocessed text, means for extracting images from the collected posts and analyzing their features, means for predicting the poster's personality based on the analysis results, and means for providing the predicted personality information to the device. This allows the user to grasp the specific personality of the attacker, making it possible to reduce anxiety and stress.

[0096] A "malicious reply" is a comment or message sent on social media with the intention of attacking or offending others.

[0097] "Devices" is a general term for devices, machines, and related systems and software used by users to access SNS.

[0098] A "poster" is a person or account that posts comments, messages, images, etc. on social media.

[0099] "Past posts" refer to comments, messages, images, and other output made on social media within a specified period.

[0100] "Text" refers to posts and comments containing written information made on social media.

[0101] "Preprocessing" is the process of removing unnecessary elements (e.g., hashtags and links) from text information and converting it into a format suitable for analysis.

[0102] A "frequent word" is a word that occurs many times within a particular text dataset.

[0103] A "key phrase" is a phrase or word that has an important meaning within text data.

[0104] "Analysis" is the process of analyzing text and image data and extracting meaningful information.

[0105] "Features" are important patterns or characteristics extracted from image data.

[0106] A "personal profile" is the result of predicting an individual's attribute information, such as age, gender, occupation, and hobbies and preferences.

[0107] "Information" refers to data and knowledge such as predictions and analysis results of a person's profile.

[0108] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the profile of the attacker and displays this information to the user, thereby reducing anxiety. This system uses the user's device, server, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the profile of the attacker.

[0109] System Configuration

[0110] 1. User Device

[0111] The user device is a device that allows users to use SNS. It has the function of receiving kusorip messages and sending their content and the poster's ID to a server. When a user receives a kusorip message on SNS, the device sends the content and the poster's ID to the server using an HTTPS request.

[0112] 2. Server

[0113] The server uses SNS APIs to collect data on the attacker's past posts based on the Kuso-Reply information received from the user's device. The collected data is subjected to text analysis and image analysis to extract frequently occurring words and image features. It also uses a machine learning model to predict the attacker's profile and sends the results to the user's device. Specifically, the server preprocesses the text data using "NLTK" and extracts important key phrases using natural language processing technology. It also performs image analysis using "OpenCV" and "TensorFlow" to extract important features.

[0114] 3. Social Media API

[0115] The SNS API is an application programming interface provided by SNS platforms, and is used by the server to collect data on past posts by attackers. Authentication is performed using the SNS API authentication token, and post data from a specified period is obtained.

[0116] Specific example of system operation

[0117] Suppose a user receives a malicious reply saying, "Your opinion is off the mark." The user's device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API. The collected text data is preprocessed using NLTK to extract frequently occurring words and key phrases. At the same time, the collected image data is analyzed using OpenCV or TensorFlow to extract features. The results are integrated, and a machine learning model (e.g., SciKit-Learn) is used to predict a profile of the attacker, such as "male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." Finally, the prediction result is sent to the user's device, and the user is presented with a standard profile of the attacker.

[0118] Prompt Sentence Examples

[0119] An example of a prompt would be:

[0120] I received a shitty reply saying, "Your opinion is off the mark." Please predict and provide a profile of the attacker.

[0121] The above-mentioned method allows users to get a clear picture of the attacker, which can reduce anxiety and stress.

[0122] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0123] Step 1: Receiving a Shit Reply

[0124] A user receives a malicious reply (kuso-repu) on social media. For example, the reply is, "Your opinion is irrelevant."

[0125] Input: Shitty reply on social media and poster ID.

[0126] The device temporarily stores the content of the KusoRip and the poster ID in its internal memory, then sends the content of the KusoRip and the poster ID to the server using an HTTPS request.

[0127] Output: The content of the shit reply sent to the server and the poster ID.

[0128] Step 2: Identifying the attacker and gathering data

[0129] The server checks the content of the kusorip received from the device and the poster ID.

[0130] Input: The content of the shitty reply sent from the device and the poster ID.

[0131] The server uses the SNS API to collect the attacker's posting data for the past year. The server authenticates using the SNS API's authentication token and sends an API request. This request includes the poster ID and the data collection period (e.g., the past year).

[0132] Output: Posted data (text and images) by attackers collected on the server over the past year.

[0133] Step 3: Extract and analyze text data

[0134] The server extracts the text from the collected post data and preprocesses the text of the post to remove hashtags and links.

[0135] Input: Attacker's past posting data (text).

[0136] The server uses NLTK to preprocess the text data, tokenizing it, removing stop words, and performing word frequency analysis, using natural language processing techniques to extract important frequently occurring words and key phrases.

[0137] Output: Preprocessed text data and analysis results (frequent words and key phrases).

[0138] Step 4: Extraction and analysis of image data

[0139] The server extracts the image portion from the collected submission data, extracts image features using computer vision and deep learning technology, and assigns tags such as characters and symbols to them.

[0140] Input: Attacker's past posting data (images).

[0141] The server uses OpenCV and TensorFlow to analyze the image and extract important features, such as identifying and labeling anime characters or specific symbols.

[0142] Output: Analyzed image data and feature information.

[0143] Step 5: Data integration for persona prediction

[0144] The server combines the results of text analysis and image analysis to create a comprehensive dataset.

[0145] Input: Text analysis results and image analysis results.

[0146] The server inputs this dataset into a machine learning model to predict the attacker's personality, such as age, gender, occupation, hobbies, etc. For example, the server inputs the dataset into a machine learning model trained using "SciKit-Learn" to obtain a predicted personality profile.

[0147] Output: Predicted attacker profile (age, gender, occupation, hobbies, etc.).

[0148] Step 6: Generate and deliver personas

[0149] The server generates a standard profile of the attacker based on the predictions of the machine learning model, and the profile information is organized in a format that is easy for users to understand.

[0150] Input: Prediction result (attacker profile).

[0151] The server generates the person profile information in JSON format and sends it to the user device using HTTPS.

[0152] The user device analyzes the persona information received from the server and displays it on the user interface. For example, it displays information such as, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0153] Output: Attacker profile information displayed in the user interface.

[0154] In this way, this system can reduce anxiety and stress by presenting a specific profile of the attacker in response to a malicious reply received by the user.

[0155] (Application example 1)

[0156] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0157] In online social networking services (SNS), users often receive malicious replies (kuso-replies) from other users. These replies can cause psychological stress and anxiety to the recipient. Therefore, there is a need for a method to reduce users' anxiety and provide a safe SNS environment by revealing the causes of these replies and the identity of the people behind them.

[0158] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0159] In this invention, the server includes a means for extracting frequently occurring words from past posts, a means for analyzing the extracted frequently occurring words, and a means for extracting images from past posts. This makes it possible to analyze the poster's personality and predict persona information. Furthermore, the prediction includes a means for using a generative AI model and prompt sentences, making it possible to improve the accuracy of the analysis results. This makes it possible to clarify the identity of the person behind malicious replies to users, allowing users to use SNS without feeling anxious.

[0160] A "terminal" is a device through which a user receives malicious replies and transmits the information to a server.

[0161] The "server" is a computer system that receives information about the poster of a malicious reply sent from a user's terminal, collects and analyzes the poster's past posting data, and provides persona information.

[0162] "Method for extracting frequently occurring words" is a method for selecting frequently used words from posted data.

[0163] "Means for analyzing frequently occurring words" is a method for analyzing extracted words and finding their meanings and relationships.

[0164] "Means for extracting images" refers to a method for selecting image media from posted data.

[0165] The "means for analyzing images" is a method for analyzing the extracted images and recognizing their contents and features.

[0166] A "means for predicting persona" is a method for inferring characteristics such as the poster's age, gender, hobbies and preferences based on collected and analyzed data.

[0167] "Using a generative AI model" means using a model generated using artificial intelligence technology.

[0168] "Prompt methods" are methods that use sentences to provide a generative AI model with a specific question and context to get an answer.

[0169] "Information Provision Interface" means the application programming interface (API) that the server uses to collect data from the social networking platform.

[0170] The system based on this invention reduces anxiety when a user receives a malicious reply (kuso-reply) on social media by predicting the identity of the attacker and providing this information to the user. The system uses the user's device, server, and social media API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[0171] System Configuration

[0172] 1. User Device

[0173] It is a device that allows users to use social networking services, and when it receives a shitty reply, it sends the content and the poster's ID to the server.

[0174] 2. Server

[0175] The server uses SNS APIs to collect data on the attacker's past posts based on the Kuso-Reply information received from the user's device. The collected data is then subjected to text and image analysis. Natural language processing (NLP) technology is used for text analysis, and computer vision and deep learning technology are used for image analysis. Furthermore, a generative AI model is used to predict the attacker's persona. A prompt sentence is used to make highly accurate predictions.

[0176] 3. Social Media API

[0177] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[0178] Program processing example

[0179] Receiving and sending shit replies

[0180] The user receives a kusorip message on a social networking site and enters the message's content and the poster's ID into their device, which then sends this information to the server.

[0181] Attacker Identification and Data Collection

[0182] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's posting data for the past year.

[0183] Text data extraction and analysis

[0184] The server extracts text from the collected post data and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques, such as NLTK and spaCy libraries.

[0185] Image data extraction and analysis

[0186] The server extracts image data from the collected submissions and uses computer vision technology to extract image features, using deep learning technologies such as TensorFlow and PyTorch.

[0187] Persona Prediction

[0188] The server integrates the results of text analysis and image analysis and predicts the attacker's persona based on a generative AI model. The prompt sentence is used as input to the model, and persona information is predicted with high accuracy.

[0189] Providing persona information

[0190] The predicted persona information is sent to the user's device and displayed to the user, allowing the user to understand the profile of the attacker and reduce anxiety.

[0191] Specific examples

[0192] When a user receives a crappy reply saying, "Your opinion is completely meaningless," the device sends the content of the crappy reply and the poster's ID to the server. The server collects the poster's posting data from the past year via SNS APIs and analyzes the text data and image data. For example, if "games" and "anime" are detected as frequently occurring words and image analysis reveals that specific character images appear frequently, the generative AI model is used to predict a persona such as "male in his 20s, likes anime and games, unemployed or working part-time." The prediction results are then sent to the user's device and displayed to the user.

[0193] Example prompt sentence:

[0194] Please predict the attacker's persona based on the following text analysis and image analysis results. Text analysis results: ['game', 'anime'], Image analysis results: ['anime character']

[0195] The above is a specific embodiment for carrying out the invention.

[0196] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0197] Step 1:

[0198] The user receives a kusorip message and enters its contents and the poster's ID into the device.

[0199] When a user receives a malicious reply (kuso-reply) on social media, they enter the content and the sender's ID into a dedicated application. This entered data (the content of the kuso-reply and the poster's ID) is saved on the device.

[0200] Step 2:

[0201] The device sends the content of the Kuso-Reply and the poster ID to the server.

[0202] The device collects the text of the entered kusorip and the poster ID and sends it to the server. This is done using a method such as an HTTP POST request. The input is the kusorip and the poster ID, and the output is a status message confirming that this information has been sent to the server.

[0203] Step 3:

[0204] The server uses the SNS API to collect the attacker's past posting data.

[0205] The server uses the received poster ID to collect the attacker's past posting data via the SNS platform's API. The collected data includes text posts, images, videos, etc., covering the past year. The input is the poster ID, and the output is the collected posting data.

[0206] Step 4:

[0207] Extract the text data collected by the server

[0208] The server extracts text parts from the collected submission data using regular expressions and NLP preprocessing techniques. The input is the submission data, and the output is the extracted text data.

[0209] Step 5:

[0210] The server analyzes the text data using NLP techniques.

[0211] The server preprocesses the extracted text data to identify frequently occurring words and key phrases. The libraries used are NLTK and spaCy. The input is text data, and the output is frequently occurring words and key phrases. Specific operations include tokenization and stop word removal.

[0212] Step 6:

[0213] Extracting image data collected by the server

[0214] The server extracts images from the collected submission data by filtering the data type. The input is the submission data, and the output is the extracted image data.

[0215] Step 7:

[0216] The server analyzes the image data using image analysis technology.

[0217] The server analyzes the extracted image data using computer vision techniques, such as deep learning frameworks like TensorFlow and PyTorch. The input is image data, and the output is image features and labels. Specific operations include object detection and image classification.

[0218] Step 8:

[0219] The server integrates the results of text analysis and image analysis and predicts the persona using a generative AI model.

[0220] The server uses a dataset that integrates the results of text analysis and image analysis, and inputs a prompt into the generative AI model to predict the attacker's persona. The input is the integrated dataset and prompt, and the output is the persona prediction result. Example prompt: "Please predict the attacker's persona based on the following text analysis and image analysis results. Text analysis result: ['Game', 'Anime'], Image analysis result: ['Anime character']"

[0221] Step 9:

[0222] The server sends the predicted persona information to the user device.

[0223] The server sends the predicted persona information to the user's device. This is done via an HTTP POST request, etc. The input is the persona prediction result, and the output is the confirmation status of the transmission to the user's device.

[0224] Step 10:

[0225] The user device displays the persona information to the user.

[0226] The user device displays the received persona information on the user screen. This allows the user to understand the profile of the attacker and reduce anxiety. The input is persona information, and the output is the displayed persona information. Specifically, it operates by displaying text and images on the interface.

[0227] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0228] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the attacker's personality and further recognizes the user's emotions and adjusts the presentation method to reduce anxiety. This system uses the user's device, server, emotion engine, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[0229] System Configuration

[0230] 1. User Device

[0231] It is a device that allows users to use SNS and can receive Kuso-Reply messages. It also has the function of sending the content of the Kuso-Reply message and the poster's ID to the server. It also has the function of providing the user's voice and facial expressions to the emotion engine.

[0232] 2. Server

[0233] The server uses SNS APIs to collect attacker's past posting data based on the information on kusoripulp received from the user's device and the user's emotional data. The collected data is subjected to text and image analysis to extract frequently occurring words and image features. Furthermore, a machine learning model is used to predict the attacker's persona and the results are sent to the user's device.

[0234] 3. Emotion Engine

[0235] The emotion engine collects and analyzes the user's voice and facial expression data to recognize the user's emotional state. It also predicts emotions based on the user's past posts and behavioral history, and adjusts the way persona information is presented.

[0236] 4. Social Media API

[0237] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[0238] Program processing

[0239] 1. Receiving shitty replies

[0240] A user receives a malicious reply on a social networking site and checks its content and the poster's ID.

[0241] The device sends the content of the shit reply and the poster ID to the server.

[0242] 2. Attacker Identification and Data Collection

[0243] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0244] 3. Extracting and analyzing text data

[0245] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques.

[0246] 4. Image Data Extraction and Analysis

[0247] The server extracts image data from the collected submissions, extracts image features using image analysis technology (e.g., computer vision and deep learning), and assigns tags such as characters and symbols to the images.

[0248] 5. Emotion recognition

[0249] The device continuously collects the user's voice and facial expression data and sends it to the emotion engine.

[0250] The emotion engine analyzes this data to recognize the user's emotional state and predicts their emotions based on their past posts and behavioral history.

[0251] 6. Data integration for persona prediction

[0252] The server combines the results of text analysis and image analysis to create a comprehensive dataset, which is then fed into a machine learning model to predict the attacker's age, gender, occupation, hobbies, and preferences.

[0253] 7. Persona generation and provision

[0254] The server generates a standard persona for an attacker based on the predictions of the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[0255] The emotion engine adjusts the presentation of persona information based on the user's perceived emotions. For example, if the user is under a lot of stress, it will display softer expressions and encouraging words.

[0256] 8. Displaying the results

[0257] The server transmits the generated persona information to the user's terminal.

[0258] The device then presents the received persona information to the user. For example, information such as "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time," is displayed in a format that is adjusted to reflect the user's emotional state.

[0259] Specific examples

[0260] Suppose a user receives a malicious reply saying, "You're such an idiot." The device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API and analyzes the text and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict a persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." During this time, the device sends the user's voice and facial expression data to an emotion engine, which then recognizes the user's stress level. The emotion engine then adjusts the persona information and sends it from the server to the device. Finally, the persona information is presented in a way that takes into account the user's psychological state.

[0261] The above is a specific example of how to implement the present invention, which allows users to understand the personality of the sender of a malicious reply and reduce anxiety. By incorporating an emotion engine, information is provided that takes into consideration the user's psychological state.

[0262] The processing flow will be explained below.

[0263] Step 1:

[0264] A user receives a malicious reply (kuso-reply) on a social networking site, which appears as a comment on a specific post.

[0265] Step 2:

[0266] The device obtains the content of the Kuso-Reply and the poster's ID, and sends this to the system's server, including the text content of the Kuso-Reply and the poster's account ID.

[0267] Step 3:

[0268] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0269] Step 4:

[0270] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and creates a clean dataset.

[0271] Step 5:

[0272] The server uses natural language processing (NLP) technology to analyze the extracted text data. This analysis identifies frequently occurring words and key phrases. Specifically, it uses morphological analysis and TF-IDF (Term Frequency-Inverse Document Frequency) techniques.

[0273] Step 6:

[0274] The server extracts the image portion from the collected post data, which includes all image files included in each post.

[0275] Step 7:

[0276] The server uses image analysis technology (e.g., computer vision and deep learning) to analyze the features of the extracted images. Specifically, it performs object recognition and assigns tags such as characters and symbols.

[0277] Step 8:

[0278] The server combines the results of the text analysis and the image analysis, creating a comprehensive dataset.

[0279] Step 9:

[0280] The server then inputs the combined data set into a machine learning model to predict the attacker's persona, including attributes such as age, gender, occupation, and hobbies.

[0281] Step 10:

[0282] The server generates a standard persona for an attacker based on the predictions obtained from the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[0283] Step 11:

[0284] The device continuously collects the user's voice and facial expression data and sends it to the emotion engine.

[0285] Step 12:

[0286] The emotion engine analyzes this data to recognize the user's emotional state and predicts their emotions based on their past posts and behavioral history.

[0287] Step 13:

[0288] The emotion engine adjusts the presentation of persona information based on the user's emotional state. For example, if a user is under a lot of stress, it will display softer language and encouraging words.

[0289] Step 14:

[0290] The server transmits the generated persona information to the user's terminal.

[0291] Step 15:

[0292] The device presents the received persona information to the user. This is displayed on the screen in a format that is easy for the user to confirm. For example, it might say, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0293] Through these steps, users can understand the profile of the attacker and reduce their anxiety.By incorporating an emotion engine, information is provided that takes into account the user's psychological state.

[0294] Example 2

[0295] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0296] On modern social media platforms, malicious replies (kusoripu) can cause psychological stress to users. However, there is no way for users who receive such replies to understand who is attacking them, and there is no way to provide appropriate information depending on their psychological state, making it difficult to alleviate their anxiety and stress.

[0297] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for collecting past posts by posters of malicious replies, means for extracting frequently occurring words from past posts, means for analyzing the extracted images, means for predicting the poster's persona based on the analysis results, an emotion engine for collecting and analyzing user emotion data, means for adjusting the predicted persona information based on the user's emotional state, and means for providing the adjusted persona information to the terminal. This makes it easier for the user to grasp the profile of the attacker and further enables the provision of appropriate information based on the user's psychological state.

[0298] A "malicious reply" is a comment or message on a social networking site that contains offensive, derogatory, or unpleasant content directed at a user.

[0299] A "terminal" is a device that allows a user to use a social networking service and has the function of receiving malicious replies and sending their contents and the poster's ID to the server.

[0300] A "server" is a computer system that receives data sent from a user's device, collects the necessary data through the SNS API, and performs analytical processing.

[0301] "Past posts" refers to all content posted by an attacker on social media within a specific period of time.

[0302] "Frequent words" are keywords and phrases that appear frequently and are extracted from past posts.

[0303] An "extraction method" is a set of algorithms or techniques used to collect and analyze specific data (e.g., text or images).

[0304] "Means for analysis" refers to methods and techniques for analyzing the extracted data and extracting useful information and features.

[0305] A "persona" is a predicted profile of an attacker based on the analysis results, including their age, gender, occupation, hobbies, and preferences.

[0306] An "emotion engine" is a system or technology that collects and analyzes a user's emotional data to recognize the user's psychological state.

[0307] "SNS API" is an application programming interface provided by an SNS platform and is an interface used to collect past posting data.

[0308] A "machine learning model" is an algorithm that learns using large amounts of data and uses the learning results to make predictions and analyze new data.

[0309] "Means for adjusting based on the user's emotional state" refers to techniques or methods for changing the method or content of information presentation depending on the user's psychological state.

[0310] "Means for providing" refers to communication and interface technologies for displaying analysis results and predicted information on a user terminal.

[0311] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the identity of the attacker and further recognizes the user's emotions and adjusts the presentation method to reduce anxiety. This system utilizes the user's device, server, emotion engine, and social networking site API, and is implemented in the following specific steps.

[0312] First, when a user receives a malicious reply on a social networking site, the device sends the content of the kusoriply and the poster's ID to a server. The device can be a smartphone or a PC, and the social networking application runs as the communication software.

[0313] The server receives the Kuso-Reply information sent from the device and uses the SNS API to collect the poster's past posting data. The collected data is classified into text data and image data, and each type of data is analyzed. For example, the text data is analyzed using natural language processing (NLP) technology to extract frequently occurring words and key phrases, and the image data is analyzed using computer vision and deep learning technology to analyze the image features.

[0314] Next, the emotion engine collects the user's voice and facial expression data in real time to recognize the user's emotional state. This data includes the user's facial expressions and tone of voice collected using a camera and microphone. The emotion engine also takes into account the user's past posts and behavioral history to predict the user's emotional state.

[0315] The server combines the results of text analysis and image analysis to generate an integrated dataset. This dataset is then input into a machine learning model to predict the attacker's persona, including their age, gender, occupation, and hobbies and interests. Machine learning models such as multilayer perceptrons and convolutional neural networks (CNNs) are used.

[0316] After obtaining the persona prediction results, the emotion engine adjusts the way the persona information is presented based on the user's emotional state. For example, if the user is under a lot of stress, the explanation will be changed to a more gentle expression or words of encouragement will be added.

[0317] Finally, the server sends the adjusted persona information to the user's device, which receives it and presents it to the user. For example, information such as "According to an analysis of this user, it appears that he is a man in his 20s who likes anime and is currently unemployed or working part-time" is displayed in a format that reflects the user's emotional state.

[0318] Specific examples

[0319] If a user receives a malicious reply such as "You're such an idiot," the device sends the content of the kusoriply and the poster's ID to the server. The server uses a social media API to collect the poster's posting data from the past year and analyzes the text and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict a persona of "male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." During this time, the device sends the user's voice and facial expression data to an emotion engine, which then recognizes the user's stress level. The persona information is then adjusted by the emotion engine and sent from the server to the device. Finally, the persona information is presented in a form that takes into account the user's psychological state.

[0320] Prompt Sentence Examples

[0321] Below are some example prompts to input to a generative AI model:

[0322] "We want to predict the personality of the person who posts malicious replies (kuso-replies) on social media that cause anxiety in users. At the same time, we want to design a method of providing information that recognizes the user's emotional state and can alleviate that anxiety. An example of a kuso-reply is: 'You're such an idiot.'"

[0323] Based on these prompts, the generative AI model will generate an appropriate persona and process emotions.

[0324] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0325] Step 1: Receiving a Shit Reply

[0326] The user receives malicious replies on social media such as "You're such an idiot."

[0327] Input: Malicious reply on a social media app

[0328] Output: Malicious reply content and poster ID

[0329] Specific behavior: When a user opens a social media application on their smartphone or computer, they will receive notifications and direct messages containing malicious replies.

[0330] Step 2: Send the content of the kusorip and the poster ID

[0331] The device automatically sends the content of the malicious reply received and the poster ID to the server.

[0332] Input: Malicious reply content and poster ID

[0333] Output: The malicious reply sent to the server and its poster ID.

[0334] Specific operation: The content of the shitty reply, "You are truly an idiot," and the poster's ID are sent to the server in the device's background process.

[0335] Step 3: Identifying the attacker and gathering data

[0336] The server collects posting data for the past year from the Shit Reply poster ID received using the SNS API.

[0337] Input: Shit reply poster ID

[0338] Output: Post data from the past year

[0339] What happens: The server sends a request to the SNS API to retrieve post data associated with the poster ID, e.g., tweets, posts, and comments from the past year.

[0340] Step 4: Extract and analyze text data

[0341] The server extracts the text portion from the collected posting data, performs preprocessing, and then analyzes it using natural language processing (NLP) technology.

[0342] Input: Post data from the past year

[0343] Output: Frequent words and key phrases

[0344] What it does: It removes hashtags and links, then performs morphological analysis on the text to extract frequently occurring words and key phrases. For example, it identifies keywords like "anime," "games," and "stress."

[0345] Step 5: Extract and analyze image data

[0346] The server extracts image portions from the collected posting data and extracts features using image analysis technology.

[0347] Input: Post data from the past year

[0348] Output: Image feature information

[0349] What it does: Uses computer vision technology to identify and tag characters and symbols from images. Example: Detecting the anime character "Naruto."

[0350] Step 6: Emotion Recognition

[0351] The device continuously transmits the user's voice and facial expression data to the emotion engine.

[0352] Input: User's voice data and facial expression data

[0353] Output: Voice and facial expression data sent to the emotion engine

[0354] Specific operation: When a user uses a social networking app, the camera and microphone are used to detect facial expressions and tone of voice. This data is sent to the emotion engine in real time.

[0355] The emotion engine analyzes the received data and recognizes the user's emotional state.

[0356] Input: Voice data and facial expression data

[0357] Output: User's emotional state

[0358] Specific behavior: If the user exhibits a sad facial expression or an irritated tone of voice, the emotion engine detects a stressed state.

[0359] Step 7: Data integration for persona prediction

[0360] The server combines the results of text analysis and image analysis to create a comprehensive dataset.

[0361] Input: Text analysis results and image analysis results

[0362] Output: Unified dataset

[0363] Specific operation: Combine text and image features into a single dataset and provide it to a machine learning model. Example: Combine text keywords and image tag information.

[0364] Step 8: Generate and deliver personas

[0365] The server generates a standard persona of an attacker from the prediction results of the machine learning model.

[0366] Input: Integrated dataset

[0367] Output: Persona information

[0368] Specific operation: The machine learning model outputs the following persona information: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation."

[0369] The emotional engine adjusts how persona information is presented based on the user's emotional state.

[0370] Input: User's emotional state and persona information

[0371] Output: Adjusted persona information

[0372] Specific actions: For users who are under a lot of stress, add soft words and words of encouragement. For example, "This user is a man in his 20s who loves anime. He may be unemployed or working part-time, but you'll be fine. Don't let his opinions affect you too much."

[0373] Step 9: View the results

[0374] The server transmits the adjusted persona information to the user's terminal.

[0375] Input: Adjusted persona information

[0376] Output: Data sent to the terminal

[0377] Specific operation: The server sends persona information to the user's device and issues a display instruction.

[0378] The terminal displays the received persona information to the user.

[0379] Input: Persona information sent from the server

[0380] Output: Data displayed to the user

[0381] Specific operation: The device displays a message on the screen saying, "This user is a man in his 20s who likes anime and is currently unemployed or working part-time." It also displays words of encouragement based on the user's stress level.

[0382] The above is the specific processing flow of this system's program. By including the specific operations at each step, it is clear how the data is processed when a user receives a 'kusorip' and what information is presented as the final result. In addition, by incorporating an emotion engine, it is possible to provide information that takes into account the user's psychological state.

[0383] (Application example 2)

[0384] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0385] In recent years, the spread of social networking sites has increased the risk of receiving malicious replies (kuso-replies). In such situations, many users experience psychological stress and anxiety. Furthermore, because they do not know the sender of the malicious replies, users are likely to be placed in unpleasant situations. Current social networking sites lack the ability to recognize users' emotions and respond appropriately. For this reason, a system that can reduce users' psychological burden is needed.

[0386] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0387] In this invention, the server includes a means for extracting frequently occurring words from past posts, a means for analyzing the extracted frequently occurring words, and a means for extracting images from past posts. This makes it possible to reduce the psychological burden on users who receive malicious replies. It also includes a means for predicting personas using a generative AI model and a means for recognizing the user's emotional state and adjusting the presentation method of persona information, making it possible to provide attacker information in a format tailored to the user. Furthermore, by including a means for transmitting the content of the kusorip and the poster ID to the server and a means for transmitting the user's voice and facial expression data to the emotion engine, it is possible to more accurately grasp the user's emotions and take appropriate action.

[0388] "Hateful replies" refer to comments or messages that are offensive, insulting, or inappropriate towards others on social media or other platforms.

[0389] "Terminal" refers to a device used by a user to use an SNS, including a smartphone, tablet, or PC.

[0390] A "server" is a device that stores, processes, and transmits data over a computer network.

[0391] "Past posts" refers to all messages, comments, images, videos, and other posts made by a particular user in the past on a social media platform.

[0392] A "frequent word" refers to a word that appears more frequently than other words in the text data to be analyzed.

[0393] "Extraction means" refers to a process or device that extracts specific data (text, images, audio, etc.) from other data.

[0394] "Analysis means" refers to the process or device used to analyze the extracted data in detail and understand its meaning and characteristics.

[0395] A "persona" is a virtual image of a person created based on the poster's characteristics, including attributes such as age, gender, occupation, and hobbies.

[0396] An "emotion engine" refers to software or hardware that recognizes and analyzes a user's emotional state from voice, facial expression, and other data.

[0397] "Generative AI model" refers to an artificial intelligence model used to generate specific patterns or information from given data.

[0398] A "prompt" is an instruction given to a generative AI model to perform a specific task.

[0399] The system based on the present invention, "KusoReplyGuard," predicts the identity of the attacker when a malicious reply is received on a social networking site, and provides information according to the user's emotional state, thereby reducing anxiety. A specific embodiment of this system will be described below.

[0400] System Configuration

[0401] The system consists of four main components:

[0402] 1. User Device

[0403] This is a device that users use to access social media. It can be a smartphone, tablet, or PC. The device receives the Kuso-Reply and sends the content and poster ID to a server. In addition, the device has the ability to collect the user's voice and facial expression data and send it to the emotion engine.

[0404] 2. Server

[0405] The server uses SNS APIs to collect the attacker's past posting data based on the Kuso-Reply information and poster ID received from the user's device. The collected data is subjected to text analysis and image analysis to extract frequently occurring words and image features. The analysis results are used to predict the attacker's persona using a generative AI model.

[0406] 3. Emotion Engine

[0407] The emotion engine analyzes the user's voice and facial expression data to recognize the user's emotional state. It also predicts the user's emotions based on the user's past posts and behavioral history. This information is used to adjust the way the persona information is presented.

[0408] 4. Social Media API

[0409] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[0410] Program processing explanation

[0411] The server first collects the attacker's past posting data using the social networking service API, then performs text and image analysis to extract frequently used words and image features.

[0412] The server then combines this information and inputs it into a generative AI model, which predicts a persona for the attacker, including their age, gender, occupation, hobbies, and preferences.

[0413] The server then transmits this persona information to the user's device. At the same time, the device transmits the user's emotional data to the emotion engine. The emotion engine analyzes the user's emotional state and adjusts the way the persona information is presented. For example, if the user is feeling stressed, the persona information is presented in a softer manner.

[0414] Specific examples

[0415] Consider the case where User A receives a malicious reply on social media saying, "You're an idiot." User A's device sends the content of this shitty reply and the poster's ID to a server. The server uses the social media API to collect the poster's past posting data and analyzes the text data and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a generative AI model to predict the persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation."

[0416] During this time, the device sends user A's voice and facial expression data to the emotion engine, which then recognizes user A's stress level. The persona information is then adjusted by the emotion engine and sent from the server to the device. Finally, persona information is provided in a form that takes user A's psychological state into consideration.

[0417] Prompt Sentence Examples

[0418] Examples of prompts that can be input to a generative AI model include:

[0419] "To reduce the anxiety of users who receive malicious replies, please generate a persona for the attacker. The data to be analyzed is the text data 'I love anime, I bought a new gachapon' and the image data 'animes_yokai.jpg'."

[0420] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0421] Step 1:

[0422] A user receives a malicious reply (kusoripu) on social media.

[0423] Input: Malicious replies on social media and their poster IDs

[0424] Output: Kuso-reply content and poster ID

[0425] How it works: The user checks the message on the social networking site, and the device obtains the content of the shitty reply and the poster's ID.

[0426] Step 2:

[0427] The device sends the content of the shit reply and the poster ID to the server.

[0428] Input: Kuso reply content and poster ID

[0429] Output: The content of the kusorip sent to the server and the poster ID

[0430] How it works: The communication module in the device sends the content of the shitty reply and the poster's ID to the server.

[0431] Step 3:

[0432] The server uses the SNS API to collect the poster's past posting data.

[0433] Input: Poster ID

[0434] Output: Collected past post data (text, images)

[0435] How it works: The server uses the SNS API to identify past posts associated with the poster ID and collects post data for a certain period of time.

[0436] Step 4:

[0437] The text portion is extracted from the posted data collected by the server and text analysis is performed.

[0438] Input: Text portion of past post data

[0439] Output: Extracted frequent words and key phrases

[0440] How it works: A natural language processing (NLP) engine on the server analyzes text data and extracts frequently occurring words and key phrases.

[0441] Step 5:

[0442] The image portion is extracted from the posted data collected by the server and image analysis is performed.

[0443] Input: Image portion of past post data

[0444] Output: Extracted image features (character, symbol tags)

[0445] How it works: The image analysis engine on the server analyzes the image data, extracts features, and tags them.

[0446] Step 6:

[0447] The server combines the results of text analysis and image analysis and inputs them into a generative AI model.

[0448] Input: Text analysis results (frequent words, key phrases), image analysis results (features, tags)

[0449] Output: Predicted persona information

[0450] How it works: The server integrates the analysis results and inputs them as prompts into the generative AI model, which then predicts the persona.

[0451] Step 7:

[0452] The server transmits the generated persona information to the user's terminal.

[0453] Input: Generated persona information

[0454] Output: Persona information sent to the device

[0455] Operation: The server sends persona information to the device and prepares it for display.

[0456] Step 8:

[0457] The device sends the user's voice and facial expression data to the emotion engine.

[0458] Input: User's voice data, facial expression data

[0459] Output: User emotion data sent to the emotion engine

[0460] Operation: The device's sensors capture the user's voice and facial expressions and send them to the emotion engine.

[0461] Step 9:

[0462] The emotion engine analyzes the user's emotional state and generates emotion data.

[0463] Input: Voice data, facial expression data

[0464] Output: User's emotional state (stress level, type of emotion)

[0465] How it works: The emotion engine analyzes voice and facial expression data in real time to identify the user's emotional state.

[0466] Step 10:

[0467] An emotional engine adjusts the presentation of persona information based on the user's emotional state.

[0468] Input: User's emotional state, predicted persona information

[0469] Output: Adjusted persona information

[0470] How it works: The emotion engine takes into account the user's emotional state and adjusts the way persona information is presented appropriately. For example, if a user is stressed, it will explain things in a gentler way.

[0471] Step 11:

[0472] The device displays the adjusted persona information to the user.

[0473] Input: Adjusted persona information

[0474] Output: Persona information displayed to the user

[0475] Action: The device uses a display device to present the adjusted persona information to the user in an appropriate format.

[0476] In this way, the system provides information to reduce anxiety when users receive malicious replies and responds according to the user's emotional state.

[0477] 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.

[0478] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0479] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0480] [Second embodiment]

[0481] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0482] 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.

[0483] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0484] 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.

[0485] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0486] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0487] 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.

[0488] 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.

[0489] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0490] 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.

[0491] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0492] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0493] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the profile of the attacker and displays it to the user, thereby reducing anxiety. This system uses the user's device, server, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[0494] System Configuration

[0495] 1. User Device

[0496] It is a device that allows users to use SNS and can receive Kuso-Replies. It also has the function of sending the content of Kuso-Replies and the poster ID to the server.

[0497] 2. Server

[0498] The server uses SNS APIs to collect data on attackers' past posts based on the Kuso-Reply information received from the user's device. The collected data is then subjected to text and image analysis to extract frequently occurring words and image features. Furthermore, a machine learning model is used to predict the attacker's persona, and the results are sent to the user's device.

[0499] 3. Social Media API

[0500] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[0501] Program processing

[0502] 1. Receiving shitty replies

[0503] A user receives a shitty reply on social media and checks its content and the poster's ID.

[0504] The device sends the content of the shit reply and the poster ID to the server.

[0505] 2. Attacker Identification and Data Collection

[0506] The server receives the ID of the user who posted the kusorip and uses the SNS API to collect the attacker's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0507] 3. Extracting and analyzing text data

[0508] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques.

[0509] 4. Image Data Extraction and Analysis

[0510] The server extracts image data from the collected submissions, extracts image features using image analysis technology (e.g., computer vision and deep learning), and assigns tags such as characters and symbols to the images.

[0511] 5. Data integration for persona prediction

[0512] The server combines the results of text analysis and image analysis to create a comprehensive dataset, which is then fed into a machine learning model to predict the attacker's age, gender, occupation, hobbies, and preferences.

[0513] 6. Persona generation and provision

[0514] The server generates a standard persona of an attacker based on the prediction results of the machine learning model. The generated persona information is structured in a format that is easy for users to understand.

[0515] The device provides the user with the persona information received from the server and displays information such as, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0516] Specific examples

[0517] Let's say a user receives a malicious reply saying, "Your opinion is off the mark." The device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API and analyzes the text data and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict the persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." The prediction result is then sent to the user's device, and the attacker's standard persona information is presented to the user.

[0518] The above is a specific embodiment for carrying out the present invention, which allows the user to understand the personality of the sender of the malicious reply and reduce anxiety.

[0519] The processing flow will be explained below.

[0520] Step 1:

[0521] A user receives a malicious reply on a social networking site, which appears as a reply to a specific post.

[0522] Step 2:

[0523] The device obtains the content of the Kuso-Reply and the poster's ID, and sends this to the system's server, including the text content of the Kuso-Reply and the poster's account ID.

[0524] Step 3:

[0525] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0526] Step 4:

[0527] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and creates a clean dataset.

[0528] Step 5:

[0529] The server uses natural language processing (NLP) technology to analyze the extracted text data. This analysis identifies frequently occurring words and key phrases. Specifically, it uses morphological analysis and TF-IDF (Term Frequency-Inverse Document Frequency) techniques.

[0530] Step 6:

[0531] The server extracts the image portion from the collected post data, which includes all image files included in each post.

[0532] Step 7:

[0533] The server uses image analysis technology (e.g., computer vision and deep learning) to analyze the features of the extracted images. Specifically, it performs object recognition and assigns tags such as characters and symbols.

[0534] Step 8:

[0535] The server combines the results of the text analysis and the image analysis, creating a comprehensive dataset.

[0536] Step 9:

[0537] The server then inputs the combined data set into a machine learning model to predict the attacker's persona, including attributes such as age, gender, occupation, and hobbies.

[0538] Step 10:

[0539] The server generates a standard persona for an attacker based on the predictions obtained from the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[0540] Step 11:

[0541] The server transmits the generated persona information to the user's terminal.

[0542] Step 12:

[0543] The device presents the received persona information to the user. This is displayed on the screen in a format that is easy for the user to confirm. For example, it might say, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0544] By going through these steps, users can understand the profile of the attacker and reduce their anxiety.

[0545] Example 1

[0546] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0547] Users who receive malicious replies (kuso-replies) on social media often feel anxious and stressed. However, typical countermeasures include ignoring the replies or blocking the user, which does not fundamentally alleviate the anxiety. Furthermore, there are limited ways to specifically identify the attacker, leaving users plagued by fear of the so-called "invisible enemy." The present invention aims to address these issues by predicting the attacker's profile and presenting it to users, thereby providing them with specific information and reducing their anxiety.

[0548] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0549] In this invention, the server includes a device for receiving malicious replies, a device for collecting past posts by the poster of the malicious replies, means for extracting text from the collected past posts and performing preprocessing, means for analyzing frequently occurring words and key phrases from the preprocessed text, means for extracting images from the collected posts and analyzing their features, means for predicting the poster's personality based on the analysis results, and means for providing the predicted personality information to the device. This allows the user to grasp the specific personality of the attacker, making it possible to reduce anxiety and stress.

[0550] A "malicious reply" is a comment or message sent on social media with the intention of attacking or offending others.

[0551] "Devices" is a general term for devices, machines, and related systems and software used by users to access SNS.

[0552] A "poster" is a person or account that posts comments, messages, images, etc. on social media.

[0553] "Past posts" refer to comments, messages, images, and other output made on social media within a specified period.

[0554] "Text" refers to posts and comments containing written information made on social media.

[0555] "Preprocessing" is the process of removing unnecessary elements (e.g., hashtags and links) from text information and converting it into a format suitable for analysis.

[0556] A "frequent word" is a word that occurs many times within a particular text dataset.

[0557] A "key phrase" is a phrase or word that has an important meaning within text data.

[0558] "Analysis" is the process of analyzing text and image data and extracting meaningful information.

[0559] "Features" are important patterns or characteristics extracted from image data.

[0560] A "personal profile" is the result of predicting an individual's attribute information, such as age, gender, occupation, and hobbies and preferences.

[0561] "Information" refers to data and knowledge such as predictions and analysis results of a person's profile.

[0562] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the profile of the attacker and displays this information to the user, thereby reducing anxiety. This system uses the user's device, server, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the profile of the attacker.

[0563] System Configuration

[0564] 1. User Device

[0565] The user device is a device that allows users to use SNS. It has the function of receiving kusorip messages and sending their content and the poster's ID to a server. When a user receives a kusorip message on SNS, the device sends the content and the poster's ID to the server using an HTTPS request.

[0566] 2. Server

[0567] The server uses SNS APIs to collect data on the attacker's past posts based on the Kuso-Reply information received from the user's device. The collected data is subjected to text analysis and image analysis to extract frequently occurring words and image features. It also uses a machine learning model to predict the attacker's profile and sends the results to the user's device. Specifically, the server preprocesses the text data using "NLTK" and extracts important key phrases using natural language processing technology. It also performs image analysis using "OpenCV" and "TensorFlow" to extract important features.

[0568] 3. Social Media API

[0569] The SNS API is an application programming interface provided by SNS platforms, and is used by the server to collect data on past posts by attackers. Authentication is performed using the SNS API authentication token, and post data from a specified period is obtained.

[0570] Specific example of system operation

[0571] Suppose a user receives a malicious reply saying, "Your opinion is off the mark." The user's device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API. The collected text data is preprocessed using NLTK to extract frequently occurring words and key phrases. At the same time, the collected image data is analyzed using OpenCV or TensorFlow to extract features. The results are integrated, and a machine learning model (e.g., SciKit-Learn) is used to predict a profile of the attacker, such as "male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." Finally, the prediction result is sent to the user's device, and the user is presented with a standard profile of the attacker.

[0572] Prompt Sentence Examples

[0573] An example of a prompt would be:

[0574] I received a shitty reply saying, "Your opinion is off the mark." Please predict and provide a profile of the attacker.

[0575] The above-mentioned method allows users to get a clear picture of the attacker, which can reduce anxiety and stress.

[0576] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0577] Step 1: Receiving a Shit Reply

[0578] A user receives a malicious reply (kuso-repu) on social media. For example, the reply is, "Your opinion is irrelevant."

[0579] Input: Shitty reply on social media and poster ID.

[0580] The device temporarily stores the content of the KusoRip and the poster ID in its internal memory, then sends the content of the KusoRip and the poster ID to the server using an HTTPS request.

[0581] Output: The content of the shit reply sent to the server and the poster ID.

[0582] Step 2: Identifying the attacker and gathering data

[0583] The server checks the content of the kusorip received from the device and the poster ID.

[0584] Input: The content of the shitty reply sent from the device and the poster ID.

[0585] The server uses the SNS API to collect the attacker's posting data for the past year. The server authenticates using the SNS API's authentication token and sends an API request. This request includes the poster ID and the data collection period (e.g., the past year).

[0586] Output: Posted data (text and images) by attackers collected on the server over the past year.

[0587] Step 3: Extract and analyze text data

[0588] The server extracts the text from the collected post data and preprocesses the text of the post to remove hashtags and links.

[0589] Input: Attacker's past posting data (text).

[0590] The server uses NLTK to preprocess the text data, tokenizing it, removing stop words, and performing word frequency analysis, using natural language processing techniques to extract important frequently occurring words and key phrases.

[0591] Output: Preprocessed text data and analysis results (frequent words and key phrases).

[0592] Step 4: Extraction and analysis of image data

[0593] The server extracts the image portion from the collected submission data, extracts image features using computer vision and deep learning technology, and assigns tags such as characters and symbols to the images.

[0594] Input: Attacker's past posting data (images).

[0595] The server uses OpenCV and TensorFlow to analyze the image and extract important features, such as identifying and labeling anime characters or specific symbols.

[0596] Output: Analyzed image data and feature information.

[0597] Step 5: Data integration for persona prediction

[0598] The server combines the results of text analysis and image analysis to create a comprehensive dataset.

[0599] Input: Text analysis results and image analysis results.

[0600] The server inputs this dataset into a machine learning model to predict the attacker's personality, such as age, gender, occupation, hobbies, etc. For example, the server inputs the dataset into a machine learning model trained using "SciKit-Learn" to obtain a predicted personality profile.

[0601] Output: Predicted attacker profile (age, gender, occupation, hobbies, etc.).

[0602] Step 6: Generate and deliver personas

[0603] The server generates a standard profile of the attacker based on the predictions of the machine learning model, and the profile information is organized in a format that is easy for users to understand.

[0604] Input: Prediction result (attacker profile).

[0605] The server generates the person profile information in JSON format and sends it to the user device using HTTPS.

[0606] The user device analyzes the persona information received from the server and displays it on the user interface. For example, it displays information such as, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0607] Output: Attacker profile information displayed in the user interface.

[0608] In this way, this system can reduce anxiety and stress by presenting a specific profile of the attacker in response to a malicious reply received by the user.

[0609] (Application example 1)

[0610] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0611] In online social networking services (SNS), users often receive malicious replies (kuso-replies) from other users. These replies can cause psychological stress and anxiety to the recipient. Therefore, there is a need for a method to reduce users' anxiety and provide a safe SNS environment by revealing the causes of these replies and the identity of the people behind them.

[0612] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0613] In this invention, the server includes a means for extracting frequently occurring words from past posts, a means for analyzing the extracted frequently occurring words, and a means for extracting images from past posts. This makes it possible to analyze the poster's personality and predict persona information. Furthermore, the prediction includes a means for using a generative AI model and prompt sentences, making it possible to improve the accuracy of the analysis results. This makes it possible to clarify the identity of the person behind malicious replies to users, allowing users to use SNS without feeling anxious.

[0614] A "terminal" is a device through which a user receives malicious replies and transmits the information to a server.

[0615] The "server" is a computer system that receives information about the poster of a malicious reply sent from a user's terminal, collects and analyzes the poster's past posting data, and provides persona information.

[0616] "Method for extracting frequently occurring words" is a method for selecting frequently used words from posted data.

[0617] "Means for analyzing frequently occurring words" is a method for analyzing extracted words and finding their meanings and relationships.

[0618] "Means for extracting images" refers to a method for selecting image media from posted data.

[0619] The "means for analyzing images" is a method for analyzing the extracted images and recognizing their contents and features.

[0620] A "means for predicting persona" is a method for inferring characteristics such as the poster's age, gender, hobbies and preferences based on collected and analyzed data.

[0621] "Using a generative AI model" means using a model generated using artificial intelligence technology.

[0622] "Prompt methods" are methods that use sentences to provide a generative AI model with a specific question and context to get an answer.

[0623] "Information Provision Interface" means the application programming interface (API) that the server uses to collect data from the social networking platform.

[0624] The system based on this invention reduces anxiety when a user receives a malicious reply (kuso-reply) on social media by predicting the identity of the attacker and providing this information to the user. The system uses the user's device, server, and social media API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[0625] System Configuration

[0626] 1. User Device

[0627] It is a device that allows users to use social networking services, and when it receives a shitty reply, it sends the content and the poster's ID to the server.

[0628] 2. Server

[0629] The server uses SNS APIs to collect data on the attacker's past posts based on the Kuso-Reply information received from the user's device. The collected data is then subjected to text and image analysis. Natural language processing (NLP) technology is used for text analysis, and computer vision and deep learning technology are used for image analysis. Furthermore, a generative AI model is used to predict the attacker's persona. A prompt sentence is used to make highly accurate predictions.

[0630] 3. Social Media API

[0631] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[0632] Program processing example

[0633] Receiving and sending shit replies

[0634] The user receives a kusorip message on a social networking site and enters the message's content and the poster's ID into their device, which then sends this information to the server.

[0635] Attacker Identification and Data Collection

[0636] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's posting data for the past year.

[0637] Text data extraction and analysis

[0638] The server extracts text from the collected post data and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques, such as NLTK and spaCy libraries.

[0639] Image data extraction and analysis

[0640] The server extracts image data from the collected submissions and uses computer vision technology to extract image features, using deep learning technologies such as TensorFlow and PyTorch.

[0641] Persona Prediction

[0642] The server integrates the results of text analysis and image analysis and predicts the attacker's persona based on a generative AI model. The prompt sentence is used as input to the model, and persona information is predicted with high accuracy.

[0643] Providing persona information

[0644] The predicted persona information is sent to the user's device and displayed to the user, allowing the user to understand the profile of the attacker and reduce anxiety.

[0645] Specific examples

[0646] When a user receives a crappy reply saying, "Your opinion is completely meaningless," the device sends the content of the crappy reply and the poster's ID to the server. The server collects the poster's posting data from the past year via SNS APIs and analyzes the text data and image data. For example, if "games" and "anime" are detected as frequently occurring words and image analysis reveals that specific character images appear frequently, the generative AI model is used to predict a persona such as "male in his 20s, likes anime and games, unemployed or working part-time." The prediction results are then sent to the user's device and displayed to the user.

[0647] Example prompt sentence:

[0648] Please predict the attacker's persona based on the following text analysis and image analysis results. Text analysis results: ['game', 'anime'], Image analysis results: ['anime character']

[0649] The above is a specific embodiment for carrying out the invention.

[0650] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0651] Step 1:

[0652] The user receives a kusorip message and enters its contents and the poster's ID into the device.

[0653] When a user receives a malicious reply (kuso-reply) on social media, they enter the content and the sender's ID into a dedicated application. This entered data (the content of the kuso-reply and the poster's ID) is saved on the device.

[0654] Step 2:

[0655] The device sends the content of the Kuso-Reply and the poster ID to the server.

[0656] The device collects the text of the entered kusorip and the poster ID and sends it to the server. This is done using a method such as an HTTP POST request. The input is the kusorip and the poster ID, and the output is a status message confirming that this information has been sent to the server.

[0657] Step 3:

[0658] The server uses the SNS API to collect the attacker's past posting data.

[0659] The server uses the received poster ID to collect the attacker's past posting data via the SNS platform's API. The collected data includes text posts, images, videos, etc., covering the past year. The input is the poster ID, and the output is the collected posting data.

[0660] Step 4:

[0661] Extract the text data collected by the server

[0662] The server extracts text parts from the collected submission data using regular expressions and NLP preprocessing techniques. The input is the submission data, and the output is the extracted text data.

[0663] Step 5:

[0664] The server analyzes the text data using NLP techniques.

[0665] The server preprocesses the extracted text data to identify frequently occurring words and key phrases. The libraries used are NLTK and spaCy. The input is text data, and the output is frequently occurring words and key phrases. Specific operations include tokenization and stop word removal.

[0666] Step 6:

[0667] Extracting image data collected by the server

[0668] The server extracts images from the collected submission data by filtering the data type. The input is the submission data, and the output is the extracted image data.

[0669] Step 7:

[0670] The server analyzes the image data using image analysis technology.

[0671] The server analyzes the extracted image data using computer vision techniques, such as deep learning frameworks like TensorFlow and PyTorch. The input is image data, and the output is image features and labels. Specific operations include object detection and image classification.

[0672] Step 8:

[0673] The server integrates the results of text analysis and image analysis and predicts the persona using a generative AI model.

[0674] The server uses a dataset that integrates the results of text analysis and image analysis, and inputs a prompt into the generative AI model to predict the attacker's persona. The input is the integrated dataset and prompt, and the output is the persona prediction result. Example prompt: "Please predict the attacker's persona based on the following text analysis and image analysis results. Text analysis result: ['Game', 'Anime'], Image analysis result: ['Anime character']"

[0675] Step 9:

[0676] The server sends the predicted persona information to the user device.

[0677] The server sends the predicted persona information to the user's device. This is done via an HTTP POST request, etc. The input is the persona prediction result, and the output is the confirmation status of the transmission to the user's device.

[0678] Step 10:

[0679] The user device displays the persona information to the user.

[0680] The user device displays the received persona information on the user screen. This allows the user to understand the profile of the attacker and reduce anxiety. The input is persona information, and the output is the displayed persona information. Specifically, it operates by displaying text and images on the interface.

[0681] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0682] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the attacker's personality and further recognizes the user's emotions and adjusts the presentation method to reduce anxiety. This system uses the user's device, server, emotion engine, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[0683] System Configuration

[0684] 1. User Device

[0685] It is a device that allows users to use SNS and can receive Kuso-Reply messages. It also has the function of sending the content of the Kuso-Reply message and the poster's ID to the server. It also has the function of providing the user's voice and facial expressions to the emotion engine.

[0686] 2. Server

[0687] The server uses SNS APIs to collect attacker's past posting data based on the information on kusoripulp received from the user's device and the user's emotional data. The collected data is subjected to text and image analysis to extract frequently occurring words and image features. Furthermore, a machine learning model is used to predict the attacker's persona and the results are sent to the user's device.

[0688] 3. Emotion Engine

[0689] The emotion engine collects and analyzes the user's voice and facial expression data to recognize the user's emotional state. It also predicts emotions based on the user's past posts and behavioral history, and adjusts the way persona information is presented.

[0690] 4. Social Media API

[0691] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[0692] Program processing

[0693] 1. Receiving shitty replies

[0694] A user receives a malicious reply on a social networking site and checks its content and the poster's ID.

[0695] The device sends the content of the shit reply and the poster ID to the server.

[0696] 2. Attacker Identification and Data Collection

[0697] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0698] 3. Extracting and analyzing text data

[0699] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques.

[0700] 4. Image Data Extraction and Analysis

[0701] The server extracts image data from the collected submissions, extracts image features using image analysis technology (e.g., computer vision and deep learning), and assigns tags such as characters and symbols to the images.

[0702] 5. Emotion recognition

[0703] The device continuously collects the user's voice and facial expression data and sends it to the emotion engine.

[0704] The emotion engine analyzes this data to recognize the user's emotional state and predicts their emotions based on their past posts and behavioral history.

[0705] 6. Data integration for persona prediction

[0706] The server combines the results of text analysis and image analysis to create a comprehensive dataset, which is then fed into a machine learning model to predict the attacker's age, gender, occupation, hobbies, and preferences.

[0707] 7. Persona generation and provision

[0708] The server generates a standard persona for an attacker based on the predictions of the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[0709] The emotion engine adjusts the presentation of persona information based on the user's perceived emotions. For example, if the user is under a lot of stress, it will display softer expressions and encouraging words.

[0710] 8. Displaying the results

[0711] The server transmits the generated persona information to the user's terminal.

[0712] The device then presents the received persona information to the user. For example, information such as "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time," is displayed in a format that is adjusted to reflect the user's emotional state.

[0713] Specific examples

[0714] Suppose a user receives a malicious reply saying, "You're such an idiot." The device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API and analyzes the text and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict a persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." During this time, the device sends the user's voice and facial expression data to an emotion engine, which then recognizes the user's stress level. The emotion engine then adjusts the persona information and sends it from the server to the device. Finally, the persona information is presented in a way that takes into account the user's psychological state.

[0715] The above is a specific example of how to implement the present invention, which allows users to understand the personality of the sender of a malicious reply and reduce anxiety. By incorporating an emotion engine, information is provided that takes into consideration the user's psychological state.

[0716] The processing flow will be explained below.

[0717] Step 1:

[0718] A user receives a malicious reply (kuso-reply) on a social networking site, which appears as a comment on a specific post.

[0719] Step 2:

[0720] The device obtains the content of the Kuso-Reply and the poster's ID, and sends this to the system's server, including the text content of the Kuso-Reply and the poster's account ID.

[0721] Step 3:

[0722] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0723] Step 4:

[0724] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and creates a clean dataset.

[0725] Step 5:

[0726] The server uses natural language processing (NLP) technology to analyze the extracted text data. This analysis identifies frequently occurring words and key phrases. Specifically, it uses morphological analysis and TF-IDF (Term Frequency-Inverse Document Frequency) techniques.

[0727] Step 6:

[0728] The server extracts the image portion from the collected post data, which includes all image files included in each post.

[0729] Step 7:

[0730] The server uses image analysis technology (e.g., computer vision and deep learning) to analyze the features of the extracted images. Specifically, it performs object recognition and assigns tags such as characters and symbols.

[0731] Step 8:

[0732] The server combines the results of the text analysis and the image analysis, creating a comprehensive dataset.

[0733] Step 9:

[0734] The server then inputs the combined data set into a machine learning model to predict the attacker's persona, including attributes such as age, gender, occupation, and hobbies.

[0735] Step 10:

[0736] The server generates a standard persona for an attacker based on the predictions obtained from the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[0737] Step 11:

[0738] The device continuously collects the user's voice and facial expression data and sends it to the emotion engine.

[0739] Step 12:

[0740] The emotion engine analyzes this data to recognize the user's emotional state and predicts their emotions based on their past posts and behavioral history.

[0741] Step 13:

[0742] The emotion engine adjusts the presentation of persona information based on the user's emotional state. For example, if a user is under a lot of stress, it will display softer language and encouraging words.

[0743] Step 14:

[0744] The server transmits the generated persona information to the user's terminal.

[0745] Step 15:

[0746] The device presents the received persona information to the user. This is displayed on the screen in a format that is easy for the user to confirm. For example, it might say, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0747] Through these steps, users can understand the profile of the attacker and reduce their anxiety. By incorporating an emotion engine, information is provided that takes into account the user's psychological state.

[0748] Example 2

[0749] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0750] On modern social media platforms, malicious replies (kusoripu) can cause psychological stress to users. However, there is no way for users who receive such replies to understand who is attacking them, and there is no way to provide appropriate information depending on their psychological state, making it difficult to alleviate their anxiety and stress.

[0751] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for collecting past posts by posters of malicious replies, means for extracting frequently occurring words from past posts, means for analyzing the extracted images, means for predicting the poster's persona based on the analysis results, an emotion engine for collecting and analyzing user emotion data, means for adjusting the predicted persona information based on the user's emotional state, and means for providing the adjusted persona information to the terminal. This makes it easier for the user to grasp the profile of the attacker and further enables the provision of appropriate information based on the user's psychological state.

[0752] A "malicious reply" is a comment or message on a social networking site that contains offensive, derogatory, or unpleasant content directed at a user.

[0753] A "terminal" is a device that a user uses to access a social networking service, and has the function of receiving malicious replies and sending their contents and the poster's ID to a server.

[0754] A "server" is a computer system that receives data sent from a user's device, collects the necessary data through the SNS API, and performs analytical processing.

[0755] "Past posts" refers to all content posted by an attacker on social media within a specific period of time.

[0756] "Frequent words" are keywords and phrases that appear frequently and are extracted from past posts.

[0757] An "extraction method" is a set of algorithms or techniques used to collect and analyze specific data (e.g., text or images).

[0758] "Means for analysis" refers to methods and techniques for analyzing the extracted data and extracting useful information and features.

[0759] A "persona" is a predicted profile of an attacker based on the analysis results, including their age, gender, occupation, hobbies, and preferences.

[0760] An "emotion engine" is a system or technology that collects and analyzes a user's emotional data to recognize the user's psychological state.

[0761] "SNS API" is an application programming interface provided by an SNS platform and is an interface used to collect past posting data.

[0762] A "machine learning model" is an algorithm that learns using large amounts of data and uses the learning results to make predictions and analyze new data.

[0763] "Means for adjusting based on the user's emotional state" refers to techniques or methods for changing the method or content of information presentation depending on the user's psychological state.

[0764] "Means for providing" refers to communication and interface technologies for displaying analysis results and predicted information on a user terminal.

[0765] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the identity of the attacker and further recognizes the user's emotions and adjusts the presentation method to reduce anxiety. This system utilizes the user's device, server, emotion engine, and social networking site API, and is implemented in the following specific steps.

[0766] First, when a user receives a malicious reply on a social networking site, the device sends the content of the kusoriply and the poster's ID to a server. The device can be a smartphone or a PC, and the social networking application runs as the communication software.

[0767] The server receives the Kuso-Reply information sent from the device and uses the SNS API to collect the poster's past posting data. The collected data is classified into text data and image data, and each type of data is analyzed. For example, the text data is analyzed using natural language processing (NLP) technology to extract frequently occurring words and key phrases, and the image data is analyzed using computer vision and deep learning technology to analyze the image features.

[0768] Next, the emotion engine collects the user's voice and facial expression data in real time to recognize the user's emotional state. This data includes the user's facial expressions and tone of voice collected using a camera and microphone. The emotion engine also takes into account the user's past posts and behavioral history to predict the user's emotional state.

[0769] The server combines the results of text analysis and image analysis to generate an integrated dataset. This dataset is then input into a machine learning model to predict the attacker's persona, including their age, gender, occupation, and hobbies and interests. Machine learning models such as multilayer perceptrons and convolutional neural networks (CNNs) are used.

[0770] After obtaining the persona prediction results, the emotion engine adjusts the way the persona information is presented based on the user's emotional state. For example, if the user is under a lot of stress, the explanation will be changed to a more gentle expression or words of encouragement will be added.

[0771] Finally, the server sends the adjusted persona information to the user's device, which receives it and presents it to the user. For example, information such as "According to an analysis of this user, it appears that he is a man in his 20s who likes anime and is currently unemployed or working part-time" is displayed in a format that reflects the user's emotional state.

[0772] Specific examples

[0773] If a user receives a malicious reply such as "You're such an idiot," the device sends the content of the kusoriply and the poster's ID to the server. The server uses a social media API to collect the poster's posting data from the past year and analyzes the text and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict a persona of "male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." During this time, the device sends the user's voice and facial expression data to an emotion engine, which then recognizes the user's stress level. The persona information is then adjusted by the emotion engine and sent from the server to the device. Finally, the persona information is presented in a form that takes into account the user's psychological state.

[0774] Prompt Sentence Examples

[0775] Below are some example prompts to input to a generative AI model:

[0776] "We want to predict the personality of the person who posts malicious replies (kuso-replies) on social media that cause anxiety in users. At the same time, we want to design a method of providing information that recognizes the user's emotional state and can alleviate that anxiety. An example of a kuso-reply is: 'You're such an idiot.'"

[0777] Based on these prompts, the generative AI model will generate an appropriate persona and process emotions.

[0778] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0779] Step 1: Receiving a Shit Reply

[0780] The user receives malicious replies on social media such as "You're such an idiot."

[0781] Input: Malicious reply on a social media app

[0782] Output: Malicious reply content and poster ID

[0783] Specific behavior: When a user opens a social media application on their smartphone or computer, they will receive notifications and direct messages containing malicious replies.

[0784] Step 2: Send the content of the kusorip and the poster ID

[0785] The device automatically sends the content of the malicious reply received and the poster ID to the server.

[0786] Input: Malicious reply content and poster ID

[0787] Output: The malicious reply sent to the server and its poster ID.

[0788] Specific operation: The content of the shitty reply, "You are truly an idiot," and the poster's ID are sent to the server in the device's background process.

[0789] Step 3: Identifying the attacker and gathering data

[0790] The server collects posting data for the past year from the Shit Reply poster ID received using the SNS API.

[0791] Input: Shit reply poster ID

[0792] Output: Post data from the past year

[0793] What happens: The server sends a request to the SNS API to retrieve post data associated with the poster ID, e.g., tweets, posts, and comments from the past year.

[0794] Step 4: Extract and analyze text data

[0795] The server extracts the text portion from the collected posting data, performs preprocessing, and then analyzes it using natural language processing (NLP) technology.

[0796] Input: Post data from the past year

[0797] Output: Frequent words and key phrases

[0798] What it does: It removes hashtags and links, then performs morphological analysis on the text to extract frequently occurring words and key phrases. For example, it identifies keywords like "anime," "games," and "stress."

[0799] Step 5: Extract and analyze image data

[0800] The server extracts image portions from the collected posting data and extracts features using image analysis technology.

[0801] Input: Post data from the past year

[0802] Output: Image feature information

[0803] What it does: Uses computer vision technology to identify and tag characters and symbols from images. Example: Detecting the anime character "Naruto."

[0804] Step 6: Emotion Recognition

[0805] The device continuously transmits the user's voice and facial expression data to the emotion engine.

[0806] Input: User's voice data and facial expression data

[0807] Output: Voice and facial expression data sent to the emotion engine

[0808] Specific operation: When a user uses a social networking app, the camera and microphone are used to detect facial expressions and tone of voice. This data is sent to the emotion engine in real time.

[0809] The emotion engine analyzes the received data and recognizes the user's emotional state.

[0810] Input: Voice data and facial expression data

[0811] Output: User's emotional state

[0812] Specific behavior: If the user exhibits a sad facial expression or an irritated tone of voice, the emotion engine detects a stressed state.

[0813] Step 7: Data integration for persona prediction

[0814] The server combines the results of text analysis and image analysis to create a comprehensive dataset.

[0815] Input: Text analysis results and image analysis results

[0816] Output: Unified dataset

[0817] Specific operation: Combine text and image features into a single dataset and provide it to a machine learning model. Example: Combine text keywords and image tag information.

[0818] Step 8: Generate and deliver personas

[0819] The server generates a standard persona of an attacker from the prediction results of the machine learning model.

[0820] Input: Integrated dataset

[0821] Output: Persona information

[0822] Specific operation: The machine learning model outputs the following persona information: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation."

[0823] The emotional engine adjusts how persona information is presented based on the user's emotional state.

[0824] Input: User's emotional state and persona information

[0825] Output: Adjusted persona information

[0826] Specific actions: For users who are under a lot of stress, add soft words and words of encouragement. For example, "This user is a man in his 20s who loves anime. He may be unemployed or working part-time, but you'll be fine. Don't let his opinions affect you too much."

[0827] Step 9: View the results

[0828] The server transmits the adjusted persona information to the user's terminal.

[0829] Input: Adjusted persona information

[0830] Output: Data sent to the terminal

[0831] Specific operation: The server sends persona information to the user's device and issues a display instruction.

[0832] The terminal displays the received persona information to the user.

[0833] Input: Persona information sent from the server

[0834] Output: Data displayed to the user

[0835] Specific operation: The device displays a message on the screen stating, "This user is a male in his 20s who likes anime and is currently unemployed or possibly working part-time." It also displays words of encouragement based on the user's stress level.

[0836] The above is the specific processing flow of this system's program. By including the specific operations at each step, it is clear how the data is processed when a user receives a 'kusorip' and what information is presented as the final result. In addition, by incorporating an emotion engine, it is possible to provide information that takes into account the user's psychological state.

[0837] (Application example 2)

[0838] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0839] In recent years, the spread of social networking sites has increased the risk of receiving malicious replies (kuso-replies). In such situations, many users experience psychological stress and anxiety. Furthermore, because they do not know the sender of the malicious replies, users are likely to be placed in unpleasant situations. Current social networking sites lack the ability to recognize users' emotions and respond appropriately. For this reason, a system that can reduce users' psychological burden is needed.

[0840] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0841] In this invention, the server includes a means for extracting frequently occurring words from past posts, a means for analyzing the extracted frequently occurring words, and a means for extracting images from past posts. This makes it possible to reduce the psychological burden on users who receive malicious replies. It also includes a means for predicting personas using a generative AI model and a means for recognizing the user's emotional state and adjusting the presentation method of persona information, making it possible to provide attacker information in a format tailored to the user. Furthermore, by including a means for transmitting the content of the kusorip and the poster ID to the server and a means for transmitting the user's voice and facial expression data to the emotion engine, it is possible to more accurately grasp the user's emotions and take appropriate action.

[0842] "Hateful replies" refer to comments or messages that are offensive, insulting, or inappropriate towards others on social media or other platforms.

[0843] "Terminal" refers to a device used by a user to use an SNS, including a smartphone, tablet, or PC.

[0844] A "server" is a device that stores, processes, and transmits data over a computer network.

[0845] "Past posts" refers to all messages, comments, images, videos, and other posts made by a particular user in the past on a social media platform.

[0846] A "frequent word" refers to a word that appears more frequently than other words in the text data to be analyzed.

[0847] "Extraction means" refers to a process or device that extracts specific data (text, images, audio, etc.) from other data.

[0848] "Analysis means" refers to the process or device used to analyze the extracted data in detail and understand its meaning and characteristics.

[0849] A "persona" is a virtual image of a person created based on the poster's characteristics, including attributes such as age, gender, occupation, and hobbies.

[0850] An "emotion engine" refers to software or hardware that recognizes and analyzes a user's emotional state from voice, facial expression, and other data.

[0851] "Generative AI model" refers to an artificial intelligence model used to generate specific patterns or information from given data.

[0852] A "prompt" is an instruction given to a generative AI model to perform a specific task.

[0853] The system based on the present invention, "KusoReplyGuard," predicts the identity of the attacker when a malicious reply is received on a social networking site, and provides information according to the user's emotional state, thereby reducing anxiety. A specific embodiment of this system will be described below.

[0854] System Configuration

[0855] The system consists of four main components:

[0856] 1. User Device

[0857] This is a device that users use to access social media. It can be a smartphone, tablet, or PC. The device receives the Kuso-Reply and sends the content and poster ID to a server. In addition, the device has the ability to collect the user's voice and facial expression data and send it to the emotion engine.

[0858] 2. Server

[0859] The server uses SNS APIs to collect the attacker's past posting data based on the Kuso-Reply information and poster ID received from the user's device. The collected data is subjected to text analysis and image analysis to extract frequently occurring words and image features. The analysis results are used to predict the attacker's persona using a generative AI model.

[0860] 3. Emotion Engine

[0861] The emotion engine analyzes the user's voice and facial expression data to recognize the user's emotional state. It also predicts the user's emotions based on the user's past posts and behavioral history. This information is used to adjust the way the persona information is presented.

[0862] 4. Social Media API

[0863] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[0864] Program processing explanation

[0865] The server first collects the attacker's past posting data using the social networking service API, then performs text and image analysis to extract frequently used words and image features.

[0866] The server then combines this information and inputs it into a generative AI model, which predicts a persona for the attacker, including their age, gender, occupation, hobbies, and preferences.

[0867] The server then transmits this persona information to the user's device. At the same time, the device transmits the user's emotional data to the emotion engine. The emotion engine analyzes the user's emotional state and adjusts the way the persona information is presented. For example, if the user is feeling stressed, the persona information is presented in a softer manner.

[0868] Specific examples

[0869] Consider the case where User A receives a malicious reply on social media saying, "You're an idiot." User A's device sends the content of this shitty reply and the poster's ID to a server. The server uses the social media API to collect the poster's past posting data and analyzes the text data and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a generative AI model to predict the persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation."

[0870] During this time, the device sends user A's voice and facial expression data to the emotion engine, which then recognizes user A's stress level. The persona information is then adjusted by the emotion engine and sent from the server to the device. Finally, persona information is provided in a form that takes user A's psychological state into consideration.

[0871] Prompt Sentence Examples

[0872] Examples of prompts that can be input to a generative AI model include:

[0873] "To reduce the anxiety of users who receive malicious replies, please generate a persona for the attacker. The data to be analyzed is the text data 'I love anime, I bought a new gachapon' and the image data 'animes_yokai.jpg'."

[0874] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0875] Step 1:

[0876] A user receives a malicious reply (kusoripu) on social media.

[0877] Input: Malicious replies on social media and their poster IDs

[0878] Output: Kuso-reply content and poster ID

[0879] How it works: The user checks the message on the social networking site, and the device obtains the content of the shitty reply and the poster's ID.

[0880] Step 2:

[0881] The device sends the content of the shit reply and the poster ID to the server.

[0882] Input: Kuso reply content and poster ID

[0883] Output: The content of the kusorip sent to the server and the poster ID

[0884] How it works: The communication module in the device sends the content of the shitty reply and the poster's ID to the server.

[0885] Step 3:

[0886] The server uses the SNS API to collect the poster's past posting data.

[0887] Input: Poster ID

[0888] Output: Collected past post data (text, images)

[0889] How it works: The server uses the SNS API to identify past posts associated with the poster ID and collects post data for a certain period of time.

[0890] Step 4:

[0891] The text portion is extracted from the posted data collected by the server and text analysis is performed.

[0892] Input: Text portion of past post data

[0893] Output: Extracted frequent words and key phrases

[0894] How it works: A natural language processing (NLP) engine on the server analyzes text data and extracts frequently occurring words and key phrases.

[0895] Step 5:

[0896] The image portion is extracted from the posted data collected by the server and image analysis is performed.

[0897] Input: Image portion of past post data

[0898] Output: Extracted image features (character, symbol tags)

[0899] How it works: The image analysis engine on the server analyzes the image data, extracts features, and tags them.

[0900] Step 6:

[0901] The server combines the results of text analysis and image analysis and inputs them into a generative AI model.

[0902] Input: Text analysis results (frequent words, key phrases), image analysis results (features, tags)

[0903] Output: Predicted persona information

[0904] How it works: The server integrates the analysis results and inputs them as prompts into the generative AI model, which then predicts the persona.

[0905] Step 7:

[0906] The server transmits the generated persona information to the user's terminal.

[0907] Input: Generated persona information

[0908] Output: Persona information sent to the device

[0909] Operation: The server sends persona information to the device and prepares it for display.

[0910] Step 8:

[0911] The device sends the user's voice and facial expression data to the emotion engine.

[0912] Input: User's voice data, facial expression data

[0913] Output: User emotion data sent to the emotion engine

[0914] Operation: The device's sensors capture the user's voice and facial expressions and send them to the emotion engine.

[0915] Step 9:

[0916] The emotion engine analyzes the user's emotional state and generates emotion data.

[0917] Input: Voice data, facial expression data

[0918] Output: User's emotional state (stress level, type of emotion)

[0919] How it works: The emotion engine analyzes voice and facial expression data in real time to identify the user's emotional state.

[0920] Step 10:

[0921] An emotional engine adjusts the presentation of persona information based on the user's emotional state.

[0922] Input: User's emotional state, predicted persona information

[0923] Output: Adjusted persona information

[0924] How it works: The emotion engine takes into account the user's emotional state and adjusts the way persona information is presented appropriately. For example, if a user is stressed, it will explain things in a gentler way.

[0925] Step 11:

[0926] The device displays the adjusted persona information to the user.

[0927] Input: Adjusted persona information

[0928] Output: Persona information displayed to the user

[0929] Action: The device uses a display device to present the adjusted persona information to the user in an appropriate format.

[0930] In this way, the system provides information to reduce anxiety when users receive malicious replies and responds according to the user's emotional state.

[0931] 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.

[0932] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0933] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0934] [Third embodiment]

[0935] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0936] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0937] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0938] 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.

[0939] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0940] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0941] 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.

[0942] 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.

[0943] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0944] 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.

[0945] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0946] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0947] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the profile of the attacker and displays it to the user, thereby reducing anxiety. This system uses the user's device, server, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[0948] System Configuration

[0949] 1. User Device

[0950] It is a device that allows users to use SNS and can receive Kuso-Replies. It also has the function of sending the content of Kuso-Replies and the poster ID to the server.

[0951] 2. Server

[0952] The server uses SNS APIs to collect data on attackers' past posts based on the Kuso-Reply information received from the user's device. The collected data is then subjected to text and image analysis to extract frequently occurring words and image features. Furthermore, a machine learning model is used to predict the attacker's persona, and the results are sent to the user's device.

[0953] 3. Social Media API

[0954] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[0955] Program processing

[0956] 1. Receiving shitty replies

[0957] A user receives a shitty reply on social media and checks its content and the poster's ID.

[0958] The device sends the content of the shit reply and the poster ID to the server.

[0959] 2. Attacker Identification and Data Collection

[0960] The server receives the ID of the user who posted the kusorip and uses the SNS API to collect the attacker's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0961] 3. Extracting and analyzing text data

[0962] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques.

[0963] 4. Image Data Extraction and Analysis

[0964] The server extracts image data from the collected submissions, extracts image features using image analysis technology (e.g., computer vision and deep learning), and assigns tags such as characters and symbols to the images.

[0965] 5. Data integration for persona prediction

[0966] The server combines the results of text analysis and image analysis to create a comprehensive dataset, which is then fed into a machine learning model to predict the attacker's age, gender, occupation, hobbies, and preferences.

[0967] 6. Persona generation and provision

[0968] The server generates a standard persona of an attacker based on the prediction results of the machine learning model. The generated persona information is structured in a format that is easy for users to understand.

[0969] The device provides the user with the persona information received from the server and displays information such as, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0970] Specific examples

[0971] Let's say a user receives a malicious reply saying, "Your opinion is off the mark." The device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API and analyzes the text data and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict the persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." The prediction result is then sent to the user's device, and the attacker's standard persona information is presented to the user.

[0972] The above is a specific embodiment for carrying out the present invention, which allows the user to understand the personality of the sender of the malicious reply and reduce anxiety.

[0973] The processing flow will be explained below.

[0974] Step 1:

[0975] A user receives a malicious reply on a social networking site, which appears as a reply to a specific post.

[0976] Step 2:

[0977] The device obtains the content of the Kuso-Reply and the poster's ID, and sends this to the system's server, including the text content of the Kuso-Reply and the poster's account ID.

[0978] Step 3:

[0979] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[0980] Step 4:

[0981] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and creates a clean dataset.

[0982] Step 5:

[0983] The server uses natural language processing (NLP) technology to analyze the extracted text data. This analysis identifies frequently occurring words and key phrases. Specifically, it uses morphological analysis and TF-IDF (Term Frequency-Inverse Document Frequency) techniques.

[0984] Step 6:

[0985] The server extracts the image portion from the collected post data, which includes all image files included in each post.

[0986] Step 7:

[0987] The server uses image analysis technology (e.g., computer vision and deep learning) to analyze the features of the extracted images. Specifically, it performs object recognition and assigns tags such as characters and symbols.

[0988] Step 8:

[0989] The server combines the results of the text analysis and the image analysis, creating a comprehensive dataset.

[0990] Step 9:

[0991] The server then inputs the combined data set into a machine learning model to predict the attacker's persona, including attributes such as age, gender, occupation, and hobbies.

[0992] Step 10:

[0993] The server generates a standard persona for an attacker based on the predictions obtained from the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[0994] Step 11:

[0995] The server transmits the generated persona information to the user's terminal.

[0996] Step 12:

[0997] The device presents the received persona information to the user. This is displayed on the screen in a format that is easy for the user to confirm. For example, it might say, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[0998] By going through these steps, users can understand the profile of the attacker and reduce their anxiety.

[0999] Example 1

[1000] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1001] Users who receive malicious replies (kuso-replies) on social media often feel anxious and stressed. However, typical countermeasures include ignoring the replies or blocking the user, which does not fundamentally alleviate the anxiety. Furthermore, there are limited ways to specifically identify the attacker, leaving users plagued by fear of the so-called "invisible enemy." The present invention aims to address these issues by predicting the attacker's profile and presenting it to users, thereby providing them with specific information and reducing their anxiety.

[1002] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1003] In this invention, the server includes a device for receiving malicious replies, a device for collecting past posts by the poster of the malicious replies, means for extracting text from the collected past posts and performing preprocessing, means for analyzing frequently occurring words and key phrases from the preprocessed text, means for extracting images from the collected posts and analyzing their features, means for predicting the poster's personality based on the analysis results, and means for providing the predicted personality information to the device. This allows the user to grasp the specific personality of the attacker, making it possible to reduce anxiety and stress.

[1004] A "malicious reply" is a comment or message sent on social media with the intention of attacking or offending others.

[1005] "Devices" is a general term for devices, machines, and related systems and software used by users to access SNS.

[1006] A "poster" is a person or account that posts comments, messages, images, etc. on social media.

[1007] "Past posts" refer to comments, messages, images, and other output made on social media within a specified period.

[1008] "Text" refers to posts and comments containing written information made on social media.

[1009] "Preprocessing" is the process of removing unnecessary elements (e.g., hashtags and links) from text information and converting it into a format suitable for analysis.

[1010] A "frequent word" is a word that occurs many times within a particular text dataset.

[1011] A "key phrase" is a phrase or word that has an important meaning within text data.

[1012] "Analysis" is the process of analyzing text and image data and extracting meaningful information.

[1013] "Features" are important patterns or characteristics extracted from image data.

[1014] A "personal profile" is the result of predicting an individual's attribute information, such as age, gender, occupation, and hobbies and preferences.

[1015] "Information" refers to data and knowledge such as predictions and analysis results of a person's profile.

[1016] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the profile of the attacker and displays this information to the user, thereby reducing anxiety. This system uses the user's device, server, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the profile of the attacker.

[1017] System Configuration

[1018] 1. User Device

[1019] The user device is a device that allows users to use SNS. It has the function of receiving kusorip messages and sending their content and the poster's ID to a server. When a user receives a kusorip message on SNS, the device sends the content and the poster's ID to the server using an HTTPS request.

[1020] 2. Server

[1021] The server uses SNS APIs to collect data on the attacker's past posts based on the Kuso-Reply information received from the user's device. The collected data is subjected to text analysis and image analysis to extract frequently occurring words and image features. It also uses a machine learning model to predict the attacker's profile and sends the results to the user's device. Specifically, the server preprocesses the text data using "NLTK" and extracts important key phrases using natural language processing technology. It also performs image analysis using "OpenCV" and "TensorFlow" to extract important features.

[1022] 3. Social Media API

[1023] The SNS API is an application programming interface provided by SNS platforms, and is used by the server to collect data on past posts by attackers. Authentication is performed using the SNS API authentication token, and post data from a specified period is obtained.

[1024] Specific example of system operation

[1025] Suppose a user receives a malicious reply saying, "Your opinion is off the mark." The user's device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API. The collected text data is preprocessed using NLTK to extract frequently occurring words and key phrases. At the same time, the collected image data is analyzed using OpenCV or TensorFlow to extract features. The results are integrated, and a machine learning model (e.g., SciKit-Learn) is used to predict a profile of the attacker, such as "male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." Finally, the prediction result is sent to the user's device, and the user is presented with a standard profile of the attacker.

[1026] Prompt Sentence Examples

[1027] An example of a prompt would be:

[1028] I received a shitty reply saying, "Your opinion is off the mark." Please predict and provide a profile of the attacker.

[1029] The above-mentioned method allows users to get a clear picture of the attacker, which can reduce anxiety and stress.

[1030] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1031] Step 1: Receiving a Shit Reply

[1032] A user receives a malicious reply (kuso-repu) on social media. For example, the reply is, "Your opinion is irrelevant."

[1033] Input: Shitty reply on social media and poster ID.

[1034] The device temporarily stores the content of the KusoRip and the poster ID in its internal memory, then sends the content of the KusoRip and the poster ID to the server using an HTTPS request.

[1035] Output: The content of the shit reply sent to the server and the poster ID.

[1036] Step 2: Identifying the attacker and gathering data

[1037] The server checks the content of the kusorip received from the device and the poster ID.

[1038] Input: The content of the shitty reply sent from the device and the poster ID.

[1039] The server uses the SNS API to collect the attacker's posting data for the past year. The server authenticates using the SNS API's authentication token and sends an API request. This request includes the poster ID and the data collection period (e.g., the past year).

[1040] Output: Posted data (text and images) by attackers collected on the server over the past year.

[1041] Step 3: Extract and analyze text data

[1042] The server extracts the text from the collected post data and preprocesses the text of the post to remove hashtags and links.

[1043] Input: Attacker's past posting data (text).

[1044] The server uses NLTK to preprocess the text data, tokenizing it, removing stop words, and performing word frequency analysis, using natural language processing techniques to extract important frequently occurring words and key phrases.

[1045] Output: Preprocessed text data and analysis results (frequent words and key phrases).

[1046] Step 4: Extraction and analysis of image data

[1047] The server extracts the image portion from the collected submission data, extracts image features using computer vision and deep learning technology, and assigns tags such as characters and symbols to them.

[1048] Input: Attacker's past posting data (images).

[1049] The server uses OpenCV and TensorFlow to analyze the image and extract important features, such as identifying and labeling anime characters or specific symbols.

[1050] Output: Analyzed image data and feature information.

[1051] Step 5: Data integration for persona prediction

[1052] The server combines the results of text analysis and image analysis to create a comprehensive dataset.

[1053] Input: Text analysis results and image analysis results.

[1054] The server inputs this dataset into a machine learning model to predict the attacker's personality, such as age, gender, occupation, hobbies, etc. For example, the server inputs the dataset into a machine learning model trained using "SciKit-Learn" to obtain a predicted personality profile.

[1055] Output: Predicted attacker profile (age, gender, occupation, hobbies, etc.).

[1056] Step 6: Generate and deliver personas

[1057] The server generates a standard profile of the attacker based on the predictions of the machine learning model, and the profile information is organized in a format that is easy for users to understand.

[1058] Input: Prediction result (attacker profile).

[1059] The server generates the person profile information in JSON format and sends it to the user device using HTTPS.

[1060] The user device analyzes the persona information received from the server and displays it on the user interface. For example, it displays information such as, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[1061] Output: Attacker profile information displayed in the user interface.

[1062] In this way, this system can reduce anxiety and stress by presenting a specific profile of the attacker in response to a malicious reply received by the user.

[1063] (Application example 1)

[1064] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1065] In online social networking services (SNS), users often receive malicious replies (kuso-replies) from other users. These replies can cause psychological stress and anxiety to the recipient. Therefore, there is a need for a method to reduce users' anxiety and provide a safe SNS environment by revealing the causes of these replies and the identity of the people behind them.

[1066] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1067] In this invention, the server includes a means for extracting frequently occurring words from past posts, a means for analyzing the extracted frequently occurring words, and a means for extracting images from past posts. This makes it possible to analyze the poster's personality and predict persona information. Furthermore, the prediction includes a means for using a generative AI model and prompt sentences, making it possible to improve the accuracy of the analysis results. This makes it possible to clarify the identity of the person behind malicious replies to users, allowing users to use SNS without feeling anxious.

[1068] A "terminal" is a device through which a user receives malicious replies and transmits the information to a server.

[1069] The "server" is a computer system that receives information about the poster of a malicious reply sent from a user's terminal, collects and analyzes the poster's past posting data, and provides persona information.

[1070] "Method for extracting frequently occurring words" is a method for selecting frequently used words from posted data.

[1071] "Means for analyzing frequently occurring words" is a method for analyzing extracted words and finding their meanings and relationships.

[1072] "Means for extracting images" refers to a method for selecting image media from posted data.

[1073] The "means for analyzing images" is a method for analyzing the extracted images and recognizing their contents and features.

[1074] A "means for predicting persona" is a method for inferring characteristics such as the poster's age, gender, hobbies and preferences based on collected and analyzed data.

[1075] "Using a generative AI model" means using a model generated using artificial intelligence technology.

[1076] "Prompt methods" are methods that use sentences to provide a generative AI model with a specific question and context to get an answer.

[1077] "Information Provision Interface" means the application programming interface (API) that the server uses to collect data from the social networking platform.

[1078] The system based on this invention reduces anxiety when a user receives a malicious reply (kuso-reply) on social media by predicting the profile of the attacker and providing this information to the user. The system uses the user's device, server, and social media API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[1079] System Configuration

[1080] 1. User Device

[1081] It is a device that allows users to use social networking services, and when it receives a shitty reply, it sends the content and the poster's ID to the server.

[1082] 2. Server

[1083] The server uses SNS APIs to collect data on the attacker's past posts based on the Kuso-Reply information received from the user's device. The collected data is then subjected to text and image analysis. Natural language processing (NLP) technology is used for text analysis, and computer vision and deep learning technology are used for image analysis. Furthermore, a generative AI model is used to predict the attacker's persona. A prompt sentence is used to make highly accurate predictions.

[1084] 3. Social Media API

[1085] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[1086] Program processing example

[1087] Receiving and sending shit replies

[1088] The user receives a kusorip message on a social networking site and enters the message's content and the poster's ID into their device, which then sends this information to the server.

[1089] Attacker Identification and Data Collection

[1090] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's posting data for the past year.

[1091] Text data extraction and analysis

[1092] The server extracts text from the collected post data and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques, such as NLTK and spaCy libraries.

[1093] Image data extraction and analysis

[1094] The server extracts image data from the collected submissions and uses computer vision technology to extract image features, using deep learning technologies such as TensorFlow and PyTorch.

[1095] Persona Prediction

[1096] The server integrates the results of text analysis and image analysis and predicts the attacker's persona based on a generative AI model. The prompt sentence is used as input to the model, and persona information is predicted with high accuracy.

[1097] Providing persona information

[1098] The predicted persona information is sent to the user's device and displayed to the user, allowing the user to understand the profile of the attacker and reduce anxiety.

[1099] Specific examples

[1100] When a user receives a crappy reply saying, "Your opinion is completely meaningless," the device sends the content of the crappy reply and the poster's ID to the server. The server collects the poster's posting data from the past year via SNS APIs and analyzes the text data and image data. For example, if "games" and "anime" are detected as frequently occurring words and image analysis reveals that specific character images appear frequently, the generative AI model is used to predict a persona such as "male in his 20s, likes anime and games, unemployed or working part-time." The prediction results are then sent to the user's device and displayed to the user.

[1101] Example prompt sentence:

[1102] Please predict the attacker's persona based on the following text analysis and image analysis results. Text analysis results: ['game', 'anime'], Image analysis results: ['anime character']

[1103] The above is a specific embodiment for carrying out the invention.

[1104] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1105] Step 1:

[1106] The user receives a kusorip message and enters its contents and the poster's ID into the device.

[1107] When a user receives a malicious reply (kuso-reply) on social media, they enter the content and the sender's ID into a dedicated application. This entered data (the content of the kuso-reply and the poster's ID) is saved on the device.

[1108] Step 2:

[1109] The device sends the content of the Kuso-Reply and the poster ID to the server.

[1110] The device collects the text of the entered kusorip and the poster ID and sends it to the server. This is done using a method such as an HTTP POST request. The input is the kusorip and the poster ID, and the output is a status message confirming that this information has been sent to the server.

[1111] Step 3:

[1112] The server uses the SNS API to collect the attacker's past posting data.

[1113] The server uses the received poster ID to collect the attacker's past posting data via the SNS platform's API. The collected data includes text posts, images, videos, etc., covering the past year. The input is the poster ID, and the output is the collected posting data.

[1114] Step 4:

[1115] Extract the text data collected by the server

[1116] The server extracts text parts from the collected submission data using regular expressions and NLP preprocessing techniques. The input is the submission data, and the output is the extracted text data.

[1117] Step 5:

[1118] The server analyzes the text data using NLP techniques.

[1119] The server preprocesses the extracted text data to identify frequently occurring words and key phrases. The libraries used are NLTK and spaCy. The input is text data, and the output is frequently occurring words and key phrases. Specific operations include tokenization and stop word removal.

[1120] Step 6:

[1121] Extracting image data collected by the server

[1122] The server extracts images from the collected submission data by filtering the data type. The input is the submission data, and the output is the extracted image data.

[1123] Step 7:

[1124] The server analyzes the image data using image analysis technology.

[1125] The server analyzes the extracted image data using computer vision techniques, such as deep learning frameworks like TensorFlow and PyTorch. The input is image data, and the output is image features and labels. Specific operations include object detection and image classification.

[1126] Step 8:

[1127] The server integrates the results of text analysis and image analysis and predicts the persona using a generative AI model.

[1128] The server uses a dataset that integrates the results of text analysis and image analysis, and inputs a prompt into the generative AI model to predict the attacker's persona. The input is the integrated dataset and prompt, and the output is the persona prediction result. Example prompt: "Please predict the attacker's persona based on the following text analysis and image analysis results. Text analysis result: ['Game', 'Anime'], Image analysis result: ['Anime character']"

[1129] Step 9:

[1130] The server sends the predicted persona information to the user device.

[1131] The server sends the predicted persona information to the user's device. This is done via an HTTP POST request, etc. The input is the persona prediction result, and the output is the confirmation status of the transmission to the user's device.

[1132] Step 10:

[1133] The user device displays the persona information to the user.

[1134] The user device displays the received persona information on the user screen. This allows the user to understand the profile of the attacker and reduce anxiety. The input is persona information, and the output is the displayed persona information. Specifically, it operates by displaying text and images on the interface.

[1135] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1136] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the attacker's personality and further recognizes the user's emotions and adjusts the presentation method to reduce anxiety. This system uses the user's device, server, emotion engine, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[1137] System Configuration

[1138] 1. User Device

[1139] It is a device that allows users to use SNS and can receive Kuso-Reply messages. It also has the function of sending the content of the Kuso-Reply message and the poster's ID to the server. It also has the function of providing the user's voice and facial expressions to the emotion engine.

[1140] 2. Server

[1141] The server uses SNS APIs to collect attacker's past posting data based on the information on kusoripulp received from the user's device and the user's emotional data. The collected data is subjected to text and image analysis to extract frequently occurring words and image features. Furthermore, a machine learning model is used to predict the attacker's persona and the results are sent to the user's device.

[1142] 3. Emotion Engine

[1143] The emotion engine collects and analyzes the user's voice and facial expression data to recognize the user's emotional state. It also predicts emotions based on the user's past posts and behavioral history, and adjusts the way persona information is presented.

[1144] 4. Social Media API

[1145] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[1146] Program processing

[1147] 1. Receiving shitty replies

[1148] A user receives a malicious reply on a social networking site and checks its content and the poster's ID.

[1149] The device sends the content of the shit reply and the poster ID to the server.

[1150] 2. Attacker Identification and Data Collection

[1151] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[1152] 3. Extracting and analyzing text data

[1153] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques.

[1154] 4. Image Data Extraction and Analysis

[1155] The server extracts image data from the collected submissions, extracts image features using image analysis technology (e.g., computer vision and deep learning), and assigns tags such as characters and symbols to the images.

[1156] 5. Emotion recognition

[1157] The device continuously collects the user's voice and facial expression data and sends it to the emotion engine.

[1158] The emotion engine analyzes this data to recognize the user's emotional state and predicts their emotions based on their past posts and behavioral history.

[1159] 6. Data integration for persona prediction

[1160] The server combines the results of text analysis and image analysis to create a comprehensive dataset, which is then fed into a machine learning model to predict the attacker's age, gender, occupation, hobbies, and preferences.

[1161] 7. Persona generation and provision

[1162] The server generates a standard persona for an attacker based on the predictions of the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[1163] The emotion engine adjusts the presentation of persona information based on the user's perceived emotions. For example, if the user is under a lot of stress, it will display softer expressions and encouraging words.

[1164] 8. Displaying the results

[1165] The server transmits the generated persona information to the user's terminal.

[1166] The device then presents the received persona information to the user. For example, information such as "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time," is displayed in a format that is adjusted to reflect the user's emotional state.

[1167] Specific examples

[1168] Suppose a user receives a malicious reply saying, "You're such an idiot." The device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API and analyzes the text and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict a persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." During this time, the device sends the user's voice and facial expression data to an emotion engine, which then recognizes the user's stress level. The emotion engine then adjusts the persona information and sends it from the server to the device. Finally, the persona information is presented in a way that takes into account the user's psychological state.

[1169] The above is a specific example of how to implement the present invention, which allows users to understand the personality of the sender of a malicious reply and reduce anxiety. By incorporating an emotion engine, information is provided that takes into consideration the user's psychological state.

[1170] The processing flow will be explained below.

[1171] Step 1:

[1172] A user receives a malicious reply (kuso-reply) on a social networking site, which appears as a comment on a specific post.

[1173] Step 2:

[1174] The device obtains the content of the Kuso-Reply and the poster's ID, and sends this to the system's server, including the text content of the Kuso-Reply and the poster's account ID.

[1175] Step 3:

[1176] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[1177] Step 4:

[1178] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and creates a clean dataset.

[1179] Step 5:

[1180] The server uses natural language processing (NLP) technology to analyze the extracted text data. This analysis identifies frequently occurring words and key phrases. Specifically, it uses morphological analysis and TF-IDF (Term Frequency-Inverse Document Frequency) techniques.

[1181] Step 6:

[1182] The server extracts the image portion from the collected post data, which includes all image files included in each post.

[1183] Step 7:

[1184] The server uses image analysis technology (e.g., computer vision and deep learning) to analyze the features of the extracted images. Specifically, it performs object recognition and assigns tags such as characters and symbols.

[1185] Step 8:

[1186] The server combines the results of the text analysis and the image analysis, creating a comprehensive dataset.

[1187] Step 9:

[1188] The server then inputs the combined data set into a machine learning model to predict the attacker's persona, including attributes such as age, gender, occupation, and hobbies.

[1189] Step 10:

[1190] The server generates a standard persona for an attacker based on the predictions obtained from the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[1191] Step 11:

[1192] The device continuously collects the user's voice and facial expression data and sends it to the emotion engine.

[1193] Step 12:

[1194] The emotion engine analyzes this data to recognize the user's emotional state and predicts their emotions based on their past posts and behavioral history.

[1195] Step 13:

[1196] The emotion engine adjusts the presentation of persona information based on the user's emotional state. For example, if a user is under a lot of stress, it will display softer language and encouraging words.

[1197] Step 14:

[1198] The server transmits the generated persona information to the user's terminal.

[1199] Step 15:

[1200] The device presents the received persona information to the user. This is displayed on the screen in a format that is easy for the user to confirm. For example, it might say, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[1201] Through these steps, users can understand the profile of the attacker and reduce their anxiety. By incorporating an emotion engine, information is provided that takes into account the user's psychological state.

[1202] Example 2

[1203] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1204] On modern social media platforms, malicious replies (kusoripu) can cause psychological stress to users. However, there is no way for users who receive such replies to understand who is attacking them, and there is no way to provide appropriate information depending on their psychological state, making it difficult to alleviate their anxiety and stress.

[1205] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for collecting past posts by posters of malicious replies, means for extracting frequently occurring words from past posts, means for analyzing the extracted images, means for predicting the poster's persona based on the analysis results, an emotion engine for collecting and analyzing user emotion data, means for adjusting the predicted persona information based on the user's emotional state, and means for providing the adjusted persona information to the terminal. This makes it easier for the user to grasp the profile of the attacker and further enables the provision of appropriate information based on the user's psychological state.

[1206] A "malicious reply" is a comment or message on a social networking site that contains offensive, derogatory, or unpleasant content directed at a user.

[1207] A "terminal" is a device that allows a user to use a social networking service and has the function of receiving malicious replies and sending their contents and the poster's ID to the server.

[1208] A "server" is a computer system that receives data sent from a user's device, collects the necessary data through the SNS API, and performs analytical processing.

[1209] "Past posts" refers to all content posted by an attacker on social media within a specific period of time.

[1210] "Frequent words" are keywords and phrases that appear frequently and are extracted from past posts.

[1211] An "extraction method" is a set of algorithms or techniques used to collect and analyze specific data (e.g., text or images).

[1212] "Means for analysis" refers to methods and techniques for analyzing the extracted data and extracting useful information and features.

[1213] A "persona" is a predicted profile of an attacker based on the analysis results, including their age, gender, occupation, hobbies, and preferences.

[1214] An "emotion engine" is a system or technology that collects and analyzes a user's emotional data to recognize the user's psychological state.

[1215] "SNS API" is an application programming interface provided by an SNS platform and is an interface used to collect past posting data.

[1216] A "machine learning model" is an algorithm that learns using large amounts of data and uses the learning results to make predictions and analyze new data.

[1217] "Means for adjusting based on the user's emotional state" refers to techniques or methods for changing the method or content of information presentation depending on the user's psychological state.

[1218] "Means for providing" refers to communication and interface technologies for displaying analysis results and predicted information on a user terminal.

[1219] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the identity of the attacker and further recognizes the user's emotions and adjusts the presentation method to reduce anxiety. This system utilizes the user's device, server, emotion engine, and social networking site API, and is implemented in the following specific steps.

[1220] First, when a user receives a malicious reply on a social networking site, the device sends the content of the kusoriply and the poster's ID to a server. The device can be a smartphone or a PC, and the social networking application runs as the communication software.

[1221] The server receives the Kuso-Reply information sent from the device and uses the SNS API to collect the poster's past posting data. The collected data is classified into text data and image data, and each type of data is analyzed. For example, the text data is analyzed using natural language processing (NLP) technology to extract frequently occurring words and key phrases, and the image data is analyzed using computer vision and deep learning technology to analyze the image features.

[1222] Next, the emotion engine collects the user's voice and facial expression data in real time to recognize the user's emotional state. This data includes the user's facial expressions and tone of voice collected using a camera and microphone. The emotion engine also takes into account the user's past posts and behavioral history to predict the user's emotional state.

[1223] The server combines the results of text analysis and image analysis to generate an integrated dataset. This dataset is then input into a machine learning model to predict the attacker's persona, including their age, gender, occupation, and hobbies and interests. Machine learning models such as multilayer perceptrons and convolutional neural networks (CNNs) are used.

[1224] After obtaining the persona prediction results, the emotion engine adjusts the way the persona information is presented based on the user's emotional state. For example, if the user is under a lot of stress, the explanation will be changed to a more gentle expression or words of encouragement will be added.

[1225] Finally, the server sends the adjusted persona information to the user's device, which receives it and presents it to the user. For example, information such as "According to an analysis of this user, it appears that he is a man in his 20s who likes anime and is currently unemployed or working part-time" is displayed in a format that reflects the user's emotional state.

[1226] Specific examples

[1227] If a user receives a malicious reply such as "You're such an idiot," the device sends the content of the kusoriply and the poster's ID to the server. The server uses a social media API to collect the poster's posting data from the past year and analyzes the text and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict a persona of "male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." During this time, the device sends the user's voice and facial expression data to an emotion engine, which then recognizes the user's stress level. The persona information is then adjusted by the emotion engine and sent from the server to the device. Finally, the persona information is presented in a form that takes into account the user's psychological state.

[1228] Prompt Sentence Examples

[1229] Below are some example prompts to input to a generative AI model:

[1230] "We want to predict the personality of the person who posts malicious replies (kuso-replies) on social media that cause anxiety in users. At the same time, we want to design a method of providing information that recognizes the user's emotional state and can alleviate that anxiety. An example of a kuso-reply is: 'You're such an idiot.'"

[1231] Based on these prompts, the generative AI model will generate an appropriate persona and process emotions.

[1232] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1233] Step 1: Receiving a Shit Reply

[1234] The user receives malicious replies on social media such as "You're such an idiot."

[1235] Input: Malicious reply on a social media app

[1236] Output: Malicious reply content and poster ID

[1237] Specific behavior: When a user opens a social media application on their smartphone or computer, they will receive notifications and direct messages containing malicious replies.

[1238] Step 2: Send the content of the kusorip and the poster ID

[1239] The device automatically sends the content of the malicious reply received and the poster ID to the server.

[1240] Input: Malicious reply content and poster ID

[1241] Output: The malicious reply sent to the server and its poster ID.

[1242] Specific operation: The content of the shitty reply, "You are truly an idiot," and the poster's ID are sent to the server in the device's background process.

[1243] Step 3: Identifying the attacker and gathering data

[1244] The server collects posting data for the past year from the Shit Reply poster ID received using the SNS API.

[1245] Input: Shit reply poster ID

[1246] Output: Post data from the past year

[1247] What happens: The server sends a request to the SNS API to retrieve post data associated with the poster ID, e.g., tweets, posts, and comments from the past year.

[1248] Step 4: Extract and analyze text data

[1249] The server extracts the text portion from the collected posting data, performs preprocessing, and then analyzes it using natural language processing (NLP) technology.

[1250] Input: Post data from the past year

[1251] Output: Frequent words and key phrases

[1252] What it does: It removes hashtags and links, then performs morphological analysis on the text to extract frequently occurring words and key phrases. For example, it identifies keywords like "anime," "games," and "stress."

[1253] Step 5: Extract and analyze image data

[1254] The server extracts image portions from the collected posting data and extracts features using image analysis technology.

[1255] Input: Post data from the past year

[1256] Output: Image feature information

[1257] What it does: Uses computer vision technology to identify and tag characters and symbols from images. Example: Detecting the anime character "Naruto."

[1258] Step 6: Emotion Recognition

[1259] The device continuously transmits the user's voice and facial expression data to the emotion engine.

[1260] Input: User's voice data and facial expression data

[1261] Output: Voice and facial expression data sent to the emotion engine

[1262] Specific operation: When a user uses a social networking app, the camera and microphone are used to detect facial expressions and tone of voice. This data is sent to the emotion engine in real time.

[1263] The emotion engine analyzes the received data and recognizes the user's emotional state.

[1264] Input: Voice data and facial expression data

[1265] Output: User's emotional state

[1266] Specific behavior: If the user exhibits a sad facial expression or an irritated tone of voice, the emotion engine detects a stressed state.

[1267] Step 7: Data integration for persona prediction

[1268] The server combines the results of text analysis and image analysis to create a comprehensive dataset.

[1269] Input: Text analysis results and image analysis results

[1270] Output: Unified dataset

[1271] Specific operation: Combine text and image features into a single dataset and provide it to a machine learning model. Example: Combine text keywords and image tag information.

[1272] Step 8: Generate and deliver personas

[1273] The server generates a standard persona of an attacker from the prediction results of the machine learning model.

[1274] Input: Integrated dataset

[1275] Output: Persona information

[1276] Specific operation: The machine learning model outputs the following persona information: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation."

[1277] The emotional engine adjusts how persona information is presented based on the user's emotional state.

[1278] Input: User's emotional state and persona information

[1279] Output: Adjusted persona information

[1280] Specific actions: For users who are under a lot of stress, add soft words and words of encouragement. For example, "This user is a man in his 20s who loves anime. He may be unemployed or working part-time, but you'll be fine. Don't let his opinions affect you too much."

[1281] Step 9: View the results

[1282] The server transmits the adjusted persona information to the user's terminal.

[1283] Input: Adjusted persona information

[1284] Output: Data sent to the terminal

[1285] Specific operation: The server sends persona information to the user's device and issues a display instruction.

[1286] The terminal displays the received persona information to the user.

[1287] Input: Persona information sent from the server

[1288] Output: Data displayed to the user

[1289] Specific operation: The device displays a message on the screen stating, "This user is a male in his 20s who likes anime and is currently unemployed or possibly working part-time." It also displays words of encouragement based on the user's stress level.

[1290] The above is the specific processing flow of this system's program. By including the specific operations at each step, it is clear how the data is processed when a user receives a 'kusorip' and what information is presented as the final result. In addition, by incorporating an emotion engine, it is possible to provide information that takes into account the user's psychological state.

[1291] (Application example 2)

[1292] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1293] In recent years, the spread of social networking sites has increased the risk of receiving malicious replies (kuso-replies). In such situations, many users experience psychological stress and anxiety. Furthermore, because they do not know the sender of the malicious replies, users are likely to be placed in unpleasant situations. Current social networking sites lack the ability to recognize users' emotions and respond appropriately. For this reason, a system that can reduce users' psychological burden is needed.

[1294] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1295] In this invention, the server includes a means for extracting frequently occurring words from past posts, a means for analyzing the extracted frequently occurring words, and a means for extracting images from past posts. This makes it possible to reduce the psychological burden on users who receive malicious replies. It also includes a means for predicting personas using a generative AI model and a means for recognizing the user's emotional state and adjusting the presentation method of persona information, making it possible to provide attacker information in a format tailored to the user. Furthermore, by including a means for transmitting the content of the kusorip and the poster ID to the server and a means for transmitting the user's voice and facial expression data to the emotion engine, it is possible to more accurately grasp the user's emotions and take appropriate action.

[1296] "Hateful replies" refer to comments or messages that are offensive, insulting, or inappropriate towards others on social media or other platforms.

[1297] "Terminal" refers to a device used by a user to use an SNS, including a smartphone, tablet, or PC.

[1298] A "server" is a device that stores, processes, and transmits data over a computer network.

[1299] "Past posts" refers to all messages, comments, images, videos, and other posts made by a particular user in the past on a social media platform.

[1300] A "frequent word" refers to a word that appears more frequently than other words in the text data to be analyzed.

[1301] "Extraction means" refers to a process or device that extracts specific data (text, images, audio, etc.) from other data.

[1302] "Analysis means" refers to the process or device used to analyze the extracted data in detail and understand its meaning and characteristics.

[1303] A "persona" is a virtual image of a person created based on the poster's characteristics, including attributes such as age, gender, occupation, and hobbies.

[1304] An "emotion engine" refers to software or hardware that recognizes and analyzes a user's emotional state from voice, facial expression, and other data.

[1305] "Generative AI model" refers to an artificial intelligence model used to generate specific patterns or information from given data.

[1306] A "prompt" is an instruction given to a generative AI model to perform a specific task.

[1307] The system based on the present invention, "KusoReplyGuard," predicts the profile of the attacker when a malicious reply is received on a social networking site, and provides information according to the user's emotional state, thereby reducing anxiety. A specific embodiment of this system will be described below.

[1308] System Configuration

[1309] The system consists of four main components:

[1310] 1. User Device

[1311] This is a device that users use to access social media. It can be a smartphone, tablet, or PC. The device receives the Kuso-Reply and sends the content and poster ID to a server. In addition, the device has the ability to collect the user's voice and facial expression data and send it to the emotion engine.

[1312] 2. Server

[1313] The server uses SNS APIs to collect the attacker's past posting data based on the Kuso-Reply information and poster ID received from the user's device. The collected data is subjected to text analysis and image analysis to extract frequently occurring words and image features. The analysis results are used to predict the attacker's persona using a generative AI model.

[1314] 3. Emotion Engine

[1315] The emotion engine analyzes the user's voice and facial expression data to recognize the user's emotional state. It also predicts the user's emotions based on the user's past posts and behavioral history. This information is used to adjust the way the persona information is presented.

[1316] 4. Social Media API

[1317] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[1318] Program processing explanation

[1319] The server first collects the attacker's past posting data using the social networking service API, then performs text and image analysis to extract frequently used words and image features.

[1320] The server then combines this information and feeds it into a generative AI model, which predicts a persona for the attacker, including their age, gender, occupation, hobbies, and interests.

[1321] The server then transmits this persona information to the user's device. At the same time, the device transmits the user's emotional data to the emotion engine. The emotion engine analyzes the user's emotional state and adjusts the way the persona information is presented. For example, if the user is feeling stressed, the persona information is presented in a softer manner.

[1322] Specific examples

[1323] Consider the case where User A receives a malicious reply on social media saying, "You're an idiot." User A's device sends the content of this shitty reply and the poster's ID to a server. The server uses the social media API to collect the poster's past posting data and analyzes the text data and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a generative AI model to predict the persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation."

[1324] During this time, the device sends user A's voice and facial expression data to the emotion engine, which then recognizes user A's stress level. The persona information is then adjusted by the emotion engine and sent from the server to the device. Finally, persona information is provided in a form that takes user A's psychological state into consideration.

[1325] Prompt Sentence Examples

[1326] Examples of prompts that can be input to a generative AI model include:

[1327] "To reduce the anxiety of users who receive malicious replies, please generate a persona for the attacker. The data to be analyzed is the text data 'I love anime, I bought a new gachapon' and the image data 'animes_yokai.jpg'."

[1328] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1329] Step 1:

[1330] A user receives a malicious reply (kusoripu) on social media.

[1331] Input: Malicious replies on social media and their poster IDs

[1332] Output: Kuso-reply content and poster ID

[1333] How it works: The user checks the message on the social networking site, and the device obtains the content of the shitty reply and the poster's ID.

[1334] Step 2:

[1335] The device sends the content of the shit reply and the poster ID to the server.

[1336] Input: Kuso reply content and poster ID

[1337] Output: The content of the kusorip sent to the server and the poster ID

[1338] How it works: The communication module in the device sends the content of the shitty reply and the poster's ID to the server.

[1339] Step 3:

[1340] The server uses the SNS API to collect the poster's past posting data.

[1341] Input: Poster ID

[1342] Output: Collected past post data (text, images)

[1343] How it works: The server uses the SNS API to identify past posts associated with the poster ID and collects post data for a certain period of time.

[1344] Step 4:

[1345] The text portion is extracted from the posted data collected by the server and text analysis is performed.

[1346] Input: Text portion of past post data

[1347] Output: Extracted frequent words and key phrases

[1348] How it works: A natural language processing (NLP) engine on the server analyzes text data and extracts frequently occurring words and key phrases.

[1349] Step 5:

[1350] The image portion is extracted from the posted data collected by the server and image analysis is performed.

[1351] Input: Image portion of past post data

[1352] Output: Extracted image features (character, symbol tags)

[1353] How it works: The image analysis engine on the server analyzes the image data, extracts features, and tags them.

[1354] Step 6:

[1355] The server combines the results of text analysis and image analysis and inputs them into a generative AI model.

[1356] Input: Text analysis results (frequent words, key phrases), image analysis results (features, tags)

[1357] Output: Predicted persona information

[1358] How it works: The server integrates the analysis results and inputs them as prompts into the generative AI model, which then predicts the persona.

[1359] Step 7:

[1360] The server transmits the generated persona information to the user's terminal.

[1361] Input: Generated persona information

[1362] Output: Persona information sent to the device

[1363] Operation: The server sends persona information to the device and prepares it for display.

[1364] Step 8:

[1365] The device sends the user's voice and facial expression data to the emotion engine.

[1366] Input: User's voice data, facial expression data

[1367] Output: User emotion data sent to the emotion engine

[1368] Operation: The device's sensors capture the user's voice and facial expressions and send them to the emotion engine.

[1369] Step 9:

[1370] The emotion engine analyzes the user's emotional state and generates emotion data.

[1371] Input: Voice data, facial expression data

[1372] Output: User's emotional state (stress level, type of emotion)

[1373] How it works: The emotion engine analyzes voice and facial expression data in real time to identify the user's emotional state.

[1374] Step 10:

[1375] An emotional engine adjusts the presentation of persona information based on the user's emotional state.

[1376] Input: User's emotional state, predicted persona information

[1377] Output: Adjusted persona information

[1378] How it works: The emotion engine takes into account the user's emotional state and adjusts the way persona information is presented appropriately. For example, if a user is stressed, it will explain things in a gentler way.

[1379] Step 11:

[1380] The device displays the adjusted persona information to the user.

[1381] Input: Adjusted persona information

[1382] Output: Persona information displayed to the user

[1383] Action: The device uses a display device to present the adjusted persona information to the user in an appropriate format.

[1384] In this way, the system provides information to reduce anxiety when users receive malicious replies and responds according to the user's emotional state.

[1385] 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.

[1386] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1387] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1388] [Fourth embodiment]

[1389] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1390] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1391] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1392] 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.

[1393] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1394] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1395] 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.

[1396] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1397] 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.

[1398] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1399] 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.

[1400] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1401] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1402] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the profile of the attacker and displays it to the user, thereby reducing anxiety. This system uses the user's device, server, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[1403] System Configuration

[1404] 1. User Device

[1405] It is a device that allows users to use SNS and can receive Kuso-Replies. It also has the function of sending the content of Kuso-Replies and the poster ID to the server.

[1406] 2. Server

[1407] The server uses SNS APIs to collect data on attackers' past posts based on the Kuso-Reply information received from the user's device. The collected data is then subjected to text and image analysis to extract frequently occurring words and image features. Furthermore, a machine learning model is used to predict the attacker's persona, and the results are sent to the user's device.

[1408] 3. Social Media API

[1409] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[1410] Program processing

[1411] 1. Receiving shitty replies

[1412] A user receives a shitty reply on social media and checks its content and the poster's ID.

[1413] The device sends the content of the shit reply and the poster ID to the server.

[1414] 2. Attacker Identification and Data Collection

[1415] The server receives the ID of the user who posted the kusorip and uses the SNS API to collect the attacker's past posting data. The collection range is limited to a specified period (e.g., the past year).

[1416] 3. Extracting and analyzing text data

[1417] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques.

[1418] 4. Image Data Extraction and Analysis

[1419] The server extracts image data from the collected submissions, extracts image features using image analysis technology (e.g., computer vision and deep learning), and assigns tags such as characters and symbols to the images.

[1420] 5. Data integration for persona prediction

[1421] The server combines the results of text analysis and image analysis to create a comprehensive dataset, which is then fed into a machine learning model to predict the attacker's age, gender, occupation, hobbies, and preferences.

[1422] 6. Persona generation and provision

[1423] The server generates a standard persona of an attacker based on the prediction results of the machine learning model. The generated persona information is structured in a format that is easy for users to understand.

[1424] The device provides the user with the persona information received from the server and displays information such as, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[1425] Specific examples

[1426] Let's say a user receives a malicious reply saying, "Your opinion is off the mark." The device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API and analyzes the text data and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict the persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." The prediction result is then sent to the user's device, and the attacker's standard persona information is presented to the user.

[1427] The above is a specific embodiment for carrying out the present invention, which allows the user to understand the personality of the sender of the malicious reply and reduce anxiety.

[1428] The processing flow will be explained below.

[1429] Step 1:

[1430] A user receives a malicious reply on a social networking site, which appears as a reply to a specific post.

[1431] Step 2:

[1432] The device obtains the content of the Kuso-Reply and the poster's ID, and sends this to the system's server, including the text content of the Kuso-Reply and the poster's account ID.

[1433] Step 3:

[1434] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[1435] Step 4:

[1436] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and creates a clean dataset.

[1437] Step 5:

[1438] The server uses natural language processing (NLP) technology to analyze the extracted text data. This analysis identifies frequently occurring words and key phrases. Specifically, it uses morphological analysis and TF-IDF (Term Frequency-Inverse Document Frequency) techniques.

[1439] Step 6:

[1440] The server extracts the image portion from the collected post data, which includes all image files included in each post.

[1441] Step 7:

[1442] The server uses image analysis technology (e.g., computer vision and deep learning) to analyze the features of the extracted images. Specifically, it performs object recognition and assigns tags such as characters and symbols.

[1443] Step 8:

[1444] The server combines the results of the text analysis and the image analysis, creating a comprehensive dataset.

[1445] Step 9:

[1446] The server then inputs the combined data set into a machine learning model to predict the attacker's persona, including attributes such as age, gender, occupation, and hobbies.

[1447] Step 10:

[1448] The server generates a standard persona for an attacker based on the predictions obtained from the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[1449] Step 11:

[1450] The server transmits the generated persona information to the user's terminal.

[1451] Step 12:

[1452] The device presents the received persona information to the user. This is displayed on the screen in a format that is easy for the user to confirm. For example, it might say, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[1453] By going through these steps, users can understand the profile of the attacker and reduce their anxiety.

[1454] Example 1

[1455] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1456] Users who receive malicious replies (kuso-replies) on social media often feel anxious and stressed. However, typical countermeasures include ignoring the replies or blocking the user, which does not fundamentally alleviate the anxiety. Furthermore, there are limited ways to specifically identify the attacker, leaving users plagued by fear of the so-called "invisible enemy." The present invention aims to address these issues by predicting the attacker's profile and presenting it to users, thereby providing them with specific information and reducing their anxiety.

[1457] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1458] In this invention, the server includes a device for receiving malicious replies, a device for collecting past posts by the poster of the malicious replies, means for extracting text from the collected past posts and performing preprocessing, means for analyzing frequently occurring words and key phrases from the preprocessed text, means for extracting images from the collected posts and analyzing their features, means for predicting the poster's personality based on the analysis results, and means for providing the predicted personality information to the device. This allows the user to grasp the specific personality of the attacker, making it possible to reduce anxiety and stress.

[1459] A "malicious reply" is a comment or message sent on social media with the intention of attacking or offending others.

[1460] "Devices" is a general term for devices, machines, and related systems and software used by users to access SNS.

[1461] A "poster" is a person or account that posts comments, messages, images, etc. on social media.

[1462] "Past posts" refer to comments, messages, images, and other output made on social media within a specified period.

[1463] "Text" refers to posts and comments containing written information made on social media.

[1464] "Preprocessing" is the process of removing unnecessary elements (e.g., hashtags and links) from text information and converting it into a format suitable for analysis.

[1465] A "frequent word" is a word that occurs many times within a particular text dataset.

[1466] A "key phrase" is a phrase or word that has an important meaning within text data.

[1467] "Analysis" is the process of analyzing text and image data and extracting meaningful information.

[1468] "Features" are important patterns or characteristics extracted from image data.

[1469] A "personal profile" is the result of predicting an individual's attribute information, such as age, gender, occupation, and hobbies and preferences.

[1470] "Information" refers to data and knowledge such as predictions and analysis results of a person's profile.

[1471] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the profile of the attacker and displays this information to the user, thereby reducing anxiety. This system uses the user's device, server, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the profile of the attacker.

[1472] System Configuration

[1473] 1. User Device

[1474] The user device is a device that allows users to use SNS. It has the function of receiving kusorip messages and sending their content and the poster's ID to a server. When a user receives a kusorip message on SNS, the device sends the content and the poster's ID to the server using an HTTPS request.

[1475] 2. Server

[1476] The server uses SNS APIs to collect data on the attacker's past posts based on the Kuso-Reply information received from the user's device. The collected data is subjected to text analysis and image analysis to extract frequently occurring words and image features. It also uses a machine learning model to predict the attacker's profile and sends the results to the user's device. Specifically, the server preprocesses the text data using "NLTK" and extracts important key phrases using natural language processing technology. It also performs image analysis using "OpenCV" and "TensorFlow" to extract important features.

[1477] 3. Social Media API

[1478] The SNS API is an application programming interface provided by SNS platforms, and is used by the server to collect data on past posts by attackers. Authentication is performed using the SNS API authentication token, and post data from a specified period is obtained.

[1479] Specific example of system operation

[1480] Suppose a user receives a malicious reply saying, "Your opinion is off the mark." The user's device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API. The collected text data is preprocessed using NLTK to extract frequently occurring words and key phrases. At the same time, the collected image data is analyzed using OpenCV or TensorFlow to extract features. The results are integrated, and a machine learning model (e.g., SciKit-Learn) is used to predict a profile of the attacker, such as "male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." Finally, the prediction result is sent to the user's device, and the user is presented with a standard profile of the attacker.

[1481] Prompt Sentence Examples

[1482] An example of a prompt would be:

[1483] I received a shitty reply saying, "Your opinion is off the mark." Please predict and provide a profile of the attacker.

[1484] The above-mentioned method allows users to get a clear picture of the attacker, which can reduce anxiety and stress.

[1485] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1486] Step 1: Receiving a Shit Reply

[1487] A user receives a malicious reply (kuso-repu) on social media. For example, the reply is, "Your opinion is irrelevant."

[1488] Input: Shitty reply on social media and poster ID.

[1489] The device temporarily stores the content of the KusoRip and the poster ID in its internal memory, then sends the content of the KusoRip and the poster ID to the server using an HTTPS request.

[1490] Output: The content of the shit reply sent to the server and the poster ID.

[1491] Step 2: Identifying the attacker and gathering data

[1492] The server checks the content of the kusorip received from the device and the poster ID.

[1493] Input: The content of the shitty reply sent from the device and the poster ID.

[1494] The server uses the SNS API to collect the attacker's posting data for the past year. The server authenticates using the SNS API's authentication token and sends an API request. This request includes the poster ID and the data collection period (e.g., the past year).

[1495] Output: Posted data (text and images) by attackers collected on the server over the past year.

[1496] Step 3: Extract and analyze text data

[1497] The server extracts the text from the collected post data and preprocesses the text of the post to remove hashtags and links.

[1498] Input: Attacker's past posting data (text).

[1499] The server uses NLTK to preprocess the text data, tokenizing it, removing stop words, and performing word frequency analysis, using natural language processing techniques to extract important frequently occurring words and key phrases.

[1500] Output: Preprocessed text data and analysis results (frequent words and key phrases).

[1501] Step 4: Extraction and analysis of image data

[1502] The server extracts the image portion from the collected submission data, extracts image features using computer vision and deep learning technology, and assigns tags such as characters and symbols to them.

[1503] Input: Attacker's past posting data (images).

[1504] The server uses OpenCV and TensorFlow to analyze the image and extract important features, such as identifying and labeling anime characters or specific symbols.

[1505] Output: Analyzed image data and feature information.

[1506] Step 5: Data integration for persona prediction

[1507] The server combines the results of text analysis and image analysis to create a comprehensive dataset.

[1508] Input: Text analysis results and image analysis results.

[1509] The server inputs this dataset into a machine learning model to predict the attacker's personality, such as age, gender, occupation, hobbies, etc. For example, the server inputs the dataset into a machine learning model trained using "SciKit-Learn" to obtain a predicted personality profile.

[1510] Output: Predicted attacker profile (age, gender, occupation, hobbies, etc.).

[1511] Step 6: Generate and deliver personas

[1512] The server generates a standard profile of the attacker based on the predictions of the machine learning model, and the profile information is organized in a format that is easy for users to understand.

[1513] Input: Prediction result (attacker profile).

[1514] The server generates the person profile information in JSON format and sends it to the user device using HTTPS.

[1515] The user device analyzes the persona information received from the server and displays it on the user interface. For example, it displays information such as, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[1516] Output: Attacker profile information displayed in the user interface.

[1517] In this way, this system can reduce anxiety and stress by presenting a specific profile of the attacker in response to a malicious reply received by the user.

[1518] (Application example 1)

[1519] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1520] In online social networking services (SNS), users often receive malicious replies (kuso-replies) from other users. These replies can cause psychological stress and anxiety to the recipient. Therefore, there is a need for a method to reduce users' anxiety and provide a safe SNS environment by revealing the causes of these replies and the identity of the people behind them.

[1521] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1522] In this invention, the server includes a means for extracting frequently occurring words from past posts, a means for analyzing the extracted frequently occurring words, and a means for extracting images from past posts. This makes it possible to analyze the poster's personality and predict persona information. Furthermore, the prediction includes a means for using a generative AI model and prompt sentences, making it possible to improve the accuracy of the analysis results. This makes it possible to clarify the identity of the person behind malicious replies to users, allowing users to use SNS without feeling anxious.

[1523] A "terminal" is a device through which a user receives malicious replies and transmits the information to a server.

[1524] The "server" is a computer system that receives information about the poster of a malicious reply sent from a user's terminal, collects and analyzes the poster's past posting data, and provides persona information.

[1525] "Method for extracting frequently occurring words" is a method for selecting frequently used words from posted data.

[1526] "Means for analyzing frequently occurring words" is a method for analyzing extracted words and finding their meanings and relationships.

[1527] "Means for extracting images" refers to a method for selecting image media from posted data.

[1528] The "means for analyzing images" is a method for analyzing the extracted images and recognizing their contents and features.

[1529] A "means for predicting persona" is a method for inferring characteristics such as the poster's age, gender, hobbies and preferences based on collected and analyzed data.

[1530] "Using a generative AI model" means using a model generated using artificial intelligence technology.

[1531] "Prompt methods" are methods that use sentences to provide a generative AI model with a specific question and context to get an answer.

[1532] "Information Provision Interface" means the application programming interface (API) that the server uses to collect data from the social networking platform.

[1533] The system based on this invention reduces anxiety when a user receives a malicious reply (kuso-reply) on social media by predicting the identity of the attacker and providing this information to the user. The system uses the user's device, server, and social media API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[1534] System Configuration

[1535] 1. User Device

[1536] It is a device that allows users to use social networking services, and when it receives a shitty reply, it sends the content and the poster's ID to the server.

[1537] 2. Server

[1538] The server uses SNS APIs to collect data on the attacker's past posts based on the Kuso-Reply information received from the user's device. The collected data is then subjected to text and image analysis. Natural language processing (NLP) technology is used for text analysis, and computer vision and deep learning technology are used for image analysis. Furthermore, a generative AI model is used to predict the attacker's persona. A prompt sentence is used to make highly accurate predictions.

[1539] 3. Social Media API

[1540] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[1541] Program processing example

[1542] Receiving and sending shit replies

[1543] The user receives a kusorip message on a social networking site and enters the message's content and the poster's ID into their device, which then sends this information to the server.

[1544] Attacker Identification and Data Collection

[1545] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's posting data for the past year.

[1546] Text data extraction and analysis

[1547] The server extracts text from the collected post data and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques, such as NLTK and spaCy libraries.

[1548] Image data extraction and analysis

[1549] The server extracts image data from the collected submissions and uses computer vision technology to extract image features, using deep learning technologies such as TensorFlow and PyTorch.

[1550] Persona Prediction

[1551] The server integrates the results of text analysis and image analysis and predicts the attacker's persona based on a generative AI model. The prompt sentence is used as input to the model, and persona information is predicted with high accuracy.

[1552] Providing persona information

[1553] The predicted persona information is sent to the user's device and displayed to the user, allowing the user to understand the profile of the attacker and reduce anxiety.

[1554] Specific examples

[1555] When a user receives a crappy reply saying, "Your opinion is completely meaningless," the device sends the content of the crappy reply and the poster's ID to the server. The server collects the poster's posting data from the past year via SNS APIs and analyzes the text data and image data. For example, if "games" and "anime" are detected as frequently occurring words and image analysis reveals that specific character images appear frequently, the generative AI model is used to predict a persona such as "male in his 20s, likes anime and games, unemployed or working part-time." The prediction results are then sent to the user's device and displayed to the user.

[1556] Example prompt sentence:

[1557] Please predict the attacker's persona based on the following text analysis and image analysis results. Text analysis results: ['game', 'anime'], Image analysis results: ['anime character']

[1558] The above is a specific embodiment for carrying out the invention.

[1559] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1560] Step 1:

[1561] The user receives a kusorip message and enters its contents and the poster's ID into the device.

[1562] When a user receives a malicious reply (kuso-reply) on social media, they enter the content and the sender's ID into a dedicated application. This entered data (the content of the kuso-reply and the poster's ID) is saved on the device.

[1563] Step 2:

[1564] The device sends the content of the Kuso-Reply and the poster ID to the server.

[1565] The device collects the text of the entered kusorip and the poster ID and sends it to the server. This is done using a method such as an HTTP POST request. The input is the kusorip and the poster ID, and the output is a status message confirming that this information has been sent to the server.

[1566] Step 3:

[1567] The server uses the SNS API to collect the attacker's past posting data.

[1568] The server uses the received poster ID to collect the attacker's past posting data via the SNS platform's API. The collected data includes text posts, images, videos, etc., covering the past year. The input is the poster ID, and the output is the collected posting data.

[1569] Step 4:

[1570] Extract the text data collected by the server

[1571] The server extracts text parts from the collected submission data using regular expressions and NLP preprocessing techniques. The input is the submission data, and the output is the extracted text data.

[1572] Step 5:

[1573] The server analyzes the text data using NLP techniques.

[1574] The server preprocesses the extracted text data to identify frequently occurring words and key phrases. The libraries used are NLTK and spaCy. The input is text data, and the output is frequently occurring words and key phrases. Specific operations include tokenization and stop word removal.

[1575] Step 6:

[1576] Extracting image data collected by the server

[1577] The server extracts images from the collected submission data by filtering the data type. The input is the submission data, and the output is the extracted image data.

[1578] Step 7:

[1579] The server analyzes the image data using image analysis technology.

[1580] The server analyzes the extracted image data using computer vision techniques, such as deep learning frameworks like TensorFlow and PyTorch. The input is image data, and the output is image features and labels. Specific operations include object detection and image classification.

[1581] Step 8:

[1582] The server integrates the results of text analysis and image analysis and predicts the persona using a generative AI model.

[1583] The server uses a dataset that integrates the results of text analysis and image analysis, and inputs a prompt into the generative AI model to predict the attacker's persona. The input is the integrated dataset and prompt, and the output is the persona prediction result. Example prompt: "Please predict the attacker's persona based on the following text analysis and image analysis results. Text analysis result: ['Game', 'Anime'], Image analysis result: ['Anime character']"

[1584] Step 9:

[1585] The server sends the predicted persona information to the user device.

[1586] The server sends the predicted persona information to the user's device. This is done via an HTTP POST request, etc. The input is the persona prediction result, and the output is the confirmation status of the transmission to the user's device.

[1587] Step 10:

[1588] The user device displays the persona information to the user.

[1589] The user device displays the received persona information on the user screen. This allows the user to understand the profile of the attacker and reduce anxiety. The input is persona information, and the output is the displayed persona information. Specifically, it operates by displaying text and images on the interface.

[1590] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1591] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the attacker's personality and further recognizes the user's emotions and adjusts the presentation method to reduce anxiety. This system uses the user's device, server, emotion engine, and social networking site API to collect past posting data, analyze frequently used words and images, and predict the attacker's persona.

[1592] System Configuration

[1593] 1. User Device

[1594] It is a device that allows users to use SNS and can receive Kuso-Reply messages. It also has the function of sending the content of the Kuso-Reply message and the poster's ID to the server. It also has the function of providing the user's voice and facial expressions to the emotion engine.

[1595] 2. Server

[1596] The server uses SNS APIs to collect attacker's past posting data based on the information on kusoripulp received from the user's device and the user's emotional data. The collected data is subjected to text and image analysis to extract frequently occurring words and image features. Furthermore, a machine learning model is used to predict the attacker's persona and the results are sent to the user's device.

[1597] 3. Emotion Engine

[1598] The emotion engine collects and analyzes the user's voice and facial expression data to recognize the user's emotional state. It also predicts emotions based on the user's past posts and behavioral history, and adjusts the way persona information is presented.

[1599] 4. Social Media API

[1600] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[1601] Program processing

[1602] 1. Receiving shitty replies

[1603] A user receives a malicious reply on a social networking site and checks its content and the poster's ID.

[1604] The device sends the content of the shit reply and the poster ID to the server.

[1605] 2. Attacker Identification and Data Collection

[1606] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[1607] 3. Extracting and analyzing text data

[1608] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and analyzes frequently occurring words and key phrases using natural language processing (NLP) techniques.

[1609] 4. Image Data Extraction and Analysis

[1610] The server extracts image data from the collected submissions, extracts image features using image analysis technology (e.g., computer vision and deep learning), and assigns tags such as characters and symbols to the images.

[1611] 5. Emotion recognition

[1612] The device continuously collects the user's voice and facial expression data and sends it to the emotion engine.

[1613] The emotion engine analyzes this data to recognize the user's emotional state and predicts their emotions based on their past posts and behavioral history.

[1614] 6. Data integration for persona prediction

[1615] The server combines the results of text analysis and image analysis to create a comprehensive dataset, which is then fed into a machine learning model to predict the attacker's age, gender, occupation, hobbies, and preferences.

[1616] 7. Persona generation and provision

[1617] The server generates a standard persona for an attacker based on the predictions of the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[1618] The emotion engine adjusts the presentation of persona information based on the user's perceived emotions. For example, if the user is under a lot of stress, it will display softer expressions and encouraging words.

[1619] 8. Displaying the results

[1620] The server transmits the generated persona information to the user's terminal.

[1621] The device then presents the received persona information to the user. For example, information such as "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time," is displayed in a format that is adjusted to reflect the user's emotional state.

[1622] Specific examples

[1623] Suppose a user receives a malicious reply saying, "You're such an idiot." The device sends the content of this shitty reply and the poster's ID to the server. The server collects the poster's posting data from the past year via a social media API and analyzes the text and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict a persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." During this time, the device sends the user's voice and facial expression data to an emotion engine, which then recognizes the user's stress level. The emotion engine then adjusts the persona information and sends it from the server to the device. Finally, the persona information is presented in a way that takes into account the user's psychological state.

[1624] The above is a specific example of how to implement the present invention, which allows users to understand the personality of the sender of a malicious reply and reduce anxiety. By incorporating an emotion engine, information is provided that takes into consideration the user's psychological state.

[1625] The processing flow will be explained below.

[1626] Step 1:

[1627] A user receives a malicious reply (kuso-reply) on a social networking site, which appears as a comment on a specific post.

[1628] Step 2:

[1629] The device obtains the content of the Kuso-Reply and the poster's ID, and sends this to the system's server, including the text content of the Kuso-Reply and the poster's account ID.

[1630] Step 3:

[1631] The server receives the poster ID of the Kuso-Reply and uses the SNS API to collect the poster's past posting data. The collection range is limited to a specified period (e.g., the past year).

[1632] Step 4:

[1633] The server extracts text from the collected posts, preprocesses the text (e.g., removes hashtags and links), and creates a clean dataset.

[1634] Step 5:

[1635] The server uses natural language processing (NLP) technology to analyze the extracted text data. This analysis identifies frequently occurring words and key phrases. Specifically, it uses morphological analysis and TF-IDF (Term Frequency-Inverse Document Frequency) techniques.

[1636] Step 6:

[1637] The server extracts the image portion from the collected post data, which includes all image files included in each post.

[1638] Step 7:

[1639] The server uses image analysis technology (e.g., computer vision and deep learning) to analyze the features of the extracted images. Specifically, it performs object recognition and assigns tags such as characters and symbols.

[1640] Step 8:

[1641] The server combines the results of the text analysis and the image analysis, creating a comprehensive dataset.

[1642] Step 9:

[1643] The server then inputs the combined data set into a machine learning model to predict the attacker's persona, including attributes such as age, gender, occupation, and hobbies.

[1644] Step 10:

[1645] The server generates a standard persona for an attacker based on the predictions obtained from the machine learning model, such as a "male in his 20s, anime fan, unemployed or working part-time, with a strong sense of social isolation."

[1646] Step 11:

[1647] The device continuously collects the user's voice and facial expression data and sends it to the emotion engine.

[1648] Step 12:

[1649] The emotion engine analyzes this data to recognize the user's emotional state and predicts their emotions based on their past posts and behavioral history.

[1650] Step 13:

[1651] The emotion engine adjusts the presentation of persona information based on the user's emotional state. For example, if a user is under a lot of stress, it will display softer language and encouraging words.

[1652] Step 14:

[1653] The server transmits the generated persona information to the user's terminal.

[1654] Step 15:

[1655] The device presents the received persona information to the user. This is displayed on the screen in a format that is easy for the user to confirm. For example, it might say, "According to an analysis of this user, he is a man in his 20s who likes anime and is likely currently unemployed or working part-time."

[1656] Through these steps, users can understand the profile of the attacker and reduce their anxiety. By incorporating an emotion engine, information is provided that takes into account the user's psychological state.

[1657] Example 2

[1658] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1659] On modern social media platforms, malicious replies (kusoripu) can cause psychological stress to users. However, there is no way for users who receive such replies to understand who is attacking them, and there is no way to provide appropriate information depending on their psychological state, making it difficult to alleviate their anxiety and stress.

[1660] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for collecting past posts by posters of malicious replies, means for extracting frequently occurring words from past posts, means for analyzing the extracted images, means for predicting the poster's persona based on the analysis results, an emotion engine for collecting and analyzing user emotion data, means for adjusting the predicted persona information based on the user's emotional state, and means for providing the adjusted persona information to the terminal. This makes it easier for the user to grasp the profile of the attacker and further enables the provision of appropriate information based on the user's psychological state.

[1661] A "malicious reply" is a comment or message on a social networking site that contains offensive, derogatory, or unpleasant content directed at a user.

[1662] A "terminal" is a device that allows a user to use a social networking service and has the function of receiving malicious replies and sending their contents and the poster's ID to the server.

[1663] A "server" is a computer system that receives data sent from a user's device, collects the necessary data through the SNS API, and performs analytical processing.

[1664] "Past posts" refers to all content posted by an attacker on social media within a specific period of time.

[1665] "Frequent words" are keywords and phrases that appear frequently and are extracted from past posts.

[1666] An "extraction method" is a set of algorithms or techniques used to collect and analyze specific data (e.g., text or images).

[1667] "Means for analysis" refers to methods and techniques for analyzing the extracted data and extracting useful information and features.

[1668] A "persona" is a predicted profile of an attacker based on the analysis results, including their age, gender, occupation, hobbies, and preferences.

[1669] An "emotion engine" is a system or technology that collects and analyzes a user's emotional data to recognize the user's psychological state.

[1670] "SNS API" is an application programming interface provided by an SNS platform and is an interface used to collect past posting data.

[1671] A "machine learning model" is an algorithm that learns using large amounts of data and uses the learning results to make predictions and analyze new data.

[1672] "Means for adjusting based on the user's emotional state" refers to techniques or methods for changing the method or content of information presentation depending on the user's psychological state.

[1673] "Means for providing" refers to communication and interface technologies for displaying analysis results and predicted information on a user terminal.

[1674] When a user receives a malicious reply (kuso-reply) on a social networking site, the system based on this invention predicts the identity of the attacker and further recognizes the user's emotions and adjusts the presentation method to reduce anxiety. This system utilizes the user's device, server, emotion engine, and social networking site API, and is implemented in the following specific steps.

[1675] First, when a user receives a malicious reply on a social networking site, the device sends the content of the kusoriply and the poster's ID to a server. The device can be a smartphone or a PC, and the social networking application runs as the communication software.

[1676] The server receives the Kuso-Reply information sent from the device and uses the SNS API to collect the poster's past posting data. The collected data is classified into text data and image data, and each type of data is analyzed. For example, the text data is analyzed using natural language processing (NLP) technology to extract frequently occurring words and key phrases, and the image data is analyzed using computer vision and deep learning technology to analyze the image features.

[1677] Next, the emotion engine collects the user's voice and facial expression data in real time to recognize the user's emotional state. This data includes the user's facial expressions and tone of voice collected using a camera and microphone. The emotion engine also takes into account the user's past posts and behavioral history to predict the user's emotional state.

[1678] The server combines the results of text analysis and image analysis to generate an integrated dataset. This dataset is then input into a machine learning model to predict the attacker's persona, including their age, gender, occupation, and hobbies and interests. Machine learning models such as multilayer perceptrons and convolutional neural networks (CNNs) are used.

[1679] After obtaining the persona prediction results, the emotion engine adjusts the way the persona information is presented based on the user's emotional state. For example, if the user is under a lot of stress, the explanation will be changed to a more gentle expression or words of encouragement will be added.

[1680] Finally, the server sends the adjusted persona information to the user's device, which receives it and presents it to the user. For example, information such as "According to an analysis of this user, it appears that he is a man in his 20s who likes anime and is currently unemployed or working part-time" is displayed in a format that reflects the user's emotional state.

[1681] Specific examples

[1682] If a user receives a malicious reply such as "You're such an idiot," the device sends the content of the kusoriply and the poster's ID to the server. The server uses a social media API to collect the poster's posting data from the past year and analyzes the text and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a machine learning model to predict a persona of "male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation." During this time, the device sends the user's voice and facial expression data to an emotion engine, which then recognizes the user's stress level. The persona information is then adjusted by the emotion engine and sent from the server to the device. Finally, the persona information is presented in a form that takes into account the user's psychological state.

[1683] Prompt Sentence Examples

[1684] Below are some example prompts to input to a generative AI model:

[1685] "We want to predict the personality of the person who posts malicious replies (kuso-replies) on social media that cause anxiety in users. At the same time, we want to design a method of providing information that recognizes the user's emotional state and can alleviate that anxiety. An example of a kuso-reply is: 'You're such an idiot.'"

[1686] Based on these prompts, the generative AI model will generate an appropriate persona and process emotions.

[1687] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1688] Step 1: Receiving a Shit Reply

[1689] The user receives malicious replies on social media such as "You're such an idiot."

[1690] Input: Malicious reply on a social media app

[1691] Output: Malicious reply content and poster ID

[1692] Specific behavior: When a user opens a social media application on their smartphone or computer, they will receive notifications and direct messages containing malicious replies.

[1693] Step 2: Send the content of the kusorip and the poster ID

[1694] The device automatically sends the content of the malicious reply received and the poster ID to the server.

[1695] Input: Malicious reply content and poster ID

[1696] Output: The malicious reply sent to the server and its poster ID.

[1697] Specific operation: The content of the shitty reply, "You are truly an idiot," and the poster's ID are sent to the server in the device's background process.

[1698] Step 3: Identifying the attacker and gathering data

[1699] The server collects posting data for the past year from the Shit Reply poster ID received using the SNS API.

[1700] Input: Shit reply poster ID

[1701] Output: Post data from the past year

[1702] What happens: The server sends a request to the SNS API to retrieve post data associated with the poster ID, e.g., tweets, posts, and comments from the past year.

[1703] Step 4: Extract and analyze text data

[1704] The server extracts the text portion from the collected posting data, performs preprocessing, and then analyzes it using natural language processing (NLP) technology.

[1705] Input: Post data from the past year

[1706] Output: Frequent words and key phrases

[1707] What it does: It removes hashtags and links, then performs morphological analysis on the text to extract frequently occurring words and key phrases. For example, it identifies keywords like "anime," "games," and "stress."

[1708] Step 5: Extract and analyze image data

[1709] The server extracts image portions from the collected posting data and extracts features using image analysis technology.

[1710] Input: Post data from the past year

[1711] Output: Image feature information

[1712] What it does: Uses computer vision technology to identify and tag characters and symbols from images. Example: Detecting the anime character "Naruto."

[1713] Step 6: Emotion Recognition

[1714] The device continuously transmits the user's voice and facial expression data to the emotion engine.

[1715] Input: User's voice data and facial expression data

[1716] Output: Voice and facial expression data sent to the emotion engine

[1717] Specific operation: When a user uses a social networking app, the camera and microphone are used to detect facial expressions and tone of voice. This data is sent to the emotion engine in real time.

[1718] The emotion engine analyzes the received data and recognizes the user's emotional state.

[1719] Input: Voice data and facial expression data

[1720] Output: User's emotional state

[1721] Specific behavior: If the user exhibits a sad facial expression or an irritated tone of voice, the emotion engine detects a stressed state.

[1722] Step 7: Data integration for persona prediction

[1723] The server combines the results of text analysis and image analysis to create a comprehensive dataset.

[1724] Input: Text analysis results and image analysis results

[1725] Output: Unified dataset

[1726] Specific operation: Combine text and image features into a single dataset and provide it to a machine learning model. Example: Combine text keywords and image tag information.

[1727] Step 8: Generate and deliver personas

[1728] The server generates a standard persona of an attacker from the prediction results of the machine learning model.

[1729] Input: Integrated dataset

[1730] Output: Persona information

[1731] Specific operation: The machine learning model outputs the following persona information: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation."

[1732] The emotional engine adjusts how persona information is presented based on the user's emotional state.

[1733] Input: User's emotional state and persona information

[1734] Output: Adjusted persona information

[1735] Specific actions: For users who are under a lot of stress, add soft words and words of encouragement. For example, "This user is a man in his 20s who loves anime. He may be unemployed or working part-time, but you'll be fine. Don't let his opinions affect you too much."

[1736] Step 9: View the results

[1737] The server transmits the adjusted persona information to the user's terminal.

[1738] Input: Adjusted persona information

[1739] Output: Data sent to the terminal

[1740] Specific operation: The server sends persona information to the user's device and issues a display instruction.

[1741] The terminal displays the received persona information to the user.

[1742] Input: Persona information sent from the server

[1743] Output: Data displayed to the user

[1744] Specific operation: The device displays a message on the screen stating, "This user is a male in his 20s who likes anime and is currently unemployed or possibly working part-time." It also displays words of encouragement based on the user's stress level.

[1745] The above is the specific processing flow of this system's program. By including the specific operations at each step, it is clear how the data is processed when a user receives a 'kusorip' and what information is presented as the final result. In addition, by incorporating an emotion engine, it is possible to provide information that takes into account the user's psychological state.

[1746] (Application example 2)

[1747] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1748] In recent years, the spread of social networking sites has increased the risk of receiving malicious replies (kuso-replies). In such situations, many users experience psychological stress and anxiety. Furthermore, because they do not know the sender of the malicious replies, users are likely to be placed in unpleasant situations. Current social networking sites lack the ability to recognize users' emotions and respond appropriately. For this reason, a system that can reduce users' psychological burden is needed.

[1749] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1750] In this invention, the server includes a means for extracting frequently occurring words from past posts, a means for analyzing the extracted frequently occurring words, and a means for extracting images from past posts. This makes it possible to reduce the psychological burden on users who receive malicious replies. It also includes a means for predicting personas using a generative AI model and a means for recognizing the user's emotional state and adjusting the presentation method of persona information, making it possible to provide attacker information in a format tailored to the user. Furthermore, by including a means for transmitting the content of the kusorip and the poster ID to the server and a means for transmitting the user's voice and facial expression data to the emotion engine, it is possible to more accurately grasp the user's emotions and take appropriate action.

[1751] "Hateful replies" refer to comments or messages that are offensive, insulting, or inappropriate towards others on social media or other platforms.

[1752] "Terminal" refers to a device used by a user to use an SNS, including a smartphone, tablet, or PC.

[1753] A "server" is a device that stores, processes, and transmits data over a computer network.

[1754] "Past posts" refers to all messages, comments, images, videos, and other posts made by a particular user in the past on a social media platform.

[1755] A "frequent word" refers to a word that appears more frequently than other words in the text data to be analyzed.

[1756] "Extraction means" refers to a process or device that extracts specific data (text, images, audio, etc.) from other data.

[1757] "Analysis means" refers to the process or device used to analyze the extracted data in detail and understand its meaning and characteristics.

[1758] A "persona" is a virtual image of a person created based on the poster's characteristics, including attributes such as age, gender, occupation, and hobbies.

[1759] An "emotion engine" refers to software or hardware that recognizes and analyzes a user's emotional state from voice, facial expression, and other data.

[1760] "Generative AI model" refers to an artificial intelligence model used to generate specific patterns or information from given data.

[1761] A "prompt" is an instruction given to a generative AI model to perform a specific task.

[1762] The system based on the present invention, "KusoReplyGuard," predicts the identity of the attacker when a malicious reply is received on a social networking site, and provides information according to the user's emotional state, thereby reducing anxiety. A specific embodiment of this system will be described below.

[1763] System Configuration

[1764] The system consists of four main components:

[1765] 1. User Device

[1766] This is a device that users use to access social media. It can be a smartphone, tablet, or PC. The device receives the Kuso-Reply and sends the content and poster ID to a server. In addition, the device has the ability to collect the user's voice and facial expression data and send it to the emotion engine.

[1767] 2. Server

[1768] The server uses SNS APIs to collect the attacker's past posting data based on the Kuso-Reply information and poster ID received from the user's device. The collected data is subjected to text analysis and image analysis to extract frequently occurring words and image features. The analysis results are used to predict the attacker's persona using a generative AI model.

[1769] 3. Emotion Engine

[1770] The emotion engine analyzes the user's voice and facial expression data to recognize the user's emotional state. It also predicts the user's emotions based on the user's past posts and behavioral history. This information is used to adjust the way the persona information is presented.

[1771] 4. Social Media API

[1772] It is an application programming interface provided by social networking platforms that servers use to collect attacker's past posting data.

[1773] Program processing explanation

[1774] The server first collects the attacker's past posting data using the social networking service API, then performs text and image analysis to extract frequently used words and image features.

[1775] The server then combines this information and inputs it into a generative AI model, which predicts a persona for the attacker, including their age, gender, occupation, hobbies, and preferences.

[1776] The server then transmits this persona information to the user's device. At the same time, the device transmits the user's emotional data to the emotion engine. The emotion engine analyzes the user's emotional state and adjusts the way the persona information is presented. For example, if the user is feeling stressed, the persona information is presented in a softer manner.

[1777] Specific examples

[1778] Consider the case where User A receives a malicious reply on social media saying, "You're an idiot." User A's device sends the content of this shitty reply and the poster's ID to a server. The server uses the social media API to collect the poster's past posting data and analyzes the text data and image data. From the analysis results, it detects that the word "anime" and images of specific characters appear frequently, and uses a generative AI model to predict the persona: "Male in his 20s, anime fan, unemployed or part-time worker, with a strong sense of social isolation."

[1779] During this time, the device sends user A's voice and facial expression data to the emotion engine, which then recognizes user A's stress level. The persona information is then adjusted by the emotion engine and sent from the server to the device. Finally, persona information is provided in a form that takes user A's psychological state into consideration.

[1780] Prompt Sentence Examples

[1781] Examples of prompts that can be input to a generative AI model include:

[1782] "To reduce the anxiety of users who receive malicious replies, please generate a persona for the attacker. The data to be analyzed is the text data 'I love anime, I bought a new gachapon' and the image data 'animes_yokai.jpg'."

[1783] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1784] Step 1:

[1785] A user receives a malicious reply (kusoripu) on social media.

[1786] Input: Malicious replies on social media and their poster IDs

[1787] Output: Kuso-reply content and poster ID

[1788] How it works: The user checks the message on the social networking site, and the device obtains the content of the shitty reply and the poster's ID.

[1789] Step 2:

[1790] The device sends the content of the shit reply and the poster ID to the server.

[1791] Input: Kuso reply content and poster ID

[1792] Output: The content of the kusorip sent to the server and the poster ID

[1793] How it works: The communication module in the device sends the content of the shitty reply and the poster's ID to the server.

[1794] Step 3:

[1795] The server uses the SNS API to collect the poster's past posting data.

[1796] Input: Poster ID

[1797] Output: Collected past post data (text, images)

[1798] How it works: The server uses the SNS API to identify past posts associated with the poster ID and collects post data for a certain period of time.

[1799] Step 4:

[1800] The text portion is extracted from the posted data collected by the server and text analysis is performed.

[1801] Input: Text portion of past post data

[1802] Output: Extracted frequent words and key phrases

[1803] How it works: A natural language processing (NLP) engine on the server analyzes text data and extracts frequently occurring words and key phrases.

[1804] Step 5:

[1805] The image portion is extracted from the posted data collected by the server and image analysis is performed.

[1806] Input: Image portion of past post data

[1807] Output: Extracted image features (character, symbol tags)

[1808] How it works: The image analysis engine on the server analyzes the image data, extracts features, and tags them.

[1809] Step 6:

[1810] The server combines the results of text analysis and image analysis and inputs them into a generative AI model.

[1811] Input: Text analysis results (frequent words, key phrases), image analysis results (features, tags)

[1812] Output: Predicted persona information

[1813] How it works: The server integrates the analysis results and inputs them as prompts into the generative AI model, which then predicts the persona.

[1814] Step 7:

[1815] The server transmits the generated persona information to the user's terminal.

[1816] Input: Generated persona information

[1817] Output: Persona information sent to the device

[1818] Operation: The server sends persona information to the device and prepares it for display.

[1819] Step 8:

[1820] The device sends the user's voice and facial expression data to the emotion engine.

[1821] Input: User's voice data, facial expression data

[1822] Output: User emotion data sent to the emotion engine

[1823] Operation: The device's sensors capture the user's voice and facial expressions and send them to the emotion engine.

[1824] Step 9:

[1825] The emotion engine analyzes the user's emotional state and generates emotion data.

[1826] Input: Voice data, facial expression data

[1827] Output: User's emotional state (stress level, type of emotion)

[1828] How it works: The emotion engine analyzes voice and facial expression data in real time to identify the user's emotional state.

[1829] Step 10:

[1830] An emotional engine adjusts the presentation of persona information based on the user's emotional state.

[1831] Input: User's emotional state, predicted persona information

[1832] Output: Adjusted persona information

[1833] How it works: The emotion engine takes into account the user's emotional state and adjusts the way persona information is presented appropriately. For example, if a user is stressed, it will explain things in a gentler way.

[1834] Step 11:

[1835] The device displays the adjusted persona information to the user.

[1836] Input: Adjusted persona information

[1837] Output: Persona information displayed to the user

[1838] Action: The device uses a display device to present the adjusted persona information to the user in an appropriate format.

[1839] In this way, the system provides information to reduce anxiety when users receive malicious replies and responds according to the user's emotional state.

[1840] 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.

[1841] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1842] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1843] 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.

[1844] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1845] 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.

[1846] 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).

[1847] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, motorcycles, and other devices, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1848] 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."

[1849] 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.

[1850] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1851] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1852] 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.

[1853] 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.

[1854] 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.

[1855] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1856] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1857] 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.

[1858] 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.

[1859] 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.

[1860] 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.

[1861] The following is further disclosed regarding the above embodiment.

[1862] (Claim 1)

[1863] a terminal that receives the malicious reply;

[1864] A server that collects past posts from the poster of the malicious reply;

[1865] A way to extract frequently used words from past posts,

[1866] A means for analyzing the extracted frequent words;

[1867] A way to extract images from past posts,

[1868] means for analyzing the extracted image;

[1869] A means of predicting the poster's persona based on the analysis results;

[1870] means for providing predicted persona information to a terminal;

[1871] A system including:

[1872] (Claim 2)

[1873] The system according to claim 1, wherein the collection of past posts is performed using a social networking service (SNS) API.

[1874] (Claim 3)

[1875] 10. The system of claim 1, wherein the per...

Claims

1. a terminal that receives the malicious reply; A server that collects past posts from the poster of the malicious reply; A way to extract frequently used words from past posts, A means for analyzing the extracted frequent words; A way to extract images from past posts, means for analyzing the extracted image; A means of predicting the poster's persona based on the analysis results; means for providing predicted persona information to a terminal; A system including:

2. The system according to claim 1, wherein past posts are collected using an SNS API.

3. The system of claim 1 , wherein the persona prediction uses a machine learning model.

Images