system

Abstract

A system is provided.SOLUTION: Means for periodically updating a database of trusted information sources, means for parsing and applying classification tags to the collected posts using a natural language processing engine, means for collecting posts in real-time through an API of an SNS platform, means for analyzing the pre-processed text and identifying high-risk information, means for matching the identified high-risk information against the database of trusted information sources, means for generating a warning message if a contradiction is found, means for displaying the warning message on a user terminal, and a user evaluating the validity of the warning message; A system comprising: means for providing feedback; means for analyzing collected feedback and utilizing the feedback to improve a model of an artificial intelligence engine; and means for giving a badge to a user having a high degree of contribution.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] Please write according to the format.

[0005] False information spread on social media has become a serious social problem. This false information can sometimes cause serious misunderstandings and confusion, and in some cases can have a negative impact on public safety and health. In particular, as generative AI technology evolves, the content of false information is becoming increasingly sophisticated, making it more difficult to detect. For this reason, there is a need for a highly accurate system that can efficiently compare false information with reliable information and warn users about false information. [Means for solving the problem]

[0006] In this invention, the problem of false information being spread on social media is solved by introducing the following measures.

[0007] First, we provide a means to regularly update our database of reliable information sources, thereby maintaining the most up-to-date and accurate information. Second, we use a natural language processing engine to analyze collected posts and assign classification tags. This analysis clarifies which category the content of a post belongs to.

[0008] Furthermore, we collect posts in real time through the APIs of social media platforms, and preprocess these posts to normalize and tokenize the text. Based on the analyzed text, we identify high-risk information and prioritize verification of information that requires particular attention.

[0009] The system then uses a means to check the identified high-risk information against a database of trusted information sources, and generates a warning message if a discrepancy is found, and includes a means to display the warning message on the user's terminal so that the user can confirm its contents.

[0010] The system also provides a means for users to evaluate the effectiveness of warning messages and provide feedback, allowing users to determine whether the information is accurate. This feedback is collected and analyzed to help improve the AI ​​engine's models. Furthermore, the system rewards highly contributors with badges to recognize trustworthy users and increase engagement.

[0011] By combining the above measures, the present invention effectively prevents the spread of false information on social media and promotes the dissemination and sharing of reliable information.

[0012] "Reliable sources" refers to sources that are recognized for their accuracy, such as official announcements, authoritative news sites, and academic papers.

[0013] A "database" is a collection of information organized for a specific purpose, and in the present invention refers to a collection of reliable information sources.

[0014] A "natural language processing engine" refers to software and algorithms that enable computers to understand and analyze natural human language.

[0015] "API" stands for "Application Programming Interface" and refers to an interface for sharing functions between different software.

[0016] "Preprocessing" refers to the general process of preparing text data for analysis, including text normalization and tokenization.

[0017] "High risk information" refers to information that is likely to have an adverse effect on public safety or health, or that may cause significant misunderstanding or confusion.

[0018] "Warning message" refers to a text message presented to a user to alert them to certain information.

[0019] "User terminal" refers to the device (smartphone, computer, tablet, etc.) that a user operates when using SNS.

[0020] "Feedback" refers to the evaluations and opinions that users provide to the system, and is information that is used to improve the system.

[0021] An "artificial intelligence engine" refers to software and algorithms that use learning algorithms to analyze data and automatically make inferences and predictions.

[0022] "Badge" refers to a digital certificate or mark of authentication awarded to a user for a particular action or achievement.

[0023] As mentioned above, important terms related to the present invention have the above definitions. [Brief explanation of the drawings]

[0024] [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 illustrating 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

[0025] 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.

[0026] First, the terms used in the following description will be explained.

[0027] 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).

[0028] 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.

[0029] 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.

[0030] 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.

[0031] 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."

[0032] [First embodiment]

[0033] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0034] 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.

[0035] 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).

[0036] 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.

[0037] 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.

[0038] 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.

[0039] 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.

[0040] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0041] 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.

[0042] 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.

[0043] 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.

[0044] 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."

[0045] This invention is a system for preventing the spread of false information on social media and ensuring the reliability of information. This system is realized by performing highly accurate fact-checking while referring to reliable sources of information through cooperation between servers, terminals, and users.

[0046] System Overview

[0047] This system is configured as follows:

[0048] 1. Server: Maintains a database of trusted sources, analyzes posts using an NLP engine, assesses the risk of information, and generates warning messages.

[0049] 2. Terminal: A device that allows users to use SNS and displays warning messages and sends feedback.

[0050] 3. Users: contribute to the credibility of information through social media posts and feedback.

[0051] Overview of program processing

[0052] 1. Information gathering and preparation

[0053] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database, then uses an NLP engine to analyze the new information and assign it classification tags.

[0054] 2. Collect social media posts

[0055] The server collects posts in real time through the API of the social media platform. The collected post data includes text, poster information, timestamp, etc.

[0056] 3. Analysis and evaluation of information

[0057] The server preprocesses the collected text of posts and analyzes it with an NLP engine. This analysis extracts key keywords and phrases from the posts and evaluates their risk level based on them.

[0058] 4. Fact Check

[0059] The server checks high-risk posts against a database of trusted sources, using keyword matching and similarity calculations.

[0060] 5. Generating and Displaying Warning Messages

[0061] The server generates a warning message for posts that contradict reliable information. This warning message is displayed on the device, for example, "This information contradicts official statements. Read more here: [link]."

[0062] 6. Feedback Collection and Analysis

[0063] The user can rate the effectiveness of the warning message and provide feedback, which the terminal sends to the server.

[0064] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model.

[0065] 7. Badge Awarding

[0066] The server recognizes users' contributions of expertise and awards badges for useful feedback and fact-checking activities, which are reflected in users' profiles.

[0067] Specific examples

[0068] Example 1: Medical information verification

[0069] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[0070] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[0071] 3. The server checks the post against a reliable medical information database.

[0072] 4. If the server finds an official announcement that the drug is ineffective, a warning message will be displayed on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[0073] Example 2: Verifying political information

[0074] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[0075] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[0076] 3. Users rate the effectiveness of the warning message and provide feedback.

[0077] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model.

[0078] This allows the system to effectively prevent false information from spreading on social media and promote the dissemination and sharing of reliable information.

[0079] The processing flow will be explained below.

[0080] Step 1:

[0081] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[0082] Step 2:

[0083] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[0084] Step 3:

[0085] The server preprocesses the text of the posts collected by normalizing the text data (removing special characters and extra whitespace) and tokenizing it (splitting sentences and words into individual units) to prepare it for analysis.

[0086] Step 4:

[0087] The server then analyzes the preprocessed text with an NLP engine, which extracts important keywords and phrases and analyzes the information to understand its content, particularly to identify information deemed high-risk (e.g., medical or political information).

[0088] Step 5:

[0089] The server checks identified high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[0090] Step 6:

[0091] The server will then generate a warning message if any discrepancies are found, including the specific discrepancy and a link to the authoritative source.

[0092] Step 7:

[0093] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[0094] Step 8:

[0095] Users can review warning messages and evaluate their effectiveness. They can provide feedback on whether the warning is correct by rating it as "valid" or "invalid." They can also add text comments.

[0096] Step 9:

[0097] The terminal collects feedback from the user and sends it to the server.

[0098] Step 10:

[0099] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. The model is retrained based on the feedback data, improving the accuracy of information analysis and warnings.

[0100] Step 11:

[0101] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[0102] The above steps will realize a system that can effectively detect and prevent the spread of false information on social media.

[0103] Example 1

[0104] 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."

[0105] The spread of false information on social media can have a significant impact on society. However, current systems for effectively preventing this and ensuring the reliability of information are insufficient. Automatically identifying high-risk information and quickly generating and displaying warning messages are also major challenges. Furthermore, a mechanism for continuously improving AI models using feedback is needed.

[0106] 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.

[0107] In this invention, the server includes means for periodically updating a database of trusted information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of a social media platform, means for analyzing the preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of trusted information sources, means for generating a warning message if a discrepancy is found, means for displaying the warning message on a user terminal, means for a user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine, means for awarding badges to highly contributing users, means for evaluating the risk level of the content of posts using an NLP engine, and means for retraining the NLP model using user feedback data. This enables rapid identification and prevention of false information, ensures the reliability of information, and enables continuous system improvement using feedback.

[0108] "Credible sources" refer to sources of information that are deemed to have a low risk of misinformation, such as public institutions, academic institutions, and reputable news outlets.

[0109] A "database" is an information management system for efficiently storing and retrieving collected information.

[0110] A "natural language processing engine (NLP engine)" is a computer program that analyzes human language and understands its meaning and intent.

[0111] A "classification tag" refers to a label or category assigned to data based on its characteristics.

[0112] An "SNS platform API" is a programmatic interface provided by a social networking service that allows for further data collection and manipulation.

[0113] "Preprocessed text" refers to text data that has undergone any necessary cleansing processes before analysis, such as removing special characters and stop words.

[0114] "High-risk information" refers to information that is likely to be false and carries the risk of being spread as inaccurate information.

[0115] A "warning message" is a warning message that is displayed when there is doubt about the reliability of information.

[0116] "User terminal" refers to an electronic device for using an SNS application.

[0117] "Feedback" refers to the evaluations and opinions that users provide in response to warning messages in the system.

[0118] An "artificial intelligence engine" is a computer program that analyzes and learns from data.

[0119] "Model improvement" is the process of improving the accuracy and performance of artificial intelligence or machine learning models.

[0120] A "highly contributing user" refers to a user who has made a significant contribution to the system by providing useful information or feedback.

[0121] A "badge" is a symbol or icon awarded for a particular action or contribution.

[0122] "Risk level" is an index that evaluates the likelihood that information is false.

[0123] "NLP model" refers to a machine learning model trained to perform natural language processing tasks.

[0124] "Retraining" is the training process of using new data and feedback to improve the performance of an existing model.

[0125] This invention is a system for preventing the spread of false information on social media and ensuring the reliability of information. This system is realized by performing highly accurate fact-checking while referring to reliable sources of information through the cooperation of servers, terminals, and users.

[0126] First, the server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This data collection can be done using external sources such as news APIs. This data is stored in the server's database and analyzed using a natural language processing engine (NLP engine). The NLP engine is responsible for assigning classification tags to the text data and analyzing its content.

[0127] Next, the server collects posts in real time through the APIs of social media platforms (e.g., Twitter, Facebook). It can be configured to retrieve posts related to specific keywords (e.g., "virus," "vaccine"). The collected post data includes text, author information, timestamps, and other information, and is stored in the server's database.

[0128] The posted data collected by the server is preprocessed to remove unnecessary information and noise. For example, special characters and emojis are removed from the text. This preprocessed text is then analyzed by an NLP engine to extract key keywords and phrases. This allows the risk level of the posted content to be assessed and high-risk information to be identified.

[0129] Posts that are rated high risk are checked by the server against a database of trusted sources, using keyword matching and similarity calculations to see if the content of the post matches information in the database. For example, if a post mentions a particular medication, it matches information in a medical database.

[0130] If the server finds any contradictory information, it generates a warning message that is displayed to the user via the terminal, such as "This information contradicts the official statement. For more information, please click here: [link]."

[0131] Users evaluate the effectiveness of warning messages and provide feedback. This feedback is sent to the server via their device. The server analyzes the feedback and uses it to improve the AI ​​engine's model. For example, it may retrain the NLP model based on the feedback to improve the accuracy of warning messages.

[0132] Furthermore, the server recognizes users' contributions of specialized knowledge and awards badges for effective feedback and fact-checking activities. These badges are displayed on users' profiles and are visible to other users, for example, by displaying a badge icon in the profile section of a social networking app.

[0133] As a concrete example, consider the case where a user posts information on social media that "a certain drug is effective in treating the new virus." The server analyzes the post using an NLP engine and detects the keywords "new virus" and "treatment drug." The server then compares the post with a medical information database, and if it finds an official announcement that the drug is ineffective, it displays a warning message on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[0134] Example prompt sentence:

[0135] "Is this post accurate?"

[0136] "How reliable is this information?"

[0137] Please refer to the official announcement regarding this matter.

[0138] In this way, this system effectively prevents false information from spreading on social media and provides an advanced means to ensure the reliability of information.

[0139] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0140] Step 1: Gather information and prepare

[0141] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.).

[0142] Input: Data from each source.

[0143] Data processing: The server uses news APIs and scraping technology to obtain the necessary data and stores it in a database.

[0144] Output: A stored trusted source database.

[0145] What it does: The server periodically sends API requests to gather new articles and announcements.

[0146] Step 2: Collect social media posts

[0147] The server collects posts in real time through the social media platform's API.

[0148] Input: Post data obtained from SNS (text, poster information, timestamp, etc.).

[0149] Data processing: Converting submitted data into the required format for saving in the database.

[0150] Output: Stored social media post data.

[0151] What it does: The server filters and collects posts that match specific keywords.

[0152] Step 3: Analyze the information

[0153] The server analyzes the preprocessed text and identifies high-risk information.

[0154] Input: Post data stored in the database.

[0155] Data processing: As a preprocessing step, noise is removed from the text (special characters and emojis are removed).

[0156] Output: Preprocessed and clean text data.

[0157] Specific operation: The server uses regular expressions, etc. to remove unnecessary information.

[0158] Step 4: Identify high-risk information

[0159] The server analyzes using an NLP engine to extract key keywords and phrases.

[0160] Input: Preprocessed text data.

[0161] Data calculation: An NLP engine is used to extract keywords and assess risk levels.

[0162] Output: A list of posts containing high-risk information.

[0163] What happens: The server runs an NLP model to match the extracted keywords with known risk information.

[0164] Step 5: Fact Check

[0165] The server checks the identified high-risk information against a database of trusted sources.

[0166] Input: List of posts containing high-risk information, database of trusted sources.

[0167] Data arithmetic: Use keyword matching and similarity calculations to identify information matches.

[0168] Output: A list of information containing contradictions.

[0169] What happens: For each post in the list, the server looks up the relevant information in the database and checks for any discrepancies.

[0170] Step 6: Generate and display warning messages

[0171] The server generates a warning message if it finds any discrepancies.

[0172] Input: A list of information containing conflicts.

[0173] Data calculation: Embed conflict information in warning message template.

[0174] Output: A warning message.

[0175] Specific operation: The server creates a warning message based on the template and sends it to the terminal.

[0176] Step 7: Collect and analyze feedback

[0177] The user rates the effectiveness of the warning message and provides feedback.

[0178] Input: Feedback data from users.

[0179] Data processing: Analyze the collected feedback data and use it to retrain the AI ​​model.

[0180] Output: An improved AI model.

[0181] What happens: The server analyzes the feedback data and retrains the NLP model.

[0182] Step 8: Badging

[0183] The server awards badges to users with high contributions.

[0184] Input: A list of users who have provided valid feedback or fact checks.

[0185] Data processing: Badging and updating user profiles.

[0186] Output: User profile with badges reflected.

[0187] What it does: The server adds badges to the profiles of users who provide helpful feedback.

[0188] (Application example 1)

[0189] 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."

[0190] There is a need to prevent consumer confusion and misunderstanding caused by the spread of false or unreliable information on social media and in virtual stores, and to ensure the reliability of information. Also, in virtual stores where unreliable product reviews and ratings are common, it is necessary to enable consumers to select products based on accurate information.

[0191] 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.

[0192] In this invention, the server includes: means for periodically updating a database of trusted information sources; means for analyzing collected posts using a natural language processing engine and assigning classification tags; means for collecting posts in real time through the API of a social media platform; means for analyzing the preprocessed text and identifying high-risk information; means for comparing the identified high-risk information with the database of trusted information sources; means for generating a warning message if a discrepancy is found; means for displaying the warning message on a user terminal; means for users to evaluate the effectiveness of the warning message and provide feedback; means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine; means for awarding badges to highly contributing users; means for automatically evaluating the credibility of product information and reviews in a virtual store and displaying a warning message; and means for generating a credibility evaluation prompt using a generative AI model. This prevents the spread of false information and enables consumers to make decisions based on reliable information.

[0193] A "server" is a device that processes information, manages databases, and communicates with other devices and platforms on a network.

[0194] A "reliable source" is a medium or database that provides reliable information published by public institutions or experts.

[0195] A "database" is a system that organizes and stores structured information, allowing for efficient searching and updating.

[0196] A "natural language processing engine" is software that uses algorithms and techniques to understand, analyze, and generate human language.

[0197] "Posts" are information such as text and reviews written by users on social media or in virtual stores.

[0198] A "classification tag" is an identification label that is assigned to data or information in order to effectively organize and manage it.

[0199] An "SNS platform" is an online system that provides social networking services.

[0200] "API" is an abbreviation for Application Program Interface, an interface for exchanging functions and data between different software systems.

[0201] "Preprocessed text" refers to text that has been formatted and normalized prior to data analysis and natural language processing.

[0202] "High-risk information" is content that is identified as false or unreliable information.

[0203] "Verification" means comparing data or information to identify matches or discrepancies.

[0204] A "warning message" is a notification to alert the user.

[0205] A "user terminal" is a device that allows a user to connect to the Internet and use various services.

[0206] "Feedback" refers to the evaluations and opinions that users provide about a system or service.

[0207] An "artificial intelligence engine" is a system that uses AI technology to analyze data, learn, and make decisions.

[0208] A "highly contributing user" is a user who provides useful feedback and activity within the system and is recognized.

[0209] A "badge" is a digital insignia that recognizes a user's specific activities or contributions.

[0210] A "virtual store" is an online platform that provides products and services and conducts commercial transactions over the Internet.

[0211] A "generative AI model" is a type of artificial intelligence model that uses specific algorithms to generate new data or text from existing data.

[0212] A "prompt" is input text given to a generative AI model that influences the generated output.

[0213] This invention is a system that allows consumers to easily evaluate the credibility of product information and reviews in a virtual store. The system is based on existing technology that prevents false information on social networking sites and ensures the reliability of information. The embodiments of the invention will be described from the perspectives of a server, a terminal, and a user.

[0214] server

[0215] The server has the function of regularly updating a database of reliable sources. This database collects and maintains reliable information published by public institutions and experts. The server also analyzes posts using a natural language processing engine (e.g., Spacy) and assigns classification tags. This analysis allows the content of the post to be mechanically understood. Furthermore, posts can be collected in real time through the social media platform's API, allowing constant access to the latest information. The preprocessed text identifies high-risk information and compares it with the reliable source database to assess its reliability. If a discrepancy is found, the server generates a warning message and sends it to the device.

[0216] Terminal

[0217] The terminal is a device that users use to access social networking sites and virtual stores. It includes smartphones, tablets, computers, etc. The terminal has the function of displaying warning messages sent from the server. When a user receives a warning message, they can evaluate its effectiveness and provide feedback. This feedback is sent back to the server, and the collected feedback is used to improve the AI ​​engine's model. If a user's contribution is high, they are awarded a badge, which is reflected in their profile.

[0218] User

[0219] Users use the system to browse product information and reviews in a virtual store. When a user checks a particular product review, a credibility rating is displayed in real time. For example, a review that says "This supplement can completely cure your cold" is checked against a database of trusted sources. If a discrepancy is found, a warning message is displayed, saying, "This information contradicts the official statement. For more information, click here: [Official information link]."

[0220] Examples and prompts

[0221] As a concrete example, consider what happens when a consumer sees a review that says, "This supplement can completely cure your cold." The review is analyzed on the server, and the keywords "cold" and "supplement" are extracted. After a risk assessment, the review is checked against a trusted database, and if any discrepancies are found, a warning message is generated and displayed on the device.

[0222] Prompt Sentence Examples

[0223] Please evaluate the reliability of the following text and generate a warning message if it contradicts official information:

[0224] "This supplement is said to completely cure the common cold."

[0225] This allows the system to prevent the spread of false information and allows consumers to make decisions based on reliable information.

[0226] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0227] Step 1:

[0228] The server periodically updates the database of reliable information sources. Specifically, it collects announcements from public institutions and experts via APIs and stores them in the database. The input is reliable information source data, and the output is the updated database. This ensures that the latest reliable information is always maintained.

[0229] Step 2:

[0230] The server collects posts in real time through the API of the SNS platform. Specifically, it uses the API to obtain new post data (text, poster information, timestamp, etc.). The input is the SNS post data, and the output is the collected post data. This ensures that the latest post information is always available.

[0231] Step 3:

[0232] The server preprocesses the text of the posts collected and analyzes it using a natural language processing engine (such as Spacy). Specifically, it performs text tokenization, morphological analysis, and keyword extraction. The input is the collected text data, and the output is the preprocessed text and extracted keywords. This allows the content of the posts to be mechanically understood.

[0233] Step 4:

[0234] The server analyzes the preprocessed text and identifies high-risk information. Specifically, it applies a risk assessment algorithm based on the extracted keywords to calculate a risk level. The input is the preprocessed text and keywords, and the output is a risk level. This identifies information that may be false.

[0235] Step 5:

[0236] The server checks the identified high-risk information against a database of trusted information sources. Specifically, the check is performed using keyword matching and similarity calculations. The input is the high-risk information and the database of trusted information sources, and the output is the check result (presence or absence of inconsistencies). This confirms whether the information is trustworthy.

[0237] Step 6:

[0238] If a discrepancy is found, the server generates a warning message. Specifically, it creates a message in the format "This information contradicts the official announcement. For more information, click here: [Official information link]." The input is the match result and details of the contradictory information, and the output is a warning message. This allows users to recognize misinformation.

[0239] Step 7:

[0240] The terminal displays a warning message to the user. Specifically, the warning message is displayed in a pop-up format on the page the user is viewing. The input is the warning message sent from the server, and the output is the warning message displayed on the terminal screen. This allows the user to check the reliability of the information in real time.

[0241] Step 8:

[0242] Users evaluate the effectiveness of warning messages and provide feedback. Specifically, they enter their evaluation of effectiveness and comments in a feedback form and send it to the server. The input is the user's feedback, and the output is the feedback data sent to the server. This contributes to improving the accuracy of the system.

[0243] Step 9:

[0244] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. Specifically, the feedback data is used to retrain the model and adjust parameters. The input is the collected feedback data, and the output is an improved AI model. This improves the accuracy and reliability of the system.

[0245] Step 10:

[0246] The server assigns badges to users with high contributions. Specifically, it evaluates the effectiveness of feedback and the level of contribution, and adds the badge to the user's profile. The input is the user's evaluation data, and the output is an updated user profile. This makes the user's contributions visible, improving motivation.

[0247] 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.

[0248] This invention is a system for preventing the spread of false information on social networking sites and ensuring the reliability of information. This system performs highly accurate fact-checking based on data from reliable sources and combines it with an emotion engine that recognizes the user's emotions. Specifically, with the cooperation of the server, device, and user, the system generates appropriate warning messages according to the user's emotional state, improving the accuracy of information and user acceptance.

[0249] System Overview

[0250] This system is configured as follows:

[0251] 1. Server: Maintains a database of trusted sources, analyzes posts using NLP and sentiment engines, assesses the risk of information, and generates warning messages.

[0252] 2. Terminal: A device that allows users to use SNS and displays warning messages and sends feedback.

[0253] 3. Users: contribute to the credibility of information through social media posts and feedback.

[0254] Overview of program processing

[0255] 1. Information gathering and preparation

[0256] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[0257] 2. Collect social media posts

[0258] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[0259] 3. Analysis and evaluation of information

[0260] The server preprocesses the text of collected posts and then analyzes it with an NLP engine, which extracts important keywords and phrases and analyzes them to understand the content of the information, particularly to identify information that is considered high-risk (e.g., medical or political information).

[0261] 4. Emotion analysis

[0262] The server uses an emotion engine to analyze the emotions of users' posts and reactions. This emotion analysis allows us to understand the intention and tone of the posts and evaluate the emotional state of the users.

[0263] 5. Fact Check

[0264] The server checks high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[0265] 6. Generating and Displaying Warning Messages

[0266] The server generates warning messages for posts that contradict reliable information. The content and presentation of these warning messages are tailored to the user's emotional state. For example, if a user is emotionally charged, a message urging them to stay calm is displayed.

[0267] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[0268] 7. Feedback Collection and Analysis

[0269] Users can review warning messages and evaluate their effectiveness. They can provide feedback on whether the warning is correct by rating it as "valid" or "invalid." They can also add text comments.

[0270] The terminal collects feedback from the user and sends it to the server.

[0271] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. The feedback data is used to retrain the model, improving the accuracy of information analysis and warnings. The emotion engine also analyzes the emotional tone of user feedback and uses it to further improve the model.

[0272] 8. Badging

[0273] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[0274] Specific examples

[0275] Example 1: Medical information verification and sentiment analysis

[0276] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[0277] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[0278] 3. The server checks the post against a reliable medical information database.

[0279] 4. If the server finds an official announcement that the drug is ineffective, a warning message will be displayed on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[0280] 5. The server analyzes the user's emotions using an emotion engine and issues a warning in a calm tone.

[0281] Example 2: Political information verification and feedback analysis

[0282] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[0283] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[0284] 3. Users rate the effectiveness of the warning message and provide feedback.

[0285] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model. The emotion engine also analyzes the emotional tone of the feedback to further improve the model.

[0286] This allows the system to effectively detect and prevent false information being spread on social media, while also providing appropriate warning messages based on the user's emotional state.

[0287] The processing flow will be explained below.

[0288] Step 1:

[0289] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[0290] Step 2:

[0291] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[0292] Step 3:

[0293] The server preprocesses the collected text of posts, including normalizing the text data (removing special characters and extra whitespace) and tokenizing it (splitting sentences and words into individual units) to make it analyzable.

[0294] Step 4:

[0295] The server then analyzes the preprocessed text using an NLP engine, extracting key keywords and phrases and understanding the content. This analysis identifies information that is considered particularly high-risk (e.g., medical or political information).

[0296] Step 5:

[0297] The server uses an emotion engine to analyze emotions from users' posts and reactions. The emotion engine incorporates a text analysis algorithm to identify the user's emotional state (e.g., "joy," "anger," "sadness," etc.).

[0298] Step 6:

[0299] The server checks identified high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[0300] Step 7:

[0301] If the server finds any discrepancies based on the comparison results, it generates a warning message. The content and display of this warning message are adjusted according to the user's emotional state. For example, if the user is emotionally charged, a message urging them to stay calm is generated.

[0302] Step 8:

[0303] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[0304] Step 9:

[0305] Users can review warning messages and rate their validity. Users can rate warnings as valid or invalid, and can also add text comments to provide feedback.

[0306] Step 10:

[0307] The terminal collects feedback from the user and sends it to the server.

[0308] Step 11:

[0309] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model. The feedback data is used to retrain the model, improving the accuracy of information analysis and warnings. The emotion engine also analyzes the emotional tone of the user's feedback and uses it to further improve the model.

[0310] Step 12:

[0311] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[0312] As a result, this system can effectively detect and prevent false information from spreading on social media, and provide appropriate warning messages according to the user's emotional state.

[0313] Example 2

[0314] 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."

[0315] Modern online communication platforms face the problem of easily spreading unreliable and false information. This increases the risk of users receiving incorrect information, potentially leading to social confusion and misunderstanding. Furthermore, preventing the spread of misinformation is difficult because appropriate information is not provided based on the user's emotional state. Furthermore, systems have not yet been sufficiently improved based on user feedback, necessitating improved accuracy in detecting and warning about false information.

[0316] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for periodically updating a database of reliable information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of an online platform, means for analyzing preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of reliable information sources, means for evaluating text consistency and inconsistency through keyword matching and similarity calculation, means for performing user sentiment analysis and evaluating the intention and tone of the post, means for generating a warning message and adjusting it according to the user's emotional state, means for displaying the warning message on a user terminal, means for the user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and utilizing it to improve the model of the artificial intelligence engine, and means for awarding badges to highly contributing users. This enables effective detection and prevention of false information spreading on social media and provision of appropriate information taking into account the user's emotional state.

[0317] A "trusted information source database" is a database that collects and stores publicly recognized information such as news sites, academic papers, and official announcements, and ensures the authenticity and reliability of the information.

[0318] A "natural language processing engine" is an artificial intelligence technology that analyzes text data and has functions such as understanding the meaning of language, extracting keywords, and classifying documents.

[0319] An "online platform API" is an interface for communicating with web services and applications and managing the sending and receiving of data.

[0320] "Preprocessed text" refers to text data that has undergone processing such as tokenization, stop word removal, and stemming before being analyzed by a natural language processing engine.

[0321] "High-risk information" is information that may have significant social, economic, or health impacts and should be handled with particular care.

[0322] "Keyword matching" is a technique for evaluating matches between texts, and is a technology that primarily compares based on the presence or absence and frequency of keywords.

[0323] "Similarity calculation" is a technique for evaluating the semantic similarity between texts, and is a comparison technique using mathematical indices such as cosine similarity and Jaccard coefficient.

[0324] "Sentiment analysis" is an artificial intelligence technique that analyzes the emotions and tone contained in text to assess the user's emotional state.

[0325] A "warning message" is a notification message generated to alert the user to detected high-risk information.

[0326] A "user terminal" is a device that allows a user to view information and interact with the device, including a computer, smartphone, tablet, etc.

[0327] "Feedback" refers to evaluation information and comments provided by users, and is data used to improve the system and retrain the model.

[0328] An "artificial intelligence engine" is an artificial intelligence technology that trains models and makes predictions based on feedback data and learning data.

[0329] A "badge" is a digital award given to a user as recognition for their contribution or specific activity within the system.

[0330] The present invention provides a system for preventing the spread of false information on social networking sites and ensuring the reliability of information. This system performs highly accurate fact-checking based on reliable information sources and combines it with an emotion engine that recognizes user emotions. Detailed embodiments are described below.

[0331] System configuration

[0332] This system consists of three elements: a server, a terminal, and a user.

[0333] 1. Server

[0334] The server has many roles and is responsible for the following processes:

[0335] Regularly updated database of reliable sources: The server collects data from news sites, academic papers, official announcements, etc. and updates the database. Specifically, the data collection is done using Python's BeautifulSoup library and the Scrapy framework.

[0336] Analyze collected posts using a natural language processing engine: The server analyzes the collected data using an NLP engine (for example, Google BERT or OpenAI GPT) and assigns classification tags. The collected text data is preprocessed using libraries such as NLTK or spaCy before analysis.

[0337] Collecting posts through APIs of online platforms: The server collects post data in real time from various social media platforms (e.g., Twitter API, Facebook Graph API) and stores it in a database.

[0338] Preprocessing and analysis of submitted text: The server preprocesses the collected text (tokenization, stop word removal, stemming, etc.) and analyzes it using an NLP engine.

[0339] Identifying high-risk information: The server extracts important keywords and phrases and evaluates the agreements and inconsistencies between the texts using keyword matching and similarity calculations (e.g., Cosine Similarity or Jaccard Index).

[0340] Sentiment analysis: The server uses an emotion engine (for example, IBM Watson Tone Analyzer or Microsoft Azure Text Analytics) to assess the user's emotional state and understand the intent and tone of the post.

[0341] Generate warning messages: The server checks high-risk information against trusted sources and generates warning messages if inconsistencies are found, tailoring the messages to the user's emotional state.

[0342] Feedback collection and analysis: The server collects feedback provided by users, analyzes it using an artificial intelligence engine, and uses it to improve the model.

[0343] 2. Terminal

[0344] The terminal is the device through which the user views information and interacts, and is responsible for the following processes:

[0345] Display warning message: The terminal displays the warning message sent from the server to the user. The warning message is displayed as a pop-up window.

[0346] Collecting and sending feedback: The device collects feedback from users (ratings of the effectiveness of warning messages and comments) and sends it to the server.

[0347] 3. Users

[0348] Users play a role in contributing to the credibility of information through social media posts and feedback.

[0349] Review and rate warning messages: Users can review the displayed warning messages and rate their validity as "valid" or "invalid." They can also add text comments.

[0350] Providing feedback: The user provides feedback on the effectiveness of the warning message through the terminal.

[0351] Specific examples

[0352] Example 1: Medical information verification and sentiment analysis

[0353] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[0354] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[0355] 3. The server checks the post against a reliable medical information database.

[0356] 4. If the server finds an official announcement that the drug is ineffective, it will display a warning message on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[0357] 5. The server analyzes the user's emotions using an emotion engine and issues a warning in a calm tone.

[0358] Example 2: Political information verification and feedback analysis

[0359] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[0360] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[0361] 3. Users rate the effectiveness of the warning message and provide feedback.

[0362] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model. The emotion engine also analyzes the emotional tone of the feedback to further improve the model.

[0363] Example prompts for generative AI models

[0364] "Please check whether the COVID-19 treatment information posted on social media is reliable."

[0365] "Compare this political information to see if it matches the official announcement."

[0366] "Generate appropriate warning messages based on the user's emotional state."

[0367] As described above, this system effectively detects false information being spread on social media, prevents its spread, and provides appropriate warning messages according to the user's emotional state.

[0368] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0369] Step 1:

[0370] Data collection from reliable sources

[0371] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.). Specifically, it periodically retrieves data from a specified URL using Python's BeautifulSoup library or the Scrapy framework. The input is the specified URL, and the output is the retrieved text data.

[0372] Step 2:

[0373] Update the database of collected data

[0374] The server stores the collected text data in a database (for example, MongoDB or MySQL). The input is the collected text data, and the output is the updated database. Specifically, the Python script uses the Pandas library to insert the data organized in a data frame into the database.

[0375] Step 3:

[0376] Preprocessing of submitted data

[0377] The server preprocesses the collected data using an NLP engine, performing processes such as tokenization, stop word removal, and stemming. The input is the collected text data, and the output is the preprocessed text data. Specific operations use the NLTK and spaCy libraries.

[0378] Step 4:

[0379] Analysis using natural language processing

[0380] The server inputs the preprocessed text data into an NLP engine (such as Google BERT or OpenAI GPT) to analyze and extract important keywords and phrases. The input is the preprocessed text data, and the output is a list of keywords and phrases resulting from the analysis.

[0381] Step 5:

[0382] Adding classification tags

[0383] The server assigns classification tags (e.g., "medical," "politics," "economy," etc.) to the text data based on the results of analysis by the NLP engine. The input is a list of keywords and phrases from the analysis results, and the output is text data with classification tags.

[0384] Step 6:

[0385] Real-time collection of SNS posts

[0386] The server collects social media content in real time through the API of an online platform (e.g., Twitter API, Facebook Graph API). The input is a specified API endpoint, and the output is the collected post data. Specifically, it sends an HTTP request and obtains the post data as a response.

[0387] Step 7:

[0388] Identifying high-risk information

[0389] The server preprocesses the social media posts collected and analyzes them using an NLP engine. High-risk information (e.g., "medical information" or "political information") is identified and extracted. The input is the preprocessed post data, and the output is the post data identified as high-risk. Specifically, it detects important keywords.

[0390] Step 8:

[0391] Performing sentiment analysis

[0392] The server uses an emotion engine (for example, IBM Watson Tone Analyzer or Microsoft Azure Text Analytics) to analyze emotions from the post content and reactions. The input is the post data identified as high risk, and the output is the emotion analysis results. Specifically, the server sends text data to the emotion engine's API and receives the analysis results.

[0393] Step 9:

[0394] Conducting fact-checks

[0395] The server checks high-risk posts against a database of trusted sources and uses keyword matching or similarity calculations (e.g., Cosine Similarity or Jaccard Index) to check for matches or contradictions. The input is the high-risk post data and the source database, and the output is the fact-check results. Specifically, the server applies a similarity calculation algorithm.

[0396] Step 10:

[0397] Generate a warning message

[0398] The server generates a warning message based on the fact-check results if a contradiction is found. The content and display method of this message are adjusted according to the user's emotional state. The input is the fact-check results and the emotion analysis results, and the output is the generated warning message. Specific operation involves embedding specific information into the template message.

[0399] Step 11:

[0400] Displaying a warning message

[0401] The terminal receives the warning message and displays it to the user. The input is the generated warning message, and the output is the warning message displayed on the terminal screen. As a specific operation, it executes the code that displays the popup window.

[0402] Step 12:

[0403] Collecting feedback

[0404] The user evaluates the effectiveness of the warning message and provides feedback. The input is the displayed warning message, and the output is the feedback from the user. Specifically, the user selects "enabled" or "disabled" and enters a comment.

[0405] Step 13:

[0406] Send Feedback

[0407] The terminal sends the feedback collected from the user to the server. The input is the feedback from the user, and the output is the feedback sent to the server. The specific operation is to send the feedback data using an API.

[0408] Step 14:

[0409] Feedback Analysis

[0410] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model. The input is the feedback sent to the server, and the output is the improved AI model. Specifically, the feedback data is used to retrain the model.

[0411] Step 15:

[0412] Badge Awarding

[0413] The server assigns badges to users who have contributed significantly to effective feedback and fact-checking activities. The input is feedback history, and the output is the assigned badges. Specifically, the server automatically assigns badges to users who meet certain evaluation criteria.

[0414] (Application example 2)

[0415] 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."

[0416] As the degree of freedom in the distribution of information on the Internet and social networking services (SNS) increases, the spread of false information has become a social problem. In particular, false information about high-risk topics (e.g., medical, political, economic, etc.) can have a significant impact. Conventional systems have been required to effectively prevent the spread of such false information while at the same time taking into account the user's emotions, but this has been difficult to achieve. The purpose of this invention is to solve this problem by quickly and accurately detecting false information and providing warning messages that correspond to the user's emotional state.

[0417] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for periodically updating a database of reliable information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of an SNS platform, means for analyzing preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of reliable information sources, means for generating a warning message if a discrepancy is found, means for displaying the warning message on a user terminal, means for the user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine, means for awarding badges to highly contributing users, and means for adjusting the content and display method of the warning message according to the user's emotions and displaying it on the user's device. This prevents the spread of false information and enables the provision of reliable information that takes user emotions into consideration.

[0418] A "database of trusted sources" is a regularly updated database that aggregates data collected from sources such as news sites, academic papers, and official announcements whose credibility has been confirmed.

[0419] A "natural language processing engine" is an algorithm and software that analyzes text data and performs tasks such as extracting keywords, understanding context, and tagging.

[0420] An "SNS platform API" is an application programming interface provided by a social networking service, and is a function that allows external developers to obtain and manipulate data on the SNS.

[0421] "Preprocessed text" is text data that has been converted into a form that is easier to analyze by filtering and cleaning up raw data.

[0422] "High-risk information" refers to information that is likely to contain false information and that falls into categories such as medical, political, and economic information, which may have particularly serious consequences.

[0423] A "warning message" is a message that notifies or alerts the user that there is a doubt about the accuracy or reliability of the information.

[0424] "User terminal" refers to any device used by a user to access SNS, including smartphones, tablets, and PCs.

[0425] "Feedback" refers to the evaluations and comments that users make on the system's warning messages and the information provided.

[0426] "Artificial intelligence engine" is a general term for algorithms and software that learns from collected data and improves models.

[0427] A "badge" is a symbol of recognition and appreciation given to users with specialized knowledge or who have made significant contributions to the system.

[0428] The means for adjusting the content and display method according to "emotion" is a function that generates and displays an appropriate warning message according to the user's current emotional state based on the results of emotion analysis.

[0429] This invention relates to a system for preventing the spread of false information on the Internet and social networking services (SNS) and ensuring the reliability of information. As a specific example, a reliability guide quickly and accurately detects false information and provides warning messages tailored to the user's emotional state. The following describes the components of this system and their functions.

[0430] server

[0431] The server has the following functions:

[0432] 1. Updating the database of reliable sources: Regularly collect data from reliable news sites, academic papers, official announcements, etc. and update the database. This data will be used as reliable data.

[0433] 2. Natural Language Processing Engine (NLP Engine): Analyzes collected posts and assigns classification tags based on their content. This engine extracts important keywords and phrases and identifies high-risk information.

[0434] 3. Real-time collection using SNS platform APIs: Collect user posts from SNS platforms in real time. The posts include text, author information, timestamps, etc.

[0435] 4. Pre-processed text analysis: The collected text data is pre-processed and analyzed by an NLP engine to identify information that is considered high risk.

[0436] 5. Matching high-risk information with reliable data: Identified high-risk information is matched with a database of reliable sources to identify any matches or discrepancies.

[0437] 6. Sentiment analysis engine: Analyzes emotions from user posts and reactions to evaluate the user's emotional state.

[0438] 7. Generating a warning message: If a contradiction is found, a warning message is generated according to the user's emotional state. For example, if the user is emotionally charged, a message is generated to encourage them to calm down.

[0439] 8. AI engine model improvement: Analyze user feedback to retrain the AI ​​engine model and improve the accuracy of warnings.

[0440] Terminal

[0441] The terminal has the following functions:

[0442] 1. Displaying a warning message: Displaying a warning message received from the server to the user. This includes devices such as smart glasses and smartphones.

[0443] 2. Feedback collection: Collect feedback from users about the warning message and send it to the server.

[0444] User

[0445] The user has the following capabilities:

[0446] 1. Posting to SNS: Users post information to SNS and share it with other users.

[0447] 2. Providing Feedback: You can rate the effectiveness of the warning message and provide feedback. You can also add text comments.

[0448] Specific examples

[0449] For example, if a user tries to post information on a social networking site that "a certain drug is effective in treating the new virus," the server performs the following process.

[0450] The NLP engine detects the keywords "new virus" and "treatment drug" and compares them with a reliable medical information database.

[0451] If the database contains an official statement that the drug is ineffective, the server generates a warning message stating, "This drug is said to be ineffective against the new virus. Source: [Official statement link]" and sends it to the device in a calm tone that corresponds to the user's emotional state.

[0452] The terminal displays this message to the user.

[0453] This will prevent the spread of false information and allow for the provision of information that takes into account the emotional state of the user.

[0454] Prompt Sentence Examples

[0455] "Are certain drugs effective in treating the new virus?"

[0456] As described above, the Trust Guide provides warning messages based on the user's emotional state and improves the AI ​​engine through feedback, thereby providing reliable information and preventing the spread of false information.

[0457] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0458] Step 1:

[0459] The server collects data from reliable sources and updates the database. Specifically, it obtains data from news sites, academic papers, official announcements, etc. via API and stores it in the database. This data collection occurs periodically. The input is data from reliable sources, and the output is an updated database.

[0460] Step 2:

[0461] The server uses the API of the SNS platform to collect user posts in real time. The collected data includes text, poster information, timestamps, etc. The input is the SNS data posted in real time, and the output is the collected SNS post data.

[0462] Step 3:

[0463] The server preprocesses the collected text data of SNS posts, performing noise removal, normalization, tokenization, etc. The input is the text portion of the collected SNS post data, and the output is the preprocessed text data.

[0464] Step 4:

[0465] The server analyzes the preprocessed text using a natural language processing (NLP) engine to perform entity extraction, keyword extraction, sentiment analysis, etc. The input is the preprocessed text data, and the output is the analysis results in the form of keywords, sentiment information, and tags for suspicious information.

[0466] Step 5:

[0467] The server identifies high-risk information based on the analysis results. For example, it checks whether keywords such as "new virus" or "treatment drug" are included. The input is the analysis results of the NLP engine, and the output is a flag for high-risk information.

[0468] Step 6:

[0469] The server checks the identified high-risk information against a database of reliable information sources. It checks for matches and inconsistencies to detect high-risk, unreliable information. For example, it checks for matches with an official announcement that "the specified therapeutic drug is ineffective." The input is the flag for high-risk information and the contents of the database, and the output is the matching result.

[0470] Step 7:

[0471] The server generates a warning message based on the matching results if any discrepancies are found. An emotion analysis engine is used to determine the message content according to the user's emotional state. The input is the matching results and the user's emotional state, and the output is the generated warning message.

[0472] Step 8:

[0473] The terminal displays the generated warning message to the user. On devices such as smart glasses or smartphones, the warning message is presented as a popup at the appropriate time. The input is the generated warning message, and the output is the display of the warning message to the user.

[0474] Step 9:

[0475] Users review warning messages, rate their effectiveness, and provide feedback. Users can rate warning messages as "valid" or "invalid" and add comments. The input is user feedback, and the output is the collected feedback data.

[0476] Step 10:

[0477] The terminal collects feedback from the user and sends it to the server, where the input is the collected feedback data and the output is the transmission of the feedback data to the server.

[0478] Step 11:

[0479] The server analyzes the collected feedback data and uses it to improve the artificial intelligence (AI) engine's model. The AI ​​engine retrains the model based on the feedback, improving the accuracy of information analysis and warnings. The input is the feedback data, and the output is an improved AI model.

[0480] Step 12:

[0481] The server assigns badges to users with high contributions. Users who provide reliable feedback are recognized for their contributions by being given badges. The input is the user's feedback history and contribution evaluation, and the output is the assignment of a badge.

[0482] 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.

[0483] 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.

[0484] 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.

[0485] [Second embodiment]

[0486] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0487] 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.

[0488] 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).

[0489] 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.

[0490] 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.

[0491] 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).

[0492] 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.

[0493] 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.

[0494] 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.

[0495] 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.

[0496] 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.

[0497] 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."

[0498] This invention is a system for preventing the spread of false information on social media and ensuring the reliability of information. This system is realized by performing highly accurate fact-checking while referring to reliable sources of information through cooperation between servers, terminals, and users.

[0499] System Overview

[0500] This system is configured as follows:

[0501] 1. Server: Maintains a database of trusted sources, analyzes posts using an NLP engine, assesses the risk of information, and generates warning messages.

[0502] 2. Terminal: A device that allows users to use SNS and displays warning messages and sends feedback.

[0503] 3. Users: contribute to the credibility of information through social media posts and feedback.

[0504] Overview of program processing

[0505] 1. Information gathering and preparation

[0506] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database, then uses an NLP engine to analyze the new information and assign it classification tags.

[0507] 2. Collect social media posts

[0508] The server collects posts in real time through the API of the social media platform. The collected post data includes text, poster information, timestamp, etc.

[0509] 3. Analysis and evaluation of information

[0510] The server preprocesses the collected text of posts and analyzes it with an NLP engine. This analysis extracts key keywords and phrases from the posts and evaluates their risk level based on them.

[0511] 4. Fact Check

[0512] The server checks high-risk posts against a database of trusted sources, using keyword matching and similarity calculations.

[0513] 5. Generating and Displaying Warning Messages

[0514] The server generates a warning message for posts that contradict reliable information. This warning message is displayed on the device, for example, "This information contradicts official statements. Read more here: [link]."

[0515] 6. Feedback Collection and Analysis

[0516] The user can rate the effectiveness of the warning message and provide feedback, which the terminal sends to the server.

[0517] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model.

[0518] 7. Badge Awarding

[0519] The server recognizes users' contributions of expertise and awards badges for useful feedback and fact-checking activities, which are reflected in users' profiles.

[0520] Specific examples

[0521] Example 1: Medical information verification

[0522] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[0523] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[0524] 3. The server checks the post against a reliable medical information database.

[0525] 4. If the server finds an official announcement that the drug is ineffective, a warning message will be displayed on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[0526] Example 2: Verifying political information

[0527] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[0528] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[0529] 3. Users rate the effectiveness of the warning message and provide feedback.

[0530] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model.

[0531] This allows the system to effectively prevent false information from spreading on social media and promote the dissemination and sharing of reliable information.

[0532] The processing flow will be explained below.

[0533] Step 1:

[0534] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[0535] Step 2:

[0536] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[0537] Step 3:

[0538] The server preprocesses the text of the posts collected by normalizing the text data (removing special characters and extra whitespace) and tokenizing it (splitting sentences and words into individual units) to prepare it for analysis.

[0539] Step 4:

[0540] The server then analyzes the preprocessed text with an NLP engine, which extracts important keywords and phrases and analyzes the information to understand its content, particularly to identify information deemed high-risk (e.g., medical or political information).

[0541] Step 5:

[0542] The server checks identified high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[0543] Step 6:

[0544] The server will then generate a warning message if any discrepancies are found, including the specific discrepancy and a link to the authoritative source.

[0545] Step 7:

[0546] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[0547] Step 8:

[0548] Users can review warning messages and evaluate their effectiveness. They can provide feedback on whether the warning is correct by rating it as "valid" or "invalid." They can also add text comments.

[0549] Step 9:

[0550] The terminal collects feedback from the user and sends it to the server.

[0551] Step 10:

[0552] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. The model is retrained based on the feedback data, improving the accuracy of information analysis and warnings.

[0553] Step 11:

[0554] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[0555] The above steps will realize a system that can effectively detect and prevent the spread of false information on social media.

[0556] Example 1

[0557] 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."

[0558] The spread of false information on social media can have a significant impact on society. However, current systems for effectively preventing this and ensuring the reliability of information are insufficient. Automatically identifying high-risk information and quickly generating and displaying warning messages are also major challenges. Furthermore, a mechanism for continuously improving AI models using feedback is needed.

[0559] 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.

[0560] In this invention, the server includes means for periodically updating a database of trusted information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of a social media platform, means for analyzing the preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of trusted information sources, means for generating a warning message if a discrepancy is found, means for displaying the warning message on a user terminal, means for a user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine, means for awarding badges to highly contributing users, means for evaluating the risk level of the content of posts using an NLP engine, and means for retraining the NLP model using user feedback data. This enables rapid identification and prevention of false information, ensures the reliability of information, and enables continuous system improvement using feedback.

[0561] "Credible sources" refer to sources of information that are deemed to have a low risk of misinformation, such as public institutions, academic institutions, and reputable news outlets.

[0562] A "database" is an information management system for efficiently storing and retrieving collected information.

[0563] A "natural language processing engine (NLP engine)" is a computer program that analyzes human language and understands its meaning and intent.

[0564] A "classification tag" refers to a label or category assigned to data based on its characteristics.

[0565] An "SNS platform API" is a programmatic interface provided by a social networking service that allows for further data collection and manipulation.

[0566] "Preprocessed text" refers to text data that has undergone any necessary cleansing processes before analysis, such as removing special characters and stop words.

[0567] "High-risk information" refers to information that is likely to be false and carries the risk of being spread as inaccurate information.

[0568] A "warning message" is a warning message that is displayed when there is doubt about the reliability of information.

[0569] "User terminal" refers to an electronic device for using an SNS application.

[0570] "Feedback" refers to the evaluations and opinions that users provide in response to warning messages in the system.

[0571] An "artificial intelligence engine" is a computer program that analyzes and learns from data.

[0572] "Model improvement" is the process of improving the accuracy and performance of artificial intelligence or machine learning models.

[0573] A "highly contributing user" refers to a user who has made a significant contribution to the system by providing useful information or feedback.

[0574] A "badge" is a symbol or icon awarded for a particular action or contribution.

[0575] "Risk level" is an index that evaluates the likelihood that information is false.

[0576] "NLP model" refers to a machine learning model trained to perform natural language processing tasks.

[0577] "Retraining" is the training process of using new data and feedback to improve the performance of an existing model.

[0578] This invention is a system for preventing the spread of false information on social media and ensuring the reliability of information. This system is realized by performing highly accurate fact-checking while referring to reliable sources of information through the cooperation of servers, terminals, and users.

[0579] First, the server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This data collection can be done using external sources such as news APIs. This data is stored in the server's database and analyzed using a natural language processing engine (NLP engine). The NLP engine is responsible for assigning classification tags to the text data and analyzing its content.

[0580] Next, the server collects posts in real time through the APIs of social media platforms (e.g., Twitter, Facebook). It can be configured to retrieve posts related to specific keywords (e.g., "virus," "vaccine"). The collected post data includes text, author information, timestamps, and other information, and is stored in the server's database.

[0581] The posted data collected by the server is preprocessed to remove unnecessary information and noise. For example, special characters and emojis are removed from the text. This preprocessed text is then analyzed by an NLP engine to extract key keywords and phrases. This allows the risk level of the posted content to be assessed and high-risk information to be identified.

[0582] Posts that are rated high risk are checked by the server against a database of trusted sources, using keyword matching and similarity calculations to see if the content of the post matches information in the database. For example, if a post mentions a particular medication, it matches information in a medical database.

[0583] If the server finds any contradictory information, it generates a warning message that is displayed to the user via the terminal, such as "This information contradicts the official statement. For more information, please click here: [link]."

[0584] Users evaluate the effectiveness of warning messages and provide feedback. This feedback is sent to the server via their device. The server analyzes the feedback and uses it to improve the AI ​​engine's model. For example, it may retrain the NLP model based on the feedback to improve the accuracy of warning messages.

[0585] Furthermore, the server recognizes users' contributions of specialized knowledge and awards badges for effective feedback and fact-checking activities. These badges are displayed on users' profiles and are visible to other users, for example, by displaying a badge icon in the profile section of a social networking app.

[0586] As a concrete example, consider the case where a user posts information on social media that "a certain drug is effective in treating the new virus." The server analyzes the post using an NLP engine and detects the keywords "new virus" and "treatment drug." The server then compares the post with a medical information database, and if it finds an official announcement that the drug is ineffective, it displays a warning message on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[0587] Example prompt sentence:

[0588] "Is this post accurate?"

[0589] "How reliable is this information?"

[0590] Please refer to the official announcement regarding this matter.

[0591] In this way, this system effectively prevents false information from spreading on social media and provides an advanced means to ensure the reliability of information.

[0592] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0593] Step 1: Gather information and prepare

[0594] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.).

[0595] Input: Data from each source.

[0596] Data processing: The server uses news APIs and scraping technology to obtain the necessary data and stores it in a database.

[0597] Output: A stored trusted source database.

[0598] What it does: The server periodically sends API requests to gather new articles and announcements.

[0599] Step 2: Collect social media posts

[0600] The server collects posts in real time through the social media platform's API.

[0601] Input: Post data obtained from SNS (text, poster information, timestamp, etc.).

[0602] Data processing: Converting submitted data into the required format for saving in the database.

[0603] Output: Stored social media post data.

[0604] What it does: The server filters and collects posts that match specific keywords.

[0605] Step 3: Analyze the information

[0606] The server analyzes the preprocessed text and identifies high-risk information.

[0607] Input: Post data stored in the database.

[0608] Data processing: As a preprocessing step, noise is removed from the text (special characters and emojis are removed).

[0609] Output: Preprocessed and clean text data.

[0610] Specific operation: The server uses regular expressions, etc. to remove unnecessary information.

[0611] Step 4: Identify high-risk information

[0612] The server analyzes using an NLP engine to extract key keywords and phrases.

[0613] Input: Preprocessed text data.

[0614] Data calculation: An NLP engine is used to extract keywords and assess risk levels.

[0615] Output: A list of posts containing high-risk information.

[0616] What happens: The server runs an NLP model to match the extracted keywords with known risk information.

[0617] Step 5: Fact Check

[0618] The server checks the identified high-risk information against a database of trusted sources.

[0619] Input: List of posts containing high-risk information, database of trusted sources.

[0620] Data arithmetic: Use keyword matching and similarity calculations to identify information matches.

[0621] Output: A list of information containing contradictions.

[0622] What happens: For each post in the list, the server looks up the relevant information in the database and checks for any discrepancies.

[0623] Step 6: Generate and display warning messages

[0624] The server generates a warning message if it finds any discrepancies.

[0625] Input: A list of information containing conflicts.

[0626] Data calculation: Embed conflict information in warning message template.

[0627] Output: A warning message.

[0628] Specific operation: The server creates a warning message based on the template and sends it to the terminal.

[0629] Step 7: Collect and analyze feedback

[0630] The user rates the effectiveness of the warning message and provides feedback.

[0631] Input: Feedback data from users.

[0632] Data processing: Analyze the collected feedback data and use it to retrain the AI ​​model.

[0633] Output: An improved AI model.

[0634] What happens: The server analyzes the feedback data and retrains the NLP model.

[0635] Step 8: Badging

[0636] The server awards badges to users with high contributions.

[0637] Input: A list of users who have provided valid feedback or fact checks.

[0638] Data processing: Badging and updating user profiles.

[0639] Output: User profile with badges reflected.

[0640] What it does: The server adds badges to the profiles of users who provide helpful feedback.

[0641] (Application example 1)

[0642] 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."

[0643] There is a need to prevent consumer confusion and misunderstanding caused by the spread of false or unreliable information on social media and in virtual stores, and to ensure the reliability of information. Also, in virtual stores where unreliable product reviews and ratings are common, it is necessary to enable consumers to select products based on accurate information.

[0644] 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.

[0645] In this invention, the server includes: means for periodically updating a database of trusted information sources; means for analyzing collected posts using a natural language processing engine and assigning classification tags; means for collecting posts in real time through the API of a social media platform; means for analyzing the preprocessed text and identifying high-risk information; means for comparing the identified high-risk information with the database of trusted information sources; means for generating a warning message if a discrepancy is found; means for displaying the warning message on a user terminal; means for users to evaluate the effectiveness of the warning message and provide feedback; means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine; means for awarding badges to highly contributing users; means for automatically evaluating the credibility of product information and reviews in a virtual store and displaying a warning message; and means for generating a credibility evaluation prompt using a generative AI model. This prevents the spread of false information and enables consumers to make decisions based on reliable information.

[0646] A "server" is a device that processes information, manages databases, and communicates with other devices and platforms on a network.

[0647] A "reliable source" is a medium or database that provides reliable information published by public institutions or experts.

[0648] A "database" is a system that organizes and stores structured information, allowing for efficient searching and updating.

[0649] A "natural language processing engine" is software that uses algorithms and techniques to understand, analyze, and generate human language.

[0650] "Posts" are information such as text and reviews written by users on social media or in virtual stores.

[0651] A "classification tag" is an identification label that is assigned to data or information in order to effectively organize and manage it.

[0652] An "SNS platform" is an online system that provides social networking services.

[0653] "API" is an abbreviation for Application Program Interface, an interface for exchanging functions and data between different software systems.

[0654] "Preprocessed text" refers to text that has been formatted and normalized prior to data analysis and natural language processing.

[0655] "High-risk information" is content that is identified as false or unreliable information.

[0656] "Verification" means comparing data or information to identify matches or discrepancies.

[0657] A "warning message" is a notification to alert the user.

[0658] A "user terminal" is a device that allows a user to connect to the Internet and use various services.

[0659] "Feedback" refers to the evaluations and opinions that users provide about a system or service.

[0660] An "artificial intelligence engine" is a system that uses AI technology to analyze data, learn, and make decisions.

[0661] A "highly contributing user" is a user who provides useful feedback and activity within the system and is recognized.

[0662] A "badge" is a digital insignia that recognizes a user's specific activities or contributions.

[0663] A "virtual store" is an online platform that provides products and services and conducts commercial transactions over the Internet.

[0664] A "generative AI model" is a type of artificial intelligence model that uses specific algorithms to generate new data or text from existing data.

[0665] A "prompt" is input text given to a generative AI model that influences the generated output.

[0666] This invention is a system that allows consumers to easily evaluate the credibility of product information and reviews in a virtual store. The system is based on existing technology that prevents false information on social networking sites and ensures the reliability of information. The embodiments of the invention will be described from the perspectives of a server, a terminal, and a user.

[0667] server

[0668] The server has the function of regularly updating a database of reliable sources. This database collects and maintains reliable information published by public institutions and experts. The server also analyzes posts using a natural language processing engine (e.g., Spacy) and assigns classification tags. This analysis allows the content of the post to be mechanically understood. Furthermore, posts can be collected in real time through the social media platform's API, allowing constant access to the latest information. The preprocessed text identifies high-risk information and compares it with the reliable source database to assess its reliability. If a discrepancy is found, the server generates a warning message and sends it to the device.

[0669] Terminal

[0670] The terminal is a device that users use to access social networking sites and virtual stores. It includes smartphones, tablets, computers, etc. The terminal has the function of displaying warning messages sent from the server. When a user receives a warning message, they can evaluate its effectiveness and provide feedback. This feedback is sent back to the server, and the collected feedback is used to improve the AI ​​engine's model. If a user's contribution is high, they are awarded a badge, which is reflected in their profile.

[0671] User

[0672] Users use the system to browse product information and reviews in a virtual store. When a user checks a particular product review, a credibility rating is displayed in real time. For example, a review that says "This supplement can completely cure your cold" is checked against a database of trusted sources. If a discrepancy is found, a warning message is displayed, saying, "This information contradicts the official statement. For more information, click here: [Official information link]."

[0673] Examples and prompts

[0674] As a concrete example, consider what happens when a consumer sees a review that says, "This supplement can completely cure your cold." The review is analyzed on the server, and the keywords "cold" and "supplement" are extracted. After a risk assessment, the review is checked against a trusted database, and if any discrepancies are found, a warning message is generated and displayed on the device.

[0675] Prompt Sentence Examples

[0676] Please evaluate the reliability of the following text and generate a warning message if it contradicts official information:

[0677] "This supplement is said to completely cure the common cold."

[0678] This allows the system to prevent the spread of false information and allows consumers to make decisions based on reliable information.

[0679] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0680] Step 1:

[0681] The server periodically updates the database of reliable information sources. Specifically, it collects announcements from public institutions and experts via APIs and stores them in the database. The input is reliable information source data, and the output is the updated database. This ensures that the latest reliable information is always maintained.

[0682] Step 2:

[0683] The server collects posts in real time through the API of the SNS platform. Specifically, it uses the API to obtain new post data (text, poster information, timestamp, etc.). The input is the SNS post data, and the output is the collected post data. This ensures that the latest post information is always available.

[0684] Step 3:

[0685] The server preprocesses the text of the posts collected and analyzes it using a natural language processing engine (such as Spacy). Specifically, it performs text tokenization, morphological analysis, and keyword extraction. The input is the collected text data, and the output is the preprocessed text and extracted keywords. This allows the content of the posts to be mechanically understood.

[0686] Step 4:

[0687] The server analyzes the preprocessed text and identifies high-risk information. Specifically, it applies a risk assessment algorithm based on the extracted keywords to calculate a risk level. The input is the preprocessed text and keywords, and the output is a risk level. This identifies information that may be false.

[0688] Step 5:

[0689] The server checks the identified high-risk information against a database of trusted information sources. Specifically, the check is performed using keyword matching and similarity calculations. The input is the high-risk information and the database of trusted information sources, and the output is the check result (presence or absence of inconsistencies). This confirms whether the information is trustworthy.

[0690] Step 6:

[0691] If a discrepancy is found, the server generates a warning message. Specifically, it creates a message in the format "This information contradicts the official announcement. For more information, click here: [Official information link]." The input is the match result and details of the contradictory information, and the output is a warning message. This allows users to recognize misinformation.

[0692] Step 7:

[0693] The terminal displays a warning message to the user. Specifically, the warning message is displayed in a pop-up format on the page the user is viewing. The input is the warning message sent from the server, and the output is the warning message displayed on the terminal screen. This allows the user to check the reliability of the information in real time.

[0694] Step 8:

[0695] Users evaluate the effectiveness of warning messages and provide feedback. Specifically, they enter their evaluation of effectiveness and comments in a feedback form and send it to the server. The input is the user's feedback, and the output is the feedback data sent to the server. This contributes to improving the accuracy of the system.

[0696] Step 9:

[0697] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. Specifically, the feedback data is used to retrain the model and adjust parameters. The input is the collected feedback data, and the output is an improved AI model. This improves the accuracy and reliability of the system.

[0698] Step 10:

[0699] The server assigns badges to users with high contributions. Specifically, it evaluates the effectiveness of feedback and the level of contribution, and adds the badge to the user's profile. The input is the user's evaluation data, and the output is an updated user profile. This makes the user's contributions visible, improving motivation.

[0700] 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.

[0701] This invention is a system for preventing the spread of false information on social networking sites and ensuring the reliability of information. This system performs highly accurate fact-checking based on data from reliable sources and combines it with an emotion engine that recognizes the user's emotions. Specifically, with the cooperation of the server, device, and user, the system generates appropriate warning messages according to the user's emotional state, improving the accuracy of information and user acceptance.

[0702] System Overview

[0703] This system is configured as follows:

[0704] 1. Server: Maintains a database of trusted sources, analyzes posts using NLP and sentiment engines, assesses the risk of information, and generates warning messages.

[0705] 2. Terminal: A device that allows users to use SNS and displays warning messages and sends feedback.

[0706] 3. Users: contribute to the credibility of information through social media posts and feedback.

[0707] Overview of program processing

[0708] 1. Information gathering and preparation

[0709] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[0710] 2. Collect social media posts

[0711] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[0712] 3. Analysis and evaluation of information

[0713] The server preprocesses the text of collected posts and then analyzes it with an NLP engine, which extracts important keywords and phrases and analyzes them to understand the content of the information, particularly to identify information that is considered high-risk (e.g., medical or political information).

[0714] 4. Emotion analysis

[0715] The server uses an emotion engine to analyze the emotions of users' posts and reactions. This emotion analysis allows us to understand the intention and tone of the posts and evaluate the emotional state of the users.

[0716] 5. Fact Check

[0717] The server checks high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[0718] 6. Generating and Displaying Warning Messages

[0719] The server generates warning messages for posts that contradict reliable information. The content and presentation of these warning messages are tailored to the user's emotional state. For example, if a user is emotionally charged, a message urging them to stay calm is displayed.

[0720] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[0721] 7. Feedback Collection and Analysis

[0722] Users can review warning messages and evaluate their effectiveness. They can provide feedback on whether the warning is correct by rating it as "valid" or "invalid." They can also add text comments.

[0723] The terminal collects feedback from the user and sends it to the server.

[0724] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. The feedback data is used to retrain the model, improving the accuracy of information analysis and warnings. The emotion engine also analyzes the emotional tone of user feedback and uses it to further improve the model.

[0725] 8. Badging

[0726] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[0727] Specific examples

[0728] Example 1: Medical information verification and sentiment analysis

[0729] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[0730] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[0731] 3. The server checks the post against a reliable medical information database.

[0732] 4. If the server finds an official announcement that the drug is ineffective, a warning message will be displayed on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[0733] 5. The server analyzes the user's emotions using an emotion engine and issues a warning in a calm tone.

[0734] Example 2: Political information verification and feedback analysis

[0735] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[0736] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[0737] 3. Users rate the effectiveness of the warning message and provide feedback.

[0738] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model. The emotion engine also analyzes the emotional tone of the feedback to further improve the model.

[0739] This allows the system to effectively detect and prevent false information being spread on social media, while also providing appropriate warning messages based on the user's emotional state.

[0740] The processing flow will be explained below.

[0741] Step 1:

[0742] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[0743] Step 2:

[0744] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[0745] Step 3:

[0746] The server preprocesses the collected text of posts, including normalizing the text data (removing special characters and extra whitespace) and tokenizing it (splitting sentences and words into individual units) to make it analyzable.

[0747] Step 4:

[0748] The server then analyzes the preprocessed text using an NLP engine, extracting key keywords and phrases and understanding the content. This analysis identifies information that is considered particularly high-risk (e.g., medical or political information).

[0749] Step 5:

[0750] The server uses an emotion engine to analyze emotions from users' posts and reactions. The emotion engine incorporates a text analysis algorithm to identify the user's emotional state (e.g., "joy," "anger," "sadness," etc.).

[0751] Step 6:

[0752] The server checks identified high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[0753] Step 7:

[0754] If the server finds any discrepancies based on the comparison results, it generates a warning message. The content and display of this warning message are adjusted according to the user's emotional state. For example, if the user is emotionally charged, a message urging them to stay calm is generated.

[0755] Step 8:

[0756] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[0757] Step 9:

[0758] Users can review warning messages and rate their validity. Users can rate warnings as valid or invalid, and can also add text comments to provide feedback.

[0759] Step 10:

[0760] The terminal collects feedback from the user and sends it to the server.

[0761] Step 11:

[0762] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model. The feedback data is used to retrain the model, improving the accuracy of information analysis and warnings. The emotion engine also analyzes the emotional tone of the user's feedback and uses it to further improve the model.

[0763] Step 12:

[0764] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[0765] As a result, this system can effectively detect and prevent false information from spreading on social media, and provide appropriate warning messages according to the user's emotional state.

[0766] Example 2

[0767] 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."

[0768] Modern online communication platforms face the problem of easily spreading unreliable and false information. This increases the risk of users receiving incorrect information, potentially leading to social confusion and misunderstanding. Furthermore, preventing the spread of misinformation is difficult because appropriate information is not provided based on the user's emotional state. Furthermore, systems have not yet been sufficiently improved based on user feedback, necessitating improved accuracy in detecting and warning about false information.

[0769] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for periodically updating a database of reliable information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of an online platform, means for analyzing preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of reliable information sources, means for evaluating text consistency and inconsistency through keyword matching and similarity calculation, means for performing user sentiment analysis and evaluating the intention and tone of the post, means for generating a warning message and adjusting it according to the user's emotional state, means for displaying the warning message on a user terminal, means for the user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and utilizing it to improve the model of the artificial intelligence engine, and means for awarding badges to highly contributing users. This enables effective detection and prevention of false information spreading on social media and provision of appropriate information taking into account the user's emotional state.

[0770] A "trusted information source database" is a database that collects and stores publicly recognized information such as news sites, academic papers, and official announcements, and ensures the authenticity and reliability of the information.

[0771] A "natural language processing engine" is an artificial intelligence technology that analyzes text data and has functions such as understanding the meaning of language, extracting keywords, and classifying documents.

[0772] An "online platform API" is an interface for communicating with web services and applications and managing the sending and receiving of data.

[0773] "Preprocessed text" refers to text data that has undergone processing such as tokenization, stop word removal, and stemming before being analyzed by a natural language processing engine.

[0774] "High-risk information" is information that may have significant social, economic, or health impacts and should be handled with particular care.

[0775] "Keyword matching" is a technique for evaluating matches between texts, and is a technology that primarily compares based on the presence or absence and frequency of keywords.

[0776] "Similarity calculation" is a technique for evaluating the semantic similarity between texts, and is a comparison technique using mathematical indices such as cosine similarity and Jaccard coefficient.

[0777] "Sentiment analysis" is an artificial intelligence technique that analyzes the emotions and tone contained in text to assess the user's emotional state.

[0778] A "warning message" is a notification message generated to alert the user to detected high-risk information.

[0779] A "user terminal" is a device that allows a user to view information and interact with the device, including a computer, smartphone, tablet, etc.

[0780] "Feedback" refers to evaluation information and comments provided by users, and is data used to improve the system and retrain the model.

[0781] An "artificial intelligence engine" is an artificial intelligence technology that trains models and makes predictions based on feedback data and learning data.

[0782] A "badge" is a digital award given to a user as recognition for their contribution or specific activity within the system.

[0783] The present invention provides a system for preventing the spread of false information on social networking sites and ensuring the reliability of information. This system performs highly accurate fact-checking based on reliable information sources and combines it with an emotion engine that recognizes user emotions. Detailed embodiments are described below.

[0784] System configuration

[0785] This system consists of three elements: a server, a terminal, and a user.

[0786] 1. Server

[0787] The server has many roles and is responsible for the following processes:

[0788] Regularly updated database of reliable sources: The server collects data from news sites, academic papers, official announcements, etc. and updates the database. Specifically, the data collection is done using Python's BeautifulSoup library and the Scrapy framework.

[0789] Analyze collected posts using a natural language processing engine: The server analyzes the collected data using an NLP engine (for example, Google BERT or OpenAI GPT) and assigns classification tags. The collected text data is preprocessed using libraries such as NLTK or spaCy before analysis.

[0790] Collecting posts through APIs of online platforms: The server collects post data in real time from various social media platforms (e.g., Twitter API, Facebook Graph API) and stores it in a database.

[0791] Preprocessing and analysis of submitted text: The server preprocesses the collected text (tokenization, stop word removal, stemming, etc.) and analyzes it using an NLP engine.

[0792] Identifying high-risk information: The server extracts important keywords and phrases and evaluates the agreements and inconsistencies between the texts using keyword matching and similarity calculations (e.g., Cosine Similarity or Jaccard Index).

[0793] Sentiment analysis: The server uses an emotion engine (for example, IBM Watson Tone Analyzer or Microsoft Azure Text Analytics) to assess the user's emotional state and understand the intent and tone of the post.

[0794] Generate warning messages: The server checks high-risk information against trusted sources and generates warning messages if inconsistencies are found, tailoring the messages to the user's emotional state.

[0795] Feedback collection and analysis: The server collects feedback provided by users, analyzes it using an artificial intelligence engine, and uses it to improve the model.

[0796] 2. Terminal

[0797] The terminal is the device through which the user views information and interacts, and is responsible for the following processes:

[0798] Display warning message: The terminal displays the warning message sent from the server to the user. The warning message is displayed as a pop-up window.

[0799] Collecting and sending feedback: The device collects feedback from users (ratings of the effectiveness of warning messages and comments) and sends it to the server.

[0800] 3. Users

[0801] Users play a role in contributing to the credibility of information through social media posts and feedback.

[0802] Review and rate warning messages: Users can review the displayed warning messages and rate their validity as "valid" or "invalid." They can also add text comments.

[0803] Providing feedback: The user provides feedback on the effectiveness of the warning message through the terminal.

[0804] Specific examples

[0805] Example 1: Medical information verification and sentiment analysis

[0806] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[0807] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[0808] 3. The server checks the post against a reliable medical information database.

[0809] 4. If the server finds an official announcement that the drug is ineffective, it will display a warning message on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[0810] 5. The server analyzes the user's emotions using an emotion engine and issues a warning in a calm tone.

[0811] Example 2: Political information verification and feedback analysis

[0812] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[0813] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[0814] 3. Users rate the effectiveness of the warning message and provide feedback.

[0815] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model. The emotion engine also analyzes the emotional tone of the feedback to further improve the model.

[0816] Example prompts for generative AI models

[0817] "Please check whether the COVID-19 treatment information posted on social media is reliable."

[0818] "Compare this political information to see if it matches the official announcement."

[0819] "Generate appropriate warning messages based on the user's emotional state."

[0820] As described above, this system effectively detects false information being spread on social media, prevents its spread, and provides appropriate warning messages according to the user's emotional state.

[0821] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0822] Step 1:

[0823] Data collection from reliable sources

[0824] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.). Specifically, it periodically retrieves data from a specified URL using Python's BeautifulSoup library or the Scrapy framework. The input is the specified URL, and the output is the retrieved text data.

[0825] Step 2:

[0826] Update the database of collected data

[0827] The server stores the collected text data in a database (for example, MongoDB or MySQL). The input is the collected text data, and the output is the updated database. Specifically, the Python script uses the Pandas library to insert the data organized in a data frame into the database.

[0828] Step 3:

[0829] Preprocessing of submitted data

[0830] The server preprocesses the collected data using an NLP engine, performing processes such as tokenization, stop word removal, and stemming. The input is the collected text data, and the output is the preprocessed text data. Specific operations use the NLTK and spaCy libraries.

[0831] Step 4:

[0832] Analysis using natural language processing

[0833] The server inputs the preprocessed text data into an NLP engine (such as Google BERT or OpenAI GPT) to analyze and extract important keywords and phrases. The input is the preprocessed text data, and the output is a list of keywords and phrases resulting from the analysis.

[0834] Step 5:

[0835] Adding classification tags

[0836] The server assigns classification tags (e.g., "medical," "politics," "economy," etc.) to the text data based on the results of analysis by the NLP engine. The input is a list of keywords and phrases from the analysis results, and the output is text data with classification tags.

[0837] Step 6:

[0838] Real-time collection of SNS posts

[0839] The server collects social media content in real time through the API of an online platform (e.g., Twitter API, Facebook Graph API). The input is a specified API endpoint, and the output is the collected post data. Specifically, it sends an HTTP request and obtains the post data as a response.

[0840] Step 7:

[0841] Identifying high-risk information

[0842] The server preprocesses the social media posts collected and analyzes them using an NLP engine. High-risk information (e.g., "medical information" or "political information") is identified and extracted. The input is the preprocessed post data, and the output is the post data identified as high-risk. Specifically, it detects important keywords.

[0843] Step 8:

[0844] Performing sentiment analysis

[0845] The server uses an emotion engine (for example, IBM Watson Tone Analyzer or Microsoft Azure Text Analytics) to analyze emotions from the post content and reactions. The input is the post data identified as high risk, and the output is the emotion analysis results. Specifically, the server sends text data to the emotion engine's API and receives the analysis results.

[0846] Step 9:

[0847] Conducting fact-checks

[0848] The server checks high-risk posts against a database of trusted sources and uses keyword matching or similarity calculations (e.g., Cosine Similarity or Jaccard Index) to check for matches or contradictions. The input is the high-risk post data and the source database, and the output is the fact-check results. Specifically, the server applies a similarity calculation algorithm.

[0849] Step 10:

[0850] Generate a warning message

[0851] The server generates a warning message based on the fact-check results if a contradiction is found. The content and display method of this message are adjusted according to the user's emotional state. The input is the fact-check results and the emotion analysis results, and the output is the generated warning message. Specific operation involves embedding specific information into the template message.

[0852] Step 11:

[0853] Displaying a warning message

[0854] The terminal receives the warning message and displays it to the user. The input is the generated warning message, and the output is the warning message displayed on the terminal screen. As a specific operation, it executes the code that displays the popup window.

[0855] Step 12:

[0856] Collecting feedback

[0857] The user evaluates the effectiveness of the warning message and provides feedback. The input is the displayed warning message, and the output is the feedback from the user. Specifically, the user selects "enabled" or "disabled" and enters a comment.

[0858] Step 13:

[0859] Send Feedback

[0860] The terminal sends the feedback collected from the user to the server. The input is the feedback from the user, and the output is the feedback sent to the server. The specific operation is to send the feedback data using an API.

[0861] Step 14:

[0862] Feedback Analysis

[0863] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model. The input is the feedback sent to the server, and the output is the improved AI model. Specifically, the feedback data is used to retrain the model.

[0864] Step 15:

[0865] Badge Awarding

[0866] The server assigns badges to users who have contributed significantly to effective feedback and fact-checking activities. The input is feedback history, and the output is the assigned badges. Specifically, the server automatically assigns badges to users who meet certain evaluation criteria.

[0867] (Application example 2)

[0868] 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."

[0869] As the degree of freedom in the distribution of information on the Internet and social networking services (SNS) increases, the spread of false information has become a social problem. In particular, false information about high-risk topics (e.g., medical, political, economic, etc.) can have a significant impact. Conventional systems have been required to effectively prevent the spread of such false information while at the same time taking into account the user's emotions, but this has been difficult to achieve. The purpose of this invention is to solve this problem by quickly and accurately detecting false information and providing warning messages that correspond to the user's emotional state.

[0870] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for periodically updating a database of reliable information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of an SNS platform, means for analyzing preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of reliable information sources, means for generating a warning message if a discrepancy is found, means for displaying the warning message on a user terminal, means for the user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine, means for awarding badges to highly contributing users, and means for adjusting the content and display method of the warning message according to the user's emotions and displaying it on the user's device. This prevents the spread of false information and enables the provision of reliable information that takes user emotions into consideration.

[0871] A "database of trusted sources" is a regularly updated database that aggregates data collected from sources such as news sites, academic papers, and official announcements whose credibility has been confirmed.

[0872] A "natural language processing engine" is an algorithm and software that analyzes text data and performs tasks such as extracting keywords, understanding context, and tagging.

[0873] An "SNS platform API" is an application programming interface provided by a social networking service, and is a function that allows external developers to obtain and manipulate data on the SNS.

[0874] "Preprocessed text" is text data that has been converted into a form that is easier to analyze by filtering and cleaning up raw data.

[0875] "High-risk information" refers to information that is likely to contain false information and that falls into categories such as medical, political, and economic information, which may have particularly serious consequences.

[0876] A "warning message" is a message that notifies or alerts the user that there is a doubt about the accuracy or reliability of the information.

[0877] "User terminal" refers to any device used by a user to access SNS, including smartphones, tablets, and PCs.

[0878] "Feedback" refers to the evaluations and comments that users make on the system's warning messages and the information provided.

[0879] "Artificial intelligence engine" is a general term for algorithms and software that learns from collected data and improves models.

[0880] A "badge" is a symbol of recognition and appreciation given to users with specialized knowledge or who have made significant contributions to the system.

[0881] The means for adjusting the content and display method according to "emotion" is a function that generates and displays an appropriate warning message according to the user's current emotional state based on the results of emotion analysis.

[0882] This invention relates to a system for preventing the spread of false information on the Internet and social networking services (SNS) and ensuring the reliability of information. As a specific example, a reliability guide quickly and accurately detects false information and provides warning messages tailored to the user's emotional state. The following describes the components of this system and their functions.

[0883] server

[0884] The server has the following functions:

[0885] 1. Updating the database of reliable sources: Regularly collect data from reliable news sites, academic papers, official announcements, etc. and update the database. This data will be used as reliable data.

[0886] 2. Natural Language Processing Engine (NLP Engine): Analyzes collected posts and assigns classification tags based on their content. This engine extracts important keywords and phrases and identifies high-risk information.

[0887] 3. Real-time collection using SNS platform APIs: Collect user posts from SNS platforms in real time. The posts include text, author information, timestamps, etc.

[0888] 4. Pre-processed text analysis: The collected text data is pre-processed and analyzed by an NLP engine to identify information that is considered high risk.

[0889] 5. Matching high-risk information with reliable data: Identified high-risk information is matched with a database of reliable sources to identify any matches or discrepancies.

[0890] 6. Sentiment analysis engine: Analyzes emotions from user posts and reactions to evaluate the user's emotional state.

[0891] 7. Generating a warning message: If a contradiction is found, a warning message is generated according to the user's emotional state. For example, if the user is emotionally charged, a message is generated to encourage them to calm down.

[0892] 8. AI engine model improvement: Analyze user feedback to retrain the AI ​​engine model and improve the accuracy of warnings.

[0893] Terminal

[0894] The terminal has the following functions:

[0895] 1. Displaying a warning message: Displaying a warning message received from the server to the user. This includes devices such as smart glasses and smartphones.

[0896] 2. Feedback collection: Collect feedback from users about the warning message and send it to the server.

[0897] User

[0898] The user has the following capabilities:

[0899] 1. Posting to SNS: Users post information to SNS and share it with other users.

[0900] 2. Providing Feedback: You can rate the effectiveness of the warning message and provide feedback. You can also add text comments.

[0901] Specific examples

[0902] For example, if a user tries to post information on a social networking site that "a certain drug is effective in treating the new virus," the server performs the following process.

[0903] The NLP engine detects the keywords "new virus" and "treatment drug" and compares them with a reliable medical information database.

[0904] If the database contains an official statement that the drug is ineffective, the server generates a warning message stating, "This drug is said to be ineffective against the new virus. Source: [Official statement link]" and sends it to the device in a calm tone that corresponds to the user's emotional state.

[0905] The terminal displays this message to the user.

[0906] This will prevent the spread of false information and allow for the provision of information that takes into account the emotional state of the user.

[0907] Prompt Sentence Examples

[0908] "Are certain drugs effective in treating the new virus?"

[0909] As described above, the Trust Guide provides warning messages based on the user's emotional state and improves the AI ​​engine through feedback, thereby providing reliable information and preventing the spread of false information.

[0910] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0911] Step 1:

[0912] The server collects data from reliable sources and updates the database. Specifically, it obtains data from news sites, academic papers, official announcements, etc. via API and stores it in the database. This data collection occurs periodically. The input is data from reliable sources, and the output is an updated database.

[0913] Step 2:

[0914] The server uses the API of the SNS platform to collect user posts in real time. The collected data includes text, poster information, timestamps, etc. The input is the SNS data posted in real time, and the output is the collected SNS post data.

[0915] Step 3:

[0916] The server preprocesses the collected text data of SNS posts, performing noise removal, normalization, tokenization, etc. The input is the text portion of the collected SNS post data, and the output is the preprocessed text data.

[0917] Step 4:

[0918] The server analyzes the preprocessed text using a natural language processing (NLP) engine to perform entity extraction, keyword extraction, sentiment analysis, etc. The input is the preprocessed text data, and the output is the analysis results in the form of keywords, sentiment information, and tags for suspicious information.

[0919] Step 5:

[0920] The server identifies high-risk information based on the analysis results. For example, it checks whether keywords such as "new virus" or "treatment drug" are included. The input is the analysis results of the NLP engine, and the output is a flag for high-risk information.

[0921] Step 6:

[0922] The server checks the identified high-risk information against a database of reliable information sources. It checks for matches and inconsistencies to detect high-risk, unreliable information. For example, it checks for matches with an official announcement that "the specified therapeutic drug is ineffective." The input is the flag for high-risk information and the contents of the database, and the output is the matching result.

[0923] Step 7:

[0924] The server generates a warning message based on the matching results if any discrepancies are found. An emotion analysis engine is used to determine the message content according to the user's emotional state. The input is the matching results and the user's emotional state, and the output is the generated warning message.

[0925] Step 8:

[0926] The terminal displays the generated warning message to the user. On devices such as smart glasses or smartphones, the warning message is presented as a popup at the appropriate time. The input is the generated warning message, and the output is the display of the warning message to the user.

[0927] Step 9:

[0928] Users review warning messages, rate their effectiveness, and provide feedback. Users can rate warning messages as "valid" or "invalid" and add comments. The input is user feedback, and the output is the collected feedback data.

[0929] Step 10:

[0930] The terminal collects feedback from the user and sends it to the server, where the input is the collected feedback data and the output is the transmission of the feedback data to the server.

[0931] Step 11:

[0932] The server analyzes the collected feedback data and uses it to improve the artificial intelligence (AI) engine's model. The AI ​​engine retrains the model based on the feedback, improving the accuracy of information analysis and warnings. The input is the feedback data, and the output is an improved AI model.

[0933] Step 12:

[0934] The server assigns badges to users with high contributions. Users who provide reliable feedback are recognized for their contributions by being given badges. The input is the user's feedback history and contribution evaluation, and the output is the assignment of a badge.

[0935] 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.

[0936] 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.

[0937] 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.

[0938] [Third embodiment]

[0939] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0940] 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.

[0941] 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).

[0942] 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.

[0943] 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.

[0944] 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).

[0945] 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.

[0946] 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.

[0947] 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.

[0948] 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.

[0949] 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.

[0950] 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."

[0951] This invention is a system for preventing the spread of false information on social media and ensuring the reliability of information. This system is realized by performing highly accurate fact-checking while referring to reliable sources of information through cooperation between servers, terminals, and users.

[0952] System Overview

[0953] This system is configured as follows:

[0954] 1. Server: Maintains a database of trusted sources, analyzes posts using an NLP engine, assesses the risk of information, and generates warning messages.

[0955] 2. Terminal: A device that allows users to use SNS and displays warning messages and sends feedback.

[0956] 3. Users: contribute to the credibility of information through social media posts and feedback.

[0957] Overview of program processing

[0958] 1. Information gathering and preparation

[0959] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database, then uses an NLP engine to analyze the new information and assign it classification tags.

[0960] 2. Collect social media posts

[0961] The server collects posts in real time through the API of the social media platform. The collected post data includes text, poster information, timestamp, etc.

[0962] 3. Analysis and evaluation of information

[0963] The server preprocesses the collected text of posts and analyzes it with an NLP engine. This analysis extracts key keywords and phrases from the posts and evaluates their risk level based on them.

[0964] 4. Fact Check

[0965] The server checks high-risk posts against a database of trusted sources, using keyword matching and similarity calculations.

[0966] 5. Generating and Displaying Warning Messages

[0967] The server generates a warning message for posts that contradict reliable information. This warning message is displayed on the device, for example, "This information contradicts official statements. Read more here: [link]."

[0968] 6. Feedback Collection and Analysis

[0969] The user can rate the effectiveness of the warning message and provide feedback, which the terminal sends to the server.

[0970] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model.

[0971] 7. Badge Awarding

[0972] The server recognizes users' contributions of expertise and awards badges for useful feedback and fact-checking activities, which are reflected in users' profiles.

[0973] Specific examples

[0974] Example 1: Medical information verification

[0975] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[0976] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[0977] 3. The server checks the post against a reliable medical information database.

[0978] 4. If the server finds an official announcement that the drug is ineffective, a warning message will be displayed on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[0979] Example 2: Verifying political information

[0980] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[0981] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[0982] 3. Users rate the effectiveness of the warning message and provide feedback.

[0983] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model.

[0984] This allows the system to effectively prevent false information from spreading on social media and promote the dissemination and sharing of reliable information.

[0985] The processing flow will be explained below.

[0986] Step 1:

[0987] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[0988] Step 2:

[0989] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[0990] Step 3:

[0991] The server preprocesses the text of the posts collected by normalizing the text data (removing special characters and extra whitespace) and tokenizing it (splitting sentences and words into individual units) to prepare it for analysis.

[0992] Step 4:

[0993] The server then analyzes the preprocessed text with an NLP engine, which extracts important keywords and phrases and analyzes the information to understand its content, particularly to identify information deemed high-risk (e.g., medical or political information).

[0994] Step 5:

[0995] The server checks identified high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[0996] Step 6:

[0997] The server will then generate a warning message if any discrepancies are found, including the specific discrepancy and a link to the authoritative source.

[0998] Step 7:

[0999] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[1000] Step 8:

[1001] Users can review warning messages and evaluate their effectiveness. They can provide feedback on whether the warning is correct by rating it as "valid" or "invalid." They can also add text comments.

[1002] Step 9:

[1003] The terminal collects feedback from the user and sends it to the server.

[1004] Step 10:

[1005] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. The model is retrained based on the feedback data, improving the accuracy of information analysis and warnings.

[1006] Step 11:

[1007] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[1008] The above steps will realize a system that can effectively detect and prevent the spread of false information on social media.

[1009] Example 1

[1010] 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."

[1011] The spread of false information on social media can have a significant impact on society. However, current systems for effectively preventing this and ensuring the reliability of information are insufficient. Automatically identifying high-risk information and quickly generating and displaying warning messages are also major challenges. Furthermore, a mechanism for continuously improving AI models using feedback is needed.

[1012] 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.

[1013] In this invention, the server includes means for periodically updating a database of trusted information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of a social media platform, means for analyzing the preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of trusted information sources, means for generating a warning message if a discrepancy is found, means for displaying the warning message on a user terminal, means for a user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine, means for awarding badges to highly contributing users, means for evaluating the risk level of the content of posts using an NLP engine, and means for retraining the NLP model using user feedback data. This enables rapid identification and prevention of false information, ensures the reliability of information, and enables continuous system improvement using feedback.

[1014] "Credible sources" refer to sources of information that are deemed to have a low risk of misinformation, such as public institutions, academic institutions, and reputable news outlets.

[1015] A "database" is an information management system for efficiently storing and retrieving collected information.

[1016] A "natural language processing engine (NLP engine)" is a computer program that analyzes human language and understands its meaning and intent.

[1017] A "classification tag" refers to a label or category assigned to data based on its characteristics.

[1018] An "SNS platform API" is a programmatic interface provided by a social networking service that allows for further data collection and manipulation.

[1019] "Preprocessed text" refers to text data that has undergone any necessary cleansing processes before analysis, such as removing special characters and stop words.

[1020] "High-risk information" refers to information that is likely to be false and carries the risk of being spread as inaccurate information.

[1021] A "warning message" is a warning message that is displayed when there is doubt about the reliability of information.

[1022] "User terminal" refers to an electronic device for using an SNS application.

[1023] "Feedback" refers to the evaluations and opinions that users provide in response to warning messages in the system.

[1024] An "artificial intelligence engine" is a computer program that analyzes and learns from data.

[1025] "Model improvement" is the process of improving the accuracy and performance of artificial intelligence or machine learning models.

[1026] A "highly contributing user" refers to a user who has made a significant contribution to the system by providing useful information or feedback.

[1027] A "badge" is a symbol or icon awarded for a particular action or contribution.

[1028] "Risk level" is an index that evaluates the likelihood that information is false.

[1029] "NLP model" refers to a machine learning model trained to perform natural language processing tasks.

[1030] "Retraining" is the training process of using new data and feedback to improve the performance of an existing model.

[1031] This invention is a system for preventing the spread of false information on social media and ensuring the reliability of information. This system is realized by performing highly accurate fact-checking while referring to reliable sources of information through the cooperation of servers, terminals, and users.

[1032] First, the server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This data collection can be done using external sources such as news APIs. This data is stored in the server's database and analyzed using a natural language processing engine (NLP engine). The NLP engine is responsible for assigning classification tags to the text data and analyzing its content.

[1033] Next, the server collects posts in real time through the APIs of social media platforms (e.g., Twitter, Facebook). It can be configured to retrieve posts related to specific keywords (e.g., "virus," "vaccine"). The collected post data includes text, author information, timestamps, and other information, and is stored in the server's database.

[1034] The posted data collected by the server is preprocessed to remove unnecessary information and noise. For example, special characters and emojis are removed from the text. This preprocessed text is then analyzed by an NLP engine to extract key keywords and phrases. This allows the risk level of the posted content to be assessed and high-risk information to be identified.

[1035] Posts that are rated high risk are checked by the server against a database of trusted sources, using keyword matching and similarity calculations to see if the content of the post matches information in the database. For example, if a post mentions a particular medication, it matches information in a medical database.

[1036] If the server finds any contradictory information, it generates a warning message that is displayed to the user via the terminal, such as "This information contradicts the official statement. For more information, please click here: [link]."

[1037] Users evaluate the effectiveness of warning messages and provide feedback. This feedback is sent to the server via their device. The server analyzes the feedback and uses it to improve the AI ​​engine's model. For example, it may retrain the NLP model based on the feedback to improve the accuracy of warning messages.

[1038] Furthermore, the server recognizes users' contributions of specialized knowledge and awards badges for effective feedback and fact-checking activities. These badges are displayed on users' profiles and are visible to other users, for example, by displaying a badge icon in the profile section of a social networking app.

[1039] As a concrete example, consider the case where a user posts information on social media that "a certain drug is effective in treating the new virus." The server analyzes the post using an NLP engine and detects the keywords "new virus" and "treatment drug." The server then compares the post with a medical information database, and if it finds an official announcement that the drug is ineffective, it displays a warning message on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[1040] Example prompt sentence:

[1041] "Is this post accurate?"

[1042] "How reliable is this information?"

[1043] Please refer to the official announcement regarding this matter.

[1044] In this way, this system effectively prevents false information from spreading on social media and provides an advanced means to ensure the reliability of information.

[1045] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1046] Step 1: Gather information and prepare

[1047] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.).

[1048] Input: Data from each source.

[1049] Data processing: The server uses news APIs and scraping technology to obtain the necessary data and stores it in a database.

[1050] Output: A stored trusted source database.

[1051] What it does: The server periodically sends API requests to gather new articles and announcements.

[1052] Step 2: Collect social media posts

[1053] The server collects posts in real time through the social media platform's API.

[1054] Input: Post data obtained from SNS (text, poster information, timestamp, etc.).

[1055] Data processing: Converting submitted data into the required format for saving in the database.

[1056] Output: Stored social media post data.

[1057] What it does: The server filters and collects posts that match specific keywords.

[1058] Step 3: Analyze the information

[1059] The server analyzes the preprocessed text and identifies high-risk information.

[1060] Input: Post data stored in the database.

[1061] Data processing: As a preprocessing step, noise is removed from the text (special characters and emojis are removed).

[1062] Output: Preprocessed and clean text data.

[1063] Specific operation: The server uses regular expressions, etc. to remove unnecessary information.

[1064] Step 4: Identify high-risk information

[1065] The server analyzes using an NLP engine to extract key keywords and phrases.

[1066] Input: Preprocessed text data.

[1067] Data calculation: An NLP engine is used to extract keywords and assess risk levels.

[1068] Output: A list of posts containing high-risk information.

[1069] What happens: The server runs an NLP model to match the extracted keywords with known risk information.

[1070] Step 5: Fact Check

[1071] The server checks the identified high-risk information against a database of trusted sources.

[1072] Input: List of posts containing high-risk information, database of trusted sources.

[1073] Data arithmetic: Use keyword matching and similarity calculations to identify information matches.

[1074] Output: A list of information containing contradictions.

[1075] What happens: For each post in the list, the server looks up the relevant information in the database and checks for any discrepancies.

[1076] Step 6: Generate and display warning messages

[1077] The server generates a warning message if it finds any discrepancies.

[1078] Input: A list of information containing conflicts.

[1079] Data calculation: Embed conflict information in warning message template.

[1080] Output: A warning message.

[1081] Specific operation: The server creates a warning message based on the template and sends it to the terminal.

[1082] Step 7: Collect and analyze feedback

[1083] The user rates the effectiveness of the warning message and provides feedback.

[1084] Input: Feedback data from users.

[1085] Data processing: Analyze the collected feedback data and use it to retrain the AI ​​model.

[1086] Output: An improved AI model.

[1087] What happens: The server analyzes the feedback data and retrains the NLP model.

[1088] Step 8: Badging

[1089] The server awards badges to users with high contributions.

[1090] Input: A list of users who have provided valid feedback or fact checks.

[1091] Data processing: Badging and updating user profiles.

[1092] Output: User profile with badges reflected.

[1093] What it does: The server adds badges to the profiles of users who provide helpful feedback.

[1094] (Application example 1)

[1095] 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."

[1096] There is a need to prevent consumer confusion and misunderstanding caused by the spread of false or unreliable information on social media and in virtual stores, and to ensure the reliability of information. Also, in virtual stores where unreliable product reviews and ratings are common, it is necessary to enable consumers to select products based on accurate information.

[1097] 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.

[1098] In this invention, the server includes: means for periodically updating a database of trusted information sources; means for analyzing collected posts using a natural language processing engine and assigning classification tags; means for collecting posts in real time through the API of a social media platform; means for analyzing the preprocessed text and identifying high-risk information; means for comparing the identified high-risk information with the database of trusted information sources; means for generating a warning message if a discrepancy is found; means for displaying the warning message on a user terminal; means for users to evaluate the effectiveness of the warning message and provide feedback; means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine; means for awarding badges to highly contributing users; means for automatically evaluating the credibility of product information and reviews in a virtual store and displaying a warning message; and means for generating a credibility evaluation prompt using a generative AI model. This prevents the spread of false information and enables consumers to make decisions based on reliable information.

[1099] A "server" is a device that processes information, manages databases, and communicates with other devices and platforms on a network.

[1100] A "reliable source" is a medium or database that provides reliable information published by public institutions or experts.

[1101] A "database" is a system that organizes and stores structured information, allowing for efficient searching and updating.

[1102] A "natural language processing engine" is software that uses algorithms and techniques to understand, analyze, and generate human language.

[1103] "Posts" are information such as text and reviews written by users on social media or in virtual stores.

[1104] A "classification tag" is an identification label that is assigned to data or information in order to effectively organize and manage it.

[1105] An "SNS platform" is an online system that provides social networking services.

[1106] "API" is an abbreviation for Application Program Interface, an interface for exchanging functions and data between different software systems.

[1107] "Preprocessed text" refers to text that has been formatted and normalized prior to data analysis and natural language processing.

[1108] "High-risk information" is content that is identified as false or unreliable information.

[1109] "Verification" means comparing data or information to identify matches or discrepancies.

[1110] A "warning message" is a notification to alert the user.

[1111] A "user terminal" is a device that allows a user to connect to the Internet and use various services.

[1112] "Feedback" refers to the evaluations and opinions that users provide about a system or service.

[1113] An "artificial intelligence engine" is a system that uses AI technology to analyze data, learn, and make decisions.

[1114] A "highly contributing user" is a user who provides useful feedback and activity within the system and is recognized.

[1115] A "badge" is a digital insignia that recognizes a user's specific activities or contributions.

[1116] A "virtual store" is an online platform that provides products and services and conducts commercial transactions over the Internet.

[1117] A "generative AI model" is a type of artificial intelligence model that uses specific algorithms to generate new data or text from existing data.

[1118] A "prompt" is input text given to a generative AI model that influences the generated output.

[1119] This invention is a system that allows consumers to easily evaluate the credibility of product information and reviews in a virtual store. The system is based on existing technology that prevents false information on social networking sites and ensures the reliability of information. The embodiments of the invention will be described from the perspectives of a server, a terminal, and a user.

[1120] server

[1121] The server has the function of regularly updating a database of reliable sources. This database collects and maintains reliable information published by public institutions and experts. The server also analyzes posts using a natural language processing engine (e.g., Spacy) and assigns classification tags. This analysis allows the content of the post to be mechanically understood. Furthermore, posts can be collected in real time through the social media platform's API, allowing constant access to the latest information. The preprocessed text identifies high-risk information and compares it with the reliable source database to assess its reliability. If a discrepancy is found, the server generates a warning message and sends it to the device.

[1122] Terminal

[1123] The terminal is a device that users use to access social networking sites and virtual stores. It includes smartphones, tablets, computers, etc. The terminal has the function of displaying warning messages sent from the server. When a user receives a warning message, they can evaluate its effectiveness and provide feedback. This feedback is sent back to the server, and the collected feedback is used to improve the AI ​​engine's model. If a user's contribution is high, they are awarded a badge, which is reflected in their profile.

[1124] User

[1125] Users use the system to browse product information and reviews in a virtual store. When a user checks a particular product review, a credibility rating is displayed in real time. For example, a review that says "This supplement can completely cure your cold" is checked against a database of trusted sources. If a discrepancy is found, a warning message is displayed, saying, "This information contradicts the official statement. For more information, click here: [Official information link]."

[1126] Examples and prompts

[1127] As a concrete example, consider what happens when a consumer sees a review that says, "This supplement can completely cure your cold." The review is analyzed on the server, and the keywords "cold" and "supplement" are extracted. After a risk assessment, the review is checked against a trusted database, and if any discrepancies are found, a warning message is generated and displayed on the device.

[1128] Prompt Sentence Examples

[1129] Please evaluate the reliability of the following text and generate a warning message if it contradicts official information:

[1130] "This supplement is said to completely cure the common cold."

[1131] This allows the system to prevent the spread of false information and allows consumers to make decisions based on reliable information.

[1132] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1133] Step 1:

[1134] The server periodically updates the database of reliable information sources. Specifically, it collects announcements from public institutions and experts via APIs and stores them in the database. The input is reliable information source data, and the output is the updated database. This ensures that the latest reliable information is always maintained.

[1135] Step 2:

[1136] The server collects posts in real time through the API of the SNS platform. Specifically, it uses the API to obtain new post data (text, poster information, timestamp, etc.). The input is the SNS post data, and the output is the collected post data. This ensures that the latest post information is always available.

[1137] Step 3:

[1138] The server preprocesses the text of the posts collected and analyzes it using a natural language processing engine (such as Spacy). Specifically, it performs text tokenization, morphological analysis, and keyword extraction. The input is the collected text data, and the output is the preprocessed text and extracted keywords. This allows the content of the posts to be mechanically understood.

[1139] Step 4:

[1140] The server analyzes the preprocessed text and identifies high-risk information. Specifically, it applies a risk assessment algorithm based on the extracted keywords to calculate a risk level. The input is the preprocessed text and keywords, and the output is a risk level. This identifies information that may be false.

[1141] Step 5:

[1142] The server checks the identified high-risk information against a database of trusted information sources. Specifically, the check is performed using keyword matching and similarity calculations. The input is the high-risk information and the database of trusted information sources, and the output is the check result (presence or absence of inconsistencies). This confirms whether the information is trustworthy.

[1143] Step 6:

[1144] If a discrepancy is found, the server generates a warning message. Specifically, it creates a message in the format "This information contradicts the official announcement. For more information, click here: [Official information link]." The input is the match result and details of the contradictory information, and the output is a warning message. This allows users to recognize misinformation.

[1145] Step 7:

[1146] The terminal displays a warning message to the user. Specifically, the warning message is displayed in a pop-up format on the page the user is viewing. The input is the warning message sent from the server, and the output is the warning message displayed on the terminal screen. This allows the user to check the reliability of the information in real time.

[1147] Step 8:

[1148] Users evaluate the effectiveness of warning messages and provide feedback. Specifically, they enter their evaluation of effectiveness and comments in a feedback form and send it to the server. The input is the user's feedback, and the output is the feedback data sent to the server. This contributes to improving the accuracy of the system.

[1149] Step 9:

[1150] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. Specifically, the feedback data is used to retrain the model and adjust parameters. The input is the collected feedback data, and the output is an improved AI model. This improves the accuracy and reliability of the system.

[1151] Step 10:

[1152] The server assigns badges to users with high contributions. Specifically, it evaluates the effectiveness of feedback and the level of contribution, and adds the badge to the user's profile. The input is the user's evaluation data, and the output is an updated user profile. This makes the user's contributions visible, improving motivation.

[1153] 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.

[1154] This invention is a system for preventing the spread of false information on social networking sites and ensuring the reliability of information. This system performs highly accurate fact-checking based on data from reliable sources and combines it with an emotion engine that recognizes the user's emotions. Specifically, with the cooperation of the server, device, and user, the system generates appropriate warning messages according to the user's emotional state, improving the accuracy of information and user acceptance.

[1155] System Overview

[1156] This system is configured as follows:

[1157] 1. Server: Maintains a database of trusted sources, analyzes posts using NLP and sentiment engines, assesses the risk of information, and generates warning messages.

[1158] 2. Terminal: A device that allows users to use SNS and displays warning messages and sends feedback.

[1159] 3. Users: contribute to the credibility of information through social media posts and feedback.

[1160] Overview of program processing

[1161] 1. Information gathering and preparation

[1162] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[1163] 2. Collect social media posts

[1164] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[1165] 3. Analysis and evaluation of information

[1166] The server preprocesses the text of collected posts and then analyzes it with an NLP engine, which extracts important keywords and phrases and analyzes them to understand the content of the information, particularly to identify information that is considered high-risk (e.g., medical or political information).

[1167] 4. Emotion analysis

[1168] The server uses an emotion engine to analyze the emotions of users' posts and reactions. This emotion analysis allows us to understand the intention and tone of the posts and evaluate the emotional state of the users.

[1169] 5. Fact Check

[1170] The server checks high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[1171] 6. Generating and Displaying Warning Messages

[1172] The server generates warning messages for posts that contradict reliable information. The content and presentation of these warning messages are tailored to the user's emotional state. For example, if a user is emotionally charged, a message urging them to stay calm is displayed.

[1173] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[1174] 7. Feedback Collection and Analysis

[1175] Users can review warning messages and evaluate their effectiveness. They can provide feedback on whether the warning is correct by rating it as "valid" or "invalid." They can also add text comments.

[1176] The terminal collects feedback from the user and sends it to the server.

[1177] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. The feedback data is used to retrain the model, improving the accuracy of information analysis and warnings. The emotion engine also analyzes the emotional tone of user feedback and uses it to further improve the model.

[1178] 8. Badging

[1179] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[1180] Specific examples

[1181] Example 1: Medical information verification and sentiment analysis

[1182] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[1183] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[1184] 3. The server checks the post against a reliable medical information database.

[1185] 4. If the server finds an official announcement that the drug is ineffective, a warning message will be displayed on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[1186] 5. The server analyzes the user's emotions using an emotion engine and issues a warning in a calm tone.

[1187] Example 2: Political information verification and feedback analysis

[1188] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[1189] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[1190] 3. Users rate the effectiveness of the warning message and provide feedback.

[1191] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model. The emotion engine also analyzes the emotional tone of the feedback to further improve the model.

[1192] This allows the system to effectively detect and prevent false information being spread on social media, while also providing appropriate warning messages based on the user's emotional state.

[1193] The processing flow will be explained below.

[1194] Step 1:

[1195] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[1196] Step 2:

[1197] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[1198] Step 3:

[1199] The server preprocesses the collected text of posts, including normalizing the text data (removing special characters and extra whitespace) and tokenizing it (splitting sentences and words into individual units) to make it analyzable.

[1200] Step 4:

[1201] The server then analyzes the preprocessed text using an NLP engine, extracting key keywords and phrases and understanding the content. This analysis identifies information that is considered particularly high-risk (e.g., medical or political information).

[1202] Step 5:

[1203] The server uses an emotion engine to analyze emotions from users' posts and reactions. The emotion engine incorporates a text analysis algorithm to identify the user's emotional state (e.g., "joy," "anger," "sadness," etc.).

[1204] Step 6:

[1205] The server checks identified high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[1206] Step 7:

[1207] If the server finds any discrepancies based on the comparison results, it generates a warning message. The content and display of this warning message are adjusted according to the user's emotional state. For example, if the user is emotionally charged, a message urging them to stay calm is generated.

[1208] Step 8:

[1209] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[1210] Step 9:

[1211] Users can review warning messages and rate their validity. Users can rate warnings as valid or invalid, and can also add text comments to provide feedback.

[1212] Step 10:

[1213] The terminal collects feedback from the user and sends it to the server.

[1214] Step 11:

[1215] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model. The feedback data is used to retrain the model, improving the accuracy of information analysis and warnings. The emotion engine also analyzes the emotional tone of the user's feedback and uses it to further improve the model.

[1216] Step 12:

[1217] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[1218] As a result, this system can effectively detect and prevent false information from spreading on social media, and provide appropriate warning messages according to the user's emotional state.

[1219] Example 2

[1220] 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."

[1221] Modern online communication platforms face the problem of easily spreading unreliable and false information. This increases the risk of users receiving incorrect information, potentially leading to social confusion and misunderstanding. Furthermore, preventing the spread of misinformation is difficult because appropriate information is not provided based on the user's emotional state. Furthermore, systems have not yet been sufficiently improved based on user feedback, necessitating improved accuracy in detecting and warning about false information.

[1222] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for periodically updating a database of reliable information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of an online platform, means for analyzing preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of reliable information sources, means for evaluating text consistency and inconsistency through keyword matching and similarity calculation, means for performing user sentiment analysis and evaluating the intention and tone of the post, means for generating a warning message and adjusting it according to the user's emotional state, means for displaying the warning message on a user terminal, means for the user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and utilizing it to improve the model of the artificial intelligence engine, and means for awarding badges to highly contributing users. This enables effective detection and prevention of false information spreading on social media and provision of appropriate information taking into account the user's emotional state.

[1223] A "trusted information source database" is a database that collects and stores publicly recognized information such as news sites, academic papers, and official announcements, and ensures the authenticity and reliability of the information.

[1224] A "natural language processing engine" is an artificial intelligence technology that analyzes text data and has functions such as understanding the meaning of language, extracting keywords, and classifying documents.

[1225] An "online platform API" is an interface for communicating with web services and applications and managing the sending and receiving of data.

[1226] "Preprocessed text" refers to text data that has undergone processing such as tokenization, stop word removal, and stemming before being analyzed by a natural language processing engine.

[1227] "High-risk information" is information that may have significant social, economic, or health impacts and should be handled with particular care.

[1228] "Keyword matching" is a technique for evaluating matches between texts, and is a technology that primarily compares based on the presence or absence and frequency of keywords.

[1229] "Similarity calculation" is a technique for evaluating the semantic similarity between texts, and is a comparison technique using mathematical indices such as cosine similarity and Jaccard coefficient.

[1230] "Sentiment analysis" is an artificial intelligence technique that analyzes the emotions and tone contained in text to assess the user's emotional state.

[1231] A "warning message" is a notification message generated to alert the user to detected high-risk information.

[1232] A "user terminal" is a device that allows a user to view information and interact with the device, including a computer, smartphone, tablet, etc.

[1233] "Feedback" refers to evaluation information and comments provided by users, and is data used to improve the system and retrain the model.

[1234] An "artificial intelligence engine" is an artificial intelligence technology that trains models and makes predictions based on feedback data and learning data.

[1235] A "badge" is a digital award given to a user as recognition for their contribution or specific activity within the system.

[1236] The present invention provides a system for preventing the spread of false information on social networking sites and ensuring the reliability of information. This system performs highly accurate fact-checking based on reliable information sources and combines it with an emotion engine that recognizes user emotions. Detailed embodiments are described below.

[1237] System configuration

[1238] This system consists of three elements: a server, a terminal, and a user.

[1239] 1. Server

[1240] The server has many roles and is responsible for the following processes:

[1241] Regularly updated database of reliable sources: The server collects data from news sites, academic papers, official announcements, etc. and updates the database. Specifically, the data collection is done using Python's BeautifulSoup library and the Scrapy framework.

[1242] Analyze collected posts using a natural language processing engine: The server analyzes the collected data using an NLP engine (for example, Google BERT or OpenAI GPT) and assigns classification tags. The collected text data is preprocessed using libraries such as NLTK or spaCy before analysis.

[1243] Collecting posts through APIs of online platforms: The server collects post data in real time from various social media platforms (e.g., Twitter API, Facebook Graph API) and stores it in a database.

[1244] Preprocessing and analysis of submitted text: The server preprocesses the collected text (tokenization, stop word removal, stemming, etc.) and analyzes it using an NLP engine.

[1245] Identifying high-risk information: The server extracts important keywords and phrases and evaluates the agreements and inconsistencies between the texts using keyword matching and similarity calculations (e.g., Cosine Similarity or Jaccard Index).

[1246] Sentiment analysis: The server uses an emotion engine (for example, IBM Watson Tone Analyzer or Microsoft Azure Text Analytics) to assess the user's emotional state and understand the intent and tone of the post.

[1247] Generate warning messages: The server checks high-risk information against trusted sources and generates warning messages if inconsistencies are found, tailoring the messages to the user's emotional state.

[1248] Feedback collection and analysis: The server collects feedback provided by users, analyzes it using an artificial intelligence engine, and uses it to improve the model.

[1249] 2. Terminal

[1250] The terminal is the device through which the user views information and interacts, and is responsible for the following processes:

[1251] Display warning message: The terminal displays the warning message sent from the server to the user. The warning message is displayed as a pop-up window.

[1252] Collecting and sending feedback: The device collects feedback from users (ratings of the effectiveness of warning messages and comments) and sends it to the server.

[1253] 3. Users

[1254] Users play a role in contributing to the credibility of information through social media posts and feedback.

[1255] Review and rate warning messages: Users can review the displayed warning messages and rate their validity as "valid" or "invalid." They can also add text comments.

[1256] Providing feedback: The user provides feedback on the effectiveness of the warning message through the terminal.

[1257] Specific examples

[1258] Example 1: Medical information verification and sentiment analysis

[1259] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[1260] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[1261] 3. The server checks the post against a reliable medical information database.

[1262] 4. If the server finds an official announcement that the drug is ineffective, it will display a warning message on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[1263] 5. The server analyzes the user's emotions using an emotion engine and issues a warning in a calm tone.

[1264] Example 2: Political information verification and feedback analysis

[1265] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[1266] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[1267] 3. Users rate the effectiveness of the warning message and provide feedback.

[1268] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model. The emotion engine also analyzes the emotional tone of the feedback to further improve the model.

[1269] Example prompts for generative AI models

[1270] "Please check whether the COVID-19 treatment information posted on social media is reliable."

[1271] "Compare this political information to see if it matches the official announcement."

[1272] "Generate appropriate warning messages based on the user's emotional state."

[1273] As described above, this system effectively detects false information being spread on social media, prevents its spread, and provides appropriate warning messages according to the user's emotional state.

[1274] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1275] Step 1:

[1276] Data collection from reliable sources

[1277] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.). Specifically, it periodically retrieves data from a specified URL using Python's BeautifulSoup library or the Scrapy framework. The input is the specified URL, and the output is the retrieved text data.

[1278] Step 2:

[1279] Update the database of collected data

[1280] The server stores the collected text data in a database (for example, MongoDB or MySQL). The input is the collected text data, and the output is the updated database. Specifically, the Python script uses the Pandas library to insert the data organized in a data frame into the database.

[1281] Step 3:

[1282] Preprocessing of submitted data

[1283] The server preprocesses the collected data using an NLP engine, performing processes such as tokenization, stop word removal, and stemming. The input is the collected text data, and the output is the preprocessed text data. Specific operations use the NLTK and spaCy libraries.

[1284] Step 4:

[1285] Analysis using natural language processing

[1286] The server inputs the preprocessed text data into an NLP engine (such as Google BERT or OpenAI GPT) to analyze and extract important keywords and phrases. The input is the preprocessed text data, and the output is a list of keywords and phrases resulting from the analysis.

[1287] Step 5:

[1288] Adding classification tags

[1289] The server assigns classification tags (e.g., "medical," "politics," "economy," etc.) to the text data based on the results of analysis by the NLP engine. The input is a list of keywords and phrases from the analysis results, and the output is text data with classification tags.

[1290] Step 6:

[1291] Real-time collection of SNS posts

[1292] The server collects social media content in real time through the API of an online platform (e.g., Twitter API, Facebook Graph API). The input is a specified API endpoint, and the output is the collected post data. Specifically, it sends an HTTP request and obtains the post data as a response.

[1293] Step 7:

[1294] Identifying high-risk information

[1295] The server preprocesses the social media posts collected and analyzes them using an NLP engine. High-risk information (e.g., "medical information" or "political information") is identified and extracted. The input is the preprocessed post data, and the output is the post data identified as high-risk. Specifically, it detects important keywords.

[1296] Step 8:

[1297] Performing sentiment analysis

[1298] The server uses an emotion engine (for example, IBM Watson Tone Analyzer or Microsoft Azure Text Analytics) to analyze emotions from the post content and reactions. The input is the post data identified as high risk, and the output is the emotion analysis results. Specifically, the server sends text data to the emotion engine's API and receives the analysis results.

[1299] Step 9:

[1300] Conducting fact-checks

[1301] The server checks high-risk posts against a database of trusted sources and uses keyword matching or similarity calculations (e.g., Cosine Similarity or Jaccard Index) to check for matches or contradictions. The input is the high-risk post data and the source database, and the output is the fact-check results. Specifically, the server applies a similarity calculation algorithm.

[1302] Step 10:

[1303] Generate a warning message

[1304] The server generates a warning message based on the fact-check results if a contradiction is found. The content and display method of this message are adjusted according to the user's emotional state. The input is the fact-check results and the emotion analysis results, and the output is the generated warning message. Specific operation involves embedding specific information into the template message.

[1305] Step 11:

[1306] Displaying a warning message

[1307] The terminal receives the warning message and displays it to the user. The input is the generated warning message, and the output is the warning message displayed on the terminal screen. As a specific operation, it executes the code that displays the popup window.

[1308] Step 12:

[1309] Collecting feedback

[1310] The user evaluates the effectiveness of the warning message and provides feedback. The input is the displayed warning message, and the output is the feedback from the user. Specifically, the user selects "enabled" or "disabled" and enters a comment.

[1311] Step 13:

[1312] Send Feedback

[1313] The terminal sends the feedback collected from the user to the server. The input is the feedback from the user, and the output is the feedback sent to the server. The specific operation is to send the feedback data using an API.

[1314] Step 14:

[1315] Feedback Analysis

[1316] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model. The input is the feedback sent to the server, and the output is the improved AI model. Specifically, the feedback data is used to retrain the model.

[1317] Step 15:

[1318] Badge Awarding

[1319] The server assigns badges to users who have contributed significantly to effective feedback and fact-checking activities. The input is feedback history, and the output is the assigned badges. Specifically, the server automatically assigns badges to users who meet certain evaluation criteria.

[1320] (Application example 2)

[1321] 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."

[1322] As the degree of freedom in the distribution of information on the Internet and social networking services (SNS) increases, the spread of false information has become a social problem. In particular, false information about high-risk topics (e.g., medical, political, economic, etc.) can have a significant impact. Conventional systems have been required to effectively prevent the spread of such false information while at the same time taking into account the user's emotions, but this has been difficult to achieve. The purpose of this invention is to solve this problem by quickly and accurately detecting false information and providing warning messages that correspond to the user's emotional state.

[1323] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for periodically updating a database of reliable information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of an SNS platform, means for analyzing preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of reliable information sources, means for generating a warning message if a discrepancy is found, means for displaying the warning message on a user terminal, means for the user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine, means for awarding badges to highly contributing users, and means for adjusting the content and display method of the warning message according to the user's emotions and displaying it on the user's device. This prevents the spread of false information and enables the provision of reliable information that takes user emotions into consideration.

[1324] A "database of trusted sources" is a regularly updated database that aggregates data collected from sources such as news sites, academic papers, and official announcements whose credibility has been confirmed.

[1325] A "natural language processing engine" is an algorithm and software that analyzes text data and performs tasks such as extracting keywords, understanding context, and tagging.

[1326] An "SNS platform API" is an application programming interface provided by a social networking service, and is a function that allows external developers to obtain and manipulate data on the SNS.

[1327] "Preprocessed text" is text data that has been converted into a form that is easier to analyze by filtering and cleaning up raw data.

[1328] "High-risk information" refers to information that is likely to contain false information and that falls into categories such as medical, political, and economic information, which may have particularly serious consequences.

[1329] A "warning message" is a message that notifies or alerts the user that there is a doubt about the accuracy or reliability of the information.

[1330] "User terminal" refers to any device used by a user to access SNS, including smartphones, tablets, and PCs.

[1331] "Feedback" refers to the evaluations and comments that users make on the system's warning messages and the information provided.

[1332] "Artificial intelligence engine" is a general term for algorithms and software that learns from collected data and improves models.

[1333] A "badge" is a symbol of recognition and appreciation given to users with specialized knowledge or who have made significant contributions to the system.

[1334] The means for adjusting the content and display method according to "emotion" is a function that generates and displays an appropriate warning message according to the user's current emotional state based on the results of emotion analysis.

[1335] This invention relates to a system for preventing the spread of false information on the Internet and social networking services (SNS) and ensuring the reliability of information. As a specific example, a reliability guide quickly and accurately detects false information and provides warning messages tailored to the user's emotional state. The following describes the components of this system and their functions.

[1336] server

[1337] The server has the following functions:

[1338] 1. Updating the database of reliable sources: Regularly collect data from reliable news sites, academic papers, official announcements, etc. and update the database. This data will be used as reliable data.

[1339] 2. Natural Language Processing Engine (NLP Engine): Analyzes collected posts and assigns classification tags based on their content. This engine extracts important keywords and phrases and identifies high-risk information.

[1340] 3. Real-time collection using SNS platform APIs: Collect user posts from SNS platforms in real time. The posts include text, author information, timestamps, etc.

[1341] 4. Pre-processed text analysis: The collected text data is pre-processed and analyzed by an NLP engine to identify information that is considered high risk.

[1342] 5. Matching high-risk information with reliable data: Identified high-risk information is matched with a database of reliable sources to identify any matches or discrepancies.

[1343] 6. Sentiment analysis engine: Analyzes emotions from user posts and reactions to evaluate the user's emotional state.

[1344] 7. Generating a warning message: If a contradiction is found, a warning message is generated according to the user's emotional state. For example, if the user is emotionally charged, a message is generated to encourage them to calm down.

[1345] 8. AI engine model improvement: Analyze user feedback to retrain the AI ​​engine model and improve the accuracy of warnings.

[1346] Terminal

[1347] The terminal has the following functions:

[1348] 1. Displaying a warning message: Displaying a warning message received from the server to the user. This includes devices such as smart glasses and smartphones.

[1349] 2. Feedback collection: Collect feedback from users about the warning message and send it to the server.

[1350] User

[1351] The user has the following capabilities:

[1352] 1. Posting to SNS: Users post information to SNS and share it with other users.

[1353] 2. Providing Feedback: You can rate the effectiveness of the warning message and provide feedback. You can also add text comments.

[1354] Specific examples

[1355] For example, if a user tries to post information on a social networking site that "a certain drug is effective in treating the new virus," the server performs the following process.

[1356] The NLP engine detects the keywords "new virus" and "treatment drug" and compares them with a reliable medical information database.

[1357] If the database contains an official statement that the drug is ineffective, the server generates a warning message stating, "This drug is said to be ineffective against the new virus. Source: [Official statement link]" and sends it to the device in a calm tone that corresponds to the user's emotional state.

[1358] The terminal displays this message to the user.

[1359] This will prevent the spread of false information and allow for the provision of information that takes into account the emotional state of the user.

[1360] Prompt Sentence Examples

[1361] "Are certain drugs effective in treating the new virus?"

[1362] As described above, the Trust Guide provides warning messages based on the user's emotional state and improves the AI ​​engine through feedback, thereby providing reliable information and preventing the spread of false information.

[1363] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1364] Step 1:

[1365] The server collects data from reliable sources and updates the database. Specifically, it obtains data from news sites, academic papers, official announcements, etc. via API and stores it in the database. This data collection occurs periodically. The input is data from reliable sources, and the output is an updated database.

[1366] Step 2:

[1367] The server uses the API of the SNS platform to collect user posts in real time. The collected data includes text, poster information, timestamps, etc. The input is the SNS data posted in real time, and the output is the collected SNS post data.

[1368] Step 3:

[1369] The server preprocesses the collected text data of SNS posts, performing noise removal, normalization, tokenization, etc. The input is the text portion of the collected SNS post data, and the output is the preprocessed text data.

[1370] Step 4:

[1371] The server analyzes the preprocessed text using a natural language processing (NLP) engine to perform entity extraction, keyword extraction, sentiment analysis, etc. The input is the preprocessed text data, and the output is the analysis results in the form of keywords, sentiment information, and tags for suspicious information.

[1372] Step 5:

[1373] The server identifies high-risk information based on the analysis results. For example, it checks whether keywords such as "new virus" or "treatment drug" are included. The input is the analysis results of the NLP engine, and the output is a flag for high-risk information.

[1374] Step 6:

[1375] The server checks the identified high-risk information against a database of reliable information sources. It checks for matches and inconsistencies to detect high-risk, unreliable information. For example, it checks for matches with an official announcement that "the specified therapeutic drug is ineffective." The input is the flag for high-risk information and the contents of the database, and the output is the matching result.

[1376] Step 7:

[1377] The server generates a warning message based on the matching results if any discrepancies are found. An emotion analysis engine is used to determine the message content according to the user's emotional state. The input is the matching results and the user's emotional state, and the output is the generated warning message.

[1378] Step 8:

[1379] The terminal displays the generated warning message to the user. On devices such as smart glasses or smartphones, the warning message is presented as a popup at the appropriate time. The input is the generated warning message, and the output is the display of the warning message to the user.

[1380] Step 9:

[1381] Users review warning messages, rate their effectiveness, and provide feedback. Users can rate warning messages as "valid" or "invalid" and add comments. The input is user feedback, and the output is the collected feedback data.

[1382] Step 10:

[1383] The terminal collects feedback from the user and sends it to the server, where the input is the collected feedback data and the output is the transmission of the feedback data to the server.

[1384] Step 11:

[1385] The server analyzes the collected feedback data and uses it to improve the artificial intelligence (AI) engine's model. The AI ​​engine retrains the model based on the feedback, improving the accuracy of information analysis and warnings. The input is the feedback data, and the output is an improved AI model.

[1386] Step 12:

[1387] The server assigns badges to users with high contributions. Users who provide reliable feedback are recognized for their contributions by being given badges. The input is the user's feedback history and contribution evaluation, and the output is the assignment of a badge.

[1388] 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.

[1389] 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.

[1390] 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.

[1391] [Fourth embodiment]

[1392] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1393] 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.

[1394] 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).

[1395] 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.

[1396] 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.

[1397] 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).

[1398] 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.

[1399] 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.

[1400] 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.

[1401] 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.

[1402] 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.

[1403] 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.

[1404] 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."

[1405] This invention is a system for preventing the spread of false information on social media and ensuring the reliability of information. This system is realized by performing highly accurate fact-checking while referring to reliable sources of information through cooperation between servers, terminals, and users.

[1406] System Overview

[1407] This system is configured as follows:

[1408] 1. Server: Maintains a database of trusted sources, analyzes posts using an NLP engine, assesses the risk of information, and generates warning messages.

[1409] 2. Terminal: A device that allows users to use SNS and displays warning messages and sends feedback.

[1410] 3. Users: contribute to the credibility of information through social media posts and feedback.

[1411] Overview of program processing

[1412] 1. Information gathering and preparation

[1413] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database, then uses an NLP engine to analyze the new information and assign it classification tags.

[1414] 2. Collect social media posts

[1415] The server collects posts in real time through the API of the social media platform. The collected post data includes text, poster information, timestamp, etc.

[1416] 3. Analysis and evaluation of information

[1417] The server preprocesses the collected text of posts and analyzes it with an NLP engine. This analysis extracts key keywords and phrases from the posts and evaluates their risk level based on them.

[1418] 4. Fact Check

[1419] The server checks high-risk posts against a database of trusted sources, using keyword matching and similarity calculations.

[1420] 5. Generating and Displaying Warning Messages

[1421] The server generates a warning message for posts that contradict reliable information. This warning message is displayed on the device, for example, "This information contradicts official statements. Read more here: [link]."

[1422] 6. Feedback Collection and Analysis

[1423] The user can rate the effectiveness of the warning message and provide feedback, which the terminal sends to the server.

[1424] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model.

[1425] 7. Badge Awarding

[1426] The server recognizes users' contributions of expertise and awards badges for useful feedback and fact-checking activities, which are reflected in users' profiles.

[1427] Specific examples

[1428] Example 1: Medical information verification

[1429] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[1430] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[1431] 3. The server checks the post against a reliable medical information database.

[1432] 4. If the server finds an official announcement that the drug is ineffective, a warning message will be displayed on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[1433] Example 2: Verifying political information

[1434] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[1435] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[1436] 3. Users rate the effectiveness of the warning message and provide feedback.

[1437] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model.

[1438] This allows the system to effectively prevent false information from spreading on social media and promote the dissemination and sharing of reliable information.

[1439] The processing flow will be explained below.

[1440] Step 1:

[1441] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[1442] Step 2:

[1443] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[1444] Step 3:

[1445] The server preprocesses the text of the posts collected by normalizing the text data (removing special characters and extra whitespace) and tokenizing it (splitting sentences and words into individual units) to prepare it for analysis.

[1446] Step 4:

[1447] The server then analyzes the preprocessed text with an NLP engine, which extracts important keywords and phrases and analyzes the information to understand its content, particularly to identify information deemed high-risk (e.g., medical or political information).

[1448] Step 5:

[1449] The server checks identified high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[1450] Step 6:

[1451] The server will then generate a warning message if any discrepancies are found, including the specific discrepancy and a link to the authoritative source.

[1452] Step 7:

[1453] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[1454] Step 8:

[1455] Users can review warning messages and evaluate their effectiveness. They can provide feedback on whether the warning is correct by rating it as "valid" or "invalid." They can also add text comments.

[1456] Step 9:

[1457] The terminal collects feedback from the user and sends it to the server.

[1458] Step 10:

[1459] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. The model is retrained based on the feedback data, improving the accuracy of information analysis and warnings.

[1460] Step 11:

[1461] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[1462] The above steps will realize a system that can effectively detect and prevent the spread of false information on social media.

[1463] Example 1

[1464] 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."

[1465] The spread of false information on social media can have a significant impact on society. However, current systems for effectively preventing this and ensuring the reliability of information are insufficient. Automatically identifying high-risk information and quickly generating and displaying warning messages are also major challenges. Furthermore, a mechanism for continuously improving AI models using feedback is needed.

[1466] 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.

[1467] In this invention, the server includes means for periodically updating a database of trusted information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of a social media platform, means for analyzing the preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of trusted information sources, means for generating a warning message if a discrepancy is found, means for displaying the warning message on a user terminal, means for a user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine, means for awarding badges to highly contributing users, means for evaluating the risk level of the content of posts using an NLP engine, and means for retraining the NLP model using user feedback data. This enables rapid identification and prevention of false information, ensures the reliability of information, and enables continuous system improvement using feedback.

[1468] "Credible sources" refer to sources of information that are deemed to have a low risk of misinformation, such as public institutions, academic institutions, and reputable news outlets.

[1469] A "database" is an information management system for efficiently storing and retrieving collected information.

[1470] A "natural language processing engine (NLP engine)" is a computer program that analyzes human language and understands its meaning and intent.

[1471] A "classification tag" refers to a label or category assigned to data based on its characteristics.

[1472] An "SNS platform API" is a programmatic interface provided by a social networking service that allows for further data collection and manipulation.

[1473] "Preprocessed text" refers to text data that has undergone any necessary cleansing processes before analysis, such as removing special characters and stop words.

[1474] "High-risk information" refers to information that is likely to be false and carries the risk of being spread as inaccurate information.

[1475] A "warning message" is a warning message that is displayed when there is doubt about the reliability of information.

[1476] "User terminal" refers to an electronic device for using an SNS application.

[1477] "Feedback" refers to the evaluations and opinions that users provide in response to warning messages in the system.

[1478] An "artificial intelligence engine" is a computer program that analyzes and learns from data.

[1479] "Model improvement" is the process of improving the accuracy and performance of artificial intelligence or machine learning models.

[1480] A "highly contributing user" refers to a user who has made a significant contribution to the system by providing useful information or feedback.

[1481] A "badge" is a symbol or icon awarded for a particular action or contribution.

[1482] "Risk level" is an index that evaluates the likelihood that information is false.

[1483] "NLP model" refers to a machine learning model trained to perform natural language processing tasks.

[1484] "Retraining" is the training process of using new data and feedback to improve the performance of an existing model.

[1485] This invention is a system for preventing the spread of false information on social media and ensuring the reliability of information. This system is realized by performing highly accurate fact-checking while referring to reliable sources of information through the cooperation of servers, terminals, and users.

[1486] First, the server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This data collection can be done using external sources such as news APIs. This data is stored in the server's database and analyzed using a natural language processing engine (NLP engine). The NLP engine is responsible for assigning classification tags to the text data and analyzing its content.

[1487] Next, the server collects posts in real time through the APIs of social media platforms (e.g., Twitter, Facebook). It can be configured to retrieve posts related to specific keywords (e.g., "virus," "vaccine"). The collected post data includes text, author information, timestamps, and other information, and is stored in the server's database.

[1488] The posted data collected by the server is preprocessed to remove unnecessary information and noise. For example, special characters and emojis are removed from the text. This preprocessed text is then analyzed by an NLP engine to extract key keywords and phrases. This allows the risk level of the posted content to be assessed and high-risk information to be identified.

[1489] Posts that are rated high risk are checked by the server against a database of trusted sources, using keyword matching and similarity calculations to see if the content of the post matches information in the database. For example, if a post mentions a particular medication, it matches information in a medical database.

[1490] If the server finds any contradictory information, it generates a warning message that is displayed to the user via the terminal, such as "This information contradicts the official statement. For more information, please click here: [link]."

[1491] Users evaluate the effectiveness of warning messages and provide feedback. This feedback is sent to the server via their device. The server analyzes the feedback and uses it to improve the AI ​​engine's model. For example, it may retrain the NLP model based on the feedback to improve the accuracy of warning messages.

[1492] Furthermore, the server recognizes users' contributions of specialized knowledge and awards badges for effective feedback and fact-checking activities. These badges are displayed on users' profiles and are visible to other users, for example, by displaying a badge icon in the profile section of a social networking app.

[1493] As a concrete example, consider the case where a user posts information on social media that "a certain drug is effective in treating the new virus." The server analyzes the post using an NLP engine and detects the keywords "new virus" and "treatment drug." The server then compares the post with a medical information database, and if it finds an official announcement that the drug is ineffective, it displays a warning message on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[1494] Example prompt sentence:

[1495] "Is this post accurate?"

[1496] "How reliable is this information?"

[1497] Please refer to the official announcement regarding this matter.

[1498] In this way, this system effectively prevents false information from spreading on social media and provides an advanced means to ensure the reliability of information.

[1499] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1500] Step 1: Gather information and prepare

[1501] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.).

[1502] Input: Data from each source.

[1503] Data processing: The server uses news APIs and scraping technology to obtain the necessary data and stores it in a database.

[1504] Output: A stored trusted source database.

[1505] What it does: The server periodically sends API requests to gather new articles and announcements.

[1506] Step 2: Collect social media posts

[1507] The server collects posts in real time through the social media platform's API.

[1508] Input: Post data obtained from SNS (text, poster information, timestamp, etc.).

[1509] Data processing: Converting submitted data into the required format for saving in the database.

[1510] Output: Stored social media post data.

[1511] What it does: The server filters and collects posts that match specific keywords.

[1512] Step 3: Analyze the information

[1513] The server analyzes the preprocessed text and identifies high-risk information.

[1514] Input: Post data stored in the database.

[1515] Data processing: As a preprocessing step, noise is removed from the text (special characters and emojis are removed).

[1516] Output: Preprocessed and clean text data.

[1517] Specific operation: The server uses regular expressions, etc. to remove unnecessary information.

[1518] Step 4: Identify high-risk information

[1519] The server analyzes using an NLP engine to extract key keywords and phrases.

[1520] Input: Preprocessed text data.

[1521] Data calculation: An NLP engine is used to extract keywords and assess risk levels.

[1522] Output: A list of posts containing high-risk information.

[1523] What happens: The server runs an NLP model to match the extracted keywords with known risk information.

[1524] Step 5: Fact Check

[1525] The server checks the identified high-risk information against a database of trusted sources.

[1526] Input: List of posts containing high-risk information, database of trusted sources.

[1527] Data arithmetic: Use keyword matching and similarity calculations to identify information matches.

[1528] Output: A list of information containing contradictions.

[1529] What happens: For each post in the list, the server looks up the relevant information in the database and checks for any discrepancies.

[1530] Step 6: Generate and display warning messages

[1531] The server generates a warning message if it finds any discrepancies.

[1532] Input: A list of information containing conflicts.

[1533] Data calculation: Embed conflict information in warning message template.

[1534] Output: A warning message.

[1535] Specific operation: The server creates a warning message based on the template and sends it to the terminal.

[1536] Step 7: Collect and analyze feedback

[1537] The user rates the effectiveness of the warning message and provides feedback.

[1538] Input: Feedback data from users.

[1539] Data processing: Analyze the collected feedback data and use it to retrain the AI ​​model.

[1540] Output: An improved AI model.

[1541] What happens: The server analyzes the feedback data and retrains the NLP model.

[1542] Step 8: Badging

[1543] The server awards badges to users with high contributions.

[1544] Input: A list of users who have provided valid feedback or fact checks.

[1545] Data processing: Badging and updating user profiles.

[1546] Output: User profile with badges reflected.

[1547] What it does: The server adds badges to the profiles of users who provide helpful feedback.

[1548] (Application example 1)

[1549] 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."

[1550] There is a need to prevent consumer confusion and misunderstanding caused by the spread of false or unreliable information on social media and in virtual stores, and to ensure the reliability of information. Also, in virtual stores where unreliable product reviews and ratings are common, it is necessary to enable consumers to select products based on accurate information.

[1551] 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.

[1552] In this invention, the server includes: means for periodically updating a database of trusted information sources; means for analyzing collected posts using a natural language processing engine and assigning classification tags; means for collecting posts in real time through the API of a social media platform; means for analyzing the preprocessed text and identifying high-risk information; means for comparing the identified high-risk information with the database of trusted information sources; means for generating a warning message if a discrepancy is found; means for displaying the warning message on a user terminal; means for users to evaluate the effectiveness of the warning message and provide feedback; means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine; means for awarding badges to highly contributing users; means for automatically evaluating the credibility of product information and reviews in a virtual store and displaying a warning message; and means for generating a credibility evaluation prompt using a generative AI model. This prevents the spread of false information and enables consumers to make decisions based on reliable information.

[1553] A "server" is a device that processes information, manages databases, and communicates with other devices and platforms on a network.

[1554] A "reliable source" is a medium or database that provides reliable information published by public institutions or experts.

[1555] A "database" is a system that organizes and stores structured information, allowing for efficient searching and updating.

[1556] A "natural language processing engine" is software that uses algorithms and techniques to understand, analyze, and generate human language.

[1557] "Posts" are information such as text and reviews written by users on social media or in virtual stores.

[1558] A "classification tag" is an identification label that is assigned to data or information in order to effectively organize and manage it.

[1559] An "SNS platform" is an online system that provides social networking services.

[1560] "API" is an abbreviation for Application Program Interface, an interface for exchanging functions and data between different software systems.

[1561] "Preprocessed text" refers to text that has been formatted and normalized prior to data analysis and natural language processing.

[1562] "High-risk information" is content that is identified as false or unreliable information.

[1563] "Verification" means comparing data or information to identify matches or discrepancies.

[1564] A "warning message" is a notification to alert the user.

[1565] A "user terminal" is a device that allows a user to connect to the Internet and use various services.

[1566] "Feedback" refers to the evaluations and opinions that users provide about a system or service.

[1567] An "artificial intelligence engine" is a system that uses AI technology to analyze data, learn, and make decisions.

[1568] A "highly contributing user" is a user who provides useful feedback and activity within the system and is recognized.

[1569] A "badge" is a digital insignia that recognizes a user's specific activities or contributions.

[1570] A "virtual store" is an online platform that provides products and services and conducts commercial transactions over the Internet.

[1571] A "generative AI model" is a type of artificial intelligence model that uses specific algorithms to generate new data or text from existing data.

[1572] A "prompt" is input text given to a generative AI model that influences the generated output.

[1573] This invention is a system that allows consumers to easily evaluate the credibility of product information and reviews in a virtual store. The system is based on existing technology that prevents false information on social networking sites and ensures the reliability of information. The embodiments of the invention will be described from the perspectives of a server, a terminal, and a user.

[1574] server

[1575] The server has the function of regularly updating a database of reliable sources. This database collects and maintains reliable information published by public institutions and experts. The server also analyzes posts using a natural language processing engine (e.g., Spacy) and assigns classification tags. This analysis allows the content of the post to be mechanically understood. Furthermore, posts can be collected in real time through the social media platform's API, allowing constant access to the latest information. The preprocessed text identifies high-risk information and compares it with the reliable source database to assess its reliability. If a discrepancy is found, the server generates a warning message and sends it to the device.

[1576] Terminal

[1577] The terminal is a device that users use to access social networking sites and virtual stores. It includes smartphones, tablets, computers, etc. The terminal has the function of displaying warning messages sent from the server. When a user receives a warning message, they can evaluate its effectiveness and provide feedback. This feedback is sent back to the server, and the collected feedback is used to improve the AI ​​engine's model. If a user's contribution is high, they are awarded a badge, which is reflected in their profile.

[1578] User

[1579] Users use the system to browse product information and reviews in a virtual store. When a user checks a particular product review, a credibility rating is displayed in real time. For example, a review that says "This supplement can completely cure your cold" is checked against a database of trusted sources. If a discrepancy is found, a warning message is displayed, saying, "This information contradicts the official statement. For more information, click here: [Official information link]."

[1580] Examples and prompts

[1581] As a concrete example, consider what happens when a consumer sees a review that says, "This supplement can completely cure your cold." The review is analyzed on the server, and the keywords "cold" and "supplement" are extracted. After a risk assessment, the review is checked against a trusted database, and if any discrepancies are found, a warning message is generated and displayed on the device.

[1582] Prompt Sentence Examples

[1583] Please evaluate the reliability of the following text and generate a warning message if it contradicts official information:

[1584] "This supplement is said to completely cure the common cold."

[1585] This allows the system to prevent the spread of false information and allows consumers to make decisions based on reliable information.

[1586] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1587] Step 1:

[1588] The server periodically updates the database of reliable information sources. Specifically, it collects announcements from public institutions and experts via APIs and stores them in the database. The input is reliable information source data, and the output is the updated database. This ensures that the latest reliable information is always maintained.

[1589] Step 2:

[1590] The server collects posts in real time through the API of the SNS platform. Specifically, it uses the API to obtain new post data (text, poster information, timestamp, etc.). The input is the SNS post data, and the output is the collected post data. This ensures that the latest post information is always available.

[1591] Step 3:

[1592] The server preprocesses the text of the posts collected and analyzes it using a natural language processing engine (such as Spacy). Specifically, it performs text tokenization, morphological analysis, and keyword extraction. The input is the collected text data, and the output is the preprocessed text and extracted keywords. This allows the content of the posts to be mechanically understood.

[1593] Step 4:

[1594] The server analyzes the preprocessed text and identifies high-risk information. Specifically, it applies a risk assessment algorithm based on the extracted keywords to calculate a risk level. The input is the preprocessed text and keywords, and the output is a risk level. This identifies information that may be false.

[1595] Step 5:

[1596] The server checks the identified high-risk information against a database of trusted information sources. Specifically, the check is performed using keyword matching and similarity calculations. The input is the high-risk information and the database of trusted information sources, and the output is the check result (presence or absence of inconsistencies). This confirms whether the information is trustworthy.

[1597] Step 6:

[1598] If a discrepancy is found, the server generates a warning message. Specifically, it creates a message in the format "This information contradicts the official announcement. For more information, click here: [Official information link]." The input is the match result and details of the contradictory information, and the output is a warning message. This allows users to recognize misinformation.

[1599] Step 7:

[1600] The terminal displays a warning message to the user. Specifically, the warning message is displayed in a pop-up format on the page the user is viewing. The input is the warning message sent from the server, and the output is the warning message displayed on the terminal screen. This allows the user to check the reliability of the information in real time.

[1601] Step 8:

[1602] Users evaluate the effectiveness of warning messages and provide feedback. Specifically, they enter their evaluation of effectiveness and comments in a feedback form and send it to the server. The input is the user's feedback, and the output is the feedback data sent to the server. This contributes to improving the accuracy of the system.

[1603] Step 9:

[1604] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. Specifically, the feedback data is used to retrain the model and adjust parameters. The input is the collected feedback data, and the output is an improved AI model. This improves the accuracy and reliability of the system.

[1605] Step 10:

[1606] The server assigns badges to users with high contributions. Specifically, it evaluates the effectiveness of feedback and the level of contribution, and adds the badge to the user's profile. The input is the user's evaluation data, and the output is an updated user profile. This makes the user's contributions visible, improving motivation.

[1607] 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.

[1608] This invention is a system for preventing the spread of false information on social networking sites and ensuring the reliability of information. This system performs highly accurate fact-checking based on data from reliable sources and combines it with an emotion engine that recognizes the user's emotions. Specifically, with the cooperation of the server, device, and user, the system generates appropriate warning messages according to the user's emotional state, improving the accuracy of information and user acceptance.

[1609] System Overview

[1610] This system is configured as follows:

[1611] 1. Server: Maintains a database of trusted sources, analyzes posts using NLP and sentiment engines, assesses the risk of information, and generates warning messages.

[1612] 2. Terminal: A device that allows users to use SNS and displays warning messages and sends feedback.

[1613] 3. Users: contribute to the credibility of information through social media posts and feedback.

[1614] Overview of program processing

[1615] 1. Information gathering and preparation

[1616] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[1617] 2. Collect social media posts

[1618] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[1619] 3. Analysis and evaluation of information

[1620] The server preprocesses the text of collected posts and then analyzes it with an NLP engine, which extracts important keywords and phrases and analyzes them to understand the content of the information, particularly to identify information that is considered high-risk (e.g., medical or political information).

[1621] 4. Emotion analysis

[1622] The server uses an emotion engine to analyze the emotions of users' posts and reactions. This emotion analysis allows us to understand the intention and tone of the posts and evaluate the emotional state of the users.

[1623] 5. Fact Check

[1624] The server checks high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[1625] 6. Generating and Displaying Warning Messages

[1626] The server generates warning messages for posts that contradict reliable information. The content and presentation of these warning messages are tailored to the user's emotional state. For example, if a user is emotionally charged, a message urging them to stay calm is displayed.

[1627] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[1628] 7. Feedback Collection and Analysis

[1629] Users can review warning messages and evaluate their effectiveness. They can provide feedback on whether the warning is correct by rating it as "valid" or "invalid." They can also add text comments.

[1630] The terminal collects feedback from the user and sends it to the server.

[1631] The server analyzes the collected feedback and uses it to improve the AI ​​engine model. The feedback data is used to retrain the model, improving the accuracy of information analysis and warnings. The emotion engine also analyzes the emotional tone of user feedback and uses it to further improve the model.

[1632] 8. Badging

[1633] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[1634] Specific examples

[1635] Example 1: Medical information verification and sentiment analysis

[1636] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[1637] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[1638] 3. The server checks the post against a reliable medical information database.

[1639] 4. If the server finds an official announcement that the drug is ineffective, a warning message will be displayed on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[1640] 5. The server analyzes the user's emotions using an emotion engine and issues a warning in a calm tone.

[1641] Example 2: Political information verification and feedback analysis

[1642] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[1643] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[1644] 3. Users rate the effectiveness of the warning message and provide feedback.

[1645] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model. The emotion engine also analyzes the emotional tone of the feedback to further improve the model.

[1646] This allows the system to effectively detect and prevent false information being spread on social media, while also providing appropriate warning messages based on the user's emotional state.

[1647] The processing flow will be explained below.

[1648] Step 1:

[1649] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.) and updates the database. This collected data is analyzed by an NLP engine and assigned classification tags (e.g., "medical," "politics," "economics," etc.).

[1650] Step 2:

[1651] The server collects SNS content posted by users in real time through the SNS platform's API. The collected posting data includes text, poster information, timestamps, etc.

[1652] Step 3:

[1653] The server preprocesses the collected text of posts, including normalizing the text data (removing special characters and extra whitespace) and tokenizing it (splitting sentences and words into individual units) to make it analyzable.

[1654] Step 4:

[1655] The server then analyzes the preprocessed text using an NLP engine, extracting key keywords and phrases and understanding the content. This analysis identifies information that is considered particularly high-risk (e.g., medical or political information).

[1656] Step 5:

[1657] The server uses an emotion engine to analyze emotions from users' posts and reactions. The emotion engine incorporates a text analysis algorithm to identify the user's emotional state (e.g., "joy," "anger," "sadness," etc.).

[1658] Step 6:

[1659] The server checks identified high-risk posts against a database of trusted sources, using keyword matching and similarity calculations to check for matches and contradictions with trusted information.

[1660] Step 7:

[1661] If the server finds any discrepancies based on the comparison results, it generates a warning message. The content and display of this warning message are adjusted according to the user's emotional state. For example, if the user is emotionally charged, a message urging them to stay calm is generated.

[1662] Step 8:

[1663] The device receives the warning message and displays it to the user. The warning message is displayed in a pop-up format when the user views a social media post or attempts to post a new one.

[1664] Step 9:

[1665] Users can review warning messages and rate their validity. Users can rate warnings as valid or invalid, and can also add text comments to provide feedback.

[1666] Step 10:

[1667] The terminal collects feedback from the user and sends it to the server.

[1668] Step 11:

[1669] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model. The feedback data is used to retrain the model, improving the accuracy of information analysis and warnings. The emotion engine also analyzes the emotional tone of the user's feedback and uses it to further improve the model.

[1670] Step 12:

[1671] The server will award badges to users with expert knowledge who provide useful feedback and fact-checking activities, allowing high-contribution users to be recognized and trusted by other users.

[1672] As a result, this system can effectively detect and prevent false information from spreading on social media, and provide appropriate warning messages according to the user's emotional state.

[1673] Example 2

[1674] 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."

[1675] Modern online communication platforms face the problem of easily spreading unreliable and false information. This increases the risk of users receiving incorrect information, potentially leading to social confusion and misunderstanding. Furthermore, preventing the spread of misinformation is difficult because appropriate information is not provided based on the user's emotional state. Furthermore, systems have not yet been sufficiently improved based on user feedback, necessitating improved accuracy in detecting and warning about false information.

[1676] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for periodically updating a database of reliable information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of an online platform, means for analyzing preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of reliable information sources, means for evaluating text consistency and inconsistency through keyword matching and similarity calculation, means for performing user sentiment analysis and evaluating the intention and tone of the post, means for generating a warning message and adjusting it according to the user's emotional state, means for displaying the warning message on a user terminal, means for the user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and utilizing it to improve the model of the artificial intelligence engine, and means for awarding badges to highly contributing users. This enables effective detection and prevention of false information spreading on social media and provision of appropriate information taking into account the user's emotional state.

[1677] A "trusted information source database" is a database that collects and stores publicly recognized information such as news sites, academic papers, and official announcements, and ensures the authenticity and reliability of the information.

[1678] A "natural language processing engine" is an artificial intelligence technology that analyzes text data and has functions such as understanding the meaning of language, extracting keywords, and classifying documents.

[1679] An "online platform API" is an interface for communicating with web services and applications and managing the sending and receiving of data.

[1680] "Preprocessed text" refers to text data that has undergone processing such as tokenization, stop word removal, and stemming before being analyzed by a natural language processing engine.

[1681] "High-risk information" is information that may have significant social, economic, or health impacts and should be handled with particular care.

[1682] "Keyword matching" is a technique for evaluating matches between texts, and is a technology that primarily compares based on the presence or absence and frequency of keywords.

[1683] "Similarity calculation" is a technique for evaluating the semantic similarity between texts, and is a comparison technique using mathematical indices such as cosine similarity and Jaccard coefficient.

[1684] "Sentiment analysis" is an artificial intelligence technique that analyzes the emotions and tone contained in text to assess the user's emotional state.

[1685] A "warning message" is a notification message generated to alert the user to detected high-risk information.

[1686] A "user terminal" is a device that allows a user to view information and interact with the device, including a computer, smartphone, tablet, etc.

[1687] "Feedback" refers to evaluation information and comments provided by users, and is data used to improve the system and retrain the model.

[1688] An "artificial intelligence engine" is an artificial intelligence technology that trains models and makes predictions based on feedback data and learning data.

[1689] A "badge" is a digital award given to a user as recognition for their contribution or specific activity within the system.

[1690] The present invention provides a system for preventing the spread of false information on social networking sites and ensuring the reliability of information. This system performs highly accurate fact-checking based on reliable information sources and combines it with an emotion engine that recognizes user emotions. Detailed embodiments are described below.

[1691] System configuration

[1692] This system consists of three elements: a server, a terminal, and a user.

[1693] 1. Server

[1694] The server has many roles and is responsible for the following processes:

[1695] Regularly updated database of reliable sources: The server collects data from news sites, academic papers, official announcements, etc. and updates the database. Specifically, the data collection is done using Python's BeautifulSoup library and the Scrapy framework.

[1696] Analyze collected posts using a natural language processing engine: The server analyzes the collected data using an NLP engine (for example, Google BERT or OpenAI GPT) and assigns classification tags. The collected text data is preprocessed using libraries such as NLTK or spaCy before analysis.

[1697] Collecting posts through APIs of online platforms: The server collects post data in real time from various social media platforms (e.g., Twitter API, Facebook Graph API) and stores it in a database.

[1698] Preprocessing and analysis of submitted text: The server preprocesses the collected text (tokenization, stop word removal, stemming, etc.) and analyzes it using an NLP engine.

[1699] Identifying high-risk information: The server extracts important keywords and phrases and evaluates the agreements and inconsistencies between the texts using keyword matching and similarity calculations (e.g., Cosine Similarity or Jaccard Index).

[1700] Sentiment analysis: The server uses an emotion engine (for example, IBM Watson Tone Analyzer or Microsoft Azure Text Analytics) to assess the user's emotional state and understand the intent and tone of the post.

[1701] Generate warning messages: The server checks high-risk information against trusted sources and generates warning messages if inconsistencies are found, tailoring the messages to the user's emotional state.

[1702] Feedback collection and analysis: The server collects feedback provided by users, analyzes it using an artificial intelligence engine, and uses it to improve the model.

[1703] 2. Terminal

[1704] The terminal is the device through which the user views information and interacts, and is responsible for the following processes:

[1705] Display warning message: The terminal displays the warning message sent from the server to the user. The warning message is displayed as a pop-up window.

[1706] Collecting and sending feedback: The device collects feedback from users (ratings of the effectiveness of warning messages and comments) and sends it to the server.

[1707] 3. Users

[1708] Users play a role in contributing to the credibility of information through social media posts and feedback.

[1709] Review and rate warning messages: Users can review the displayed warning messages and rate their validity as "valid" or "invalid." They can also add text comments.

[1710] Providing feedback: The user provides feedback on the effectiveness of the warning message through the terminal.

[1711] Specific examples

[1712] Example 1: Medical information verification and sentiment analysis

[1713] 1. A user attempts to post information on social media that a certain drug is effective in treating the new virus.

[1714] 2. The server analyzes the post using an NLP engine to detect the keywords "new virus" and "treatment drug."

[1715] 3. The server checks the post against a reliable medical information database.

[1716] 4. If the server finds an official announcement that the drug is ineffective, it will display a warning message on the device stating, "This drug is said to be ineffective against the new virus. Source: [Official announcement link]."

[1717] 5. The server analyzes the user's emotions using an emotion engine and issues a warning in a calm tone.

[1718] Example 2: Political information verification and feedback analysis

[1719] 1. A user sees a post that says "Party X has announced new policy Y" and that information contradicts a trusted database.

[1720] 2. A warning message appears on your device saying, "This information does not match the official announcement. For more information, please see: [database link]."

[1721] 3. Users rate the effectiveness of the warning message and provide feedback.

[1722] 4. The server analyzes the feedback and uses it to improve the accuracy of the AI ​​engine's model. The emotion engine also analyzes the emotional tone of the feedback to further improve the model.

[1723] Example prompts for generative AI models

[1724] "Please check whether the COVID-19 treatment information posted on social media is reliable."

[1725] "Compare this political information to see if it matches the official announcement."

[1726] "Generate appropriate warning messages based on the user's emotional state."

[1727] As described above, this system effectively detects false information being spread on social media, prevents its spread, and provides appropriate warning messages according to the user's emotional state.

[1728] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1729] Step 1:

[1730] Data collection from reliable sources

[1731] The server periodically collects data from reliable sources (news sites, academic papers, official announcements, etc.). Specifically, it periodically retrieves data from a specified URL using Python's BeautifulSoup library or the Scrapy framework. The input is the specified URL, and the output is the retrieved text data.

[1732] Step 2:

[1733] Update the database of collected data

[1734] The server stores the collected text data in a database (for example, MongoDB or MySQL). The input is the collected text data, and the output is the updated database. Specifically, the Python script uses the Pandas library to insert the data organized in a data frame into the database.

[1735] Step 3:

[1736] Preprocessing of submitted data

[1737] The server preprocesses the collected data using an NLP engine, performing processes such as tokenization, stop word removal, and stemming. The input is the collected text data, and the output is the preprocessed text data. Specific operations use the NLTK and spaCy libraries.

[1738] Step 4:

[1739] Analysis using natural language processing

[1740] The server inputs the preprocessed text data into an NLP engine (such as Google BERT or OpenAI GPT) to analyze and extract important keywords and phrases. The input is the preprocessed text data, and the output is a list of keywords and phrases resulting from the analysis.

[1741] Step 5:

[1742] Adding classification tags

[1743] The server assigns classification tags (e.g., "medical," "politics," "economy," etc.) to the text data based on the results of analysis by the NLP engine. The input is a list of keywords and phrases from the analysis results, and the output is text data with classification tags.

[1744] Step 6:

[1745] Real-time collection of SNS posts

[1746] The server collects social media content in real time through the API of an online platform (e.g., Twitter API, Facebook Graph API). The input is a specified API endpoint, and the output is the collected post data. Specifically, it sends an HTTP request and obtains the post data as a response.

[1747] Step 7:

[1748] Identifying high-risk information

[1749] The server preprocesses the social media posts collected and analyzes them using an NLP engine. High-risk information (e.g., "medical information" or "political information") is identified and extracted. The input is the preprocessed post data, and the output is the post data identified as high-risk. Specifically, it detects important keywords.

[1750] Step 8:

[1751] Performing sentiment analysis

[1752] The server uses an emotion engine (for example, IBM Watson Tone Analyzer or Microsoft Azure Text Analytics) to analyze emotions from the post content and reactions. The input is the post data identified as high risk, and the output is the emotion analysis results. Specifically, the server sends text data to the emotion engine's API and receives the analysis results.

[1753] Step 9:

[1754] Conducting fact-checks

[1755] The server checks high-risk posts against a database of trusted sources and uses keyword matching or similarity calculations (e.g., Cosine Similarity or Jaccard Index) to check for matches or contradictions. The input is the high-risk post data and the source database, and the output is the fact-check results. Specifically, the server applies a similarity calculation algorithm.

[1756] Step 10:

[1757] Generate a warning message

[1758] The server generates a warning message based on the fact-check results if a contradiction is found. The content and display method of this message are adjusted according to the user's emotional state. The input is the fact-check results and the emotion analysis results, and the output is the generated warning message. Specific operation involves embedding specific information into the template message.

[1759] Step 11:

[1760] Displaying a warning message

[1761] The terminal receives the warning message and displays it to the user. The input is the generated warning message, and the output is the warning message displayed on the terminal screen. As a specific operation, it executes the code that displays the popup window.

[1762] Step 12:

[1763] Collecting feedback

[1764] The user evaluates the effectiveness of the warning message and provides feedback. The input is the displayed warning message, and the output is the feedback from the user. Specifically, the user selects "enabled" or "disabled" and enters a comment.

[1765] Step 13:

[1766] Send Feedback

[1767] The terminal sends the feedback collected from the user to the server. The input is the feedback from the user, and the output is the feedback sent to the server. The specific operation is to send the feedback data using an API.

[1768] Step 14:

[1769] Feedback Analysis

[1770] The server analyzes the collected feedback and uses it to improve the AI ​​engine's model. The input is the feedback sent to the server, and the output is the improved AI model. Specifically, the feedback data is used to retrain the model.

[1771] Step 15:

[1772] Badge Awarding

[1773] The server assigns badges to users who have contributed significantly to effective feedback and fact-checking activities. The input is feedback history, and the output is the assigned badges. Specifically, the server automatically assigns badges to users who meet certain evaluation criteria.

[1774] (Application example 2)

[1775] 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."

[1776] As the degree of freedom in the distribution of information on the Internet and social networking services (SNS) increases, the spread of false information has become a social problem. In particular, false information about high-risk topics (e.g., medical, political, economic, etc.) can have a significant impact. Conventional systems have been required to effectively prevent the spread of such false information while at the same time taking into account the user's emotions, but this has been difficult to achieve. The purpose of this invention is to solve this problem by quickly and accurately detecting false information and providing warning messages that correspond to the user's emotional state.

[1777] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for periodically updating a database of reliable information sources, means for analyzing collected posts using a natural language processing engine and assigning classification tags, means for collecting posts in real time through the API of an SNS platform, means for analyzing preprocessed text and identifying high-risk information, means for comparing the identified high-risk information with the database of reliable information sources, means for generating a warning message if a discrepancy is found, means for displaying the warning message on a user terminal, means for the user to evaluate the effectiveness of the warning message and provide feedback, means for analyzing the collected feedback and using it to improve the model of the artificial intelligence engine, means for awarding badges to highly contributing users, and means for adjusting the content and display method of the warning message according to the user's emotions and displaying it on the user's device. This prevents the spread of false information and enables the provision of reliable information that takes user emotions into consideration.

[1778] A "database of trusted sources" is a regularly updated database that aggregates data collected from sources such as news sites, academic papers, and official announcements whose credibility has been confirmed.

[1779] A "natural language processing engine" is an algorithm and software that analyzes text data and performs tasks such as extracting keywords, understanding context, and tagging.

[1780] An "SNS platform API" is an application programming interface provided by a social networking service, and is a function that allows external developers to obtain and manipulate data on the SNS.

[1781] "Preprocessed text" is text data that has been converted into a form that is easier to analyze by filtering and cleaning up raw data.

[1782] "High-risk information" refers to information that is likely to contain false information and that falls into categories such as medical, political, and economic information, which may have particularly serious consequences.

[1783] A "warning message" is a message that notifies or alerts the user that there is a doubt about the accuracy or reliability of the information.

[1784] "User terminal" refers to any device used by a user to access SNS, including smartphones, tablets, and PCs.

[1785] "Feedback" refers to the evaluations and comments that users make on the system's warning messages and the information provided.

[1786] "Artificial intelligence engine" is a general term for algorithms and software that learns from collected data and improves models.

[1787] A "badge" is a symbol of recognition and appreciation given to users with specialized knowledge or who have made significant contributions to the system.

[1788] The means for adjusting the content and display method according to "emotion" is a function that generates and displays an appropriate warning message according to the user's current emotional state based on the results of emotion analysis.

[1789] This invention relates to a system for preventing the spread of false information on the Internet and social networking services (SNS) and ensuring the reliability of information. As a specific example, a reliability guide quickly and accurately detects false information and provides warning messages tailored to the user's emotional state. The following describes the components of this system and their functions.

[1790] server

[1791] The server has the following functions:

[1792] 1. Updating the database of reliable sources: Regularly collect data from reliable news sites, academic papers, official announcements, etc. and update the database. This data will be used as reliable data.

[1793] 2. Natural Language Processing Engine (NLP Engine): Analyzes collected posts and assigns classification tags based on their content. This engine extracts important keywords and phrases and identifies high-risk information.

[1794] 3. Real-time collection using SNS platform APIs: Collect user posts from SNS platforms in real time. The posts include text, author information, timestamps, etc.

[1795] 4. Pre-processed text analysis: The collected text data is pre-processed and analyzed by an NLP engine to identify information that is considered high risk.

[1796] 5. Matching high-risk information with reliable data: Identified high-risk information is matched with a database of reliable sources to identify any matches or discrepancies.

[1797] 6. Sentiment analysis engine: Analyzes emotions from user posts and reactions to evaluate the user's emotional state.

[1798] 7. Generating a warning message: If a contradiction is found, a warning message is generated according to the user's emotional state. For example, if the user is emotionally charged, a message is generated to encourage them to calm down.

[1799] 8. AI engine model improvement: Analyze user feedback to retrain the AI ​​e...

Claims

1. A means of regularly updating the database of reliable sources of information; A means for analyzing collected posts using a natural language processing engine and assigning classification tags; A means of collecting posts in real time through the API of social media platforms, a means for analyzing the preprocessed text to identify high-risk information; a means for cross-checking identified high-risk information with a database of trusted sources; a means for generating a warning message if a discrepancy is found; means for displaying a warning message on a user terminal; a means for the user to rate the effectiveness of the warning message and provide feedback; A means to analyze the collected feedback and use it to improve the AI ​​engine's model; and means for awarding badges to highly contributing users.

2. The system according to claim 1, wherein posts on SNS are analyzed using a natural language processing engine to identify high-risk information.

3. 10. The system of claim 1, wherein the system checks the identified high-risk information against a database of trusted sources and generates a warning message if a discrepancy is found.

Images