System

Abstract

A system is provided.SOLUTION: A system comprising: means for obtaining information data from social media; means for pre-processing the obtained information data; means for determining authenticity of information using a machine learning model based on the pre-processed data; means for filtering determined false information; and means for notifying a user of the filtered information.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] In today's social media environment, users are exposed to the risk of false information and fake news. This can lead to fraud and crime, as well as misjudgment in times of disaster. This risk is particularly serious for elderly people and families with children. For this reason, it is urgent to provide an environment where users can use social media with peace of mind. [Means for solving the problem]

[0005] To solve this problem, the present invention provides the following means. First, a means for acquiring information data from social media is provided. Next, a means for preprocessing the acquired information data is introduced. A means for determining the authenticity of information using a machine learning model based on the preprocessed data is added. Next, a means for filtering out false information that has been determined to be false is provided, and finally, a means for notifying the user of the filtered information is provided. In this way, a system is provided that allows users to use social media with peace of mind.

[0006] "Social media" refers to online platforms that allow users to share information and communicate with each other.

[0007] "Information data" refers to all digital content posted on social media, including text, images, videos, and links.

[0008] "Means of acquisition" refers to the methods and processes for collecting information data using social media APIs and scraping technology.

[0009] "Preprocessing" refers to the process of analyzing collected information data, removing unnecessary information, and converting it into a form that is easy for machine learning models to handle.

[0010] A "machine learning model" refers to an algorithm or software that learns from past data and makes predictions or classifications for new data.

[0011] "Means for determining the truth of information" refers to a method that uses a machine learning model to determine whether information data is true or false as an evaluation score.

[0012] "Measures for filtering false information" refers to the process of removing or blocking determined false information from being displayed to users.

[0013] "Means for notifying" refers to methods and systems for communicating filtering results and other important information to users in real time.

[0014] "User" refers to any individual or legal entity that uses the System to browse social media.

[0015] "Device" refers to the electronic device used by a User, such as a smartphone, tablet, or computer.

[0016] "Server" refers to a remote computer system that performs processing such as collecting information data, preprocessing, running machine learning models, and filtering. [Brief explanation of the drawings]

[0017] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10]1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0018] 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.

[0019] First, the terms used in the following description will be explained.

[0020] 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).

[0021] 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.

[0022] 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.

[0023] 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.

[0024] 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."

[0025] [First embodiment]

[0026] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0027] 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.

[0028] 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).

[0029] 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.

[0030] 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.

[0031] 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.

[0032] 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.

[0033] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0034] 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.

[0035] 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.

[0036] 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.

[0037] 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."

[0038] The present invention relates to a system for filtering false information on social media, which mainly consists of a server, a terminal, and a user. This system acquires information data from social media, preprocesses it, determines the authenticity of the information using a machine learning model, filters the false information, and notifies the user of the filtering results.

[0039] Program processing (natural language explanation)

[0040] 1. Data Collection

[0041] The server obtains information data from social media platforms (e.g., popular social networking sites) using APIs. API stands for Application Programming Interface, and refers to the rules and protocols that allow software to communicate with other software.

[0042] The device acquires the posting data of the SNS accessed by the user in real time and sends it to the server.

[0043] 2. Data Preprocessing

[0044] The server analyzes the acquired information data and performs tokenization, which is the process of dividing text data into words and phrases.

[0045] Additionally, the server removes unnecessary vocabulary and stop words (e.g., "the" and "and" in English) and normalizes the data. Normalization is the process of unifying data by performing stemming and lameness.

[0046] 3. Learning and judgment using AI models

[0047] The server then inputs the preprocessed data into a machine learning model that learns the characteristics of false information based on past data.

[0048] The server's AI module then scores the information based on newly acquired data. For example, if it calculates that a particular news article has a 0.9 probability of being true, the article is deemed to be highly reliable.

[0049] 4. Filtering and Notifications

[0050] The server filters out information data that is determined to be false and does not send it to the user's terminal.

[0051] The device will notify the user of the filtering results in the form of a pop-up or push notification.

[0052] Users can use social media with peace of mind.

[0053] Specific examples

[0054] Example 1: Filtering news articles

[0055] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server retrieves the data and performs preprocessing. It then uses a machine learning model to determine whether the news article is authentic or not. For example, if the news article is determined to be unreliable with a score of 0.8, the server filters the article and does not display it on the user's device. The device then notifies the user that it has been filtered.

[0056] Example 2: Filtering fraudulent ads

[0057] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud, with a score of 0.85, the server filters the ad and does not display it on the user's device. The device then notifies the user that the ad has been filtered.

[0058] In this way, the present invention provides an environment in which users can use social media with peace of mind.

[0059] The processing flow will be explained below.

[0060] Step 1: Data collection

[0061] The server retrieves information data from the social media platform via API, including user posts, comments, news articles, advertisements, etc.

[0062] The device automatically captures the social media posts that the user views or interacts with and sends the data to the server in real time.

[0063] Step 2: Preprocessing

[0064] The server receives the acquired information data and performs text analysis, specifically, using tokenization to divide the text into words and phrases.

[0065] The server removes stop words (e.g., common conjunctions and prepositions) and performs stemming to improve the accuracy of the data.

[0066] Step 3: Training the machine learning model

[0067] The server matches the pre-processed data with existing datasets and uses machine learning models to learn new patterns, including spam filtering, fake news detection, and fraudulent ad detection.

[0068] The server evaluates the accuracy of the trained model and adjusts parameters as needed.

[0069] Step 4: Determine the information

[0070] The server feeds new data into a machine learning model in real time to determine whether the information is true or false. For example, a particular news article may be identified as fake news with a probability of 0.7.

[0071] The server outputs the information judgment result as a score.

[0072] Step 5: Filtering

[0073] The server filters data determined to be false and does not send it to the user's device. For example, data determined to be fraudulent advertisements will not be displayed to the user.

[0074] The server stores the filtered data and uses it as learning data at a later date.

[0075] Step 6: Notifications and Display

[0076] The device will notify the user of the filtering results, for example by displaying a pop-up notification such as "This news article was not displayed due to low reliability."

[0077] Users will receive notifications and can use social media with peace of mind.

[0078] Step 7: Reassess and improve

[0079] The server periodically re-evaluates the accuracy of the filtering and retrains the model, allowing it to keep up with the latest disinformation trends.

[0080] The server receives feedback from users and uses it to improve the system.

[0081] Through the above steps, the present invention provides an environment in which users can use social media with peace of mind.

[0082] Example 1

[0083] 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."

[0084] There is a lot of false information and fraudulent advertisements on social media, making it difficult for users to quickly obtain reliable information. There is also a lack of effective measures to prevent incorrect decisions and actions based on false information. For this reason, it is necessary to provide an environment where users can use social media with peace of mind.

[0085] 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.

[0086] In this invention, the server includes a means for acquiring information data from social media, a means for preprocessing the acquired information data, and a means for determining the authenticity of the information using a machine learning model based on the preprocessed data, thereby making it possible to filter out false information and provide reliable information.

[0087] "Social media" refers to a platform on the Internet where users can disseminate, share, and interact with information.

[0088] "Information data" refers to data including content such as text, images, and videos posted on social media.

[0089] "Preprocessing" refers to the process of analyzing, dividing, and organizing acquired information data, and converting it into a format that is easy for the machine learning model to process.

[0090] "Tokenization" refers to the process of dividing text data into smaller units such as words or phrases.

[0091] "Stop words" refer to common words (e.g., the, and, etc.) that are not necessary for semantic analysis of a sentence in natural language processing, and the purpose is to remove them.

[0092] "Normalization" refers to the process of converting data into a standard form using stemming and lambdaization.

[0093] "Stemming" refers to a technique for extracting only the stem portion of a word.

[0094] "Lametization" refers to the process of converting words into their base forms.

[0095] A "machine learning model" refers to a system that includes algorithms that learn from past data and make predictions or classifications for new data.

[0096] "Scoring" refers to the process by which a machine learning model quantifies the reliability and veracity of information data.

[0097] "Filtering" refers to the process of filtering out unnecessary or inappropriate data based on specific conditions.

[0098] "Notification" refers to the process of informing users of filtered results and important information.

[0099] A "pop-up" is a small window that suddenly appears on a user's screen and is used to display specific information or notifications.

[0100] "Push notification" refers to a form of notification that an application or system automatically sends to a user's device.

[0101] "Database" refers to a system that systematically stores and manages collected data and results.

[0102] This invention provides a method for filtering false information on social media using a system consisting mainly of a server, a terminal, and a user. This system uses a combination of specific software and hardware to perform a series of processes including data collection, data preprocessing, AI-based judgment, filtering, and notification.

[0103] Data collection

[0104] The server retrieves information data from social media platforms using APIs. For example, to collect data from Twitter, the Python library "Tweepy" is used. The server uses "Tweepy" to periodically collect post data related to specific keywords and store it in a database.

[0105] The device collects data posted to the social media sites accessed by the user in real time and sends it to the server. This process is achieved using technologies such as JavaScript (registered trademark)'s "Fetch API" and Swift. The device sends the posted data to the server in response to user actions, and the data is collected on the server side.

[0106] Data Preprocessing

[0107] The server analyzes the collected information data and performs tokenization using the Python library NLTK, which splits the text into words and stores them in a database.

[0108] The server then removes unnecessary vocabulary and stop words from the tokenized data, performs stemming, and lambdaization, and standardizes the data using libraries like NLTK and SpaCy, converting the data into a format suitable for processing by machine learning models.

[0109] Learning and judgment by AI model

[0110] The server then feeds the pre-processed data into machine learning models, specifically pre-trained BERT and GPT models, which excel at natural language processing and learn the characteristics of misinformation from the data.

[0111] The server's AI module scores the authenticity of the information based on the newly input data, and the results are stored in a database. For example, if the AI ​​module calculates that a particular news article has a 0.9 probability of being true, the article is deemed to be highly reliable.

[0112] Filtering and Notifications

[0113] The server filters out false information based on the scoring results of the machine learning model, and if the score falls below a certain threshold, the information is prevented from being sent to the user's device.

[0114] The device will notify the user of the filtering results. By notifying the user using a pop-up or push notification, the user can check in real time that the information has been filtered. For example, if a particular news article is deemed to be unreliable and filtered, the device will notify the user of the result.

[0115] Specific examples

[0116] 1. Filtering news articles

[0117] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server retrieves the data and performs preprocessing. It then uses a machine learning model to determine the authenticity of the news article. If the news article is judged to be low reliability with a score of 0.8, the server filters the article and does not display it on the user's device. The device then notifies the user that it has been filtered.

[0118] 2. Filtering fraudulent ads

[0119] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. If the ad is determined to be highly fraudulent with a score of 0.85, the server filters the ad and does not display it on the user's device. The device notifies the user that the ad has been filtered.

[0120] This system allows users to reduce the risk of false information and fraud on social media, allowing them to use only reliable information with peace of mind.

[0121] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0122] Step 1: Data collection

[0123] Specific operation:

[0124] The server retrieves information data using the API of the social media platform. Specifically, it uses the Python library "Tweepy" to collect data from Twitter. It periodically monitors specific keywords (e.g., "COVID-19" or "disaster information") and retrieves new posts.

[0125] input:

[0126] Social media platform API keys

[0127] Specific keywords to collect

[0128] Data processing:

[0129] Call API to get post data related to keywords

[0130] output:

[0131] Acquired post data (text, user information, post date and time, etc.)

[0132] Step 2: Data Preprocessing

[0133] Specific operation:

[0134] The server preprocesses the acquired information data. First, it performs tokenization, splitting the text into words using the Python library "NLTK."

[0135] Next, we remove unnecessary vocabulary and stop words, and perform stemming and lambdaization using tools like NLTK and SpaCy.

[0136] input:

[0137] Obtained post data (output from Step 1)

[0138] Data processing:

[0139] Text tokenization (splitting into words)

[0140] Stop word removal

[0141] Stemming and Lamellarization

[0142] output:

[0143] Preprocessed text data

[0144] Step 3: Learning and judgment by AI model

[0145] Specific operation:

[0146] The server then feeds the preprocessed data into a machine learning model (e.g., the BERT model), which has learned the characteristics of false information based on past data.

[0147] The server's AI module scores the truth of the information based on the newly entered data, and the scoring result (e.g., 0.0 to 1.0) is stored in a database.

[0148] input:

[0149] Preprocessed text data (output of step 2)

[0150] Data Calculation:

[0151] Judgment and scoring using machine learning models

[0152] output:

[0153] Reliability score (e.g., 0.8)

[0154] Step 4: Filtering

[0155] Specific operation:

[0156] The server filters out false information based on the scoring results. If the score is below a certain threshold (e.g., 0.5), the information is not sent to the user's device.

[0157] The filtered information is either deleted from the database or moved to another table.

[0158] input:

[0159] Determined reliability scores and corresponding information data (output of Step 3)

[0160] Data processing:

[0161] Filtering data based on scores

[0162] Save or delete filtered results

[0163] output:

[0164] Filtered and clean information data

[0165] Step 5: Notification

[0166] Specific operation:

[0167] The device will notify the user of the filtering results, using a pop-up or push notification to let the user know that false information has been filtered.

[0168] The user confirms the notification and views the clean information data.

[0169] input:

[0170] Filtered and clean information data (output of step 4)

[0171] Data processing:

[0172] Generate and send notifications

[0173] output:

[0174] Notification of filtering results

[0175] (Application example 1)

[0176] 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."

[0177] Traditional social media platforms require users to independently determine the authenticity of the information they access. However, in today's information-intensive world, there is a growing risk that users will be misled by unreliable information, leading to the spread of inaccurate information. Furthermore, there is a lack of systems that can effectively filter false information in real time and immediately warn users. To solve this problem, a system is needed that can detect false information with high accuracy and in real time and notify users.

[0178] 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.

[0179] In this invention, the server includes means for acquiring information data from social media, means for preprocessing the acquired information data, and means for determining the authenticity of the information using a machine learning model based on the preprocessed data. This enables a system including means for tokenizing the information data and removing unnecessary information, means for notifying the filtering results in real time, and means for issuing a warning when false information is detected.

[0180] "Social media" is a general term for online platforms that allow users to share information and communicate with each other.

[0181] "Information Data" refers to data such as text, images, videos, and links generated by users on social media.

[0182] "Preprocessing" is the process of converting acquired information data into a format that can be applied to a machine learning model through methods such as tokenization and removal of unnecessary information.

[0183] A "machine learning model" is a system that uses algorithms to learn patterns and specific characteristics based on past data to determine the authenticity of new data.

[0184] "Disinformation" is information that is intentionally or unintentionally provided in an inaccurate or misleading manner.

[0185] "Filtering" is the process of identifying and removing false information from acquired information data.

[0186] "Tokenization" is the process of dividing text data into words and phrases.

[0187] "Nutrition" is the process of removing portions of text data that are not required for analysis, based on stop words or specific patterns.

[0188] "Real-time" means that data acquisition and processing are carried out immediately, and the results are notified to the user without making them wait unintentionally.

[0189] "Notification" is an action to notify the user of filtering results or warnings.

[0190] MODE FOR CARRYING OUT THE INVENTION

[0191] To implement this invention, three main components, a server, a terminal, and a user, must cooperate. The role and processing method of each component will be specifically explained below.

[0192] server

[0193] The server plays a key role in acquiring information data from social media, preprocessing it, and then using machine learning models to determine its authenticity. Specifically, it uses the following hardware and software:

[0194] Hardware:

[0195] High-performance processor (e.g., Intel Xeon)

[0196] Large memory capacity (e.g. 64GB RAM or more)

[0197] software:

[0198] Sinatra (Ruby web framework)

[0199] NLTK library (natural language processing library)

[0200] Transformers library (including the BERT model)

[0201] The server first collects information data in real time using social media APIs. The information data then undergoes a preprocessing step, where it is tokenized and stop words are removed. Based on this preprocessed data, the reliability of the information is scored using the BERT model. Based on the results, false information is filtered and notifications are sent to the device.

[0202] Terminal

[0203] The terminal is a device such as a smartphone or tablet used by the user, and serves to notify the user of the filtering results received from the server.

[0204] Hardware:

[0205] Smartphones and tablets (e.g., iPhone, Android devices)

[0206] software:

[0207] Web browser or native application

[0208] The device displays notifications received from the server to the user as pop-up or push notifications. For example, if a user clicks on a specific article in a social media app and the information is determined to be unreliable, a warning pop-up will be displayed saying, "This news article is unreliable."

[0209] User

[0210] The user is an end user of social media. The user can operate their device and check the authenticity of the information sent from the server in real time. This significantly reduces the risk of being misled by incorrect or fake information.

[0211] Specific examples

[0212] For example, consider a case where the reliability of information is evaluated using the BERT model in a news application. When a user taps on a specific article in the news app, the server collects the article's text data, preprocesses it, and then analyzes it using the BERT model. If the analysis results indicate that the information is not reliable, a notification is sent to the device and a warning is displayed to the user.

[0213] Prompt Sentence Examples

[0214] "This is an example news article text for analysis."

[0215] In this way, the present invention can provide users with access to reliable information and protection from inaccurate information.

[0216] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0217] Step 1: Data collection

[0218] The server collects information data in real time using social media APIs. The input data is text data of the latest posts on social media and news articles. The server obtains this data via the API and sends it to the next preprocessing step.

[0219] Step 2: Data Preprocessing

[0220] The server performs preprocessing on the collected information data. Specifically, it performs tokenization to divide the text into words or phrases. It also removes stop words and eliminates unnecessary words. In this way, the information contained in the input data is made clearer and converted into a format suitable for analysis by machine learning models. The output data is preprocessed, clean text data.

[0221] Step 3: Authentication using machine learning models

[0222] The server inputs the preprocessed data into a generative AI model (e.g., BERT) to determine whether the text is true or false. The server uses a machine learning model to analyze the input data and generate a score indicating whether it is trustworthy information. Based on this scoring, the server determines whether the information is accurate or false. The output data is a reliability score and a determination result corresponding to each piece of text data.

[0223] Step 4: Filtering

[0224] The server filters out false information based on the results determined by the machine learning model. In this step, information data determined to be unreliable is removed or flagged as unreliable. The output data is filtered, reliable information data.

[0225] Step 5: Notification

[0226] The terminal receives the filtering results from the server and notifies the user. The terminal displays a warning popup for unreliable information and displays the normal screen for reliable information. In this step, the user is prompted to take specific action. The output data is a notification message on the user's device.

[0227] Specific examples of specific actions

[0228] Data collection:

[0229] The server retrieves the news article "This is an example news article text for analysis." via the API.

[0230] Data preprocessing:

[0231] The server converts the retrieved text from "This is an example news article text for analysis." into a tokenized format such as "this example news article text analysis," and removes stop words to generate clean data.

[0232] Machine learning model for determining truth:

[0233] The server inputs the clean text data into the BERT model and scores it. For example, if the confidence score is 0.4, the information is determined to be false.

[0234] filtering:

[0235] Information with a reliability score of 0.4 is filtered as false information and users will not receive it.

[0236] notification:

[0237] Based on the filtering results received by the device from the server, a pop-up warning is displayed to the user stating, "This news article is unreliable."

[0238] 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.

[0239] The present invention relates to a system that filters false information on social media and notifies users while taking their emotions into consideration. The system mainly consists of a server, a terminal, a user, and an emotion engine. The system acquires information data from social media, preprocesses it, and then uses a machine learning model to determine whether the information is true or false. Inaccurate or harmful information is filtered, and the user is notified of the results. The emotion engine is also used to analyze the user's emotional state and adjust the notification content and filtering criteria based on this.

[0240] Program processing (natural language explanation)

[0241] 1. Data Collection

[0242] The server retrieves information data from social media platforms (e.g., general social networking sites) via APIs, including user posts, comments, news articles, advertisements, etc.

[0243] The device acquires the posting data of the SNS accessed by the user in real time and transmits the data to the server.

[0244] 2. Pretreatment

[0245] The server receives the acquired information data and performs analysis, which includes tokenization, which divides the text data into words and phrases.

[0246] The server removes unnecessary vocabulary and stop words and normalizes the data, a process that involves stemming and lameness.

[0247] 3. Learning and assessment of machine learning models

[0248] The server feeds the preprocessed data into a machine learning model that learns the characteristics of misinformation based on past data.

[0249] The server then scores the information based on the newly acquired data. For example, if a particular news article is judged to be fake news with a probability of 0.8, the article is deemed to be false information.

[0250] 4. Emotion analysis using an emotion engine

[0251] The server uses an emotion engine to analyze the emotions users express on social media, categorizing them as positive, negative, neutral, etc.

[0252] The server obtains the user's emotional state and reflects it in the filtering results and notification content.

[0253] 5. Filtering and Notifications

[0254] The server filters data determined to be false and does not send it to the user's device.

[0255] The server notifies the user of the filtering results, for example by displaying a pop-up message saying, "This news article was not displayed because it was not reliable."

[0256] The device selects an appropriate notification method depending on the user's emotional state. For example, if the emotional state is negative, the notification will be delivered in a gentler manner.

[0257] Specific examples

[0258] Example 1: News article filtering and sentiment analysis

[0259] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in reliability with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server detects that the user is feeling stressed, it will display a gentle notification such as, "This news article has been temporarily suppressed due to its low reliability."

[0260] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[0261] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[0262] In this way, the present invention is a system that provides an environment in which users can use social media with peace of mind, and at the same time provides appropriate notifications taking into account the user's emotional state.

[0263] The processing flow will be explained below.

[0264] Step 1: Data collection

[0265] The server retrieves information data from popular social media platforms via APIs, including user posts, comments, news articles, and advertisements.

[0266] The device captures the social networking site posting data that the user has viewed or operated in real time and immediately transmits that data to the server.

[0267] Step 2: Preprocessing

[0268] The server analyzes the received information data, specifically by using tokenization to divide the text data into words and phrases.

[0269] The server normalizes the data by removing stop words (e.g., common conjunctions and prepositions), stemming, and lame- tizing.

[0270] Step 3: Training the machine learning model

[0271] The server then matches the pre-processed data with existing datasets and uses machine learning models to learn new patterns, resulting in a model that can detect misinformation with high accuracy.

[0272] The server evaluates the accuracy of the model and retrains it if necessary.

[0273] Step 4: Determine the information

[0274] The server feeds new data into the machine learning model in real time. For example, if a particular news article has a 0.75 probability of being fake news, the article is flagged as false.

[0275] The server outputs the information judgment results as a score, which is used as a basis for filtering decisions.

[0276] Step 5: Sentiment Analysis

[0277] The server uses an emotion engine to analyze the user's emotions, for example by analyzing text and behavioral data to determine what emotions (positive, negative, neutral) the user is feeling.

[0278] The server stores the analysis results in a database, making them available for subsequent processing.

[0279] Step 6: Filtering and Notifications

[0280] The server filters data determined to be false and does not send it to the user's device.

[0281] The server generates appropriate notification content based on the results of emotion analysis, depending on the user's emotional state. For example, if the user is feeling anxious, the server will send a notification in a calm manner.

[0282] The device will notify the user of the filtering results, for example, in the form of a pop-up message saying, "This news article was not displayed due to its low reliability."

[0283] Step 7: Reassess and improve

[0284] The server periodically re-evaluates the accuracy of filtering and sentiment analysis and re-trains the model if necessary.

[0285] The server collects feedback from users and uses it to improve the system.

[0286] Specific examples

[0287] Example 1: News article filtering and sentiment analysis

[0288] When a user clicks on a specific news article on a social networking site, the device immediately sends the news article data to a server. The server preprocesses the data and uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in credibility with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server detects that the user is feeling stressed, it may display a gentle notification such as, "This news article is low in credibility, so we have temporarily withheld it from view."

[0289] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[0290] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[0291] In this way, the present invention is a system that provides an environment in which users can use social media with peace of mind, and at the same time provides appropriate notifications taking into account the user's emotional state.

[0292] Example 2

[0293] 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."

[0294] There is a lot of information circulating on social media, including fake and incorrect information. This fake information can cause confusion and anxiety among users. Furthermore, if information is notified without taking into account the user's emotional state, it can cause further stress and discomfort. To solve this situation, a notification system that can effectively filter fake information and take into account the user's emotional state is needed.

[0295] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0296] In this invention, the server includes means for acquiring information data from social media, means for preprocessing the acquired information data, means for determining the authenticity of the information using a machine learning model based on the preprocessed data, means for filtering the determined false information, means for analyzing the emotional state of the user, means for adjusting filtering criteria and notification content based on the analysis results, and means for notifying the user of the filtered information. This makes it possible to efficiently filter false information and provide appropriate notifications that take the user's emotions into consideration.

[0297] "Social media" is a platform on the Internet that allows users to share information and communicate.

[0298] "Information data" refers to all information, including text data and media data such as user posts, comments, news articles, and advertisements.

[0299] "Preprocessing" refers to analyzing the acquired information data and performing tokenization and normalization processing.

[0300] "Tokenization" is the process of dividing text data into words and phrases.

[0301] "Normalization" is the process of stemming, lamellarizing, and removing unnecessary vocabulary and stop words from data.

[0302] A "machine learning model" is an algorithm that learns from past data and analyzes and classifies new data.

[0303] An "emotion engine" is a software tool for analyzing a user's emotional state, categorizing it as positive, negative, neutral, etc.

[0304] "Filtering" is the process of sorting information data based on specific criteria to remove unwanted or harmful information.

[0305] "Notifications" are messages or alerts that communicate filtering results or other information to the user.

[0306] MODE FOR CARRYING OUT THE INVENTION

[0307] The present invention relates to a system that filters false information on social media and notifies users while taking their emotions into consideration. The system mainly consists of a server, a terminal, a user, and an emotion engine. The system acquires information data from social media, preprocesses it, and then uses a machine learning model to determine the authenticity of the information. Inaccurate or harmful information is filtered, and the user is notified of the results. The emotion engine is also used to analyze the user's emotional state and adjust the notification content and filtering criteria based on this.

[0308] Server Features

[0309] The server periodically retrieves information data from social media platforms via API. This information includes user posts, comments, news articles, advertisements, etc. The retrieved information data is preprocessed and analyzed. Preprocessing includes tokenization (dividing text data into words and phrases), stemming, lameness (standardizing word forms), and removal of unnecessary vocabulary and stop words.

[0310] The preprocessed data is then fed into a machine learning model (e.g., BERT or RoBERTa). This model has learned the characteristics of false information based on past data, and scores new data for authenticity. Data that is determined to be false is then filtered out.

[0311] The server also uses an emotion engine (e.g., IBM Watson® Emotion Analysis) to analyze the user's emotional state. Based on the results of the emotion analysis, the filtering criteria and notification content are adjusted. For example, if the user is in a negative emotional state, the notification is adjusted to a milder tone.

[0312] Device Features

[0313] The device collects data on social media posts accessed by users in real time and sends it to a server. When a user clicks on a specific news article or advertisement, the device also immediately sends that information to the server.

[0314] The device also notifies the user of the filtering results. This notification is adjusted based on the analysis results of the emotion engine. For example, if the user's emotional state is negative, the notification content will be milder.

[0315] Specific examples

[0316] Example 1: News article filtering and sentiment analysis

[0317] When a user clicks on a specific news article on a social media site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in credibility with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the notification content accordingly. For example, if the server detects that the user is feeling stressed, it may display a gentle notification such as, "This news article has been temporarily suppressed due to its low credibility."

[0318] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[0319] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[0320] Example of input prompt for generative AI model

[0321] Example prompt 1:

[0322] "Please analyze the reliability of this news article."

[0323] Example prompt 2:

[0324] "Determine whether this ad is potentially fraudulent."

[0325] Example prompt 3:

[0326] "Analyze the sentiment of your users' latest posts and tailor your notifications based on that."

[0327] As described above, the present invention is a system that provides an environment in which users can use social media with peace of mind, effectively filters out false information, and provides appropriate notifications based on the user's emotions.

[0328] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0329] Step 1:

[0330] The server retrieves information data from social media platforms via APIs. As input, it receives data provided by the social media APIs (user posts, comments, news articles, advertisements, etc.). As output, the raw data is stored on the server.

[0331] Step 2:

[0332] The device collects data on posts from social media sites accessed by users in real time and sends that data to a server. As input, it receives user behavior data (such as clicked news articles or advertisements). As output, this information is transferred to the server.

[0333] Step 3:

[0334] The server analyzes the received information data and performs tokenization. As input, it takes raw data and splits it into words and phrases. As output, it generates tokenized data.

[0335] Step 4:

[0336] The server normalizes the data by stemming, lameness, and removing unnecessary vocabulary and stop words. As input, it takes tokenized data and performs normalization on it. As output, it gets clean, normalized data.

[0337] Step 5:

[0338] The server inputs the preprocessed data into a machine learning model (e.g., BERT or RoBERTa). As input, it receives the normalized data and feeds it into the machine learning model. As output, it obtains a truth score for each piece of data.

[0339] Step 6:

[0340] The server determines whether the information is true or false based on the output from the machine learning model. As input, it receives the score returned from the machine learning model, sets a threshold, and determines whether the information is true or false. The output is classified into data that is determined to be false and other data.

[0341] Step 7:

[0342] The device monitors users' latest posts and comments and sends the data to the server. As input, it collects data about users' emotions. As output, it transfers the data to the server.

[0343] Step 8:

[0344] The server uses an emotion engine to analyze the user's emotional state. It receives user posts and comment data as input and performs emotion analysis. The output is an emotion classification result such as positive, negative, or neutral.

[0345] Step 9:

[0346] The server filters data that is determined to be false. As input, it receives data that has been determined to be true or false, and filters out data that it recognizes as false. As output, it obtains a filtered, safe dataset.

[0347] Step 10:

[0348] The server notifies the user of the filtering results. As input, it receives the filtering result data and converts it into a notification message. As output, it generates the notification message.

[0349] Step 11:

[0350] The device selects an appropriate notification method according to the user's emotional state. As input, it receives the emotion analysis results and filtering results and determines the notification method. As output, the adjusted notification message is displayed to the user.

[0351] (Application example 2)

[0352] 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."

[0353] A lot of information is circulating on social media, but some of it is false or harmful, which increases the risk of users acting on incorrect information. This can also worsen users' emotional state or make them feel anxious. Therefore, there is a need for a system that can filter false information and provide appropriate notifications based on the user's emotional state.

[0354] 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 acquiring information data from social media, means for preprocessing the acquired information data, means for determining the authenticity of the information using a machine learning model based on the preprocessed data, means for filtering the determined false information, means for analyzing the emotional state of the user, means for adjusting the filtering result and notification content based on the emotional state, and means for notifying the user of the filtered information. This not only filters false information but also enables appropriate notification that takes into account the emotional state of the user.

[0355] "Social media" is an online platform that allows users to share information and communicate.

[0356] "Information data" refers to digital content such as text, images, videos, comments, news articles, and advertisements posted on social media.

[0357] "Preprocessing" refers to the initial data processing performed on acquired information data, and includes tokenization, removal of unnecessary words, data normalization, and the like.

[0358] A "machine learning model" is an algorithm or method that learns from past data and makes predictions and classifications for new data.

[0359] A "means for determining the authenticity of information" is a process or device that uses a machine learning model to determine whether the information is accurate based on preprocessed information data.

[0360] "Filtering" is the process of sorting data based on specific criteria to remove irrelevant or unwanted data.

[0361] An "emotional state" refers to the mental and emotional state a user is in at a particular moment and can be classified as positive, negative, neutral, etc.

[0362] An "emotion engine" is an algorithm or software that analyzes a user's emotional state from their posts and actions.

[0363] "Notifications" are messages or alerts provided to users based on filtered information or the results.

[0364] The present invention relates to a system for filtering false information on social media and notifying users in consideration of their feelings. Specific embodiments are described below.

[0365] The system mainly consists of a server, a device, a user, and an emotion engine. The server obtains information data from social media, preprocesses it, and then uses a machine learning model to determine whether the information is true or false. Inaccurate or harmful information is filtered out, and the user is notified of the results. The emotion engine also analyzes the user's emotional state and adjusts the notification content and filtering criteria based on this.

[0366] Hardware and software used

[0367] Hardware

[0368] Server: High-performance computers (e.g., cloud servers or dedicated physical servers) are used.

[0369] Terminal: The user's smartphone, tablet, or personal computer is used as the terminal.

[0370] software

[0371] Social Media APIs: APIs for retrieving data from social media (e.g., Twitter API, Facebook Graph API).

[0372] Text processing libraries: Libraries such as NLTK and spaCy are used to perform preprocessing for natural language processing.

[0373] Machine learning models: Machine learning algorithms such as RandomForestClassifier are used to identify false information.

[0374] Sentiment Analysis Engine: Sentiment analysis tools such as TextBlob and VADER are used.

[0375] Details of data processing and calculation

[0376] The server retrieves information data (posts, comments, news articles, advertisements, etc.) from social media via APIs. The retrieved information data is pre-processed on the server, which includes tokenization, stop word removal, stemming, lameness, etc.

[0377] The preprocessed data is then fed into a machine learning model to determine whether the information is true or false. For example, a RandomForest Classifier trained on past data is used. If the information is determined to be false, the data is filtered out.

[0378] Meanwhile, the emotion engine analyzes the user's emotional state. This analysis is done using emotion analysis tools such as TextBlob and VADER. The user's emotional state is classified as positive, negative, neutral, etc.

[0379] The notification content is adjusted based on the filtering results and the user's emotional state. For example, if the user's emotional state is negative, the notification will be softened.

[0380] Specific examples

[0381] 1. News article filtering and sentiment analysis

[0382] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be unreliable, the server filters it out and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server determines that the user is feeling stressed, it will send a gentle notification such as, "This news article is unreliable, so we have temporarily withheld it from display."

[0383] 2. Fraudulent Ad Filtering and Sentiment Analysis

[0384] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high probability of being fraudulent, the server filters the ad and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[0385] Prompt Sentence Examples

[0386] Write example code for a system that filters misinformation on social media and tailors notifications based on the user's emotional state.

[0387] Specifically, it will use Python and include the following functionality:

[0388] Data collection through social media APIs

[0389] Preprocessing of text data (tokenization, stop word removal, stemming, etc.)

[0390] Detecting false information using machine learning models

[0391] Emotion analysis using an emotion engine

[0392] User notification based on judgment results and emotional state

[0393] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0394] Step 1: Data collection

[0395] The server obtains information data from social media using social media APIs (e.g., Twitter API, Facebook Graph API). When a user clicks on a specific post or advertisement on the device, that information data is also sent to the server. The input is the data from the social media post or advertisement, and the output is raw data for preprocessing.

[0396] Step 2: Preprocessing

[0397] The server preprocesses the acquired information data. This preprocessing involves tokenization, which divides the text data into tokens, removing unnecessary vocabulary and stop words, stemming, lameness (standardizing word forms), etc. The input is raw text data, and the output is normalized text data.

[0398] Step 3: Authentication using machine learning models

[0399] The server inputs the preprocessed data into a machine learning model (e.g., RandomForestClassifier) ​​to determine whether the information is true or false. Specifically, the text data is converted into a feature vector, which is then input into a model that has learned the characteristics of false information and scored. The input is normalized text data, and the output is a determination of whether the information is false or not.

[0400] Step 4: Filtering misinformation

[0401] The server filters data determined to be false by the machine learning model. In this step, information determined to be false does not proceed to the next processing step and is processed so that it is not displayed to the user. The input is the truth determination result, and the output is the filtered information data.

[0402] Step 5: Sentiment Analysis

[0403] The server uses an emotion engine (e.g., TextBlob, VADER) to analyze the user's emotional state. It performs sentiment analysis on the text data of user posts and clicked ads, and classifies them as positive, negative, neutral, etc. The input is the user's text data, and the output is the user's emotional state.

[0404] Step 6: Adjust your notifications

[0405] The server adjusts the notification content and filtering criteria based on the emotion analysis results. For example, if the user is in a negative emotional state, the notification will be sent with a more gentle tone. The input is the emotion analysis results and filtering results, and the output is the final notification content sent to the user.

[0406] Step 7: Notification of filtering results

[0407] The terminal displays the notification content sent from the server to the user. For example, it displays a pop-up message saying, "This news article was not displayed due to low reliability." The input is the notification content sent from the server, and the output is the notification displayed on the user's smartphone or personal computer.

[0408] 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.

[0409] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0410] 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.

[0411] [Second embodiment]

[0412] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0413] 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.

[0414] 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).

[0415] 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.

[0416] 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.

[0417] 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).

[0418] 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.

[0419] 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.

[0420] 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.

[0421] 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.

[0422] 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.

[0423] 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."

[0424] The present invention relates to a system for filtering false information on social media, which mainly consists of a server, a terminal, and a user. This system acquires information data from social media, preprocesses it, determines the authenticity of the information using a machine learning model, filters the false information, and notifies the user of the filtering results.

[0425] Program processing (natural language explanation)

[0426] 1. Data Collection

[0427] The server obtains information data from social media platforms (e.g., popular social networking sites) using APIs. API stands for Application Programming Interface, and refers to the rules and protocols that allow software to communicate with other software.

[0428] The device acquires the posting data of the SNS accessed by the user in real time and sends it to the server.

[0429] 2. Data Preprocessing

[0430] The server analyzes the acquired information data and performs tokenization, which is the process of dividing text data into words and phrases.

[0431] Additionally, the server removes unnecessary vocabulary and stop words (e.g., "the" and "and" in English) and normalizes the data. Normalization is the process of unifying data by performing stemming and lameness.

[0432] 3. Learning and judgment using AI models

[0433] The server then inputs the preprocessed data into a machine learning model that learns the characteristics of false information based on past data.

[0434] The server's AI module then scores the information based on newly acquired data. For example, if it calculates that a particular news article has a 0.9 probability of being true, the article is deemed to be highly reliable.

[0435] 4. Filtering and Notifications

[0436] The server filters out information data that is determined to be false and does not send it to the user's terminal.

[0437] The device will notify the user of the filtering results in the form of a pop-up or push notification.

[0438] Users can use social media with peace of mind.

[0439] Specific examples

[0440] Example 1: Filtering news articles

[0441] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server retrieves the data and performs preprocessing. It then uses a machine learning model to determine whether the news article is authentic or not. For example, if the news article is determined to be unreliable with a score of 0.8, the server filters the article and does not display it on the user's device. The device then notifies the user that it has been filtered.

[0442] Example 2: Filtering fraudulent ads

[0443] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud, with a score of 0.85, the server filters the ad and does not display it on the user's device. The device then notifies the user that the ad has been filtered.

[0444] In this way, the present invention provides an environment in which users can use social media with peace of mind.

[0445] The processing flow will be explained below.

[0446] Step 1: Data collection

[0447] The server retrieves information data from the social media platform via API, including user posts, comments, news articles, advertisements, etc.

[0448] The device automatically captures the social media posts that the user views or interacts with and sends the data to the server in real time.

[0449] Step 2: Preprocessing

[0450] The server receives the acquired information data and performs text analysis, specifically, using tokenization to divide the text into words and phrases.

[0451] The server removes stop words (e.g., common conjunctions and prepositions) and performs stemming to improve the accuracy of the data.

[0452] Step 3: Training the machine learning model

[0453] The server matches the pre-processed data with existing datasets and uses machine learning models to learn new patterns, including spam filtering, fake news detection, and fraudulent ad detection.

[0454] The server evaluates the accuracy of the trained model and adjusts parameters as needed.

[0455] Step 4: Determine the information

[0456] The server feeds new data into a machine learning model in real time to determine whether the information is true or false. For example, a particular news article may be identified as fake news with a probability of 0.7.

[0457] The server outputs the information judgment result as a score.

[0458] Step 5: Filtering

[0459] The server filters data determined to be false and does not send it to the user's device. For example, data determined to be fraudulent advertisements will not be displayed to the user.

[0460] The server stores the filtered data and uses it as learning data at a later date.

[0461] Step 6: Notifications and Display

[0462] The device will notify the user of the filtering results, for example by displaying a pop-up notification such as "This news article was not displayed due to low reliability."

[0463] Users will receive notifications and can use social media with peace of mind.

[0464] Step 7: Reassess and improve

[0465] The server periodically re-evaluates the accuracy of the filtering and retrains the model, allowing it to keep up with the latest disinformation trends.

[0466] The server receives feedback from users and uses it to improve the system.

[0467] Through the above steps, the present invention provides an environment in which users can use social media with peace of mind.

[0468] Example 1

[0469] 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."

[0470] There is a lot of false information and fraudulent advertisements on social media, making it difficult for users to quickly obtain reliable information. There is also a lack of effective measures to prevent incorrect decisions and actions based on false information. For this reason, it is necessary to provide an environment where users can use social media with peace of mind.

[0471] 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.

[0472] In this invention, the server includes a means for acquiring information data from social media, a means for preprocessing the acquired information data, and a means for determining the authenticity of the information using a machine learning model based on the preprocessed data, thereby making it possible to filter out false information and provide reliable information.

[0473] "Social media" refers to a platform on the Internet where users can disseminate, share, and interact with information.

[0474] "Information data" refers to data including content such as text, images, and videos posted on social media.

[0475] "Preprocessing" refers to the process of analyzing, dividing, and organizing acquired information data, and converting it into a format that is easy for the machine learning model to process.

[0476] "Tokenization" refers to the process of dividing text data into smaller units such as words or phrases.

[0477] "Stop words" refer to common words (e.g., the, and, etc.) that are not necessary for semantic analysis of a sentence in natural language processing, and the purpose is to remove them.

[0478] "Normalization" refers to the process of converting data into a standard form using stemming and lambdaization.

[0479] "Stemming" refers to a technique for extracting only the stem portion of a word.

[0480] "Lametization" refers to the process of converting words into their base forms.

[0481] A "machine learning model" refers to a system that includes algorithms that learn from past data and make predictions or classifications for new data.

[0482] "Scoring" refers to the process by which a machine learning model quantifies the reliability and veracity of information data.

[0483] "Filtering" refers to the process of filtering out unnecessary or inappropriate data based on specific conditions.

[0484] "Notification" refers to the process of informing users of filtered results and important information.

[0485] A "pop-up" is a small window that suddenly appears on a user's screen and is used to display specific information or notifications.

[0486] "Push notification" refers to a form of notification that an application or system automatically sends to a user's device.

[0487] "Database" refers to a system that systematically stores and manages collected data and results.

[0488] This invention provides a method for filtering false information on social media using a system consisting mainly of a server, a terminal, and a user. This system uses a combination of specific software and hardware to perform a series of processes including data collection, data preprocessing, AI-based judgment, filtering, and notification.

[0489] Data collection

[0490] The server retrieves information data from social media platforms using APIs. For example, to collect data from Twitter, the Python library "Tweepy" is used. The server uses "Tweepy" to periodically collect post data related to specific keywords and store it in a database.

[0491] The device collects data on posts from social media sites accessed by the user in real time and sends it to the server. This process is achieved using technologies such as JavaScript's "Fetch API" and Swift. The device sends the post data to the server in response to user actions, and the data is collected on the server side.

[0492] Data Preprocessing

[0493] The server analyzes the collected information data and performs tokenization using the Python library NLTK, which splits the text into words and stores them in a database.

[0494] The server then removes unnecessary vocabulary and stop words from the tokenized data, performs stemming, and lambdaization, and standardizes the data using libraries like NLTK and SpaCy, converting the data into a format suitable for processing by machine learning models.

[0495] Learning and judgment by AI model

[0496] The server then feeds the pre-processed data into machine learning models, specifically pre-trained BERT and GPT models, which excel at natural language processing and learn the characteristics of misinformation from the data.

[0497] The server's AI module scores the authenticity of the information based on the newly input data, and the results are stored in a database. For example, if the AI ​​module calculates that a particular news article has a 0.9 probability of being true, the article is deemed to be highly reliable.

[0498] Filtering and Notifications

[0499] The server filters out false information based on the scoring results of the machine learning model, and if the score falls below a certain threshold, the information is prevented from being sent to the user's device.

[0500] The device will notify the user of the filtering results. By notifying the user using a pop-up or push notification, the user can check in real time that the information has been filtered. For example, if a particular news article is deemed to be unreliable and filtered, the device will notify the user of the result.

[0501] Specific examples

[0502] 1. Filtering news articles

[0503] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server retrieves the data and performs preprocessing. It then uses a machine learning model to determine the authenticity of the news article. If the news article is judged to be low reliability with a score of 0.8, the server filters the article and does not display it on the user's device. The device then notifies the user that it has been filtered.

[0504] 2. Filtering fraudulent ads

[0505] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. If the ad is determined to be highly fraudulent with a score of 0.85, the server filters the ad and does not display it on the user's device. The device notifies the user that the ad has been filtered.

[0506] This system allows users to reduce the risk of false information and fraud on social media, allowing them to use only reliable information with peace of mind.

[0507] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0508] Step 1: Data collection

[0509] Specific operation:

[0510] The server retrieves information data using the API of the social media platform. Specifically, it uses the Python library "Tweepy" to collect data from Twitter. It periodically monitors specific keywords (e.g., "COVID-19" or "disaster information") and retrieves new posts.

[0511] input:

[0512] Social media platform API keys

[0513] Specific keywords to collect

[0514] Data processing:

[0515] Call API to get post data related to keywords

[0516] output:

[0517] Acquired post data (text, user information, post date and time, etc.)

[0518] Step 2: Data Preprocessing

[0519] Specific operation:

[0520] The server preprocesses the acquired information data. First, it performs tokenization, splitting the text into words using the Python library "NLTK."

[0521] Next, we remove unnecessary vocabulary and stop words, and perform stemming and lambdaization using tools like NLTK and SpaCy.

[0522] input:

[0523] Obtained post data (output from Step 1)

[0524] Data processing:

[0525] Text tokenization (splitting into words)

[0526] Stop word removal

[0527] Stemming and Lamellarization

[0528] output:

[0529] Preprocessed text data

[0530] Step 3: Learning and judgment by AI model

[0531] Specific operation:

[0532] The server then feeds the preprocessed data into a machine learning model (e.g., the BERT model), which has learned the characteristics of false information based on past data.

[0533] The server's AI module scores the truth of the information based on the newly entered data, and the scoring result (e.g., 0.0 to 1.0) is stored in a database.

[0534] input:

[0535] Preprocessed text data (output of step 2)

[0536] Data Calculation:

[0537] Judgment and scoring using machine learning models

[0538] output:

[0539] Reliability score (e.g., 0.8)

[0540] Step 4: Filtering

[0541] Specific operation:

[0542] The server filters out false information based on the scoring results. If the score is below a certain threshold (e.g., 0.5), the information is not sent to the user's device.

[0543] The filtered information is either deleted from the database or moved to another table.

[0544] input:

[0545] Determined reliability scores and corresponding information data (output of Step 3)

[0546] Data processing:

[0547] Filtering data based on scores

[0548] Save or delete filtered results

[0549] output:

[0550] Filtered and clean information data

[0551] Step 5: Notification

[0552] Specific operation:

[0553] The device will notify the user of the filtering results, using a pop-up or push notification to let the user know that false information has been filtered.

[0554] The user confirms the notification and views the clean information data.

[0555] input:

[0556] Filtered and clean information data (output of step 4)

[0557] Data processing:

[0558] Generate and send notifications

[0559] output:

[0560] Notification of filtering results

[0561] (Application example 1)

[0562] 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."

[0563] Traditional social media platforms require users to independently determine the authenticity of the information they access. However, in today's information-intensive world, there is a growing risk that users will be misled by unreliable information, leading to the spread of inaccurate information. Furthermore, there is a lack of systems that can effectively filter false information in real time and immediately warn users. To solve this problem, a system is needed that can detect false information with high accuracy and in real time and notify users.

[0564] 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.

[0565] In this invention, the server includes means for acquiring information data from social media, means for preprocessing the acquired information data, and means for determining the authenticity of the information using a machine learning model based on the preprocessed data. This enables a system including means for tokenizing the information data and removing unnecessary information, means for notifying the filtering results in real time, and means for issuing a warning when false information is detected.

[0566] "Social media" is a general term for online platforms that allow users to share information and communicate with each other.

[0567] "Information Data" refers to data such as text, images, videos, and links generated by users on social media.

[0568] "Preprocessing" is the process of converting acquired information data into a format that can be applied to a machine learning model through methods such as tokenization and removal of unnecessary information.

[0569] A "machine learning model" is a system that uses algorithms to learn patterns and specific characteristics based on past data to determine the authenticity of new data.

[0570] "Disinformation" is information that is intentionally or unintentionally provided in an inaccurate or misleading manner.

[0571] "Filtering" is the process of identifying and removing false information from acquired information data.

[0572] "Tokenization" is the process of dividing text data into words and phrases.

[0573] "Nutrition" is the process of removing portions of text data that are not required for analysis, based on stop words or specific patterns.

[0574] "Real-time" means that data acquisition and processing are carried out immediately, and the results are notified to the user without making them wait unintentionally.

[0575] "Notification" is an action to notify the user of filtering results or warnings.

[0576] MODE FOR CARRYING OUT THE INVENTION

[0577] To implement this invention, three main components, a server, a terminal, and a user, must cooperate. The role and processing method of each component will be specifically explained below.

[0578] server

[0579] The server plays a key role in acquiring information data from social media, preprocessing it, and then using machine learning models to determine its authenticity. Specifically, it uses the following hardware and software:

[0580] Hardware:

[0581] High-performance processor (e.g., Intel Xeon)

[0582] Large memory capacity (e.g. 64GB RAM or more)

[0583] software:

[0584] Sinatra (Ruby web framework)

[0585] NLTK library (natural language processing library)

[0586] Transformers library (including the BERT model)

[0587] The server first collects information data in real time using social media APIs. The information data then undergoes a preprocessing step, where it is tokenized and stop words are removed. Based on this preprocessed data, the reliability of the information is scored using the BERT model. Based on the results, false information is filtered and notifications are sent to the device.

[0588] Terminal

[0589] The terminal is a device such as a smartphone or tablet used by the user, and serves to notify the user of the filtering results received from the server.

[0590] Hardware:

[0591] Smartphones and tablets (e.g. iPhone, Android devices)

[0592] software:

[0593] Web browser or native application

[0594] The device displays notifications received from the server to the user as pop-up or push notifications. For example, if a user clicks on a specific article in a social media app and the information is determined to be unreliable, a warning pop-up will be displayed saying, "This news article is unreliable."

[0595] User

[0596] The user is an end user of social media. The user can operate their device and check the authenticity of the information sent from the server in real time. This significantly reduces the risk of being misled by incorrect or fake information.

[0597] Specific examples

[0598] For example, consider a case where the reliability of information is evaluated using the BERT model in a news application. When a user taps on a specific article in the news app, the server collects the article's text data, preprocesses it, and then analyzes it using the BERT model. If the analysis results indicate that the information is not reliable, a notification is sent to the device and a warning is displayed to the user.

[0599] Prompt Sentence Examples

[0600] "This is an example news article text for analysis."

[0601] In this way, the present invention can provide users with access to reliable information and protection from inaccurate information.

[0602] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0603] Step 1: Data collection

[0604] The server collects information data in real time using social media APIs. The input data is text data of the latest posts on social media and news articles. The server obtains this data via the API and sends it to the next preprocessing step.

[0605] Step 2: Data Preprocessing

[0606] The server performs preprocessing on the collected information data. Specifically, it performs tokenization to divide the text into words or phrases. It also removes stop words and eliminates unnecessary words. In this way, the information contained in the input data is made clearer and converted into a format suitable for analysis by machine learning models. The output data is preprocessed, clean text data.

[0607] Step 3: Authentication using machine learning models

[0608] The server inputs the preprocessed data into a generative AI model (e.g., BERT) to determine whether the text is true or false. The server uses a machine learning model to analyze the input data and generate a score indicating whether it is trustworthy information. Based on this scoring, the server determines whether the information is accurate or false. The output data is a reliability score and a determination result corresponding to each piece of text data.

[0609] Step 4: Filtering

[0610] The server filters out false information based on the results determined by the machine learning model. In this step, information data determined to be unreliable is removed or flagged as unreliable. The output data is filtered, reliable information data.

[0611] Step 5: Notification

[0612] The terminal receives the filtering results from the server and notifies the user. The terminal displays a warning popup for unreliable information and displays the normal screen for reliable information. In this step, the user is prompted to take specific action. The output data is a notification message on the user's device.

[0613] Specific examples of specific actions

[0614] Data collection:

[0615] The server retrieves the news article "This is an example news article text for analysis." via the API.

[0616] Data preprocessing:

[0617] The server converts the retrieved text from "This is an example news article text for analysis." into a tokenized format such as "this example news article text analysis," and removes stop words to generate clean data.

[0618] Machine learning model for determining truth:

[0619] The server inputs the clean text data into the BERT model and scores it. For example, if the confidence score is 0.4, the information is determined to be false.

[0620] filtering:

[0621] Information with a reliability score of 0.4 is filtered as false information and users will not receive it.

[0622] notification:

[0623] Based on the filtering results received by the device from the server, a pop-up warning is displayed to the user stating, "This news article is unreliable."

[0624] 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.

[0625] The present invention relates to a system that filters false information on social media and notifies users while taking their emotions into consideration. The system mainly consists of a server, a terminal, a user, and an emotion engine. The system acquires information data from social media, preprocesses it, and then uses a machine learning model to determine whether the information is true or false. Inaccurate or harmful information is filtered, and the user is notified of the results. The emotion engine is also used to analyze the user's emotional state and adjust the notification content and filtering criteria based on this.

[0626] Program processing (natural language explanation)

[0627] 1. Data Collection

[0628] The server retrieves information data from social media platforms (e.g., general social networking sites) via APIs, including user posts, comments, news articles, advertisements, etc.

[0629] The device acquires the posting data of the SNS accessed by the user in real time and transmits the data to the server.

[0630] 2. Pretreatment

[0631] The server receives the acquired information data and performs analysis, which includes tokenization, which divides the text data into words and phrases.

[0632] The server removes unnecessary vocabulary and stop words and normalizes the data, a process that involves stemming and lameness.

[0633] 3. Learning and assessment of machine learning models

[0634] The server feeds the preprocessed data into a machine learning model that learns the characteristics of misinformation based on past data.

[0635] The server then scores the information based on the newly acquired data. For example, if a particular news article is judged to be fake news with a probability of 0.8, the article is deemed to be false information.

[0636] 4. Emotion analysis using an emotion engine

[0637] The server uses an emotion engine to analyze the emotions users express on social media, categorizing them as positive, negative, neutral, etc.

[0638] The server obtains the user's emotional state and reflects it in the filtering results and notification content.

[0639] 5. Filtering and Notifications

[0640] The server filters data determined to be false and does not send it to the user's device.

[0641] The server notifies the user of the filtering results, for example by displaying a pop-up message saying, "This news article was not displayed because it was not reliable."

[0642] The device selects an appropriate notification method depending on the user's emotional state. For example, if the emotional state is negative, the notification will be delivered in a gentler manner.

[0643] Specific examples

[0644] Example 1: News article filtering and sentiment analysis

[0645] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in reliability with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server detects that the user is feeling stressed, it will display a gentle notification such as, "This news article has been temporarily suppressed due to its low reliability."

[0646] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[0647] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[0648] In this way, the present invention is a system that provides an environment in which users can use social media with peace of mind, and at the same time provides appropriate notifications taking into account the user's emotional state.

[0649] The processing flow will be explained below.

[0650] Step 1: Data collection

[0651] The server retrieves information data from popular social media platforms via APIs, including user posts, comments, news articles, and advertisements.

[0652] The device captures the social networking site posting data that the user has viewed or operated in real time and immediately transmits that data to the server.

[0653] Step 2: Preprocessing

[0654] The server analyzes the received information data, specifically by using tokenization to divide the text data into words and phrases.

[0655] The server normalizes the data by removing stop words (e.g., common conjunctions and prepositions), stemming, and lame- tizing.

[0656] Step 3: Training the machine learning model

[0657] The server then matches the pre-processed data with existing datasets and uses machine learning models to learn new patterns, resulting in a model that can detect misinformation with high accuracy.

[0658] The server evaluates the accuracy of the model and retrains it if necessary.

[0659] Step 4: Determine the information

[0660] The server feeds new data into the machine learning model in real time. For example, if a particular news article has a 0.75 probability of being fake news, the article is flagged as false.

[0661] The server outputs the information judgment results as a score, which is used as a basis for filtering decisions.

[0662] Step 5: Sentiment Analysis

[0663] The server uses an emotion engine to analyze the user's emotions, for example by analyzing text and behavioral data to determine what emotions (positive, negative, neutral) the user is feeling.

[0664] The server stores the analysis results in a database, making them available for subsequent processing.

[0665] Step 6: Filtering and Notifications

[0666] The server filters data determined to be false and does not send it to the user's device.

[0667] The server generates appropriate notification content based on the results of emotion analysis, depending on the user's emotional state. For example, if the user is feeling anxious, the server will send a notification in a calm manner.

[0668] The device will notify the user of the filtering results, for example, in the form of a pop-up message saying, "This news article was not displayed due to its low reliability."

[0669] Step 7: Reassess and improve

[0670] The server periodically re-evaluates the accuracy of filtering and sentiment analysis and re-trains the model if necessary.

[0671] The server collects feedback from users and uses it to improve the system.

[0672] Specific examples

[0673] Example 1: News article filtering and sentiment analysis

[0674] When a user clicks on a specific news article on a social networking site, the device immediately sends the news article data to a server. The server preprocesses the data and uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in credibility with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server detects that the user is feeling stressed, it may display a gentle notification such as, "This news article is low in credibility, so we have temporarily withheld it from view."

[0675] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[0676] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[0677] In this way, the present invention is a system that provides an environment in which users can use social media with peace of mind, and at the same time provides appropriate notifications taking into account the user's emotional state.

[0678] Example 2

[0679] 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."

[0680] There is a lot of information circulating on social media, including fake and incorrect information. This fake information can cause confusion and anxiety among users. Furthermore, if information is notified without taking into account the user's emotional state, it can cause further stress and discomfort. To solve this situation, a notification system that can effectively filter fake information and take into account the user's emotional state is needed.

[0681] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0682] In this invention, the server includes means for acquiring information data from social media, means for preprocessing the acquired information data, means for determining the authenticity of the information using a machine learning model based on the preprocessed data, means for filtering the determined false information, means for analyzing the emotional state of the user, means for adjusting filtering criteria and notification content based on the analysis results, and means for notifying the user of the filtered information. This makes it possible to efficiently filter false information and provide appropriate notifications that take the user's emotions into consideration.

[0683] "Social media" is a platform on the Internet that allows users to share information and communicate.

[0684] "Information data" refers to all information, including text data and media data such as user posts, comments, news articles, and advertisements.

[0685] "Preprocessing" refers to analyzing the acquired information data and performing tokenization and normalization processing.

[0686] "Tokenization" is the process of dividing text data into words and phrases.

[0687] "Normalization" is the process of stemming, lamellarizing, and removing unnecessary vocabulary and stop words from data.

[0688] A "machine learning model" is an algorithm that learns from past data and analyzes and classifies new data.

[0689] An "emotion engine" is a software tool for analyzing a user's emotional state, categorizing it as positive, negative, neutral, etc.

[0690] "Filtering" is the process of sorting information data based on specific criteria to remove unwanted or harmful information.

[0691] "Notifications" are messages or alerts that communicate filtering results or other information to the user.

[0692] MODE FOR CARRYING OUT THE INVENTION

[0693] The present invention relates to a system that filters false information on social media and notifies users while taking their emotions into consideration. The system mainly consists of a server, a terminal, a user, and an emotion engine. The system acquires information data from social media, preprocesses it, and then uses a machine learning model to determine the authenticity of the information. Inaccurate or harmful information is filtered, and the user is notified of the results. The emotion engine is also used to analyze the user's emotional state and adjust the notification content and filtering criteria based on this.

[0694] Server Features

[0695] The server periodically retrieves information data from social media platforms via API. This information includes user posts, comments, news articles, advertisements, etc. The retrieved information data is preprocessed and analyzed. Preprocessing includes tokenization (dividing text data into words and phrases), stemming, lameness (standardizing word forms), and removal of unnecessary vocabulary and stop words.

[0696] The preprocessed data is then fed into a machine learning model (e.g., BERT or RoBERTa). This model has learned the characteristics of false information based on past data, and scores new data for authenticity. Data that is determined to be false is then filtered out.

[0697] The server also analyzes the user's emotional state using an emotion engine (e.g., IBM Watson Emotion Analysis). Based on the results of the emotion analysis, the filtering criteria and notification content are adjusted. For example, if the user is in a negative emotional state, the notification will be adjusted to a more gentle expression.

[0698] Device Features

[0699] The device collects data on social media posts accessed by users in real time and sends it to a server. When a user clicks on a specific news article or advertisement, the device also immediately sends that information to the server.

[0700] The device also notifies the user of the filtering results. This notification is adjusted based on the analysis results of the emotion engine. For example, if the user's emotional state is negative, the notification content will be milder.

[0701] Specific examples

[0702] Example 1: News article filtering and sentiment analysis

[0703] When a user clicks on a specific news article on a social media site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in credibility with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the notification content accordingly. For example, if the server detects that the user is feeling stressed, it may display a gentle notification such as, "This news article has been temporarily suppressed due to its low credibility."

[0704] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[0705] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[0706] Example of input prompt for generative AI model

[0707] Example prompt 1:

[0708] "Please analyze the reliability of this news article."

[0709] Example prompt 2:

[0710] "Determine whether this ad is potentially fraudulent."

[0711] Example prompt 3:

[0712] "Analyze the sentiment of your users' latest posts and tailor your notifications based on that."

[0713] As described above, the present invention is a system that provides an environment in which users can use social media with peace of mind, effectively filters out false information, and provides appropriate notifications based on the user's emotions.

[0714] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0715] Step 1:

[0716] The server retrieves information data from social media platforms via APIs. As input, it receives data provided by the social media APIs (user posts, comments, news articles, advertisements, etc.). As output, the raw data is stored on the server.

[0717] Step 2:

[0718] The device collects data on posts from social media sites accessed by users in real time and sends that data to a server. As input, it receives user behavior data (such as clicked news articles or advertisements). As output, this information is transferred to the server.

[0719] Step 3:

[0720] The server analyzes the received information data and performs tokenization. As input, it takes raw data and splits it into words and phrases. As output, it generates tokenized data.

[0721] Step 4:

[0722] The server normalizes the data by stemming, lameness, and removing unnecessary vocabulary and stop words. As input, it takes tokenized data and performs normalization on it. As output, it gets clean, normalized data.

[0723] Step 5:

[0724] The server inputs the preprocessed data into a machine learning model (e.g., BERT or RoBERTa). As input, it receives the normalized data and feeds it into the machine learning model. As output, it obtains a truth score for each piece of data.

[0725] Step 6:

[0726] The server determines whether the information is true or false based on the output from the machine learning model. As input, it receives the score returned from the machine learning model, sets a threshold, and determines whether the information is true or false. The output is classified into data that is determined to be false and other data.

[0727] Step 7:

[0728] The device monitors users' latest posts and comments and sends the data to the server. As input, it collects data about users' emotions. As output, it transfers the data to the server.

[0729] Step 8:

[0730] The server uses an emotion engine to analyze the user's emotional state. It receives user posts and comment data as input and performs emotion analysis. The output is an emotion classification result such as positive, negative, or neutral.

[0731] Step 9:

[0732] The server filters data that is determined to be false. As input, it receives data that has been determined to be true or false, and filters out data that it recognizes as false. As output, it obtains a filtered, safe dataset.

[0733] Step 10:

[0734] The server notifies the user of the filtering results. As input, it receives the filtering result data and converts it into a notification message. As output, it generates the notification message.

[0735] Step 11:

[0736] The device selects an appropriate notification method according to the user's emotional state. As input, it receives the emotion analysis results and filtering results and determines the notification method. As output, the adjusted notification message is displayed to the user.

[0737] (Application example 2)

[0738] 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."

[0739] A lot of information is circulating on social media, but some of it is false or harmful, which increases the risk of users acting on incorrect information. This can also worsen users' emotional state or make them feel anxious. Therefore, there is a need for a system that can filter false information and provide appropriate notifications based on the user's emotional state.

[0740] 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 acquiring information data from social media, means for preprocessing the acquired information data, means for determining the authenticity of the information using a machine learning model based on the preprocessed data, means for filtering the determined false information, means for analyzing the emotional state of the user, means for adjusting the filtering result and notification content based on the emotional state, and means for notifying the user of the filtered information. This not only filters false information but also enables appropriate notification that takes into account the emotional state of the user.

[0741] "Social media" is an online platform that allows users to share information and communicate.

[0742] "Information data" refers to digital content such as text, images, videos, comments, news articles, and advertisements posted on social media.

[0743] "Preprocessing" refers to the initial data processing performed on acquired information data, and includes tokenization, removal of unnecessary words, data normalization, and the like.

[0744] A "machine learning model" is an algorithm or method that learns from past data and makes predictions and classifications for new data.

[0745] A "means for determining the authenticity of information" is a process or device that uses a machine learning model to determine whether the information is accurate based on preprocessed information data.

[0746] "Filtering" is the process of sorting data based on specific criteria to remove irrelevant or unwanted data.

[0747] An "emotional state" refers to the mental and emotional state a user is in at a particular moment and can be classified as positive, negative, neutral, etc.

[0748] An "emotion engine" is an algorithm or software that analyzes a user's emotional state from their posts and actions.

[0749] "Notifications" are messages or alerts provided to users based on filtered information or the results.

[0750] The present invention relates to a system for filtering false information on social media and notifying users in consideration of their feelings. Specific embodiments are described below.

[0751] The system mainly consists of a server, a device, a user, and an emotion engine. The server obtains information data from social media, preprocesses it, and then uses a machine learning model to determine whether the information is true or false. Inaccurate or harmful information is filtered out, and the user is notified of the results. The emotion engine also analyzes the user's emotional state and adjusts the notification content and filtering criteria based on this.

[0752] Hardware and software used

[0753] Hardware

[0754] Server: High-performance computers (e.g., cloud servers or dedicated physical servers) are used.

[0755] Terminal: The user's smartphone, tablet, or personal computer is used as the terminal.

[0756] software

[0757] Social Media APIs: APIs for retrieving data from social media (e.g., Twitter API, Facebook Graph API).

[0758] Text processing libraries: Libraries such as NLTK and spaCy are used to perform preprocessing for natural language processing.

[0759] Machine learning models: Machine learning algorithms such as RandomForestClassifier are used to identify false information.

[0760] Sentiment Analysis Engine: Sentiment analysis tools such as TextBlob and VADER are used.

[0761] Details of data processing and calculation

[0762] The server retrieves information data (posts, comments, news articles, advertisements, etc.) from social media via APIs. The retrieved information data is pre-processed on the server, which includes tokenization, stop word removal, stemming, lameness, etc.

[0763] The preprocessed data is then fed into a machine learning model to determine whether the information is true or false. For example, a RandomForest Classifier trained on past data is used. If the information is determined to be false, the data is filtered out.

[0764] Meanwhile, the emotion engine analyzes the user's emotional state. This analysis is done using emotion analysis tools such as TextBlob and VADER. The user's emotional state is classified as positive, negative, neutral, etc.

[0765] The notification content is adjusted based on the filtering results and the user's emotional state. For example, if the user's emotional state is negative, the notification will be softened.

[0766] Specific examples

[0767] 1. News article filtering and sentiment analysis

[0768] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be unreliable, the server filters it out and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server determines that the user is feeling stressed, it will send a gentle notification such as, "This news article is unreliable, so we have temporarily withheld it from display."

[0769] 2. Fraudulent Ad Filtering and Sentiment Analysis

[0770] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high probability of being fraudulent, the server filters the ad and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[0771] Prompt Sentence Examples

[0772] Write example code for a system that filters misinformation on social media and tailors notifications based on the user's emotional state.

[0773] Specifically, it will use Python and include the following functionality:

[0774] Data collection through social media APIs

[0775] Preprocessing of text data (tokenization, stop word removal, stemming, etc.)

[0776] Detecting false information using machine learning models

[0777] Emotion analysis using an emotion engine

[0778] User notification based on judgment results and emotional state

[0779] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0780] Step 1: Data collection

[0781] The server obtains information data from social media using social media APIs (e.g., Twitter API, Facebook Graph API). When a user clicks on a specific post or advertisement on the device, that information data is also sent to the server. The input is the data from the social media post or advertisement, and the output is raw data for preprocessing.

[0782] Step 2: Preprocessing

[0783] The server preprocesses the acquired information data. This preprocessing involves tokenization, which divides the text data into tokens, removing unnecessary vocabulary and stop words, stemming, lameness (standardizing word forms), etc. The input is raw text data, and the output is normalized text data.

[0784] Step 3: Authentication using machine learning models

[0785] The server inputs the preprocessed data into a machine learning model (e.g., RandomForestClassifier) ​​to determine whether the information is true or false. Specifically, the text data is converted into a feature vector, which is then input into a model that has learned the characteristics of false information and scored. The input is normalized text data, and the output is a determination of whether the information is false or not.

[0786] Step 4: Filtering misinformation

[0787] The server filters data determined to be false by the machine learning model. In this step, information determined to be false does not proceed to the next processing step and is processed so that it is not displayed to the user. The input is the truth determination result, and the output is the filtered information data.

[0788] Step 5: Sentiment Analysis

[0789] The server uses an emotion engine (e.g., TextBlob, VADER) to analyze the user's emotional state. It performs sentiment analysis on the text data of user posts and clicked ads, and classifies them as positive, negative, neutral, etc. The input is the user's text data, and the output is the user's emotional state.

[0790] Step 6: Adjust your notifications

[0791] The server adjusts the notification content and filtering criteria based on the emotion analysis results. For example, if the user is in a negative emotional state, the notification will be sent with a more gentle tone. The input is the emotion analysis results and filtering results, and the output is the final notification content sent to the user.

[0792] Step 7: Notification of filtering results

[0793] The terminal displays the notification content sent from the server to the user. For example, it displays a pop-up message saying, "This news article was not displayed due to low reliability." The input is the notification content sent from the server, and the output is the notification displayed on the user's smartphone or personal computer.

[0794] 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.

[0795] 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.

[0796] 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.

[0797] [Third embodiment]

[0798] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0799] 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.

[0800] 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).

[0801] 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.

[0802] 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.

[0803] 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).

[0804] 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.

[0805] 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.

[0806] 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.

[0807] 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.

[0808] 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.

[0809] 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."

[0810] The present invention relates to a system for filtering false information on social media, which mainly consists of a server, a terminal, and a user. This system acquires information data from social media, preprocesses it, determines the authenticity of the information using a machine learning model, filters the false information, and notifies the user of the filtering results.

[0811] Program processing (natural language explanation)

[0812] 1. Data Collection

[0813] The server obtains information data from social media platforms (e.g., popular social networking sites) using APIs. API stands for Application Programming Interface, and refers to the rules and protocols that allow software to communicate with other software.

[0814] The device acquires the posting data of the SNS accessed by the user in real time and sends it to the server.

[0815] 2. Data Preprocessing

[0816] The server analyzes the acquired information data and performs tokenization, which is the process of dividing text data into words and phrases.

[0817] Additionally, the server removes unnecessary vocabulary and stop words (e.g., "the" and "and" in English) and normalizes the data. Normalization is the process of unifying data by performing stemming and lameness.

[0818] 3. Learning and judgment using AI models

[0819] The server then inputs the preprocessed data into a machine learning model that learns the characteristics of false information based on past data.

[0820] The server's AI module then scores the information based on newly acquired data. For example, if it calculates that a particular news article has a 0.9 probability of being true, the article is deemed to be highly reliable.

[0821] 4. Filtering and Notifications

[0822] The server filters out information data that is determined to be false and does not send it to the user's terminal.

[0823] The device will notify the user of the filtering results in the form of a pop-up or push notification.

[0824] Users can use social media with peace of mind.

[0825] Specific examples

[0826] Example 1: Filtering news articles

[0827] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server retrieves the data and performs preprocessing. It then uses a machine learning model to determine whether the news article is authentic or not. For example, if the news article is determined to be unreliable with a score of 0.8, the server filters the article and does not display it on the user's device. The device then notifies the user that it has been filtered.

[0828] Example 2: Filtering fraudulent ads

[0829] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud, with a score of 0.85, the server filters the ad and does not display it on the user's device. The device then notifies the user that the ad has been filtered.

[0830] In this way, the present invention provides an environment in which users can use social media with peace of mind.

[0831] The processing flow will be explained below.

[0832] Step 1: Data collection

[0833] The server retrieves information data from the social media platform via API, including user posts, comments, news articles, advertisements, etc.

[0834] The device automatically captures the social media posts that the user views or interacts with and sends the data to the server in real time.

[0835] Step 2: Preprocessing

[0836] The server receives the acquired information data and performs text analysis, specifically, using tokenization to divide the text into words and phrases.

[0837] The server removes stop words (e.g., common conjunctions and prepositions) and performs stemming to improve the accuracy of the data.

[0838] Step 3: Training the machine learning model

[0839] The server matches the pre-processed data with existing datasets and uses machine learning models to learn new patterns, including spam filtering, fake news detection, and fraudulent ad detection.

[0840] The server evaluates the accuracy of the trained model and adjusts parameters as needed.

[0841] Step 4: Determine the information

[0842] The server feeds new data into a machine learning model in real time to determine whether the information is true or false. For example, a particular news article may be identified as fake news with a probability of 0.7.

[0843] The server outputs the information judgment result as a score.

[0844] Step 5: Filtering

[0845] The server filters data determined to be false and does not send it to the user's device. For example, data determined to be fraudulent advertisements will not be displayed to the user.

[0846] The server stores the filtered data and uses it as learning data at a later date.

[0847] Step 6: Notifications and Display

[0848] The device will notify the user of the filtering results, for example by displaying a pop-up notification such as "This news article was not displayed due to low reliability."

[0849] Users will receive notifications and can use social media with peace of mind.

[0850] Step 7: Reassess and improve

[0851] The server periodically re-evaluates the accuracy of the filtering and retrains the model, allowing it to keep up with the latest disinformation trends.

[0852] The server receives feedback from users and uses it to improve the system.

[0853] Through the above steps, the present invention provides an environment in which users can use social media with peace of mind.

[0854] Example 1

[0855] 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."

[0856] There is a lot of false information and fraudulent advertisements on social media, making it difficult for users to quickly obtain reliable information. There is also a lack of effective measures to prevent incorrect decisions and actions based on false information. For this reason, it is necessary to provide an environment where users can use social media with peace of mind.

[0857] 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.

[0858] In this invention, the server includes a means for acquiring information data from social media, a means for preprocessing the acquired information data, and a means for determining the authenticity of the information using a machine learning model based on the preprocessed data, thereby making it possible to filter out false information and provide reliable information.

[0859] "Social media" refers to a platform on the Internet where users can disseminate, share, and interact with information.

[0860] "Information data" refers to data including content such as text, images, and videos posted on social media.

[0861] "Preprocessing" refers to the process of analyzing, dividing, and organizing acquired information data, and converting it into a format that is easy for the machine learning model to process.

[0862] "Tokenization" refers to the process of dividing text data into smaller units such as words or phrases.

[0863] "Stop words" refer to common words (e.g., the, and, etc.) that are not necessary for semantic analysis of a sentence in natural language processing, and the purpose is to remove them.

[0864] "Normalization" refers to the process of converting data into a standard form using stemming and lambdaization.

[0865] "Stemming" refers to a technique for extracting only the stem portion of a word.

[0866] "Lametization" refers to the process of converting words into their base forms.

[0867] A "machine learning model" refers to a system that includes algorithms that learn from past data and make predictions or classifications for new data.

[0868] "Scoring" refers to the process by which a machine learning model quantifies the reliability and veracity of information data.

[0869] "Filtering" refers to the process of filtering out unnecessary or inappropriate data based on specific conditions.

[0870] "Notification" refers to the process of informing users of filtered results and important information.

[0871] A "pop-up" is a small window that suddenly appears on a user's screen and is used to display specific information or notifications.

[0872] "Push notification" refers to a form of notification that an application or system automatically sends to a user's device.

[0873] "Database" refers to a system that systematically stores and manages collected data and results.

[0874] This invention provides a method for filtering false information on social media using a system consisting mainly of a server, a terminal, and a user. This system uses a combination of specific software and hardware to perform a series of processes including data collection, data preprocessing, AI-based judgment, filtering, and notification.

[0875] Data collection

[0876] The server retrieves information data from social media platforms using APIs. For example, to collect data from Twitter, the Python library "Tweepy" is used. The server uses "Tweepy" to periodically collect post data related to specific keywords and store it in a database.

[0877] The device collects data on posts from social media sites accessed by the user in real time and sends it to the server. This process is achieved using technologies such as JavaScript's "Fetch API" and Swift. The device sends the post data to the server in response to user actions, and the data is collected on the server side.

[0878] Data Preprocessing

[0879] The server analyzes the collected information data and performs tokenization using the Python library NLTK, which splits the text into words and stores them in a database.

[0880] The server then removes unnecessary vocabulary and stop words from the tokenized data, performs stemming, and lambdaization, and standardizes the data using libraries like NLTK and SpaCy, converting the data into a format suitable for processing by machine learning models.

[0881] Learning and judgment by AI model

[0882] The server then feeds the pre-processed data into machine learning models, specifically pre-trained BERT and GPT models, which excel at natural language processing and learn the characteristics of misinformation from the data.

[0883] The server's AI module scores the authenticity of the information based on the newly input data, and the results are stored in a database. For example, if the AI ​​module calculates that a particular news article has a 0.9 probability of being true, the article is deemed to be highly reliable.

[0884] Filtering and Notifications

[0885] The server filters out false information based on the scoring results of the machine learning model, and if the score falls below a certain threshold, the information is prevented from being sent to the user's device.

[0886] The device will notify the user of the filtering results. By notifying the user using a pop-up or push notification, the user can check in real time that the information has been filtered. For example, if a particular news article is deemed to be unreliable and filtered, the device will notify the user of the result.

[0887] Specific examples

[0888] 1. Filtering news articles

[0889] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server retrieves the data and performs preprocessing. It then uses a machine learning model to determine the authenticity of the news article. If the news article is judged to be low reliability with a score of 0.8, the server filters the article and does not display it on the user's device. The device then notifies the user that it has been filtered.

[0890] 2. Filtering fraudulent ads

[0891] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. If the ad is determined to be highly fraudulent with a score of 0.85, the server filters the ad and does not display it on the user's device. The device notifies the user that the ad has been filtered.

[0892] This system allows users to reduce the risk of false information and fraud on social media, allowing them to use only reliable information with peace of mind.

[0893] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0894] Step 1: Data collection

[0895] Specific operation:

[0896] The server retrieves information data using the API of the social media platform. Specifically, it uses the Python library "Tweepy" to collect data from Twitter. It periodically monitors specific keywords (e.g., "COVID-19" or "disaster information") and retrieves new posts.

[0897] input:

[0898] Social media platform API keys

[0899] Specific keywords to collect

[0900] Data processing:

[0901] Call API to get post data related to keywords

[0902] output:

[0903] Acquired post data (text, user information, post date and time, etc.)

[0904] Step 2: Data Preprocessing

[0905] Specific operation:

[0906] The server preprocesses the acquired information data. First, it performs tokenization, splitting the text into words using the Python library "NLTK."

[0907] Next, we remove unnecessary vocabulary and stop words, and perform stemming and lambdaization using tools like NLTK and SpaCy.

[0908] input:

[0909] Obtained post data (output from Step 1)

[0910] Data processing:

[0911] Text tokenization (splitting into words)

[0912] Stop word removal

[0913] Stemming and Lamellarization

[0914] output:

[0915] Preprocessed text data

[0916] Step 3: Learning and judgment by AI model

[0917] Specific operation:

[0918] The server then feeds the preprocessed data into a machine learning model (e.g., the BERT model), which has learned the characteristics of false information based on past data.

[0919] The server's AI module scores the truth of the information based on the newly entered data, and the scoring result (e.g., 0.0 to 1.0) is stored in a database.

[0920] input:

[0921] Preprocessed text data (output of step 2)

[0922] Data Calculation:

[0923] Judgment and scoring using machine learning models

[0924] output:

[0925] Reliability score (e.g., 0.8)

[0926] Step 4: Filtering

[0927] Specific operation:

[0928] The server filters out false information based on the scoring results. If the score is below a certain threshold (e.g., 0.5), the information is not sent to the user's device.

[0929] The filtered information is either deleted from the database or moved to another table.

[0930] input:

[0931] Determined reliability scores and corresponding information data (output of Step 3)

[0932] Data processing:

[0933] Filtering data based on scores

[0934] Save or delete filtered results

[0935] output:

[0936] Filtered and clean information data

[0937] Step 5: Notification

[0938] Specific operation:

[0939] The device will notify the user of the filtering results, using a pop-up or push notification to let the user know that false information has been filtered.

[0940] The user confirms the notification and views the clean information data.

[0941] input:

[0942] Filtered and clean information data (output of step 4)

[0943] Data processing:

[0944] Generate and send notifications

[0945] output:

[0946] Notification of filtering results

[0947] (Application example 1)

[0948] 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."

[0949] Traditional social media platforms require users to independently determine the authenticity of the information they access. However, in today's information-intensive world, there is a growing risk that users will be misled by unreliable information, leading to the spread of inaccurate information. Furthermore, there is a lack of systems that can effectively filter false information in real time and immediately warn users. To solve this problem, a system is needed that can detect false information with high accuracy and in real time and notify users.

[0950] 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.

[0951] In this invention, the server includes means for acquiring information data from social media, means for preprocessing the acquired information data, and means for determining the authenticity of the information using a machine learning model based on the preprocessed data. This enables a system including means for tokenizing the information data and removing unnecessary information, means for notifying the filtering results in real time, and means for issuing a warning when false information is detected.

[0952] "Social media" is a general term for online platforms that allow users to share information and communicate with each other.

[0953] "Information Data" refers to data such as text, images, videos, and links generated by users on social media.

[0954] "Preprocessing" is the process of converting acquired information data into a format that can be applied to a machine learning model through methods such as tokenization and removal of unnecessary information.

[0955] A "machine learning model" is a system that uses algorithms to learn patterns and specific characteristics based on past data to determine the authenticity of new data.

[0956] "Disinformation" is information that is intentionally or unintentionally provided in an inaccurate or misleading manner.

[0957] "Filtering" is the process of identifying and removing false information from acquired information data.

[0958] "Tokenization" is the process of dividing text data into words and phrases.

[0959] "Nutrition" is the process of removing portions of text data that are not required for analysis, based on stop words or specific patterns.

[0960] "Real-time" means that data acquisition and processing are carried out immediately, and the results are notified to the user without making them wait unintentionally.

[0961] "Notification" is an action to notify the user of filtering results or warnings.

[0962] MODE FOR CARRYING OUT THE INVENTION

[0963] To implement this invention, three main components, a server, a terminal, and a user, must cooperate. The role and processing method of each component will be specifically explained below.

[0964] server

[0965] The server plays a key role in acquiring information data from social media, preprocessing it, and then using machine learning models to determine its authenticity. Specifically, it uses the following hardware and software:

[0966] Hardware:

[0967] High-performance processor (e.g., Intel Xeon)

[0968] Large memory capacity (e.g. 64GB RAM or more)

[0969] software:

[0970] Sinatra (Ruby web framework)

[0971] NLTK library (natural language processing library)

[0972] Transformers library (including the BERT model)

[0973] The server first collects information data in real time using social media APIs. The information data then undergoes a preprocessing step, where it is tokenized and stop words are removed. Based on this preprocessed data, the reliability of the information is scored using the BERT model. Based on the results, false information is filtered and notifications are sent to the device.

[0974] Terminal

[0975] The terminal is a device such as a smartphone or tablet used by the user, and serves to notify the user of the filtering results received from the server.

[0976] Hardware:

[0977] Smartphones and tablets (e.g. iPhone, Android devices)

[0978] software:

[0979] Web browser or native application

[0980] The device displays notifications received from the server to the user as pop-up or push notifications. For example, if a user clicks on a specific article in a social media app and the information is determined to be unreliable, a warning pop-up will be displayed saying, "This news article is unreliable."

[0981] User

[0982] The user is an end user of social media. The user can operate their device and check the authenticity of the information sent from the server in real time. This significantly reduces the risk of being misled by incorrect or fake information.

[0983] Specific examples

[0984] For example, consider a case where the reliability of information is evaluated using the BERT model in a news application. When a user taps on a specific article in the news app, the server collects the article's text data, preprocesses it, and then analyzes it using the BERT model. If the analysis results indicate that the information is not reliable, a notification is sent to the device and a warning is displayed to the user.

[0985] Prompt Sentence Examples

[0986] "This is an example news article text for analysis."

[0987] In this way, the present invention can provide users with access to reliable information and protection from inaccurate information.

[0988] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0989] Step 1: Data collection

[0990] The server collects information data in real time using social media APIs. The input data is text data of the latest posts on social media and news articles. The server obtains this data via the API and sends it to the next preprocessing step.

[0991] Step 2: Data Preprocessing

[0992] The server performs preprocessing on the collected information data. Specifically, it performs tokenization to divide the text into words or phrases. It also removes stop words and eliminates unnecessary words. In this way, the information contained in the input data is made clearer and converted into a format suitable for analysis by machine learning models. The output data is preprocessed, clean text data.

[0993] Step 3: Authentication using machine learning models

[0994] The server inputs the preprocessed data into a generative AI model (e.g., BERT) to determine whether the text is true or false. The server uses a machine learning model to analyze the input data and generate a score indicating whether it is trustworthy information. Based on this scoring, the server determines whether the information is accurate or false. The output data is a reliability score and a determination result corresponding to each piece of text data.

[0995] Step 4: Filtering

[0996] The server filters out false information based on the results determined by the machine learning model. In this step, information data determined to be unreliable is removed or flagged as unreliable. The output data is filtered, reliable information data.

[0997] Step 5: Notification

[0998] The terminal receives the filtering results from the server and notifies the user. The terminal displays a warning popup for unreliable information and displays the normal screen for reliable information. In this step, the user is prompted to take specific action. The output data is a notification message on the user's device.

[0999] Specific examples of specific actions

[1000] Data collection:

[1001] The server retrieves the news article "This is an example news article text for analysis." via the API.

[1002] Data preprocessing:

[1003] The server converts the retrieved text from "This is an example news article text for analysis." into a tokenized format such as "this example news article text analysis," and removes stop words to generate clean data.

[1004] Machine learning model for determining truth:

[1005] The server inputs the clean text data into the BERT model and scores it. For example, if the confidence score is 0.4, the information is determined to be false.

[1006] filtering:

[1007] Information with a reliability score of 0.4 is filtered as false information and users will not receive it.

[1008] notification:

[1009] Based on the filtering results received by the device from the server, a pop-up warning is displayed to the user stating, "This news article is unreliable."

[1010] 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.

[1011] The present invention relates to a system that filters false information on social media and notifies users while taking their emotions into consideration. The system mainly consists of a server, a terminal, a user, and an emotion engine. The system acquires information data from social media, preprocesses it, and then uses a machine learning model to determine whether the information is true or false. Inaccurate or harmful information is filtered, and the user is notified of the results. The emotion engine is also used to analyze the user's emotional state and adjust the notification content and filtering criteria based on this.

[1012] Program processing (natural language explanation)

[1013] 1. Data Collection

[1014] The server retrieves information data from social media platforms (e.g., general social networking sites) via APIs, including user posts, comments, news articles, advertisements, etc.

[1015] The device acquires the posting data of the SNS accessed by the user in real time and transmits the data to the server.

[1016] 2. Pretreatment

[1017] The server receives the acquired information data and performs analysis, which includes tokenization, which divides the text data into words and phrases.

[1018] The server removes unnecessary vocabulary and stop words and normalizes the data, a process that involves stemming and lameness.

[1019] 3. Learning and assessment of machine learning models

[1020] The server feeds the preprocessed data into a machine learning model that learns the characteristics of misinformation based on past data.

[1021] The server then scores the information based on the newly acquired data. For example, if a particular news article is judged to be fake news with a probability of 0.8, the article is deemed to be false information.

[1022] 4. Emotion analysis using an emotion engine

[1023] The server uses an emotion engine to analyze the emotions users express on social media, categorizing them as positive, negative, neutral, etc.

[1024] The server obtains the user's emotional state and reflects it in the filtering results and notification content.

[1025] 5. Filtering and Notifications

[1026] The server filters data determined to be false and does not send it to the user's device.

[1027] The server notifies the user of the filtering results, for example by displaying a pop-up message saying, "This news article was not displayed because it was not reliable."

[1028] The device selects an appropriate notification method depending on the user's emotional state. For example, if the emotional state is negative, the notification will be delivered in a gentler manner.

[1029] Specific examples

[1030] Example 1: News article filtering and sentiment analysis

[1031] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in reliability with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server detects that the user is feeling stressed, it will display a gentle notification such as, "This news article has been temporarily suppressed due to its low reliability."

[1032] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[1033] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[1034] In this way, the present invention is a system that provides an environment in which users can use social media with peace of mind, and at the same time provides appropriate notifications taking into account the user's emotional state.

[1035] The processing flow will be explained below.

[1036] Step 1: Data collection

[1037] The server retrieves information data from popular social media platforms via APIs, including user posts, comments, news articles, and advertisements.

[1038] The device captures the social networking site posting data that the user has viewed or operated in real time and immediately transmits that data to the server.

[1039] Step 2: Preprocessing

[1040] The server analyzes the received information data, specifically by using tokenization to divide the text data into words and phrases.

[1041] The server normalizes the data by removing stop words (e.g., common conjunctions and prepositions), stemming, and lame- tizing.

[1042] Step 3: Training the machine learning model

[1043] The server then matches the pre-processed data with existing datasets and uses machine learning models to learn new patterns, resulting in a model that can detect misinformation with high accuracy.

[1044] The server evaluates the accuracy of the model and retrains it if necessary.

[1045] Step 4: Determine the information

[1046] The server feeds new data into the machine learning model in real time. For example, if a particular news article has a 0.75 probability of being fake news, the article is flagged as false.

[1047] The server outputs the information judgment results as a score, which is used as a basis for filtering decisions.

[1048] Step 5: Sentiment Analysis

[1049] The server uses an emotion engine to analyze the user's emotions, for example by analyzing text and behavioral data to determine what emotions (positive, negative, neutral) the user is feeling.

[1050] The server stores the analysis results in a database, making them available for subsequent processing.

[1051] Step 6: Filtering and Notifications

[1052] The server filters data determined to be false and does not send it to the user's device.

[1053] The server generates appropriate notification content based on the results of emotion analysis, depending on the user's emotional state. For example, if the user is feeling anxious, the server will send a notification in a calm manner.

[1054] The device will notify the user of the filtering results, for example, in the form of a pop-up message saying, "This news article was not displayed due to its low reliability."

[1055] Step 7: Reassess and improve

[1056] The server periodically re-evaluates the accuracy of filtering and sentiment analysis and re-trains the model if necessary.

[1057] The server collects feedback from users and uses it to improve the system.

[1058] Specific examples

[1059] Example 1: News article filtering and sentiment analysis

[1060] When a user clicks on a specific news article on a social networking site, the device immediately sends the news article data to a server. The server preprocesses the data and uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in credibility with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server detects that the user is feeling stressed, it may display a gentle notification such as, "This news article is low in credibility, so we have temporarily withheld it from view."

[1061] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[1062] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[1063] In this way, the present invention is a system that provides an environment in which users can use social media with peace of mind, and at the same time provides appropriate notifications taking into account the user's emotional state.

[1064] Example 2

[1065] 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."

[1066] There is a lot of information circulating on social media, including fake and incorrect information. This fake information can cause confusion and anxiety among users. Furthermore, if information is notified without taking into account the user's emotional state, it can cause further stress and discomfort. To solve this situation, a notification system that can effectively filter fake information and take into account the user's emotional state is needed.

[1067] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1068] In this invention, the server includes means for acquiring information data from social media, means for preprocessing the acquired information data, means for determining the authenticity of the information using a machine learning model based on the preprocessed data, means for filtering the determined false information, means for analyzing the emotional state of the user, means for adjusting filtering criteria and notification content based on the analysis results, and means for notifying the user of the filtered information. This makes it possible to efficiently filter false information and provide appropriate notifications that take the user's emotions into consideration.

[1069] "Social media" is a platform on the Internet that allows users to share information and communicate.

[1070] "Information data" refers to all information, including text data and media data such as user posts, comments, news articles, and advertisements.

[1071] "Preprocessing" refers to analyzing the acquired information data and performing tokenization and normalization processing.

[1072] "Tokenization" is the process of dividing text data into words and phrases.

[1073] "Normalization" is the process of stemming, lamellarizing, and removing unnecessary vocabulary and stop words from data.

[1074] A "machine learning model" is an algorithm that learns from past data and analyzes and classifies new data.

[1075] An "emotion engine" is a software tool for analyzing a user's emotional state, categorizing it as positive, negative, neutral, etc.

[1076] "Filtering" is the process of sorting information data based on specific criteria to remove unwanted or harmful information.

[1077] "Notifications" are messages or alerts that communicate filtering results or other information to the user.

[1078] MODE FOR CARRYING OUT THE INVENTION

[1079] The present invention relates to a system that filters false information on social media and notifies users while taking their emotions into consideration. The system mainly consists of a server, a terminal, a user, and an emotion engine. The system acquires information data from social media, preprocesses it, and then uses a machine learning model to determine the authenticity of the information. Inaccurate or harmful information is filtered, and the user is notified of the results. The emotion engine is also used to analyze the user's emotional state and adjust the notification content and filtering criteria based on this.

[1080] Server Features

[1081] The server periodically retrieves information data from social media platforms via API. This information includes user posts, comments, news articles, advertisements, etc. The retrieved information data is preprocessed and analyzed. Preprocessing includes tokenization (dividing text data into words and phrases), stemming, lameness (standardizing word forms), and removal of unnecessary vocabulary and stop words.

[1082] The preprocessed data is then fed into a machine learning model (e.g., BERT or RoBERTa). This model has learned the characteristics of false information based on past data, and scores new data for authenticity. Data that is determined to be false is then filtered out.

[1083] The server also analyzes the user's emotional state using an emotion engine (e.g., IBM Watson Emotion Analysis). Based on the results of the emotion analysis, the filtering criteria and notification content are adjusted. For example, if the user is in a negative emotional state, the notification will be adjusted to a more gentle expression.

[1084] Device Features

[1085] The device collects data on social media posts accessed by users in real time and sends it to a server. When a user clicks on a specific news article or advertisement, the device also immediately sends that information to the server.

[1086] The device also notifies the user of the filtering results. This notification is adjusted based on the analysis results of the emotion engine. For example, if the user's emotional state is negative, the notification content will be milder.

[1087] Specific examples

[1088] Example 1: News article filtering and sentiment analysis

[1089] When a user clicks on a specific news article on a social media site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in credibility with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the notification content accordingly. For example, if the server detects that the user is feeling stressed, it may display a gentle notification such as, "This news article has been temporarily suppressed due to its low credibility."

[1090] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[1091] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[1092] Example of input prompt for generative AI model

[1093] Example prompt 1:

[1094] "Please analyze the reliability of this news article."

[1095] Example prompt 2:

[1096] "Determine whether this ad is potentially fraudulent."

[1097] Example prompt 3:

[1098] "Analyze the sentiment of your users' latest posts and tailor your notifications based on that."

[1099] As described above, the present invention is a system that provides an environment in which users can use social media with peace of mind, effectively filters out false information, and provides appropriate notifications based on the user's emotions.

[1100] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1101] Step 1:

[1102] The server retrieves information data from social media platforms via APIs. As input, it receives data provided by the social media APIs (user posts, comments, news articles, advertisements, etc.). As output, the raw data is stored on the server.

[1103] Step 2:

[1104] The device collects data on posts from social media sites accessed by users in real time and sends that data to a server. As input, it receives user behavior data (such as clicked news articles or advertisements). As output, this information is transferred to the server.

[1105] Step 3:

[1106] The server analyzes the received information data and performs tokenization. As input, it takes raw data and splits it into words and phrases. As output, it generates tokenized data.

[1107] Step 4:

[1108] The server normalizes the data by stemming, lameness, and removing unnecessary vocabulary and stop words. As input, it takes tokenized data and performs normalization on it. As output, it gets clean, normalized data.

[1109] Step 5:

[1110] The server inputs the preprocessed data into a machine learning model (e.g., BERT or RoBERTa). As input, it receives the normalized data and feeds it into the machine learning model. As output, it obtains a truth score for each piece of data.

[1111] Step 6:

[1112] The server determines whether the information is true or false based on the output from the machine learning model. As input, it receives the score returned from the machine learning model, sets a threshold, and determines whether the information is true or false. The output is classified into data that is determined to be false and other data.

[1113] Step 7:

[1114] The device monitors users' latest posts and comments and sends the data to the server. As input, it collects data about users' emotions. As output, it transfers the data to the server.

[1115] Step 8:

[1116] The server uses an emotion engine to analyze the user's emotional state. It receives user posts and comment data as input and performs emotion analysis. The output is an emotion classification result such as positive, negative, or neutral.

[1117] Step 9:

[1118] The server filters data that is determined to be false. As input, it receives data that has been determined to be true or false, and filters out data that it recognizes as false. As output, it obtains a filtered, safe dataset.

[1119] Step 10:

[1120] The server notifies the user of the filtering results. As input, it receives the filtering result data and converts it into a notification message. As output, it generates the notification message.

[1121] Step 11:

[1122] The device selects an appropriate notification method according to the user's emotional state. As input, it receives the emotion analysis results and filtering results and determines the notification method. As output, the adjusted notification message is displayed to the user.

[1123] (Application example 2)

[1124] 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."

[1125] A lot of information is circulating on social media, but some of it is false or harmful, which increases the risk of users acting on incorrect information. This can also worsen users' emotional state or make them feel anxious. Therefore, there is a need for a system that can filter false information and provide appropriate notifications based on the user's emotional state.

[1126] 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 acquiring information data from social media, means for preprocessing the acquired information data, means for determining the authenticity of the information using a machine learning model based on the preprocessed data, means for filtering the determined false information, means for analyzing the emotional state of the user, means for adjusting the filtering result and notification content based on the emotional state, and means for notifying the user of the filtered information. This not only filters false information but also enables appropriate notification that takes into account the emotional state of the user.

[1127] "Social media" is an online platform that allows users to share information and communicate.

[1128] "Information data" refers to digital content such as text, images, videos, comments, news articles, and advertisements posted on social media.

[1129] "Preprocessing" refers to the initial data processing performed on acquired information data, and includes tokenization, removal of unnecessary words, data normalization, and the like.

[1130] A "machine learning model" is an algorithm or method that learns from past data and makes predictions and classifications for new data.

[1131] A "means for determining the authenticity of information" is a process or device that uses a machine learning model to determine whether the information is accurate based on preprocessed information data.

[1132] "Filtering" is the process of sorting data based on specific criteria to remove irrelevant or unwanted data.

[1133] An "emotional state" refers to the mental and emotional state a user is in at a particular moment and can be classified as positive, negative, neutral, etc.

[1134] An "emotion engine" is an algorithm or software that analyzes a user's emotional state from their posts and actions.

[1135] "Notifications" are messages or alerts provided to users based on filtered information or the results.

[1136] The present invention relates to a system for filtering false information on social media and notifying users in consideration of their feelings. Specific embodiments are described below.

[1137] The system mainly consists of a server, a device, a user, and an emotion engine. The server obtains information data from social media, preprocesses it, and then uses a machine learning model to determine whether the information is true or false. Inaccurate or harmful information is filtered out, and the user is notified of the results. The emotion engine also analyzes the user's emotional state and adjusts the notification content and filtering criteria based on this.

[1138] Hardware and software used

[1139] Hardware

[1140] Server: High-performance computers (e.g., cloud servers or dedicated physical servers) are used.

[1141] Terminal: The user's smartphone, tablet, or personal computer is used as the terminal.

[1142] software

[1143] Social Media APIs: APIs for retrieving data from social media (e.g., Twitter API, Facebook Graph API).

[1144] Text processing libraries: Libraries such as NLTK and spaCy are used to perform preprocessing for natural language processing.

[1145] Machine learning models: Machine learning algorithms such as RandomForestClassifier are used to identify false information.

[1146] Sentiment Analysis Engine: Sentiment analysis tools such as TextBlob and VADER are used.

[1147] Details of data processing and calculation

[1148] The server retrieves information data (posts, comments, news articles, advertisements, etc.) from social media via APIs. The retrieved information data is pre-processed on the server, which includes tokenization, stop word removal, stemming, lameness, etc.

[1149] The preprocessed data is then fed into a machine learning model to determine whether the information is true or false. For example, a RandomForest Classifier trained on past data is used. If the information is determined to be false, the data is filtered out.

[1150] Meanwhile, the emotion engine analyzes the user's emotional state. This analysis is done using emotion analysis tools such as TextBlob and VADER. The user's emotional state is classified as positive, negative, neutral, etc.

[1151] The notification content is adjusted based on the filtering results and the user's emotional state. For example, if the user's emotional state is negative, the notification will be softened.

[1152] Specific examples

[1153] 1. News article filtering and sentiment analysis

[1154] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be unreliable, the server filters it out and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server determines that the user is feeling stressed, it will send a gentle notification such as, "This news article is unreliable, so we have temporarily withheld it from display."

[1155] 2. Fraudulent Ad Filtering and Sentiment Analysis

[1156] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high probability of being fraudulent, the server filters the ad and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[1157] Prompt Sentence Examples

[1158] Write example code for a system that filters misinformation on social media and tailors notifications based on the user's emotional state.

[1159] Specifically, it will use Python and include the following functionality:

[1160] Data collection through social media APIs

[1161] Preprocessing of text data (tokenization, stop word removal, stemming, etc.)

[1162] Detecting false information using machine learning models

[1163] Emotion analysis using an emotion engine

[1164] User notification based on judgment results and emotional state

[1165] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1166] Step 1: Data collection

[1167] The server obtains information data from social media using social media APIs (e.g., Twitter API, Facebook Graph API). When a user clicks on a specific post or advertisement on the device, that information data is also sent to the server. The input is the data from the social media post or advertisement, and the output is raw data for preprocessing.

[1168] Step 2: Preprocessing

[1169] The server preprocesses the acquired information data. This preprocessing involves tokenization, which divides the text data into tokens, removing unnecessary vocabulary and stop words, stemming, lameness (standardizing word forms), etc. The input is raw text data, and the output is normalized text data.

[1170] Step 3: Authentication using machine learning models

[1171] The server inputs the preprocessed data into a machine learning model (e.g., RandomForestClassifier) ​​to determine whether the information is true or false. Specifically, the text data is converted into a feature vector, which is then input into a model that has learned the characteristics of false information and scored. The input is normalized text data, and the output is a determination of whether the information is false or not.

[1172] Step 4: Filtering misinformation

[1173] The server filters data determined to be false by the machine learning model. In this step, information determined to be false does not proceed to the next processing step and is processed so that it is not displayed to the user. The input is the truth determination result, and the output is the filtered information data.

[1174] Step 5: Sentiment Analysis

[1175] The server uses an emotion engine (e.g., TextBlob, VADER) to analyze the user's emotional state. It performs sentiment analysis on the text data of user posts and clicked ads, and classifies them as positive, negative, neutral, etc. The input is the user's text data, and the output is the user's emotional state.

[1176] Step 6: Adjust your notifications

[1177] The server adjusts the notification content and filtering criteria based on the emotion analysis results. For example, if the user is in a negative emotional state, the notification will be sent with a more gentle tone. The input is the emotion analysis results and filtering results, and the output is the final notification content sent to the user.

[1178] Step 7: Notification of filtering results

[1179] The terminal displays the notification content sent from the server to the user. For example, it displays a pop-up message saying, "This news article was not displayed due to low reliability." The input is the notification content sent from the server, and the output is the notification displayed on the user's smartphone or personal computer.

[1180] 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.

[1181] 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.

[1182] 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.

[1183] [Fourth embodiment]

[1184] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1185] 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.

[1186] 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).

[1187] 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.

[1188] 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.

[1189] 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).

[1190] 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.

[1191] 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.

[1192] 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.

[1193] 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.

[1194] 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.

[1195] 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.

[1196] 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."

[1197] The present invention relates to a system for filtering false information on social media, which mainly consists of a server, a terminal, and a user. This system acquires information data from social media, preprocesses it, determines the authenticity of the information using a machine learning model, filters the false information, and notifies the user of the filtering results.

[1198] Program processing (natural language explanation)

[1199] 1. Data Collection

[1200] The server obtains information data from social media platforms (e.g., popular social networking sites) using APIs. API stands for Application Programming Interface, and refers to the rules and protocols that allow software to communicate with other software.

[1201] The device acquires the posting data of the SNS accessed by the user in real time and sends it to the server.

[1202] 2. Data Preprocessing

[1203] The server analyzes the acquired information data and performs tokenization, which is the process of dividing text data into words and phrases.

[1204] Additionally, the server removes unnecessary vocabulary and stop words (e.g., "the" and "and" in English) and normalizes the data. Normalization is the process of unifying data by performing stemming and lameness.

[1205] 3. Learning and judgment using AI models

[1206] The server then inputs the preprocessed data into a machine learning model that learns the characteristics of false information based on past data.

[1207] The server's AI module then scores the information based on newly acquired data. For example, if it calculates that a particular news article has a 0.9 probability of being true, the article is deemed to be highly reliable.

[1208] 4. Filtering and Notifications

[1209] The server filters out information data that is determined to be false and does not send it to the user's terminal.

[1210] The device will notify the user of the filtering results in the form of a pop-up or push notification.

[1211] Users can use social media with peace of mind.

[1212] Specific examples

[1213] Example 1: Filtering news articles

[1214] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server retrieves the data and performs preprocessing. It then uses a machine learning model to determine whether the news article is authentic or not. For example, if the news article is determined to be unreliable with a score of 0.8, the server filters the article and does not display it on the user's device. The device then notifies the user that it has been filtered.

[1215] Example 2: Filtering fraudulent ads

[1216] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud, with a score of 0.85, the server filters the ad and does not display it on the user's device. The device then notifies the user that the ad has been filtered.

[1217] In this way, the present invention provides an environment in which users can use social media with peace of mind.

[1218] The processing flow will be explained below.

[1219] Step 1: Data collection

[1220] The server retrieves information data from the social media platform via API, including user posts, comments, news articles, advertisements, etc.

[1221] The device automatically captures the social media posts that the user views or interacts with and sends the data to the server in real time.

[1222] Step 2: Preprocessing

[1223] The server receives the acquired information data and performs text analysis, specifically, using tokenization to divide the text into words and phrases.

[1224] The server removes stop words (e.g., common conjunctions and prepositions) and performs stemming to improve the accuracy of the data.

[1225] Step 3: Training the machine learning model

[1226] The server matches the pre-processed data with existing datasets and uses machine learning models to learn new patterns, including spam filtering, fake news detection, and fraudulent ad detection.

[1227] The server evaluates the accuracy of the trained model and adjusts parameters as needed.

[1228] Step 4: Determine the information

[1229] The server feeds new data into a machine learning model in real time to determine whether the information is true or false. For example, a particular news article may be identified as fake news with a probability of 0.7.

[1230] The server outputs the information judgment result as a score.

[1231] Step 5: Filtering

[1232] The server filters data determined to be false and does not send it to the user's device. For example, data determined to be fraudulent advertisements will not be displayed to the user.

[1233] The server stores the filtered data and uses it as learning data at a later date.

[1234] Step 6: Notifications and Display

[1235] The device will notify the user of the filtering results, for example by displaying a pop-up notification such as "This news article was not displayed due to low reliability."

[1236] Users will receive notifications and can use social media with peace of mind.

[1237] Step 7: Reassess and improve

[1238] The server periodically re-evaluates the accuracy of the filtering and retrains the model, allowing it to keep up with the latest disinformation trends.

[1239] The server receives feedback from users and uses it to improve the system.

[1240] Through the above steps, the present invention provides an environment in which users can use social media with peace of mind.

[1241] Example 1

[1242] 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."

[1243] There is a lot of false information and fraudulent advertisements on social media, making it difficult for users to quickly obtain reliable information. There is also a lack of effective measures to prevent incorrect decisions and actions based on false information. For this reason, it is necessary to provide an environment where users can use social media with peace of mind.

[1244] 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.

[1245] In this invention, the server includes a means for acquiring information data from social media, a means for preprocessing the acquired information data, and a means for determining the authenticity of the information using a machine learning model based on the preprocessed data, thereby making it possible to filter out false information and provide reliable information.

[1246] "Social media" refers to a platform on the Internet where users can disseminate, share, and interact with information.

[1247] "Information data" refers to data including content such as text, images, and videos posted on social media.

[1248] "Preprocessing" refers to the process of analyzing, dividing, and organizing acquired information data, and converting it into a format that is easy for the machine learning model to process.

[1249] "Tokenization" refers to the process of dividing text data into smaller units such as words or phrases.

[1250] "Stop words" refer to common words (e.g., the, and, etc.) that are not necessary for semantic analysis of a sentence in natural language processing, and the purpose is to remove them.

[1251] "Normalization" refers to the process of converting data into a standard form using stemming and lambdaization.

[1252] "Stemming" refers to a technique for extracting only the stem portion of a word.

[1253] "Lametization" refers to the process of converting words into their base forms.

[1254] A "machine learning model" refers to a system that includes algorithms that learn from past data and make predictions or classifications for new data.

[1255] "Scoring" refers to the process by which a machine learning model quantifies the reliability and veracity of information data.

[1256] "Filtering" refers to the process of filtering out unnecessary or inappropriate data based on specific conditions.

[1257] "Notification" refers to the process of informing users of filtered results and important information.

[1258] A "pop-up" is a small window that suddenly appears on a user's screen and is used to display specific information or notifications.

[1259] "Push notification" refers to a form of notification that an application or system automatically sends to a user's device.

[1260] "Database" refers to a system that systematically stores and manages collected data and results.

[1261] This invention provides a method for filtering false information on social media using a system consisting mainly of a server, a terminal, and a user. This system uses a combination of specific software and hardware to perform a series of processes including data collection, data preprocessing, AI-based judgment, filtering, and notification.

[1262] Data collection

[1263] The server retrieves information data from social media platforms using APIs. For example, to collect data from Twitter, the Python library "Tweepy" is used. The server uses "Tweepy" to periodically collect post data related to specific keywords and store it in a database.

[1264] The device collects data on posts from social media sites accessed by the user in real time and sends it to the server. This process is achieved using technologies such as JavaScript's "Fetch API" and Swift. The device sends the post data to the server in response to user actions, and the data is collected on the server side.

[1265] Data Preprocessing

[1266] The server analyzes the collected information data and performs tokenization using the Python library NLTK, which splits the text into words and stores them in a database.

[1267] The server then removes unnecessary vocabulary and stop words from the tokenized data, performs stemming, and lambdaization, and standardizes the data using libraries like NLTK and SpaCy, converting the data into a format suitable for processing by machine learning models.

[1268] Learning and judgment by AI model

[1269] The server then feeds the pre-processed data into machine learning models, specifically pre-trained BERT and GPT models, which excel at natural language processing and learn the characteristics of misinformation from the data.

[1270] The server's AI module scores the authenticity of the information based on the newly input data, and the results are stored in a database. For example, if the AI ​​module calculates that a particular news article has a 0.9 probability of being true, the article is deemed to be highly reliable.

[1271] Filtering and Notifications

[1272] The server filters out false information based on the scoring results of the machine learning model, and if the score falls below a certain threshold, the information is prevented from being sent to the user's device.

[1273] The device will notify the user of the filtering results. By notifying the user using a pop-up or push notification, the user can check in real time that the information has been filtered. For example, if a particular news article is deemed to be unreliable and filtered, the device will notify the user of the result.

[1274] Specific examples

[1275] 1. Filtering news articles

[1276] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server retrieves the data and performs preprocessing. It then uses a machine learning model to determine the authenticity of the news article. If the news article is judged to be low reliability with a score of 0.8, the server filters the article and does not display it on the user's device. The device then notifies the user that it has been filtered.

[1277] 2. Filtering fraudulent ads

[1278] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. If the ad is determined to be highly fraudulent with a score of 0.85, the server filters the ad and does not display it on the user's device. The device notifies the user that the ad has been filtered.

[1279] This system allows users to reduce the risk of false information and fraud on social media, allowing them to use only reliable information with peace of mind.

[1280] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1281] Step 1: Data collection

[1282] Specific operation:

[1283] The server retrieves information data using the API of the social media platform. Specifically, it uses the Python library "Tweepy" to collect data from Twitter. It periodically monitors specific keywords (e.g., "COVID-19" or "disaster information") and retrieves new posts.

[1284] input:

[1285] Social media platform API keys

[1286] Specific keywords to collect

[1287] Data processing:

[1288] Call API to get post data related to keywords

[1289] output:

[1290] Acquired post data (text, user information, post date and time, etc.)

[1291] Step 2: Data Preprocessing

[1292] Specific operation:

[1293] The server preprocesses the acquired information data. First, it performs tokenization, splitting the text into words using the Python library "NLTK."

[1294] Next, we remove unnecessary vocabulary and stop words, and perform stemming and lambdaization using tools like NLTK and SpaCy.

[1295] input:

[1296] Obtained post data (output from Step 1)

[1297] Data processing:

[1298] Text tokenization (splitting into words)

[1299] Stop word removal

[1300] Stemming and Lamellarization

[1301] output:

[1302] Preprocessed text data

[1303] Step 3: Learning and judgment by AI model

[1304] Specific operation:

[1305] The server then feeds the preprocessed data into a machine learning model (e.g., the BERT model), which has learned the characteristics of false information based on past data.

[1306] The server's AI module scores the truth of the information based on the newly entered data, and the scoring result (e.g., 0.0 to 1.0) is stored in a database.

[1307] input:

[1308] Preprocessed text data (output of step 2)

[1309] Data Calculation:

[1310] Judgment and scoring using machine learning models

[1311] output:

[1312] Reliability score (e.g., 0.8)

[1313] Step 4: Filtering

[1314] Specific operation:

[1315] The server filters out false information based on the scoring results. If the score is below a certain threshold (e.g., 0.5), the information is not sent to the user's device.

[1316] The filtered information is either deleted from the database or moved to another table.

[1317] input:

[1318] Determined reliability scores and corresponding information data (output of Step 3)

[1319] Data processing:

[1320] Filtering data based on scores

[1321] Save or delete filtered results

[1322] output:

[1323] Filtered and clean information data

[1324] Step 5: Notification

[1325] Specific operation:

[1326] The device will notify the user of the filtering results, using a pop-up or push notification to let the user know that false information has been filtered.

[1327] The user confirms the notification and views the clean information data.

[1328] input:

[1329] Filtered and clean information data (output of step 4)

[1330] Data processing:

[1331] Generate and send notifications

[1332] output:

[1333] Notification of filtering results

[1334] (Application example 1)

[1335] 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."

[1336] Traditional social media platforms require users to independently determine the authenticity of the information they access. However, in today's information-intensive world, there is a growing risk that users will be misled by unreliable information, leading to the spread of inaccurate information. Furthermore, there is a lack of systems that can effectively filter false information in real time and immediately warn users. To solve this problem, a system is needed that can detect false information with high accuracy and in real time and notify users.

[1337] 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.

[1338] In this invention, the server includes means for acquiring information data from social media, means for preprocessing the acquired information data, and means for determining the authenticity of the information using a machine learning model based on the preprocessed data. This enables a system including means for tokenizing the information data and removing unnecessary information, means for notifying the filtering results in real time, and means for issuing a warning when false information is detected.

[1339] "Social media" is a general term for online platforms that allow users to share information and communicate with each other.

[1340] "Information Data" refers to data such as text, images, videos, and links generated by users on social media.

[1341] "Preprocessing" is the process of converting acquired information data into a format that can be applied to a machine learning model through methods such as tokenization and removal of unnecessary information.

[1342] A "machine learning model" is a system that uses algorithms to learn patterns and specific characteristics based on past data to determine the authenticity of new data.

[1343] "Disinformation" is information that is intentionally or unintentionally provided in an inaccurate or misleading manner.

[1344] "Filtering" is the process of identifying and removing false information from acquired information data.

[1345] "Tokenization" is the process of dividing text data into words and phrases.

[1346] "Nutrition" is the process of removing portions of text data that are not required for analysis, based on stop words or specific patterns.

[1347] "Real-time" means that data acquisition and processing are carried out immediately, and the results are notified to the user without making them wait unintentionally.

[1348] "Notification" is an action to notify the user of filtering results or warnings.

[1349] MODE FOR CARRYING OUT THE INVENTION

[1350] To implement this invention, three main components, a server, a terminal, and a user, must cooperate. The role and processing method of each component will be specifically explained below.

[1351] server

[1352] The server plays a key role in acquiring information data from social media, preprocessing it, and then using machine learning models to determine its authenticity. Specifically, it uses the following hardware and software:

[1353] Hardware:

[1354] High-performance processor (e.g., Intel Xeon)

[1355] Large memory capacity (e.g. 64GB RAM or more)

[1356] software:

[1357] Sinatra (Ruby web framework)

[1358] NLTK library (natural language processing library)

[1359] Transformers library (including the BERT model)

[1360] The server first collects information data in real time using social media APIs. The information data then undergoes a preprocessing step, where it is tokenized and stop words are removed. Based on this preprocessed data, the reliability of the information is scored using the BERT model. Based on the results, false information is filtered and notifications are sent to the device.

[1361] Terminal

[1362] The terminal is a device such as a smartphone or tablet used by the user, and serves to notify the user of the filtering results received from the server.

[1363] Hardware:

[1364] Smartphones and tablets (e.g. iPhone, Android devices)

[1365] software:

[1366] Web browser or native application

[1367] The device displays notifications received from the server to the user as pop-up or push notifications. For example, if a user clicks on a specific article in a social media app and the information is determined to be unreliable, a warning pop-up will be displayed saying, "This news article is unreliable."

[1368] User

[1369] The user is an end user of social media. The user can operate their device and check the authenticity of the information sent from the server in real time. This significantly reduces the risk of being misled by incorrect or fake information.

[1370] Specific examples

[1371] For example, consider a case where the reliability of information is evaluated using the BERT model in a news application. When a user taps on a specific article in the news app, the server collects the article's text data, preprocesses it, and then analyzes it using the BERT model. If the analysis results indicate that the information is not reliable, a notification is sent to the device and a warning is displayed to the user.

[1372] Prompt Sentence Examples

[1373] "This is an example news article text for analysis."

[1374] In this way, the present invention can provide users with access to reliable information and protection from inaccurate information.

[1375] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1376] Step 1: Data collection

[1377] The server collects information data in real time using social media APIs. The input data is text data of the latest posts on social media and news articles. The server obtains this data via the API and sends it to the next preprocessing step.

[1378] Step 2: Data Preprocessing

[1379] The server performs preprocessing on the collected information data. Specifically, it performs tokenization to divide the text into words or phrases. It also removes stop words and eliminates unnecessary words. In this way, the information contained in the input data is made clearer and converted into a format suitable for analysis by machine learning models. The output data is preprocessed, clean text data.

[1380] Step 3: Authentication using machine learning models

[1381] The server inputs the preprocessed data into a generative AI model (e.g., BERT) to determine whether the text is true or false. The server uses a machine learning model to analyze the input data and generate a score indicating whether it is trustworthy information. Based on this scoring, the server determines whether the information is accurate or false. The output data is a reliability score and a determination result corresponding to each piece of text data.

[1382] Step 4: Filtering

[1383] The server filters out false information based on the results determined by the machine learning model. In this step, information data determined to be unreliable is removed or flagged as unreliable. The output data is filtered, reliable information data.

[1384] Step 5: Notification

[1385] The terminal receives the filtering results from the server and notifies the user. The terminal displays a warning popup for unreliable information and displays the normal screen for reliable information. In this step, the user is prompted to take specific action. The output data is a notification message on the user's device.

[1386] Specific examples of specific actions

[1387] Data collection:

[1388] The server retrieves the news article "This is an example news article text for analysis." via the API.

[1389] Data preprocessing:

[1390] The server converts the retrieved text from "This is an example news article text for analysis." into a tokenized format such as "this example news article text analysis," and removes stop words to generate clean data.

[1391] Machine learning model for determining truth:

[1392] The server inputs the clean text data into the BERT model and scores it. For example, if the confidence score is 0.4, the information is determined to be false.

[1393] filtering:

[1394] Information with a reliability score of 0.4 is filtered as false information and users will not receive it.

[1395] notification:

[1396] Based on the filtering results received by the device from the server, a pop-up warning is displayed to the user stating, "This news article is unreliable."

[1397] 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.

[1398] The present invention relates to a system that filters false information on social media and notifies users while taking their emotions into consideration. The system mainly consists of a server, a terminal, a user, and an emotion engine. The system acquires information data from social media, preprocesses it, and then uses a machine learning model to determine whether the information is true or false. Inaccurate or harmful information is filtered, and the user is notified of the results. The emotion engine is also used to analyze the user's emotional state and adjust the notification content and filtering criteria based on this.

[1399] Program processing (natural language explanation)

[1400] 1. Data Collection

[1401] The server retrieves information data from social media platforms (e.g., general social networking sites) via APIs, including user posts, comments, news articles, advertisements, etc.

[1402] The device acquires the posting data of the SNS accessed by the user in real time and transmits the data to the server.

[1403] 2. Pretreatment

[1404] The server receives the acquired information data and performs analysis, which includes tokenization, which divides the text data into words and phrases.

[1405] The server removes unnecessary vocabulary and stop words and normalizes the data, a process that involves stemming and lameness.

[1406] 3. Learning and assessment of machine learning models

[1407] The server feeds the preprocessed data into a machine learning model that learns the characteristics of misinformation based on past data.

[1408] The server then scores the information based on the newly acquired data. For example, if a particular news article is judged to be fake news with a probability of 0.8, the article is deemed to be false information.

[1409] 4. Emotion analysis using an emotion engine

[1410] The server uses an emotion engine to analyze the emotions users express on social media, categorizing them as positive, negative, neutral, etc.

[1411] The server obtains the user's emotional state and reflects it in the filtering results and notification content.

[1412] 5. Filtering and Notifications

[1413] The server filters data determined to be false and does not send it to the user's device.

[1414] The server notifies the user of the filtering results, for example by displaying a pop-up message saying, "This news article was not displayed because it was not reliable."

[1415] The device selects an appropriate notification method depending on the user's emotional state. For example, if the emotional state is negative, the notification will be delivered in a gentler manner.

[1416] Specific examples

[1417] Example 1: News article filtering and sentiment analysis

[1418] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in reliability with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server detects that the user is feeling stressed, it will display a gentle notification such as, "This news article has been temporarily suppressed due to its low reliability."

[1419] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[1420] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[1421] In this way, the present invention is a system that provides an environment in which users can use social media with peace of mind, and at the same time provides appropriate notifications taking into account the user's emotional state.

[1422] The processing flow will be explained below.

[1423] Step 1: Data collection

[1424] The server retrieves information data from popular social media platforms via APIs, including user posts, comments, news articles, and advertisements.

[1425] The device captures the social networking site posting data that the user has viewed or operated in real time and immediately transmits that data to the server.

[1426] Step 2: Preprocessing

[1427] The server analyzes the received information data, specifically by using tokenization to divide the text data into words and phrases.

[1428] The server normalizes the data by removing stop words (e.g., common conjunctions and prepositions), stemming, and lame- tizing.

[1429] Step 3: Training the machine learning model

[1430] The server then matches the pre-processed data with existing datasets and uses machine learning models to learn new patterns, resulting in a model that can detect misinformation with high accuracy.

[1431] The server evaluates the accuracy of the model and retrains it if necessary.

[1432] Step 4: Determine the information

[1433] The server feeds new data into the machine learning model in real time. For example, if a particular news article has a 0.75 probability of being fake news, the article is flagged as false.

[1434] The server outputs the information judgment results as a score, which is used as a basis for filtering decisions.

[1435] Step 5: Sentiment Analysis

[1436] The server uses an emotion engine to analyze the user's emotions, for example by analyzing text and behavioral data to determine what emotions (positive, negative, neutral) the user is feeling.

[1437] The server stores the analysis results in a database, making them available for subsequent processing.

[1438] Step 6: Filtering and Notifications

[1439] The server filters data determined to be false and does not send it to the user's device.

[1440] The server generates appropriate notification content based on the results of emotion analysis, depending on the user's emotional state. For example, if the user is feeling anxious, the server will send a notification in a calm manner.

[1441] The device will notify the user of the filtering results, for example, in the form of a pop-up message saying, "This news article was not displayed due to its low reliability."

[1442] Step 7: Reassess and improve

[1443] The server periodically re-evaluates the accuracy of filtering and sentiment analysis and re-trains the model if necessary.

[1444] The server collects feedback from users and uses it to improve the system.

[1445] Specific examples

[1446] Example 1: News article filtering and sentiment analysis

[1447] When a user clicks on a specific news article on a social networking site, the device immediately sends the news article data to a server. The server preprocesses the data and uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in credibility with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server detects that the user is feeling stressed, it may display a gentle notification such as, "This news article is low in credibility, so we have temporarily withheld it from view."

[1448] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[1449] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[1450] In this way, the present invention is a system that provides an environment in which users can use social media with peace of mind, and at the same time provides appropriate notifications taking into account the user's emotional state.

[1451] Example 2

[1452] 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."

[1453] There is a lot of information circulating on social media, including fake and incorrect information. This fake information can cause confusion and anxiety among users. Furthermore, if information is notified without taking into account the user's emotional state, it can cause further stress and discomfort. To solve this situation, a notification system that can effectively filter fake information and take into account the user's emotional state is needed.

[1454] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1455] In this invention, the server includes means for acquiring information data from social media, means for preprocessing the acquired information data, means for determining the authenticity of the information using a machine learning model based on the preprocessed data, means for filtering the determined false information, means for analyzing the emotional state of the user, means for adjusting filtering criteria and notification content based on the analysis results, and means for notifying the user of the filtered information. This makes it possible to efficiently filter false information and provide appropriate notifications that take the user's emotions into consideration.

[1456] "Social media" is a platform on the Internet that allows users to share information and communicate.

[1457] "Information data" refers to all information, including text data and media data such as user posts, comments, news articles, and advertisements.

[1458] "Preprocessing" refers to analyzing the acquired information data and performing tokenization and normalization processing.

[1459] "Tokenization" is the process of dividing text data into words and phrases.

[1460] "Normalization" is the process of stemming, lamellarizing, and removing unnecessary vocabulary and stop words from data.

[1461] A "machine learning model" is an algorithm that learns from past data and analyzes and classifies new data.

[1462] An "emotion engine" is a software tool for analyzing a user's emotional state, categorizing it as positive, negative, neutral, etc.

[1463] "Filtering" is the process of sorting information data based on specific criteria to remove unwanted or harmful information.

[1464] "Notifications" are messages or alerts that communicate filtering results or other information to the user.

[1465] MODE FOR CARRYING OUT THE INVENTION

[1466] The present invention relates to a system that filters false information on social media and notifies users while taking their emotions into consideration. The system mainly consists of a server, a terminal, a user, and an emotion engine. The system acquires information data from social media, preprocesses it, and then uses a machine learning model to determine the authenticity of the information. Inaccurate or harmful information is filtered, and the user is notified of the results. The emotion engine is also used to analyze the user's emotional state and adjust the notification content and filtering criteria based on this.

[1467] Server Features

[1468] The server periodically retrieves information data from social media platforms via API. This information includes user posts, comments, news articles, advertisements, etc. The retrieved information data is preprocessed and analyzed. Preprocessing includes tokenization (dividing text data into words and phrases), stemming, lameness (standardizing word forms), and removal of unnecessary vocabulary and stop words.

[1469] The preprocessed data is then fed into a machine learning model (e.g., BERT or RoBERTa). This model has learned the characteristics of false information based on past data, and scores new data for authenticity. Data that is determined to be false is then filtered out.

[1470] The server also analyzes the user's emotional state using an emotion engine (e.g., IBM Watson Emotion Analysis). Based on the results of the emotion analysis, the filtering criteria and notification content are adjusted. For example, if the user is in a negative emotional state, the notification will be adjusted to a more gentle expression.

[1471] Device Features

[1472] The device collects data on social media posts accessed by users in real time and sends it to a server. When a user clicks on a specific news article or advertisement, the device also immediately sends that information to the server.

[1473] The device also notifies the user of the filtering results. This notification is adjusted based on the analysis results of the emotion engine. For example, if the user's emotional state is negative, the notification content will be milder.

[1474] Specific examples

[1475] Example 1: News article filtering and sentiment analysis

[1476] When a user clicks on a specific news article on a social media site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be low in credibility with a score of 0.75, the server filters the article and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the notification content accordingly. For example, if the server detects that the user is feeling stressed, it may display a gentle notification such as, "This news article has been temporarily suppressed due to its low credibility."

[1477] Example 2: Fraudulent Ad Filtering and Sentiment Analysis

[1478] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high likelihood of fraud with a score of 0.85, the server filters the ad and does not display it to the user. The server also analyzes the user's emotional state using an emotion engine and adjusts the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[1479] Example of input prompt for generative AI model

[1480] Example prompt 1:

[1481] "Please analyze the reliability of this news article."

[1482] Example prompt 2:

[1483] "Determine whether this ad is potentially fraudulent."

[1484] Example prompt 3:

[1485] "Analyze the sentiment of your users' latest posts and tailor your notifications based on that."

[1486] As described above, the present invention is a system that provides an environment in which users can use social media with peace of mind, effectively filters out false information, and provides appropriate notifications based on the user's emotions.

[1487] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1488] Step 1:

[1489] The server retrieves information data from social media platforms via APIs. As input, it receives data provided by the social media APIs (user posts, comments, news articles, advertisements, etc.). As output, the raw data is stored on the server.

[1490] Step 2:

[1491] The device collects data on posts from social media sites accessed by users in real time and sends that data to a server. As input, it receives user behavior data (such as clicked news articles or advertisements). As output, this information is transferred to the server.

[1492] Step 3:

[1493] The server analyzes the received information data and performs tokenization. As input, it takes raw data and splits it into words and phrases. As output, it generates tokenized data.

[1494] Step 4:

[1495] The server normalizes the data by stemming, lameness, and removing unnecessary vocabulary and stop words. As input, it takes tokenized data and performs normalization on it. As output, it gets clean, normalized data.

[1496] Step 5:

[1497] The server inputs the preprocessed data into a machine learning model (e.g., BERT or RoBERTa). As input, it receives the normalized data and feeds it into the machine learning model. As output, it obtains a truth score for each piece of data.

[1498] Step 6:

[1499] The server determines whether the information is true or false based on the output from the machine learning model. As input, it receives the score returned from the machine learning model, sets a threshold, and determines whether the information is true or false. The output is classified into data that is determined to be false and other data.

[1500] Step 7:

[1501] The device monitors users' latest posts and comments and sends the data to the server. As input, it collects data about users' emotions. As output, it transfers the data to the server.

[1502] Step 8:

[1503] The server uses an emotion engine to analyze the user's emotional state. It receives user posts and comment data as input and performs emotion analysis. The output is an emotion classification result such as positive, negative, or neutral.

[1504] Step 9:

[1505] The server filters data that is determined to be false. As input, it receives data that has been determined to be true or false, and filters out data that it recognizes as false. As output, it obtains a filtered, safe dataset.

[1506] Step 10:

[1507] The server notifies the user of the filtering results. As input, it receives the filtering result data and converts it into a notification message. As output, it generates the notification message.

[1508] Step 11:

[1509] The device selects an appropriate notification method according to the user's emotional state. As input, it receives the emotion analysis results and filtering results and determines the notification method. As output, the adjusted notification message is displayed to the user.

[1510] (Application example 2)

[1511] 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."

[1512] A lot of information is circulating on social media, but some of it is false or harmful, which increases the risk of users acting on incorrect information. This can also worsen users' emotional state or make them feel anxious. Therefore, there is a need for a system that can filter false information and provide appropriate notifications based on the user's emotional state.

[1513] 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 acquiring information data from social media, means for preprocessing the acquired information data, means for determining the authenticity of the information using a machine learning model based on the preprocessed data, means for filtering the determined false information, means for analyzing the emotional state of the user, means for adjusting the filtering result and notification content based on the emotional state, and means for notifying the user of the filtered information. This not only filters false information but also enables appropriate notification that takes into account the emotional state of the user.

[1514] "Social media" is an online platform that allows users to share information and communicate.

[1515] "Information data" refers to digital content such as text, images, videos, comments, news articles, and advertisements posted on social media.

[1516] "Preprocessing" refers to the initial data processing performed on acquired information data, and includes tokenization, removal of unnecessary words, data normalization, and the like.

[1517] A "machine learning model" is an algorithm or method that learns from past data and makes predictions and classifications for new data.

[1518] A "means for determining the authenticity of information" is a process or device that uses a machine learning model to determine whether the information is accurate based on preprocessed information data.

[1519] "Filtering" is the process of sorting data based on specific criteria to remove irrelevant or unwanted data.

[1520] An "emotional state" refers to the mental and emotional state a user is in at a particular moment and can be classified as positive, negative, neutral, etc.

[1521] An "emotion engine" is an algorithm or software that analyzes a user's emotional state from their posts and actions.

[1522] "Notifications" are messages or alerts provided to users based on filtered information or the results.

[1523] The present invention relates to a system for filtering false information on social media and notifying users in consideration of their feelings. Specific embodiments are described below.

[1524] The system mainly consists of a server, a device, a user, and an emotion engine. The server obtains information data from social media, preprocesses it, and then uses a machine learning model to determine whether the information is true or false. Inaccurate or harmful information is filtered out, and the user is notified of the results. The emotion engine also analyzes the user's emotional state and adjusts the notification content and filtering criteria based on this.

[1525] Hardware and software used

[1526] Hardware

[1527] Server: High-performance computers (e.g., cloud servers or dedicated physical servers) are used.

[1528] Terminal: The user's smartphone, tablet, or personal computer is used as the terminal.

[1529] software

[1530] Social Media APIs: APIs for retrieving data from social media (e.g., Twitter API, Facebook Graph API).

[1531] Text processing libraries: Libraries such as NLTK and spaCy are used to perform preprocessing for natural language processing.

[1532] Machine learning models: Machine learning algorithms such as RandomForestClassifier are used to identify false information.

[1533] Sentiment Analysis Engine: Sentiment analysis tools such as TextBlob and VADER are used.

[1534] Details of data processing and calculation

[1535] The server retrieves information data (posts, comments, news articles, advertisements, etc.) from social media via APIs. The retrieved information data is pre-processed on the server, which includes tokenization, stop word removal, stemming, lameness, etc.

[1536] The preprocessed data is then fed into a machine learning model to determine whether the information is true or false. For example, a RandomForest Classifier trained on past data is used. If the information is determined to be false, the data is filtered out.

[1537] Meanwhile, the emotion engine analyzes the user's emotional state. This analysis is done using emotion analysis tools such as TextBlob and VADER. The user's emotional state is classified as positive, negative, neutral, etc.

[1538] The notification content is adjusted based on the filtering results and the user's emotional state. For example, if the user's emotional state is negative, the notification will be softened.

[1539] Specific examples

[1540] 1. News article filtering and sentiment analysis

[1541] When a user clicks on a specific news article on a social networking site, the device sends the news article data to a server. The server receives the data, performs preprocessing, and then uses a machine learning model to determine the authenticity of the news article. For example, if the news article is determined to be unreliable, the server filters it out and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server determines that the user is feeling stressed, it will send a gentle notification such as, "This news article is unreliable, so we have temporarily withheld it from display."

[1542] 2. Fraudulent Ad Filtering and Sentiment Analysis

[1543] When a user clicks on an ad displayed on social media, the device sends the ad data to a server. The server preprocesses the data and uses a machine learning model to determine the likelihood of fraud. For example, if the ad is determined to have a high probability of being fraudulent, the server filters the ad and does not display it to the user. The server also uses an emotion engine to analyze the user's emotional state and adjust the content of notifications. For example, if the server determines that the user is feeling anxious, it may notify the user, "This ad was not displayed because it is unreliable. Please rest assured."

[1544] Prompt Sentence Examples

[1545] Write example code for a system that filters misinformation on social media and tailors notifications based on the user's emotional state.

[1546] Specifically, it will use Python and include the following functionality:

[1547] Data collection through social media APIs

[1548] Preprocessing of text data (tokenization, stop word removal, stemming, etc.)

[1549] Detecting false information using machine learning models

[1550] Emotion analysis using an emotion engine

[1551] User notification based on judgment results and emotional state

[1552] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1553] Step 1: Data collection

[1554] The server obtains information data from social media using social media APIs (e.g., Twitter API, Facebook Graph API). When a user clicks on a specific post or advertisement on the device, that information data is also sent to the server. The input is the data from the social media post or advertisement, and the output is raw data for preprocessing.

[1555] Step 2: Preprocessing

[1556] The server preprocesses the acquired information data. This preprocessing involves tokenization, which divides the text data into tokens, removing unnecessary vocabulary and stop words, stemming, lameness (standardizing word forms), etc. The input is raw text data, and the output is normalized text data.

[1557] Step 3: Authentication using machine learning models

[1558] The server inputs the preprocessed data into a machine learning model (e.g., RandomForestClassifier) ​​to determine whether the information is true or false. Specifically, the text data is converted into a feature vector, which is then input into a model that has learned the characteristics of false information and scored. The input is normalized text data, and the output is a determination of whether the information is false or not.

[1559] Step 4: Filtering misinformation

[1560] The server filters data determined to be false by the machine learning model. In this step, information determined to be false does not proceed to the next processing step and is processed so that it is not displayed to the user. The input is the truth determination result, and the output is the filtered information data.

[1561] Step 5: Sentiment Analysis

[1562] The server uses an emotion engine (e.g., TextBlob, VADER) to analyze the user's emotional state. It performs sentiment analysis on the text data of user posts and clicked ads, and classifies them as positive, negative, neutral, etc. The input is the user's text data, and the output is the user's emotional state.

[1563] Step 6: Adjust your notifications

[1564] The server adjusts the notification content and filtering criteria based on the emotion analysis results. For example, if the user is in a negative emotional state, the notification will be sent with a more gentle tone. The input is the emotion analysis results and filtering results, and the output is the final notification content sent to the user.

[1565] Step 7: Notification of filtering results

[1566] The terminal displays the notification content sent from the server to the user. For example, it displays a pop-up message saying, "This news article was not displayed due to low reliability." The input is the notification content sent from the server, and the output is the notification displayed on the user's smartphone or personal computer.

[1567] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1568] 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.

[1569] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1570] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1571] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1572] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1573] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1574] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1575] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1576] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1577] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1578] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1579] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1580] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1581] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1582] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1583] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1584] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1585] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1586] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1587] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1588] The following is further disclosed regarding the above embodiment.

[1589] (Claim 1)

[1590] A means for obtaining information data from social media;

[1591] means for pre-processing the acquired information data;

[1592] A means for determining the authenticity of information using a machine learning model based on preprocessed data;

[1593] a means for filtering the determined false information;

[1594] The system includes a means for notifying a user of filtered information.

[1595] (Claim 2)

[1596] 10. The system of claim 1, wherein the system obtains information data from social media in real time.

[1597] (Claim 3)

[1598] The system of claim 1, wherein the system performs tokenization and sanitization of the acquired information data.

[1599] (Claim 4)

[1600] The system of claim 1, which performs parameter adjustment and retraining of a machine learning model.

[1601] (Claim 5)

[1602] 10. The system of claim 1, wherein data determined to be false information is filtered in real time.

[1603] (Claim 6)

[1604] 10. The system of claim 1, wherein the filtered information is notified to the user in real time.

[1605] "Example 1"

[1606] (Claim 1)

[1607] A means for obtaining information data from social media;

[1608] means for pre-processing the acquired information data;

[1609] A means for determining the authenticity of information using a machine learning model based on preprocessed data;

[1610] a means for filtering the determined false information;

[1611] a means for notifying the user of the filtered information;

[1612] A means of tokenizing information data, removing unnecessary vocabulary and stop words, and normalizing the data;

[1613] A system that includes a means for a machine learning model to score the truth or falsity of information and store the results in a database.

[1614] (Claim 2)

[1615] 10. The system of claim 1, wherein the system obtains information data from social media in real time.

[1616] (Claim 3)

[1617] The system of claim 1, wherein the system performs tokenization and sanitization of the acquired information data.

[1618] (Claim 4)

[1619] The system of claim 1, wherein a machine learning model scores the truth of the information based on the acquired data.

[1620] (Claim 5)

[1621] 2. The system according to claim 1, wherein the means for notifying the user of the filtering results uses a pop-up or push notification.

[1622] "Application Example 1"

[1623] (Claim 1)

[1624] A means for obtaining information data from social media;

[1625] means for pre-processing the acquired information data;

[1626] A means for determining the authenticity of information using a machine learning model based on preprocessed data;

[1627] a means for filtering the determined false information;

[1628] a means for notifying the user of the filtered information;

[1629] A means for tokenizing and removing unnecessary information from information data;

[1630] A means to notify the filtered results in real time,

[1631] The system includes a means to issue an alert when false information is detected.

[1632] (Claim 2)

[1633] 10. The system of claim 1, wherein the system obtains information data from social media in real time and collects data.

[1634] (Claim 3)

[1635] The system according to claim 1, further comprising: pre-processing the acquired information data and performing reliability evaluation.

[1636] "Example 2: Combining Emotion Engines"

[1637] (Claim 1)

[1638] A means for obtaining information data from social media;

[1639] means for pre-processing the acquired information data;

[1640] A means for determining the authenticity of information using a machine learning model based on preprocessed data;

[1641] a means for filtering the determined false information;

[1642] means for analyzing the emotional state of a user;

[1643] means for adjusting filtering criteria and notification content based on the analysis results;

[1644] The system includes a means for notifying a user of filtered information.

[1645] (Claim 2)

[1646] 10. The system of claim 1, wherein the system obtains information data from social media in real time.

[1647] (Claim 3)

[1648] The system of claim 1, wherein the system performs tokenization and sanitization of the acquired information data.

[1649] "Application example 2 when combining emotion engines"

[1650] (Claim 1)

[1651] A means for obtaining information data from social media;

[1652] means for pre-processing the acquired information data;

[1653] A means for determining the authenticity of information using a machine learning model based on preprocessed data;

[1654] a means for filtering the determined false information;

[1655] means for analyzing the emotional state of a user;

[1656] means for adjusting filtering results and notification content based on emotional state;

[1657] The system includes a means for notifying a user of filtered information.

[1658] (Claim 2)

[1659] 10. The system of claim 1, wherein the system obtains information data from social media in real time.

[1660] (Claim 3)

[1661] The system of claim 1, wherein the system performs tokenization and sanitization of the acquired information data. [Explanation of symbols]

[1662] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A means for obtaining information data from social media; means for pre-processing the acquired information data; A means for determining the authenticity of information using a machine learning model based on preprocessed data; a means for filtering the determined false information; The system includes a means for notifying a user of filtered information.

2. The system of claim 1 , wherein the system obtains information data from social media in real time.

3. The system of claim 1 , wherein the system performs tokenization and garbage collection of the acquired information data.

4. The system of claim 1 , which performs parameter adjustment and retraining of a machine learning model.

5. 10. The system of claim 1, wherein data determined to be false information is filtered in real time.

6. 10. The system of claim 1, wherein the filtered information is notified to the user in real time.

Images