System

Abstract

A system is provided.SOLUTION: A system includes a means for collecting exhibition data, a means for analyzing the collected exhibition data, a means for detecting fraud based on the analyzed exhibition data, a means for blocking exhibition in response to the detected fraud, and a means for notifying an exhibitor and a manager when the fraud is detected.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] Fraudulent listings are common in internet auctions, damaging user trust and increasing operational costs. The listing of counterfeit branded products and the listing of large quantities of the same product are particularly problematic. A system that can efficiently and accurately detect and prevent such fraudulent listings is needed. [Means for solving the problem]

[0005] The present invention is a system including a means for collecting listing data, a means for analyzing the collected listing data, a means for detecting fraud based on the analyzed listing data, a means for blocking listings in response to detected fraud, and a means for notifying sellers and administrators when fraud is detected.

[0006] In this system, the accuracy of fraud detection is improved by further including a means for extracting product titles, descriptions, image links, and seller IDs from the collected listing data. Also, the analysis means inputs the collected listing data into a machine learning model and detects fraud based on the output, thereby achieving highly accurate fraud detection.

[0007] Specifically, the server periodically collects listing data from the Yahoo! Auctions database and trains that data using machine learning models such as the Gemini AI model to identify fraudulent listing patterns. The data is analyzed in real time each time an item is listed, and if a listing is deemed fraudulent, it is automatically blocked, a warning is sent to the seller, and details are reported to the administrator. This effectively prevents fraudulent activity in internet auctions and provides an environment where users can use the service with peace of mind.

[0008] "Listing data" refers to information such as the product title, description, image link, and seller ID provided by the seller in an Internet auction.

[0009] "Means of collection" refers to the process or functionality for periodically or real-timely retrieving listing data from the auction site's database.

[0010] "Means for analysis" refers to the process or function for analyzing collected listing data based on specific algorithms or models to assess the possibility of fraud.

[0011] "Fraud detection measures" refers to processes or functions for identifying fraudulent activities, such as counterfeit branded products or bulk listings, based on analyzed listing data.

[0012] "Blocking Measures" refers to the process or functionality used to prevent a detected fraudulent listing from being published on the auction site.

[0013] "Means of Notification" refers to the process or functionality for sending warning messages and detailed reports to sellers and administrators when fraud is detected.

[0014] A "machine learning model" is an algorithm used for data analysis, a mathematical model that learns from large amounts of data and makes predictions and decisions based on future data. [Brief explanation of the drawings]

[0015] [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

[0016] 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.

[0017] First, the terms used in the following description will be explained.

[0018] 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).

[0019] 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.

[0020] 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.

[0021] 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.

[0022] 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."

[0023] [First embodiment]

[0024] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0025] 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.

[0026] 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).

[0027] 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.

[0028] 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.

[0029] 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.

[0030] 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.

[0031] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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."

[0036] This invention is a system for detecting and preventing fraudulent listings in internet auctions, and implements a series of processes for collecting and analyzing listing data and detecting fraud. This system is mainly composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, and a warning / blocking module.

[0037] Data Collection Module

[0038] The server periodically connects to the Yahoo! Auctions database to retrieve new listing data, including product titles, descriptions, image links, seller IDs, etc. The retrieved data is temporarily stored in storage.

[0039] AI Learning Module

[0040] The server extracts listing data that is deemed fraudulent from the data stored in the storage and trains it into a machine learning model (e.g., the Gemini AI model). This training is based on a specific algorithm and is used to identify fraudulent listing patterns. Once training is complete, the model will be able to predict the likelihood of fraud for new listing data with high accuracy.

[0041] Fraud detection module

[0042] The server continuously monitors and captures new listings in real time. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. If the model's output exceeds a certain threshold, the listing is deemed fraudulent and flagged.

[0043] Warning and Blocking Module

[0044] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and the seller receives a warning message, such as "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, and the reason for the fraud.

[0045] Natural language explanation of the process

[0046] 1. Data Collection

[0047] Server: The server periodically collects new listing data from the auction database and stores it in storage.

[0048] Users: When users list items, their information is collected automatically.

[0049] 2. AI Learning

[0050] Server: The server inputs the collected data into a machine learning model to train the model on fraudulent patterns. This process identifies counterfeit branded products and bulk listing data.

[0051] 3. Real-time analysis

[0052] Server: When a new listing is made, the server immediately feeds the data into a machine learning model to analyze it for fraud.

[0053] Device: The device checks the status of the listing based on the analysis results sent from the server.

[0054] 4. Warning / Block

[0055] Server: If a listing is deemed fraudulent, the server will automatically block the listing and send a warning to the seller.

[0056] Users: Sellers receive a warning message and are informed that their problematic listings have been blocked.

[0057] Administrator: The administrator will receive a detailed report from the server and take further action if necessary.

[0058] Specific examples

[0059] Case 1: Detecting counterfeit branded products

[0060] User: Sells fake designer handbags.

[0061] Server: Collects new listings and stores them in storage.

[0062] Server: Trains the machine learning model on the collected data.

[0063] Server: The machine learning model analyzes data on newly listed handbags.

[0064] Result: If the model determines that the product is a counterfeit, the server blocks the listing and sends a warning to the seller. A detailed report is sent to the administrator.

[0065] Case 2: Large number of identical products listed

[0066] User: Listing a large number of identical items in a short period of time.

[0067] Server: Collects listing data from the auction database and stores it in storage.

[0068] Server: Trains the machine learning model on the collected data.

[0069] Server: The machine learning model analyzes data on a large number of identical product listings.

[0070] Result: If it is determined to be a bulk listing, the server will block all of the listings, send a warning message to the seller, and send a detailed report to the administrator.

[0071] This allows the system to efficiently detect and prevent fraudulent listings, providing an environment where users can use the service with peace of mind, while also enabling administrators to respond quickly.

[0072] The processing flow will be explained below.

[0073] Step 1:

[0074] The server connects to the Yahoo! Auctions database every hour to retrieve all new listing data. The retrieved data includes the product title, description, image link, and seller ID. The retrieved data is temporarily stored in storage.

[0075] Step 2:

[0076] The server extracts the product title, description, image link, and seller ID from the listing data stored in storage. It checks the data format and logs any errors.

[0077] Step 3:

[0078] The server feeds the extracted data into a machine learning model (e.g., the Gemini AI model) and the model begins the learning process, which allows the AI ​​to understand past fraudulent listing patterns and predict future fraudulent listings.

[0079] Step 4:

[0080] A user lists a new item on an auction site, and the listing is automatically added to the auction site's database.

[0081] Step 5:

[0082] When a new item is listed, the server immediately detects and collects the listing data, which is then stored in the storage.

[0083] Step 6:

[0084] The server inputs collected new listing data into a machine learning model in real time, and the AI ​​analyzes the listing for possible fraud. If the analysis result exceeds a threshold, it is flagged as fraudulent.

[0085] Step 7:

[0086] The server immediately blocks any listings that are flagged as fraudulent, changing the listing status in the database to "blocked" and prohibiting access to the product page from front-end devices.

[0087] Step 8:

[0088] The server will send a warning message to the fraudulent seller, either by email or via the notification function, informing them that "This item is not an authorized listing."

[0089] Step 9:

[0090] The server will send a detailed report to the administrator, which will include the listing ID, seller ID, reason for the fraud, and product data. The report will be displayed directly in the administrator dashboard.

[0091] This process enables the system to quickly and accurately detect fraudulent listings, allowing users to use the auction service with peace of mind.

[0092] Example 1

[0093] 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."

[0094] Fraudulent listings in internet auctions can cause a decline in credibility and harm users, so it is important to detect and deal with them quickly and accurately. However, current systems have difficulty effectively monitoring and analyzing large amounts of listing data, and fraudulent listings can be overlooked. Furthermore, it takes time to deal with fraud after detecting it, which can ultimately increase the damage. There is a need to solve these problems and provide a safer and more reliable auction environment.

[0095] 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.

[0096] In this invention, the server includes means for periodically connecting to the database to collect listing data, means for saving the collected listing data in storage, and means for inputting the saved listing data into a machine learning model to learn fraudulent patterns. This makes it possible to quickly and accurately detect and deal with fraudulent listings. In addition, new listing data can be monitored in real time and immediately input into the machine learning model to predict the possibility of fraud, enabling timely response.

[0097] "Means for periodically connecting to the database to collect listing data" refers to a technical element for accessing the auction site's database at regular intervals to obtain new listing information.

[0098] The "means for storing collected listing data in storage" refers to a data storage system or storage media for temporarily or long-term storage of acquired listing information.

[0099] The "means of inputting stored listing data into a machine learning model to learn fraudulent patterns" refers to a technical element that uses an algorithm to train a machine learning model based on accumulated listing data, thereby learning fraudulent listing patterns.

[0100] "Means for monitoring new listing data in real time" refers to the technology that instantly detects and collects newly registered listing information on the auction site.

[0101] The "means of inputting monitored listing data into a machine learning model to predict the possibility of fraud" refers to a technical element that inputs collected listing data into a trained machine learning model to predict whether a listing is fraudulent.

[0102] "Means for blocking listings determined to be fraudulent" refers to a technical element that, based on the output of a machine learning model, blocks listings that are deemed fraudulent from being published and makes them invisible on the auction site.

[0103] "Means for notifying sellers and administrators when fraud is detected" refers to a notification system that sends warnings and detailed information to sellers and site administrators when fraudulent listings are detected.

[0104] "Product Title" is a data item that indicates the name or title of the product being put up for sale.

[0105] The "description" is text data that describes detailed information about the product being sold, and explains the product's characteristics, condition, usage, etc.

[0106] "Image link" is a data item that indicates the URL or link to access the image or photo of the item being sold.

[0107] The "seller ID" is an identifier or account ID for identifying the user who put up an item for sale.

[0108] This invention is a system for detecting and preventing fraudulent listings in internet auctions, specifically for collecting and analyzing listing data from the Yahoo! Auctions database to detect and respond to fraud. This system is primarily composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, and a warning and blocking module.

[0109] Data Collection Module

[0110] The server periodically connects to the Yahoo! Auctions database to collect listing data. An API key or authentication token is used for the connection. The collected data includes the product title, description, image link, and seller ID, and this data is temporarily stored in storage. Specifically, a structured database such as MySQL or PostgreSQL is generally used.

[0111] AI Learning Module

[0112] The server extracts labeled data that is deemed to be fraudulent from the data stored in the storage and inputs it into a machine learning model to learn fraudulent patterns. The machine learning model used can be the Gemini AI model, which is widely used in industry, or another appropriate model. Python and the Scikit-learn library are often used for the learning process.

[0113] Fraud detection module

[0114] The server continuously monitors new listings in real time and detects new listing data. The detected data is immediately input into a machine learning model, which uses that data to predict the possibility of fraud. If the output exceeds a certain threshold, it is deemed fraudulent. Specifically, a Python script is used to input data into the model and obtain prediction results.

[0115] Warning and Blocking Module

[0116] Any listings that are deemed fraudulent are automatically blocked by the server and are set to not be published on the auction site. At the same time, a warning message is sent to the seller stating, "This item is not permitted for listing." A detailed report is also sent to the administrator, including the listing ID, seller ID, and reason for the fraud. Notification systems typically use email or the administrator dashboard.

[0117] Specific examples

[0118] Case 1: Detecting counterfeit branded products

[0119] User: Sells fake designer handbags.

[0120] Server: Obtains new listing data and saves it to storage.

[0121] Server: Trains machine learning models based on the stored data.

[0122] Specifically, the system extracts features such as "if the brand name is not listed on the official website, it is considered a fraudulent listing."

[0123] Server: Data on newly listed handbags is instantly fed into the model and analyzed.

[0124] Specific operations include techniques such as "detecting logo mismatches using image analysis."

[0125] Result: If the model determines that the product is a counterfeit, the server blocks the listing and sends a warning to the seller. A detailed report is sent to the administrator.

[0126] Case 2: Large number of identical products listed

[0127] User: Listing a large number of identical items in a short period of time.

[0128] Server: Collects listing data and stores it in storage.

[0129] Server: Trains the machine learning model based on the collected data.

[0130] Specifically, the system extracts features such as "detecting when the same title or image link appears frequently in a short period of time."

[0131] Server: Analyzes data on a large number of identical products listed.

[0132] Specifically, the model determines that the appearance of a certain number of identical products is fraudulent.

[0133] Result: The server will block all affected listings and send a warning message to the seller. A detailed report will be sent to the administrator.

[0134] Prompt Sentence Examples

[0135] "Use listing data collected from Yahoo! Auctions to train a machine learning model to detect fraudulent listings. Then, when new listing data is entered, analyze it for potential fraud."

[0136] This invention efficiently detects and prevents fraudulent listings, providing a safe and secure auction environment for users. It also enables administrators to take prompt and appropriate action, contributing to improving the reliability of the entire site.

[0137] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0138] Step 1:

[0139] Input: Yahoo! Auctions database connection information (API key, authentication token, etc.)

[0140] Server: The server periodically connects to the Yahoo! Auctions database to collect new listing data. Specifically, it uses the API to obtain information such as product titles, descriptions, image links, and seller IDs.

[0141] Output: Listing data (product title, description, image link, seller ID)

[0142] Step 2:

[0143] Input: Listing data collected in Step 1

[0144] Server: The server stores the acquired listing data in storage. Specifically, it stores the data in a database such as MySQL or PostgreSQL.

[0145] Output: Listing data saved in storage

[0146] Step 3:

[0147] Input: Past listing data stored in storage

[0148] Server: The server extracts labeled data that is deemed to be fraudulent listings and inputs it into a machine learning model to train it. Specifically, it uses Python and Scikit-learn to preprocess the data, extract features, and train the model.

[0149] Output: A trained machine learning model (generative AI model)

[0150] Step 4:

[0151] Input: New listing data retrieved in real time

[0152] Server: The server monitors the Yahoo! Auctions database in real time, detecting new listings and immediately feeding that data into the machine learning model.

[0153] Output: New listing data that is fed into the machine learning model

[0154] Step 5:

[0155] Input: New listing data fed into the machine learning model in Step 4

[0156] Server: The server uses machine learning models to predict the likelihood of fraud in new listings. If the prediction exceeds a certain threshold, it is deemed fraudulent. A Python script is used for the specific processing.

[0157] Output: Identification information (flag) of listing data that is determined to be fraudulent

[0158] Step 6:

[0159] Input: Listing data identified as fraudulent in Step 5

[0160] Server: The server automatically blocks listings that are deemed fraudulent, preventing them from being published on the auction site.

[0161] Output: Blocked listing data

[0162] Step 7:

[0163] Input: Identification information of the listing data that was determined to be fraudulent in Step 5

[0164] Server: Using the transmission system, send a warning message to the seller stating "This item is not an authorized listing." A detailed report is also sent to the administrator, including the listing ID, seller ID, and reason for the fraud.

[0165] Output: Warning message to seller and report to administrator

[0166] By implementing the above steps, the system can quickly and accurately detect and respond to fraudulent listings, providing a safe and reliable auction environment. Furthermore, by incorporating specific actions, the processing content of each step can be more clearly understood.

[0167] (Application example 1)

[0168] 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."

[0169] The listing of counterfeit goods, fake branded goods, and large quantities of identical goods on Internet auction and mail-order sites has become a problem. Such fraudulent listings not only undermine the trust of buyers, but also hinder sound market transactions. However, there is still no effective system for efficiently detecting and quickly dealing with these fraudulent listings. The present invention aims to provide a new method for solving these problems.

[0170] 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.

[0171] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for blocking listings in response to detected fraud, means for notifying the seller and the administrator when fraud is detected, and means for monitoring the listing data in real time via a terminal equipped with a fraudulent product listing detection application and analyzing the data using a generative AI model to evaluate the authenticity of the product, thereby enabling the listing of counterfeit products and fake brand products to be detected and prevented quickly and efficiently.

[0172] "Listing data" refers to information including the title, description, image link, seller ID, etc. of a product that a seller has listed on an online shopping site or auction site.

[0173] "Means for analyzing" refers to means for processing the collected data and conducting analysis to identify counterfeit or fake branded products.

[0174] The "means for detecting fraud" is a means for determining whether an item is a fraudulent listing based on the analyzed data.

[0175] "Methods for blocking listings" refers to removing listings that are detected as fraudulent from auction sites and mail-order sites, preventing them from being made public.

[0176] "Means of notification" refers to the means used to notify sellers and administrators when fraud is detected.

[0177] The "fraudulent product listing detection application" is a software application for detecting fraudulent product listings in real time.

[0178] A "generative AI model" is a model trained by artificial intelligence that is used to predict fraudulent product listings with high accuracy.

[0179] A "prompt" is text data to be input into the generative AI model, and includes the product title, description, image link, seller ID, etc.

[0180] This invention is a system that efficiently detects and prevents fraudulent product listings on mail order and auction sites by using a terminal and a server equipped with a fraudulent product listing detection application. Specific methods for implementing the invention are described in detail below.

[0181] Data Collection Module

[0182] The server periodically connects to the database of the mail order or auction site to retrieve new listing data. This data includes product titles, descriptions, image links, seller IDs, etc., and the retrieved data is temporarily stored in storage. Communication is via the Internet and REST API.

[0183] AI Learning Module

[0184] The server extracts listing data that is deemed fraudulent from the large amount of listing data stored in storage, and trains a generative AI model, a machine learning model. This generative AI model is used to identify fraudulent listing patterns from large amounts of data. High-performance servers and GPUs are used to train the AI ​​model.

[0185] Fraud detection module

[0186] The server continuously monitors new listings in real time and acquires data as they occur. Each time new listing data is entered, the server inputs this data into a generative AI model and analyzes the possibility of listing fraud. If the model output exceeds a certain threshold, it is deemed to be fraudulent. An example of a prompt used at this time is as follows:

[0187] Example prompt sentence:

[0188] Product Title: "Luxury brand bag"

[0189] Product Description: "Brand new, unused luxury brand bag."

[0190] Image link: "https: / / example.com / image.jpg"

[0191] Seller ID: "seller1234"

[0192] Warning and Blocking Module

[0193] The server automatically blocks listings that are deemed fraudulent. This blocking prevents the product from being displayed on online shopping or auction sites. If fraud is detected, a warning message is automatically sent to the seller. At the same time, a detailed report is sent to the administrator. This report includes the listing ID, seller ID, and the reason why the fraud was detected.

[0194] Overall system description

[0195] This system helps maintain the integrity of online shopping and auction sites. Users can purchase products with peace of mind, and administrators can quickly address fraudulent listings. The specific process involves analyzing collected listing data, predicting fraud using machine learning models, and blocking listings in real time. All of these processes are automated, increasing the efficiency and reliability of the entire system.

[0196] By utilizing the technical features of the present invention and using the above-described method to effectively prevent fraudulent product listings on mail-order sites and auction sites, it is possible to realize a highly reliable online marketplace.

[0197] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0198] Step 1:

[0199] The server periodically connects to the online shopping site's database to collect new listing data. The input is listing information received from the online shopping site's API, and the output is raw data stored in the server's storage. This data includes product titles, descriptions, image links, and seller IDs.

[0200] Step 2:

[0201] The server inputs the stored listing data into an analysis module, which converts the data into a format suitable for the generative AI model. The input here is raw data, and the output is data converted into a format that the AI ​​model can understand.

[0202] Step 3:

[0203] The server uses a generative AI model to train the converted data. It uses the fraudulent listing data that was originally accumulated and learns fraudulent patterns from newly collected data. The input of this step is the reformatted listing data, and the output is the trained AI model.

[0204] Step 4:

[0205] The server monitors new listing data in real time and inputs the collected data into the generative AI model. The input data is in the form of a prompt, and the output is a score indicating the likelihood of fraud. The prompt can be in the following form:

[0206] Product Title: "Luxury brand bag"

[0207] Product Description: "Brand new, unused luxury brand bag."

[0208] Image link: "https: / / example.com / image.jpg"

[0209] Seller ID: "seller1234"

[0210] Step 5:

[0211] The server determines the listing to be fraudulent and flags it if the score obtained from the generated AI model exceeds a certain threshold. The input is the AI ​​model score, and the output is the flagged listing data.

[0212] Step 6:

[0213] The server automatically blocks flagged listings and removes them from the online store's database. The input is the flagged listing data, and the output is the blocked listing data.

[0214] Step 7:

[0215] The server sends a warning message to the seller of any listings that are determined to be fraudulent. The message includes the content "This item has been determined to be a fraudulent listing," and at the same time, a detailed report is sent to the administrator. The input is the flagged listing data and seller information, and the output is a warning message to the seller and a report to the administrator.

[0216] 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.

[0217] This invention is a system that detects and prevents fraudulent listings in internet auctions, and by combining it with an emotion engine that recognizes user emotions, it achieves highly accurate fraudulent listing detection. This system is mainly composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, a warning / blocking module, and an emotion engine.

[0218] Data Collection Module

[0219] The server periodically connects to the internet auction database to retrieve new listing data, including product titles, descriptions, image links, and seller IDs, and temporarily stores the retrieved data in storage.

[0220] AI Learning Module

[0221] The server extracts listing data that is deemed fraudulent from the data stored in the storage and trains it into a machine learning model (e.g., an AI model). The training process is carried out so that the AI ​​can understand past fraudulent listing patterns and predict future fraudulent listings.

[0222] Fraud detection module

[0223] The server continuously monitors and captures new listings in real time. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. If the model's output exceeds a certain threshold, it is flagged as fraudulent.

[0224] Emotion Engine

[0225] The emotion engine is designed to recognize the user's emotions. It analyzes the seller's emotional state from their behavior and input and generates emotion data. Specifically, it can detect emotional states such as "anger" and "excitement."

[0226] Emotional Data Integration

[0227] The server combines the emotion data provided by the emotion engine with the results of analyzing the listing data to perform more accurate fraud detection, making it easier to identify sellers who are likely to be engaging in fraudulent activity.

[0228] Warning and Blocking Module

[0229] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and a warning message is sent to the seller. For example, the seller may receive a notification that "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, reason for fraud, and product data.

[0230] Natural language explanation of the process

[0231] 1. Data Collection

[0232] Server: The server periodically collects new listing data from the auction database and stores it in storage.

[0233] Users: When users list items, their information is collected automatically.

[0234] 2. AI Learning

[0235] Server: The server inputs the collected data into a machine learning model to train the model on fraudulent patterns. This process identifies counterfeit branded products and bulk listing data.

[0236] 3. Real-time analysis

[0237] Server: When a new listing is made, the server immediately feeds the data into a machine learning model to analyze it for fraud.

[0238] Device: The device checks the status of the listing based on the analysis results sent from the server.

[0239] 4. Acquiring Emotion Data

[0240] Server: Analyzes the seller's input and behavior, and uses the emotion engine to obtain emotion data.

[0241] Emotion engine: Detects the seller's emotional state (e.g., anger or excitement) and sends the emotional data to the server.

[0242] 5. Emotional Data Integration

[0243] Server: Integrates the analysis results of emotion data and listing data to determine whether there is any possibility of fraud.

[0244] 6. Warning Block

[0245] Server: Blocks listings that are deemed fraudulent, sends a warning to the seller, and also sends a detailed report to the administrator.

[0246] Users: Sellers receive a warning message and are informed that their problematic listings have been blocked.

[0247] Administrator: The administrator will receive a detailed report from the server and take further action if necessary.

[0248] Specific examples

[0249] Case 1: Detecting counterfeit branded products

[0250] User: Sells fake designer handbags.

[0251] Server: Collects new listings and stores them in storage.

[0252] Server: Trains the machine learning model on the collected data.

[0253] Server: The machine learning model analyzes data on newly listed handbags.

[0254] Emotion Engine: Analyzes the seller's emotional state and detects anger and excitement.

[0255] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[0256] Server: Blocks the listing and sends a warning notice to the seller. Sends a detailed report to the administrator.

[0257] Case 2: Large number of identical products listed

[0258] User: Listing a large number of identical items in a short period of time.

[0259] Server: Collects listing data from the auction database and stores it in storage.

[0260] Server: Trains the machine learning model on the collected data.

[0261] Server: The machine learning model analyzes data on a large number of identical product listings.

[0262] Emotion Engine: Analyzes the seller's emotional state and detects abnormal excitement.

[0263] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[0264] Server: Block all listings and send a warning message to the seller. Send a detailed report to the administrator.

[0265] In this way, the system can effectively prevent illegal activities and fraudulent use, providing users with a safe and secure auction service. Furthermore, the integration of emotion data can further improve the accuracy of fraud detection.

[0266] The processing flow will be explained below.

[0267] Step 1:

[0268] The server connects to the Internet auction database every hour to retrieve all new listing data. The listing data includes the product title, description, image link, and seller ID. The retrieved data is temporarily stored in storage.

[0269] Step 2:

[0270] The server extracts the product title, description, image link, and seller ID from the listing data stored in the storage. It checks the data format and logs any errors.

[0271] Step 3:

[0272] The server then inputs the extracted data into a machine learning model (e.g., an AI model) and the model begins the learning process, which allows the AI ​​to understand past fraudulent listing patterns and predict future fraudulent listings.

[0273] Step 4:

[0274] A user lists a new item on an auction site, and the listing is automatically added to the auction site's database.

[0275] Step 5:

[0276] When a new item is listed, the server immediately detects and collects the listing data, which is then stored in the storage.

[0277] Step 6:

[0278] The server inputs collected new listing data into a machine learning model in real time, and the AI ​​analyzes the listing for possible fraud. If the analysis result exceeds a threshold, it is flagged as fraudulent.

[0279] Step 7:

[0280] The server analyzes the seller's behavior and input content and acquires emotional data using an emotion engine, which detects the seller's emotional state (e.g., anger or excitement).

[0281] Step 8:

[0282] The server combines the emotion data with the results of the machine learning model analysis to perform more accurate fraud detection. If a fraud flag is raised, the server immediately blocks the listing. It changes the listing status in the database to "blocked" and prohibits access to the product page from front-end devices.

[0283] Step 9:

[0284] The server will send a warning message to the fraudulent seller, either by email or via the notification function, informing them that "This item is not an authorized listing."

[0285] Step 10:

[0286] The server will send a detailed report to the administrator, including the listing ID, seller ID, reason for the fraud, and product data. The report will be displayed on the administrator dashboard.

[0287] Through this series of processes, the system detects and prevents fraudulent listings with high accuracy and efficiency, providing an environment where users can use the auction service with peace of mind.The integration of emotional data makes it easier to identify sellers who are likely to have fraudulent intentions.

[0288] Example 2

[0289] 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."

[0290] The increase in fraudulent listings in internet auctions has created a demand for a safe and reliable trading environment. However, conventional fraud detection systems only detect fraud based on the analysis of listing data and do not take into account the emotional state of the seller, resulting in insufficient accuracy. Furthermore, there are problems with delays in appropriate response and overlooking fraud. Therefore, there is a need for a system that can more accurately detect fraud and enable prompt response by integrating not only listing data but also seller emotional data.

[0291] 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.

[0292] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for acquiring emotional data of the seller, means for integrating the acquired emotional data with the analysis means, means for blocking the listing in response to detected fraud, and means for notifying the seller and the administrator when fraud is detected. This makes it possible to detect fraudulent listings with higher accuracy and to respond quickly and appropriately.

[0293] "Listing data" refers to information such as the product title, description, image link, and seller ID provided by a seller in an Internet auction.

[0294] The "analysis means" is a function for processing collected listing data and determining whether or not there is any fraud.

[0295] The "fraud detection means" is a function for identifying fraud based on the analyzed auction data.

[0296] "Emotional data" is information indicating the emotional state of the seller obtained from their behavior and input content.

[0297] The "emotion acquisition means" is a function for acquiring emotional data of the seller.

[0298] The "data integration means" is a function for combining the analysis results of the auction data by the analysis means with emotion data.

[0299] "Blocking measures" are functions that respond to detected fraud and remove listings from the auction site.

[0300] "Notification means" is a function for notifying the seller and administrator of the relevant information when fraud is detected.

[0301] This invention is a system for detecting and preventing fraudulent listings in internet auctions. This invention is centered around a server and consists of a data collection module, an AI learning module, a fraud detection module, a warning / blocking module, and an emotion engine.

[0302] Data Collection Module

[0303] The server periodically connects to the internet auction database to retrieve new listing data. This data includes the product title, description, image link, and seller ID, and the retrieved data is temporarily stored in storage. For example, the operation is "The server sends an API request to the auction database to retrieve new listing information."

[0304] AI Learning Module

[0305] The server extracts listing data deemed fraudulent from the data stored in storage and trains it on a machine learning model (e.g., a generative AI model). During the training process, the AI ​​learns past fraudulent listing patterns and becomes capable of predicting future fraudulent listings. A high-performance server equipped with a GPU is recommended as the hardware to be used.

[0306] Fraud detection module

[0307] The server continuously monitors new listings in real time and captures that data. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. For example, if the AI ​​model exceeds a certain threshold, it flags the listing as fraudulent.

[0308] Emotion Engine

[0309] The emotion engine is designed to recognize the user's emotions. It analyzes the seller's emotional state from their behavior and input content and generates emotional data. Specifically, it can detect emotional states such as "anger" and "excitement." For example, the emotion engine works by reading emotions based on the user's input content and generating data.

[0310] Emotional Data Integration

[0311] The server combines the emotion data provided by the emotion engine with the results of analyzing the listing data to perform more accurate fraud detection, making it easier to identify sellers who are likely to be engaging in fraudulent activity.

[0312] Warning and Blocking Module

[0313] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and the seller receives a warning message, such as "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, reason for fraud, and product data.

[0314] Specific examples

[0315] Case 1: Detecting counterfeit branded products

[0316] User: Sells fake designer handbags.

[0317] Server: Collects new listing data and stores it in storage.

[0318] Server: Trains the machine learning model on the collected data.

[0319] Server: The machine learning model analyzes data on newly listed handbags.

[0320] Emotion Engine: Analyzes the seller's emotional state and detects anger and excitement.

[0321] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[0322] Server: Blocks the listing and sends a warning notice to the seller. Sends a detailed report to the administrator.

[0323] Case 2: Large number of identical products listed

[0324] User: Listing a large number of identical items in a short period of time.

[0325] Server: Collects listing data from the auction database and stores it in storage.

[0326] Server: Trains the machine learning model on the collected data.

[0327] Server: The machine learning model analyzes data on a large number of identical product listings.

[0328] Emotion Engine: Analyzes the seller's emotional state and detects abnormal excitement.

[0329] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[0330] Server: Block all listings and send a warning message to the seller. Send a detailed report to the administrator.

[0331] This allows the system to effectively prevent illegal activities and fraudulent use, providing users with a safe and secure auction service. Furthermore, the integration of emotion data can further improve the accuracy of fraud detection.

[0332] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0333] Step 1:

[0334] Data collection

[0335] Server: The server periodically connects to the auction database and collects new listing data.

[0336] Input: New listing data (item title, description, image link, seller ID) to be retrieved from the auction database.

[0337] Output: The retrieved listing data saved to storage.

[0338] Specific operation: The server sends a request to the API endpoint and saves the obtained listing data as various fields (product title, description, image link, seller ID) in the internal storage database.

[0339] Step 2:

[0340] AI Learning

[0341] Server: Extracts fraudulent listing data from the data stored in the storage and trains it using a machine learning model.

[0342] Input: Listing data saved in storage.

[0343] Output: An updated and further trained machine learning model.

[0344] What it does: The server pre-processes the stored listing data, performs feature extraction, and feeds it into a machine learning model, which learns to recognize new fraudulent listings based on data patterns.

[0345] Step 3:

[0346] Real-time analytics

[0347] Server: Monitors and validates new listing data in real time.

[0348] Input: New listing data.

[0349] Output: Judgment result of whether or not there is suspicion of fraud. Fraud flag.

[0350] What it does: The server retrieves new listing data as it arrives, feeds it into a machine learning model for real-time analysis, and flags it as fraudulent if it exceeds a certain threshold.

[0351] Step 4:

[0352] Acquiring emotion data

[0353] Server: Uses an emotion engine to obtain emotion data based on the seller's behavior and input.

[0354] Input: Seller's behavior log and input content.

[0355] Output: Emotion data (e.g., anger, excitement, etc.).

[0356] Specific operation: The server sends the seller's behavioral data to the emotion engine, which then obtains emotion data as the analysis result. At that time, natural language processing technology is used to detect signs of emotion from the text data.

[0357] Step 5:

[0358] Emotional Data Integration

[0359] Server: The results of the analysis of the listing data are combined with the emotional data to determine the likelihood of fraud.

[0360] Input: Fraud detection data and emotion data.

[0361] Output: Integrated analysis results, invalid flag.

[0362] Specific operation: The server integrates the fraud judgment results from the machine learning model with the emotion data from the emotion engine, and performs specific scoring to further increase the likelihood of fraud.

[0363] Step 6:

[0364] Warning Block

[0365] Server: Automatically blocks listings that are deemed fraudulent and sends warning messages to sellers.

[0366] Input: Combined fraud flags and listing data.

[0367] Output: Blocked listing status, warning notice to seller, detailed report to administrator.

[0368] Specific operation: The server executes the command to block the fraudulent listing that matches the conditions, sends a warning message to the seller saying "This item is not permitted for listing," and also sends a detailed report to the administrator, including the listing ID, seller ID, reason for the fraud, etc.

[0369] (Application example 2)

[0370] 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."

[0371] Conventional auction systems have difficulty detecting fraudulent listings, allowing sellers to intentionally manipulate listing information to deceive the system. Furthermore, fraud detection does not take into account the emotional state of the seller, resulting in a lack of accuracy. Furthermore, it is difficult to detect and notify fraudulent activity in real time, requiring rapid response.

[0372] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0373] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for blocking listings in response to detected fraud, means for notifying sellers and administrators when fraud is detected, means for collecting user emotion data in real time and analyzing the emotion data, and means for integrating the analysis means and the emotion data to improve fraud detection accuracy. This improves accuracy compared to conventional fraud detection methods, enabling rapid detection of fraudulent activity and countermeasures.

[0374] "Listing data" refers to information entered by a user when listing an item on an auction site, and specifically includes the item title, description, image link, seller ID, and the like.

[0375] The "collection means" refers to a device or program that has the function of acquiring the auction data from the auction site database and storing it on a server.

[0376] "Analysis means" refers to a device or program that executes a machine learning model or algorithm to assess the likelihood of fraud using collected listing data.

[0377] The "fraud detection means" is a device or program that has the function of identifying items that are likely to be fraudulent based on data analyzed using the analysis means.

[0378] "Blocking means" refers to a device or program that has the function of preventing items that are determined to be fraudulent from being published on the auction site.

[0379] The "notification means" is a device or program that has the function of sending a message to notify the seller and administrator when fraud is detected.

[0380] "Emotion data" is data that quantifies or categorizes the user's emotional state, and specifically represents emotions such as "anger," "excitement," and "fear."

[0381] The "emotion data collection means" is a device or program that has the function of collecting emotion data from the user's facial expressions, voice, etc. in real time.

[0382] An "emotion analysis engine" is an algorithm or program that analyzes collected emotional data and classifies the user's emotional state.

[0383] The "integration means" is a device or program that has the function of combining the results of the analysis of the auction data obtained from the analysis means with the emotion data to improve the accuracy of fraud detection.

[0384] This invention is a system that detects and prevents fraudulent listings based on listing data and user emotion data. This system is mainly composed of a server and has the function of detecting fraudulent listings in real time by using data from devices such as browsers and smart glasses.

[0385] Hardware and software used

[0386] Hardware

[0387] 1. Server: The central unit that performs analysis and data integration.

[0388] 2. Smart glasses: A wearable device that collects the user's emotional data (facial expressions and behavior).

[0389] 3. Storage device: A device for temporarily storing collected data.

[0390] software

[0391] 1. Data collection module: A program that periodically retrieves listing data from the auction site database.

[0392] 2. Analysis module: A machine learning model (e.g., TensorFlow) that uses the acquired data to detect fraudulent patterns.

[0393] 3. Sentiment analysis engine: A program that analyzes user emotional data (e.g., Affectiva SDK).

[0394] 4. Integration module: A program that integrates listing data and sentiment data to improve the accuracy of fraud detection.

[0395] 5. Warning and Notification Module: A program that notifies sellers and administrators when fraud is detected.

[0396] Overall system flow

[0397] 1. Data Collection

[0398] The server uses a data collection module to periodically retrieve listing data from the auction site database, as well as user emotion data obtained from the smart glasses.

[0399] 2. Data Analysis

[0400] The collected listing data is analyzed by the analysis module, which uses machine learning models to detect fraudulent patterns. At the same time, the sentiment analysis engine analyzes users' emotional data and quantifies or categorizes their emotional states.

[0401] 3. Data Integration and Fraud Detection

[0402] The integration module integrates the analysis results of the listing data with the emotion data. Using this integrated data, the server makes a decision to further increase the likelihood of fraud.

[0403] 4. Warning and Blocking

[0404] If fraud is detected, the warning and notification module will be activated and the fraudulent listing will be blocked immediately, a warning message will be sent to the seller, and a detailed report will be sent to the administrator.

[0405] Examples of concrete examples and prompts

[0406] Specific examples

[0407] Consider a case where a user is wearing smart glasses and working in a warehouse. The camera in the smart glasses monitors the user's behavior and facial expressions in real time and sends the data to a server. If the user is detected repeatedly picking up items from a specific shelf and the emotional data from that event indicates a high level of "excitement," the analysis module and emotion analysis engine will work together to assess the possibility of fraud.

[0408] Prompt Sentence Examples

[0409] By inputting the following text prompts into the generative AI model, sentiment analysis and fraud detection can be performed more accurately.

[0410] We are currently analyzing the behavior of employees wearing smart glasses. Based on the following data, we would like to infer the employee's emotional state and determine whether they are suspected of misconduct.

[0411] data:

[0412] Facial Expression Capture: Image Data

[0413] Audio Tone: Audio Data

[0414] Behavioral pattern: The behavior of repeatedly picking up products from the shelf

[0415] Rate the emotional state as "anger," "excitement," "fear," etc., and estimate the likelihood of cheating.

[0416] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0417] Step 1:

[0418] Data collection

[0419] The server uses a data collection module to connect to the auction site's database and periodically retrieve new listing data. Specifically, data such as product titles, descriptions, image links, and seller IDs are collected. At the same time, the server obtains the user's emotional data (facial expressions and voice tone) in real time from the smart glasses. The input is the auction site data and the smart glasses' sensor data, and the output is the listing data and emotional data stored in the server's storage.

[0420] Step 2:

[0421] Data storage

[0422] The server stores the collected listing data and emotion data in a storage device. Specifically, the listing data obtained from the auction site is saved in CSV format or database format, and the emotion data is recorded over time. This ensures that all data used in the subsequent analysis process is retained. The input is data from the data collection module, and the output is the data saved in the storage device.

[0423] Step 3:

[0424] Data analysis

[0425] The server analyzes the stored listing data and emotion data. The analysis module uses a machine learning model (e.g., TensorFlow) to analyze the listing data and detect fraudulent listing patterns. At the same time, the emotion analysis engine (e.g., Affectiva SDK) analyzes the user's emotion data and quantifies emotional states such as "anger," "excitement," and "fear." The input is the stored listing data and emotion data, and the output is the analysis results for fraud detection.

[0426] Step 4:

[0427] Data Integration

[0428] The server integrates the results of the analysis of the listing data and the analysis of the emotion data. The integration module combines the output of the machine learning model with the emotion data to make a comprehensive judgment to increase the likelihood of fraud. As a specific example, if a listing of a counterfeit branded product is detected and the seller is abnormally excited, the integration result will further increase the likelihood of fraud. The input is the analysis result and emotion data, and the output is the integrated fraud detection result.

[0429] Step 5:

[0430] Warn and block

[0431] If fraud is detected, the server notifies the seller and administrator via the warning and notification module. Specifically, listings determined to be fraudulent are immediately not published on the auction site, and a warning message is sent to the seller. A detailed report is sent to the administrator, including the reason for the fraud and specific listing data. The input is the consolidated fraud detection result, and the output is the warning message sent and the report to the administrator.

[0432] 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.

[0433] 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.

[0434] 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.

[0435] [Second embodiment]

[0436] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0437] 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.

[0438] 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).

[0439] 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.

[0440] 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.

[0441] 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).

[0442] 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.

[0443] 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.

[0444] 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.

[0445] 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.

[0446] 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.

[0447] 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."

[0448] This invention is a system for detecting and preventing fraudulent listings in internet auctions, and implements a series of processes for collecting and analyzing listing data and detecting fraud. This system is mainly composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, and a warning / blocking module.

[0449] Data Collection Module

[0450] The server periodically connects to the Yahoo! Auctions database to retrieve new listing data, including product titles, descriptions, image links, seller IDs, etc. The retrieved data is temporarily stored in storage.

[0451] AI Learning Module

[0452] The server extracts listing data that is deemed fraudulent from the data stored in the storage and trains it into a machine learning model (e.g., the Gemini AI model). This training is based on a specific algorithm and is used to identify fraudulent listing patterns. Once training is complete, the model will be able to predict the likelihood of fraud for new listing data with high accuracy.

[0453] Fraud detection module

[0454] The server continuously monitors and captures new listings in real time. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. If the model's output exceeds a certain threshold, the listing is deemed fraudulent and flagged.

[0455] Warning and Blocking Module

[0456] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and the seller receives a warning message, such as "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, and the reason for the fraud.

[0457] Natural language explanation of the process

[0458] 1. Data Collection

[0459] Server: The server periodically collects new listing data from the auction database and stores it in storage.

[0460] Users: When users list items, their information is collected automatically.

[0461] 2. AI Learning

[0462] Server: The server inputs the collected data into a machine learning model to train the model on fraudulent patterns. This process identifies counterfeit branded products and bulk listing data.

[0463] 3. Real-time analysis

[0464] Server: When a new listing is made, the server immediately feeds the data into a machine learning model to analyze it for fraud.

[0465] Device: The device checks the status of the listing based on the analysis results sent from the server.

[0466] 4. Warning / Block

[0467] Server: If a listing is deemed fraudulent, the server will automatically block the listing and send a warning to the seller.

[0468] Users: Sellers receive a warning message and are informed that their problematic listings have been blocked.

[0469] Administrator: The administrator will receive a detailed report from the server and take further action if necessary.

[0470] Specific examples

[0471] Case 1: Detecting counterfeit branded products

[0472] User: Sells fake designer handbags.

[0473] Server: Collects new listings and stores them in storage.

[0474] Server: Trains the machine learning model on the collected data.

[0475] Server: The machine learning model analyzes data on newly listed handbags.

[0476] Result: If the model determines that the product is a counterfeit, the server blocks the listing and sends a warning to the seller. A detailed report is sent to the administrator.

[0477] Case 2: Large number of identical products listed

[0478] User: Listing a large number of identical items in a short period of time.

[0479] Server: Collects listing data from the auction database and stores it in storage.

[0480] Server: Trains the machine learning model on the collected data.

[0481] Server: The machine learning model analyzes data on a large number of identical product listings.

[0482] Result: If it is determined to be a bulk listing, the server will block all of the listings, send a warning message to the seller, and send a detailed report to the administrator.

[0483] This allows the system to efficiently detect and prevent fraudulent listings, providing an environment where users can use the service with peace of mind, while also enabling administrators to respond quickly.

[0484] The processing flow will be explained below.

[0485] Step 1:

[0486] The server connects to the Yahoo! Auctions database every hour to retrieve all new listing data. The retrieved data includes the product title, description, image link, and seller ID. The retrieved data is temporarily stored in storage.

[0487] Step 2:

[0488] The server extracts the product title, description, image link, and seller ID from the listing data stored in storage. It checks the data format and logs any errors.

[0489] Step 3:

[0490] The server feeds the extracted data into a machine learning model (e.g., the Gemini AI model) and the model begins the learning process, which allows the AI ​​to understand past fraudulent listing patterns and predict future fraudulent listings.

[0491] Step 4:

[0492] A user lists a new item on an auction site, and the listing is automatically added to the auction site's database.

[0493] Step 5:

[0494] When a new item is listed, the server immediately detects and collects the listing data, which is then stored in the storage.

[0495] Step 6:

[0496] The server inputs collected new listing data into a machine learning model in real time, and the AI ​​analyzes the listing for possible fraud. If the analysis result exceeds a threshold, it is flagged as fraudulent.

[0497] Step 7:

[0498] The server immediately blocks any listings that are flagged as fraudulent, changing the listing status in the database to "blocked" and prohibiting access to the product page from front-end devices.

[0499] Step 8:

[0500] The server will send a warning message to the fraudulent seller, either by email or via the notification function, informing them that "This item is not an authorized listing."

[0501] Step 9:

[0502] The server will send a detailed report to the administrator, which will include the listing ID, seller ID, reason for the fraud, and product data. The report will be displayed directly in the administrator dashboard.

[0503] This process enables the system to quickly and accurately detect fraudulent listings, allowing users to use the auction service with peace of mind.

[0504] Example 1

[0505] 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."

[0506] Fraudulent listings in internet auctions can cause a decline in credibility and harm users, so it is important to detect and deal with them quickly and accurately. However, current systems have difficulty effectively monitoring and analyzing large amounts of listing data, and fraudulent listings can be overlooked. Furthermore, it takes time to deal with fraud after detecting it, which can ultimately increase the damage. There is a need to solve these problems and provide a safer and more reliable auction environment.

[0507] 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.

[0508] In this invention, the server includes means for periodically connecting to the database to collect listing data, means for saving the collected listing data in storage, and means for inputting the saved listing data into a machine learning model to learn fraudulent patterns. This makes it possible to quickly and accurately detect and deal with fraudulent listings. In addition, new listing data can be monitored in real time and immediately input into the machine learning model to predict the possibility of fraud, enabling timely response.

[0509] "Means for periodically connecting to the database to collect listing data" refers to a technical element for accessing the auction site's database at regular intervals to obtain new listing information.

[0510] The "means for storing collected listing data in storage" refers to a data storage system or storage media for temporarily or long-term storage of acquired listing information.

[0511] The "means of inputting stored listing data into a machine learning model to learn fraudulent patterns" refers to a technical element that uses an algorithm to train a machine learning model based on accumulated listing data, thereby learning fraudulent listing patterns.

[0512] "Means for monitoring new listing data in real time" refers to the technology that instantly detects and collects newly registered listing information on the auction site.

[0513] The "means of inputting monitored listing data into a machine learning model to predict the possibility of fraud" refers to a technical element that inputs collected listing data into a trained machine learning model to predict whether a listing is fraudulent.

[0514] "Means for blocking listings determined to be fraudulent" refers to a technical element that, based on the output of a machine learning model, blocks listings that are deemed fraudulent from being published and makes them invisible on the auction site.

[0515] "Means for notifying sellers and administrators when fraud is detected" refers to a notification system that sends warnings and detailed information to sellers and site administrators when fraudulent listings are detected.

[0516] "Product Title" is a data item that indicates the name or title of the product being put up for sale.

[0517] The "description" is text data that describes detailed information about the product being sold, and explains the product's characteristics, condition, usage, etc.

[0518] "Image link" is a data item that indicates the URL or link to access the image or photo of the item being sold.

[0519] The "seller ID" is an identifier or account ID for identifying the user who put up an item for sale.

[0520] This invention is a system for detecting and preventing fraudulent listings in internet auctions, specifically for collecting and analyzing listing data from the Yahoo! Auctions database to detect and respond to fraud. This system is primarily composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, and a warning and blocking module.

[0521] Data Collection Module

[0522] The server periodically connects to the Yahoo! Auctions database to collect listing data. An API key or authentication token is used for the connection. The collected data includes the product title, description, image link, and seller ID, and this data is temporarily stored in storage. Specifically, a structured database such as MySQL or PostgreSQL is generally used.

[0523] AI Learning Module

[0524] The server extracts labeled data that is deemed to be fraudulent from the data stored in the storage and inputs it into a machine learning model to learn fraudulent patterns. The machine learning model used can be the Gemini AI model, which is widely used in industry, or another appropriate model. Python and the Scikit-learn library are often used for the learning process.

[0525] Fraud detection module

[0526] The server continuously monitors new listings in real time and detects new listing data. The detected data is immediately input into a machine learning model, which uses that data to predict the possibility of fraud. If the output exceeds a certain threshold, it is deemed fraudulent. Specifically, a Python script is used to input data into the model and obtain prediction results.

[0527] Warning and Blocking Module

[0528] Any listings that are deemed fraudulent are automatically blocked by the server and are set to not be published on the auction site. At the same time, a warning message is sent to the seller stating, "This item is not permitted for listing." A detailed report is also sent to the administrator, including the listing ID, seller ID, and reason for the fraud. Notification systems typically use email or the administrator dashboard.

[0529] Specific examples

[0530] Case 1: Detecting counterfeit branded products

[0531] User: Sells fake designer handbags.

[0532] Server: Obtains new listing data and saves it to storage.

[0533] Server: Trains machine learning models based on the stored data.

[0534] Specifically, the system extracts features such as "if the brand name is not listed on the official website, it is considered a fraudulent listing."

[0535] Server: Data on newly listed handbags is instantly fed into the model and analyzed.

[0536] Specific operations include techniques such as "detecting logo mismatches using image analysis."

[0537] Result: If the model determines that the product is a counterfeit, the server blocks the listing and sends a warning to the seller. A detailed report is sent to the administrator.

[0538] Case 2: Large number of identical products listed

[0539] User: Listing a large number of identical items in a short period of time.

[0540] Server: Collects listing data and stores it in storage.

[0541] Server: Trains the machine learning model based on the collected data.

[0542] Specifically, the system extracts features such as "detecting when the same title or image link appears frequently in a short period of time."

[0543] Server: Analyzes data on a large number of identical products listed.

[0544] Specifically, the model determines that the appearance of a certain number of identical products is fraudulent.

[0545] Result: The server will block all affected listings and send a warning message to the seller. A detailed report will be sent to the administrator.

[0546] Prompt Sentence Examples

[0547] "Use listing data collected from Yahoo! Auctions to train a machine learning model to detect fraudulent listings. Then, when new listing data is entered, analyze it for potential fraud."

[0548] This invention efficiently detects and prevents fraudulent listings, providing a safe and secure auction environment for users. It also enables administrators to take prompt and appropriate action, contributing to improving the reliability of the entire site.

[0549] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0550] Step 1:

[0551] Input: Yahoo! Auctions database connection information (API key, authentication token, etc.)

[0552] Server: The server periodically connects to the Yahoo! Auctions database to collect new listing data. Specifically, it uses the API to obtain information such as product titles, descriptions, image links, and seller IDs.

[0553] Output: Listing data (product title, description, image link, seller ID)

[0554] Step 2:

[0555] Input: Listing data collected in Step 1

[0556] Server: The server stores the acquired listing data in storage. Specifically, it stores the data in a database such as MySQL or PostgreSQL.

[0557] Output: Listing data saved in storage

[0558] Step 3:

[0559] Input: Past listing data stored in storage

[0560] Server: The server extracts labeled data that is deemed to be fraudulent listings and inputs it into a machine learning model to train it. Specifically, it uses Python and Scikit-learn to preprocess the data, extract features, and train the model.

[0561] Output: A trained machine learning model (generative AI model)

[0562] Step 4:

[0563] Input: New listing data retrieved in real time

[0564] Server: The server monitors the Yahoo! Auctions database in real time, detecting new listings and immediately feeding that data into the machine learning model.

[0565] Output: New listing data that is fed into the machine learning model

[0566] Step 5:

[0567] Input: New listing data fed into the machine learning model in Step 4

[0568] Server: The server uses machine learning models to predict the likelihood of fraud in new listings. If the prediction exceeds a certain threshold, it is deemed fraudulent. A Python script is used for the specific processing.

[0569] Output: Identification information (flag) of listing data that is determined to be fraudulent

[0570] Step 6:

[0571] Input: Listing data identified as fraudulent in Step 5

[0572] Server: The server automatically blocks listings that are deemed fraudulent, preventing them from being published on the auction site.

[0573] Output: Blocked listing data

[0574] Step 7:

[0575] Input: Identification information of the listing data that was determined to be fraudulent in Step 5

[0576] Server: Using the transmission system, send a warning message to the seller stating "This item is not an authorized listing." A detailed report is also sent to the administrator, including the listing ID, seller ID, and reason for the fraud.

[0577] Output: Warning message to seller and report to administrator

[0578] By implementing the above steps, the system can quickly and accurately detect and respond to fraudulent listings, providing a safe and reliable auction environment. Furthermore, by incorporating specific actions, the processing content of each step can be more clearly understood.

[0579] (Application example 1)

[0580] 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."

[0581] The listing of counterfeit goods, fake branded goods, and large quantities of identical goods on Internet auction and mail-order sites has become a problem. Such fraudulent listings not only undermine the trust of buyers, but also hinder sound market transactions. However, there is still no effective system for efficiently detecting and quickly dealing with these fraudulent listings. The present invention aims to provide a new method for solving these problems.

[0582] 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.

[0583] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for blocking listings in response to detected fraud, means for notifying the seller and the administrator when fraud is detected, and means for monitoring the listing data in real time via a terminal equipped with a fraudulent product listing detection application and analyzing the data using a generative AI model to evaluate the authenticity of the product, thereby enabling the listing of counterfeit products and fake brand products to be detected and prevented quickly and efficiently.

[0584] "Listing data" refers to information including the title, description, image link, seller ID, etc. of a product that a seller has listed on an online shopping site or auction site.

[0585] "Means for analyzing" refers to means for processing the collected data and conducting analysis to identify counterfeit or fake branded products.

[0586] The "means for detecting fraud" is a means for determining whether an item is a fraudulent listing based on the analyzed data.

[0587] "Methods for blocking listings" refers to removing listings that are detected as fraudulent from auction sites and mail-order sites, preventing them from being made public.

[0588] "Means of notification" refers to the means used to notify sellers and administrators when fraud is detected.

[0589] The "fraudulent product listing detection application" is a software application for detecting fraudulent product listings in real time.

[0590] A "generative AI model" is a model trained by artificial intelligence that is used to predict fraudulent product listings with high accuracy.

[0591] A "prompt" is text data to be input into the generative AI model, and includes the product title, description, image link, seller ID, etc.

[0592] This invention is a system that efficiently detects and prevents fraudulent product listings on mail order and auction sites by using a terminal and a server equipped with a fraudulent product listing detection application. Specific methods for implementing the invention are described in detail below.

[0593] Data Collection Module

[0594] The server periodically connects to the database of the mail order or auction site to retrieve new listing data. This data includes product titles, descriptions, image links, seller IDs, etc., and the retrieved data is temporarily stored in storage. Communication is via the Internet and REST API.

[0595] AI Learning Module

[0596] The server extracts listing data that is deemed fraudulent from the large amount of listing data stored in storage, and trains a generative AI model, a machine learning model. This generative AI model is used to identify fraudulent listing patterns from large amounts of data. High-performance servers and GPUs are used to train the AI ​​model.

[0597] Fraud detection module

[0598] The server continuously monitors new listings in real time and acquires data as they occur. Each time new listing data is entered, the server inputs this data into a generative AI model and analyzes the possibility of listing fraud. If the model output exceeds a certain threshold, it is deemed to be fraudulent. An example of a prompt used at this time is as follows:

[0599] Example prompt sentence:

[0600] Product Title: "Luxury brand bag"

[0601] Product Description: "Brand new, unused luxury brand bag."

[0602] Image link: "https: / / example.com / image.jpg"

[0603] Seller ID: "seller1234"

[0604] Warning and Blocking Module

[0605] The server automatically blocks listings that are deemed fraudulent. This blocking prevents the product from being displayed on online shopping or auction sites. If fraud is detected, a warning message is automatically sent to the seller. At the same time, a detailed report is sent to the administrator. This report includes the listing ID, seller ID, and the reason why the fraud was detected.

[0606] Overall system description

[0607] This system helps maintain the integrity of online shopping and auction sites. Users can purchase products with peace of mind, and administrators can quickly address fraudulent listings. The specific process involves analyzing collected listing data, predicting fraud using machine learning models, and blocking listings in real time. All of these processes are automated, increasing the efficiency and reliability of the entire system.

[0608] By utilizing the technical features of the present invention and using the above-described method to effectively prevent fraudulent product listings on mail-order sites and auction sites, it is possible to realize a highly reliable online marketplace.

[0609] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0610] Step 1:

[0611] The server periodically connects to the online shopping site's database to collect new listing data. The input is listing information received from the online shopping site's API, and the output is raw data stored in the server's storage. This data includes product titles, descriptions, image links, and seller IDs.

[0612] Step 2:

[0613] The server inputs the stored listing data into an analysis module, which converts the data into a format suitable for the generative AI model. The input here is raw data, and the output is data converted into a format that the AI ​​model can understand.

[0614] Step 3:

[0615] The server uses a generative AI model to train the converted data. It uses the fraudulent listing data that was originally accumulated and learns fraudulent patterns from newly collected data. The input of this step is the reformatted listing data, and the output is the trained AI model.

[0616] Step 4:

[0617] The server monitors new listing data in real time and inputs the collected data into the generative AI model. The input data is in the form of a prompt, and the output is a score indicating the likelihood of fraud. The prompt can be in the following form:

[0618] Product Title: "Luxury brand bag"

[0619] Product Description: "Brand new, unused luxury brand bag."

[0620] Image link: "https: / / example.com / image.jpg"

[0621] Seller ID: "seller1234"

[0622] Step 5:

[0623] The server determines the listing to be fraudulent and flags it if the score obtained from the generated AI model exceeds a certain threshold. The input is the AI ​​model score, and the output is the flagged listing data.

[0624] Step 6:

[0625] The server automatically blocks flagged listings and removes them from the online store's database. The input is the flagged listing data, and the output is the blocked listing data.

[0626] Step 7:

[0627] The server sends a warning message to the seller of any listings that are determined to be fraudulent. The message includes the content "This item has been determined to be a fraudulent listing," and at the same time, a detailed report is sent to the administrator. The input is the flagged listing data and seller information, and the output is a warning message to the seller and a report to the administrator.

[0628] 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.

[0629] This invention is a system that detects and prevents fraudulent listings in internet auctions, and by combining it with an emotion engine that recognizes user emotions, it achieves highly accurate fraudulent listing detection. This system is mainly composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, a warning / blocking module, and an emotion engine.

[0630] Data Collection Module

[0631] The server periodically connects to the internet auction database to retrieve new listing data, including product titles, descriptions, image links, and seller IDs, and temporarily stores the retrieved data in storage.

[0632] AI Learning Module

[0633] The server extracts listing data that is deemed fraudulent from the data stored in the storage and trains it into a machine learning model (e.g., an AI model). The training process is carried out so that the AI ​​can understand past fraudulent listing patterns and predict future fraudulent listings.

[0634] Fraud detection module

[0635] The server continuously monitors and captures new listings in real time. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. If the model's output exceeds a certain threshold, it is flagged as fraudulent.

[0636] Emotion Engine

[0637] The emotion engine is designed to recognize the user's emotions. It analyzes the seller's emotional state from their behavior and input and generates emotion data. Specifically, it can detect emotional states such as "anger" and "excitement."

[0638] Emotional Data Integration

[0639] The server combines the emotion data provided by the emotion engine with the results of analyzing the listing data to perform more accurate fraud detection, making it easier to identify sellers who are likely to be engaging in fraudulent activity.

[0640] Warning and Blocking Module

[0641] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and a warning message is sent to the seller. For example, the seller may receive a notification that "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, reason for fraud, and product data.

[0642] Natural language explanation of the process

[0643] 1. Data Collection

[0644] Server: The server periodically collects new listing data from the auction database and stores it in storage.

[0645] Users: When users list items, their information is collected automatically.

[0646] 2. AI Learning

[0647] Server: The server inputs the collected data into a machine learning model to train the model on fraudulent patterns. This process identifies counterfeit branded products and bulk listing data.

[0648] 3. Real-time analysis

[0649] Server: When a new listing is made, the server immediately feeds the data into a machine learning model to analyze it for fraud.

[0650] Device: The device checks the status of the listing based on the analysis results sent from the server.

[0651] 4. Acquiring Emotion Data

[0652] Server: Analyzes the seller's input and behavior, and uses the emotion engine to obtain emotion data.

[0653] Emotion engine: Detects the seller's emotional state (e.g., anger or excitement) and sends the emotional data to the server.

[0654] 5. Emotional Data Integration

[0655] Server: Integrates the analysis results of emotion data and listing data to determine whether there is any possibility of fraud.

[0656] 6. Warning Block

[0657] Server: Blocks listings that are deemed fraudulent, sends a warning to the seller, and also sends a detailed report to the administrator.

[0658] Users: Sellers receive a warning message and are informed that their problematic listings have been blocked.

[0659] Administrator: The administrator will receive a detailed report from the server and take further action if necessary.

[0660] Specific examples

[0661] Case 1: Detecting counterfeit branded products

[0662] User: Sells fake designer handbags.

[0663] Server: Collects new listings and stores them in storage.

[0664] Server: Trains the machine learning model on the collected data.

[0665] Server: The machine learning model analyzes data on newly listed handbags.

[0666] Emotion Engine: Analyzes the seller's emotional state and detects anger and excitement.

[0667] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[0668] Server: Blocks the listing and sends a warning notice to the seller. Sends a detailed report to the administrator.

[0669] Case 2: Large number of identical products listed

[0670] User: Listing a large number of identical items in a short period of time.

[0671] Server: Collects listing data from the auction database and stores it in storage.

[0672] Server: Trains the machine learning model on the collected data.

[0673] Server: The machine learning model analyzes data on a large number of identical product listings.

[0674] Emotion Engine: Analyzes the seller's emotional state and detects abnormal excitement.

[0675] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[0676] Server: Block all listings and send a warning message to the seller. Send a detailed report to the administrator.

[0677] In this way, the system can effectively prevent illegal activities and fraudulent use, providing users with a safe and secure auction service. Furthermore, the integration of emotion data can further improve the accuracy of fraud detection.

[0678] The processing flow will be explained below.

[0679] Step 1:

[0680] The server connects to the Internet auction database every hour to retrieve all new listing data. The listing data includes the product title, description, image link, and seller ID. The retrieved data is temporarily stored in storage.

[0681] Step 2:

[0682] The server extracts the product title, description, image link, and seller ID from the listing data stored in the storage. It checks the data format and logs any errors.

[0683] Step 3:

[0684] The server then inputs the extracted data into a machine learning model (e.g., an AI model) and the model begins the learning process, which allows the AI ​​to understand past fraudulent listing patterns and predict future fraudulent listings.

[0685] Step 4:

[0686] A user lists a new item on an auction site, and the listing is automatically added to the auction site's database.

[0687] Step 5:

[0688] When a new item is listed, the server immediately detects and collects the listing data, which is then stored in the storage.

[0689] Step 6:

[0690] The server inputs collected new listing data into a machine learning model in real time, and the AI ​​analyzes the listing for possible fraud. If the analysis result exceeds a threshold, it is flagged as fraudulent.

[0691] Step 7:

[0692] The server analyzes the seller's behavior and input content and acquires emotional data using an emotion engine, which detects the seller's emotional state (e.g., anger or excitement).

[0693] Step 8:

[0694] The server combines the emotion data with the results of the machine learning model analysis to perform more accurate fraud detection. If a fraud flag is raised, the server immediately blocks the listing. It changes the listing status in the database to "blocked" and prohibits access to the product page from front-end devices.

[0695] Step 9:

[0696] The server will send a warning message to the fraudulent seller, either by email or via the notification function, informing them that "This item is not an authorized listing."

[0697] Step 10:

[0698] The server will send a detailed report to the administrator, including the listing ID, seller ID, reason for the fraud, and product data. The report will be displayed on the administrator dashboard.

[0699] Through this series of processes, the system detects and prevents fraudulent listings with high accuracy and efficiency, providing an environment where users can use the auction service with peace of mind.The integration of emotional data makes it easier to identify sellers who are likely to have fraudulent intentions.

[0700] Example 2

[0701] 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."

[0702] The increase in fraudulent listings in internet auctions has created a demand for a safe and reliable trading environment. However, conventional fraud detection systems only detect fraud based on the analysis of listing data and do not take into account the emotional state of the seller, resulting in insufficient accuracy. Furthermore, there are problems with delays in appropriate response and overlooking fraud. Therefore, there is a need for a system that can more accurately detect fraud and enable prompt response by integrating not only listing data but also seller emotional data.

[0703] 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.

[0704] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for acquiring emotional data of the seller, means for integrating the acquired emotional data with the analysis means, means for blocking the listing in response to detected fraud, and means for notifying the seller and the administrator when fraud is detected. This makes it possible to detect fraudulent listings with higher accuracy and to respond quickly and appropriately.

[0705] "Listing data" refers to information such as the product title, description, image link, and seller ID provided by a seller in an Internet auction.

[0706] The "analysis means" is a function for processing collected listing data and determining whether or not there is any fraud.

[0707] The "fraud detection means" is a function for identifying fraud based on the analyzed auction data.

[0708] "Emotional data" is information indicating the emotional state of the seller obtained from their behavior and input content.

[0709] The "emotion acquisition means" is a function for acquiring emotional data of the seller.

[0710] The "data integration means" is a function for combining the analysis results of the auction data by the analysis means with emotion data.

[0711] "Blocking measures" are functions that respond to detected fraud and remove listings from the auction site.

[0712] "Notification means" is a function for notifying the seller and administrator of the relevant information when fraud is detected.

[0713] This invention is a system for detecting and preventing fraudulent listings in internet auctions. This invention is centered around a server and consists of a data collection module, an AI learning module, a fraud detection module, a warning / blocking module, and an emotion engine.

[0714] Data Collection Module

[0715] The server periodically connects to the internet auction database to retrieve new listing data. This data includes the product title, description, image link, and seller ID, and the retrieved data is temporarily stored in storage. For example, the operation is "The server sends an API request to the auction database to retrieve new listing information."

[0716] AI Learning Module

[0717] The server extracts listing data deemed fraudulent from the data stored in storage and trains it on a machine learning model (e.g., a generative AI model). During the training process, the AI ​​learns past fraudulent listing patterns and becomes capable of predicting future fraudulent listings. A high-performance server equipped with a GPU is recommended as the hardware to be used.

[0718] Fraud detection module

[0719] The server continuously monitors new listings in real time and captures that data. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. For example, if the AI ​​model exceeds a certain threshold, it flags the listing as fraudulent.

[0720] Emotion Engine

[0721] The emotion engine is designed to recognize the user's emotions. It analyzes the seller's emotional state from their behavior and input content and generates emotional data. Specifically, it can detect emotional states such as "anger" and "excitement." For example, the emotion engine works by reading emotions based on the user's input content and generating data.

[0722] Emotional Data Integration

[0723] The server combines the emotion data provided by the emotion engine with the results of analyzing the listing data to perform more accurate fraud detection, making it easier to identify sellers who are likely to be engaging in fraudulent activity.

[0724] Warning and Blocking Module

[0725] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and the seller receives a warning message, such as "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, reason for fraud, and product data.

[0726] Specific examples

[0727] Case 1: Detecting counterfeit branded products

[0728] User: Sells fake designer handbags.

[0729] Server: Collects new listing data and stores it in storage.

[0730] Server: Trains the machine learning model on the collected data.

[0731] Server: The machine learning model analyzes data on newly listed handbags.

[0732] Emotion Engine: Analyzes the seller's emotional state and detects anger and excitement.

[0733] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[0734] Server: Blocks the listing and sends a warning notice to the seller. Sends a detailed report to the administrator.

[0735] Case 2: Large number of identical products listed

[0736] User: Listing a large number of identical items in a short period of time.

[0737] Server: Collects listing data from the auction database and stores it in storage.

[0738] Server: Trains the machine learning model on the collected data.

[0739] Server: The machine learning model analyzes data on a large number of identical product listings.

[0740] Emotion Engine: Analyzes the seller's emotional state and detects abnormal excitement.

[0741] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[0742] Server: Block all listings and send a warning message to the seller. Send a detailed report to the administrator.

[0743] This allows the system to effectively prevent illegal activities and fraudulent use, providing users with a safe and secure auction service. Furthermore, the integration of emotion data can further improve the accuracy of fraud detection.

[0744] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0745] Step 1:

[0746] Data collection

[0747] Server: The server periodically connects to the auction database and collects new listing data.

[0748] Input: New listing data (item title, description, image link, seller ID) to be retrieved from the auction database.

[0749] Output: The retrieved listing data saved to storage.

[0750] Specific operation: The server sends a request to the API endpoint and saves the obtained listing data as various fields (product title, description, image link, seller ID) in the internal storage database.

[0751] Step 2:

[0752] AI Learning

[0753] Server: Extracts fraudulent listing data from the data stored in the storage and trains it using a machine learning model.

[0754] Input: Listing data saved in storage.

[0755] Output: An updated and further trained machine learning model.

[0756] What it does: The server pre-processes the stored listing data, performs feature extraction, and feeds it into a machine learning model, which learns to recognize new fraudulent listings based on data patterns.

[0757] Step 3:

[0758] Real-time analytics

[0759] Server: Monitors and validates new listing data in real time.

[0760] Input: New listing data.

[0761] Output: Judgment result of whether or not there is suspicion of fraud. Fraud flag.

[0762] What it does: The server retrieves new listing data as it arrives, feeds it into a machine learning model for real-time analysis, and flags it as fraudulent if it exceeds a certain threshold.

[0763] Step 4:

[0764] Acquiring emotion data

[0765] Server: Uses an emotion engine to obtain emotion data based on the seller's behavior and input.

[0766] Input: Seller's behavior log and input content.

[0767] Output: Emotion data (e.g., anger, excitement, etc.).

[0768] Specific operation: The server sends the seller's behavioral data to the emotion engine, which then obtains emotion data as the analysis result. At that time, natural language processing technology is used to detect signs of emotion from the text data.

[0769] Step 5:

[0770] Emotional Data Integration

[0771] Server: The results of the analysis of the listing data are combined with the emotional data to determine the likelihood of fraud.

[0772] Input: Fraud detection data and emotion data.

[0773] Output: Integrated analysis results, invalid flag.

[0774] Specific operation: The server integrates the fraud judgment results from the machine learning model with the emotion data from the emotion engine, and performs specific scoring to further increase the likelihood of fraud.

[0775] Step 6:

[0776] Warning Block

[0777] Server: Automatically blocks listings that are deemed fraudulent and sends warning messages to sellers.

[0778] Input: Combined fraud flags and listing data.

[0779] Output: Blocked listing status, warning notice to seller, detailed report to administrator.

[0780] Specific operation: The server executes the command to block the fraudulent listing that matches the conditions, sends a warning message to the seller saying "This item is not permitted for listing," and also sends a detailed report to the administrator, including the listing ID, seller ID, reason for the fraud, etc.

[0781] (Application example 2)

[0782] 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."

[0783] Conventional auction systems have difficulty detecting fraudulent listings, allowing sellers to intentionally manipulate listing information to deceive the system. Furthermore, fraud detection does not take into account the emotional state of the seller, resulting in a lack of accuracy. Furthermore, it is difficult to detect and notify fraudulent activity in real time, requiring rapid response.

[0784] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0785] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for blocking listings in response to detected fraud, means for notifying sellers and administrators when fraud is detected, means for collecting user emotion data in real time and analyzing the emotion data, and means for integrating the analysis means and the emotion data to improve fraud detection accuracy. This improves accuracy compared to conventional fraud detection methods, enabling rapid detection of fraudulent activity and countermeasures.

[0786] "Listing data" refers to information entered by a user when listing an item on an auction site, and specifically includes the item title, description, image link, seller ID, and the like.

[0787] The "collection means" refers to a device or program that has the function of acquiring the auction data from the auction site database and storing it on a server.

[0788] "Analysis means" refers to a device or program that executes a machine learning model or algorithm to assess the likelihood of fraud using collected listing data.

[0789] The "fraud detection means" is a device or program that has the function of identifying items that are likely to be fraudulent based on data analyzed using the analysis means.

[0790] "Blocking means" refers to a device or program that has the function of preventing items that are determined to be fraudulent from being published on the auction site.

[0791] The "notification means" is a device or program that has the function of sending a message to notify the seller and administrator when fraud is detected.

[0792] "Emotion data" is data that quantifies or categorizes the user's emotional state, and specifically represents emotions such as "anger," "excitement," and "fear."

[0793] The "emotion data collection means" is a device or program that has the function of collecting emotion data from the user's facial expressions, voice, etc. in real time.

[0794] An "emotion analysis engine" is an algorithm or program that analyzes collected emotional data and classifies the user's emotional state.

[0795] The "integration means" is a device or program that has the function of combining the results of the analysis of the auction data obtained from the analysis means with the emotion data to improve the accuracy of fraud detection.

[0796] This invention is a system that detects and prevents fraudulent listings based on listing data and user emotion data. This system is mainly composed of a server and has the function of detecting fraudulent listings in real time by using data from devices such as browsers and smart glasses.

[0797] Hardware and software used

[0798] Hardware

[0799] 1. Server: The central unit that performs analysis and data integration.

[0800] 2. Smart glasses: A wearable device that collects the user's emotional data (facial expressions and behavior).

[0801] 3. Storage device: A device for temporarily storing collected data.

[0802] software

[0803] 1. Data collection module: A program that periodically retrieves listing data from the auction site database.

[0804] 2. Analysis module: A machine learning model (e.g., TensorFlow) that uses the acquired data to detect fraudulent patterns.

[0805] 3. Sentiment analysis engine: A program that analyzes user emotional data (e.g., Affectiva SDK).

[0806] 4. Integration module: A program that integrates listing data and sentiment data to improve the accuracy of fraud detection.

[0807] 5. Warning and Notification Module: A program that notifies sellers and administrators when fraud is detected.

[0808] Overall system flow

[0809] 1. Data Collection

[0810] The server uses a data collection module to periodically retrieve listing data from the auction site database, as well as user emotion data obtained from the smart glasses.

[0811] 2. Data Analysis

[0812] The collected listing data is analyzed by the analysis module, which uses machine learning models to detect fraudulent patterns. At the same time, the sentiment analysis engine analyzes users' emotional data and quantifies or categorizes their emotional states.

[0813] 3. Data Integration and Fraud Detection

[0814] The integration module integrates the analysis results of the listing data with the emotion data. Using this integrated data, the server makes a decision to further increase the likelihood of fraud.

[0815] 4. Warning and Blocking

[0816] If fraud is detected, the warning and notification module will be activated and the fraudulent listing will be blocked immediately, a warning message will be sent to the seller, and a detailed report will be sent to the administrator.

[0817] Examples of concrete examples and prompts

[0818] Specific examples

[0819] Consider a case where a user is wearing smart glasses and working in a warehouse. The camera in the smart glasses monitors the user's behavior and facial expressions in real time and sends the data to a server. If the user is detected repeatedly picking up items from a specific shelf and the emotional data from that event indicates a high level of "excitement," the analysis module and emotion analysis engine will work together to assess the possibility of fraud.

[0820] Prompt Sentence Examples

[0821] By inputting the following text prompts into the generative AI model, sentiment analysis and fraud detection can be performed more accurately.

[0822] We are currently analyzing the behavior of employees wearing smart glasses. Based on the following data, we would like to infer the employee's emotional state and determine whether they are suspected of misconduct.

[0823] data:

[0824] Facial Expression Capture: Image Data

[0825] Audio Tone: Audio Data

[0826] Behavioral pattern: The behavior of repeatedly picking up products from the shelf

[0827] Rate the emotional state as "anger," "excitement," "fear," etc., and estimate the likelihood of cheating.

[0828] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0829] Step 1:

[0830] Data collection

[0831] The server uses a data collection module to connect to the auction site's database and periodically retrieve new listing data. Specifically, data such as product titles, descriptions, image links, and seller IDs are collected. At the same time, the server obtains the user's emotional data (facial expressions and voice tone) in real time from the smart glasses. The input is the auction site data and the smart glasses' sensor data, and the output is the listing data and emotional data stored in the server's storage.

[0832] Step 2:

[0833] Data storage

[0834] The server stores the collected listing data and emotion data in a storage device. Specifically, the listing data obtained from the auction site is saved in CSV format or database format, and the emotion data is recorded over time. This ensures that all data used in the subsequent analysis process is retained. The input is data from the data collection module, and the output is the data saved in the storage device.

[0835] Step 3:

[0836] Data analysis

[0837] The server analyzes the stored listing data and emotion data. The analysis module uses a machine learning model (e.g., TensorFlow) to analyze the listing data and detect fraudulent listing patterns. At the same time, the emotion analysis engine (e.g., Affectiva SDK) analyzes the user's emotion data and quantifies emotional states such as "anger," "excitement," and "fear." The input is the stored listing data and emotion data, and the output is the analysis results for fraud detection.

[0838] Step 4:

[0839] Data Integration

[0840] The server integrates the results of the analysis of the listing data and the analysis of the emotion data. The integration module combines the output of the machine learning model with the emotion data to make a comprehensive judgment to increase the likelihood of fraud. As a specific example, if a listing of a counterfeit branded product is detected and the seller is abnormally excited, the integration result will further increase the likelihood of fraud. The input is the analysis result and emotion data, and the output is the integrated fraud detection result.

[0841] Step 5:

[0842] Warn and block

[0843] If fraud is detected, the server notifies the seller and administrator via the warning and notification module. Specifically, listings determined to be fraudulent are immediately not published on the auction site, and a warning message is sent to the seller. A detailed report is sent to the administrator, including the reason for the fraud and specific listing data. The input is the consolidated fraud detection result, and the output is the warning message sent and the report to the administrator.

[0844] 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.

[0845] 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.

[0846] 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.

[0847] [Third embodiment]

[0848] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0849] 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.

[0850] 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).

[0851] 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.

[0852] 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.

[0853] 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).

[0854] 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.

[0855] 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.

[0856] 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.

[0857] 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.

[0858] 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.

[0859] 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."

[0860] This invention is a system for detecting and preventing fraudulent listings in internet auctions, and implements a series of processes for collecting and analyzing listing data and detecting fraud. This system is mainly composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, and a warning / blocking module.

[0861] Data Collection Module

[0862] The server periodically connects to the Yahoo! Auctions database to retrieve new listing data, including product titles, descriptions, image links, seller IDs, etc. The retrieved data is temporarily stored in storage.

[0863] AI Learning Module

[0864] The server extracts listing data that is deemed fraudulent from the data stored in the storage and trains it into a machine learning model (e.g., the Gemini AI model). This training is based on a specific algorithm and is used to identify fraudulent listing patterns. Once training is complete, the model will be able to predict the likelihood of fraud for new listing data with high accuracy.

[0865] Fraud detection module

[0866] The server continuously monitors and captures new listings in real time. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. If the model's output exceeds a certain threshold, the listing is deemed fraudulent and flagged.

[0867] Warning and Blocking Module

[0868] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and the seller receives a warning message, such as "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, and the reason for the fraud.

[0869] Natural language explanation of the process

[0870] 1. Data Collection

[0871] Server: The server periodically collects new listing data from the auction database and stores it in storage.

[0872] Users: When users list items, their information is collected automatically.

[0873] 2. AI Learning

[0874] Server: The server inputs the collected data into a machine learning model to train the model on fraudulent patterns. This process identifies counterfeit branded products and bulk listing data.

[0875] 3. Real-time analysis

[0876] Server: When a new listing is made, the server immediately feeds the data into a machine learning model to analyze it for fraud.

[0877] Device: The device checks the status of the listing based on the analysis results sent from the server.

[0878] 4. Warning / Block

[0879] Server: If a listing is deemed fraudulent, the server will automatically block the listing and send a warning to the seller.

[0880] Users: Sellers receive a warning message and are informed that their problematic listings have been blocked.

[0881] Administrator: The administrator will receive a detailed report from the server and take further action if necessary.

[0882] Specific examples

[0883] Case 1: Detecting counterfeit branded products

[0884] User: Sells fake designer handbags.

[0885] Server: Collects new listings and stores them in storage.

[0886] Server: Trains the machine learning model on the collected data.

[0887] Server: The machine learning model analyzes data on newly listed handbags.

[0888] Result: If the model determines that the product is a counterfeit, the server blocks the listing and sends a warning to the seller. A detailed report is sent to the administrator.

[0889] Case 2: Large number of identical products listed

[0890] User: Listing a large number of identical items in a short period of time.

[0891] Server: Collects listing data from the auction database and stores it in storage.

[0892] Server: Trains the machine learning model on the collected data.

[0893] Server: The machine learning model analyzes data on a large number of identical product listings.

[0894] Result: If it is determined to be a bulk listing, the server will block all of the listings, send a warning message to the seller, and send a detailed report to the administrator.

[0895] This allows the system to efficiently detect and prevent fraudulent listings, providing an environment where users can use the service with peace of mind, while also enabling administrators to respond quickly.

[0896] The processing flow will be explained below.

[0897] Step 1:

[0898] The server connects to the Yahoo! Auctions database every hour to retrieve all new listing data. The retrieved data includes the product title, description, image link, and seller ID. The retrieved data is temporarily stored in storage.

[0899] Step 2:

[0900] The server extracts the product title, description, image link, and seller ID from the listing data stored in storage. It checks the data format and logs any errors.

[0901] Step 3:

[0902] The server feeds the extracted data into a machine learning model (e.g., the Gemini AI model) and the model begins the learning process, which allows the AI ​​to understand past fraudulent listing patterns and predict future fraudulent listings.

[0903] Step 4:

[0904] A user lists a new item on an auction site, and the listing is automatically added to the auction site's database.

[0905] Step 5:

[0906] When a new item is listed, the server immediately detects and collects the listing data, which is then stored in the storage.

[0907] Step 6:

[0908] The server inputs collected new listing data into a machine learning model in real time, and the AI ​​analyzes the listing for possible fraud. If the analysis result exceeds a threshold, it is flagged as fraudulent.

[0909] Step 7:

[0910] The server immediately blocks any listings that are flagged as fraudulent, changing the listing status in the database to "blocked" and prohibiting access to the product page from front-end devices.

[0911] Step 8:

[0912] The server will send a warning message to the fraudulent seller, either by email or via the notification function, informing them that "This item is not an authorized listing."

[0913] Step 9:

[0914] The server will send a detailed report to the administrator, which will include the listing ID, seller ID, reason for the fraud, and product data. The report will be displayed directly in the administrator dashboard.

[0915] This process enables the system to quickly and accurately detect fraudulent listings, allowing users to use the auction service with peace of mind.

[0916] Example 1

[0917] 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."

[0918] Fraudulent listings in internet auctions can cause a decline in credibility and harm users, so it is important to detect and deal with them quickly and accurately. However, current systems have difficulty effectively monitoring and analyzing large amounts of listing data, and fraudulent listings can be overlooked. Furthermore, it takes time to deal with fraud after detecting it, which can ultimately increase the damage. There is a need to solve these problems and provide a safer and more reliable auction environment.

[0919] 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.

[0920] In this invention, the server includes means for periodically connecting to the database to collect listing data, means for saving the collected listing data in storage, and means for inputting the saved listing data into a machine learning model to learn fraudulent patterns. This makes it possible to quickly and accurately detect and deal with fraudulent listings. In addition, new listing data can be monitored in real time and immediately input into the machine learning model to predict the possibility of fraud, enabling timely response.

[0921] "Means for periodically connecting to the database to collect listing data" refers to a technical element for accessing the auction site's database at regular intervals to obtain new listing information.

[0922] The "means for storing collected listing data in storage" refers to a data storage system or storage media for temporarily or long-term storage of acquired listing information.

[0923] The "means of inputting stored listing data into a machine learning model to learn fraudulent patterns" refers to a technical element that uses an algorithm to train a machine learning model based on accumulated listing data, thereby learning fraudulent listing patterns.

[0924] "Means for monitoring new listing data in real time" refers to the technology that instantly detects and collects newly registered listing information on the auction site.

[0925] The "means of inputting monitored listing data into a machine learning model to predict the possibility of fraud" refers to a technical element that inputs collected listing data into a trained machine learning model to predict whether a listing is fraudulent.

[0926] "Means for blocking listings determined to be fraudulent" refers to a technical element that, based on the output of a machine learning model, blocks listings that are deemed fraudulent from being published and makes them invisible on the auction site.

[0927] "Means for notifying sellers and administrators when fraud is detected" refers to a notification system that sends warnings and detailed information to sellers and site administrators when fraudulent listings are detected.

[0928] "Product Title" is a data item that indicates the name or title of the product being put up for sale.

[0929] The "description" is text data that describes detailed information about the product being sold, and explains the product's characteristics, condition, usage, etc.

[0930] "Image link" is a data item that indicates the URL or link to access the image or photo of the item being sold.

[0931] The "seller ID" is an identifier or account ID for identifying the user who put up an item for sale.

[0932] This invention is a system for detecting and preventing fraudulent listings in internet auctions, specifically for collecting and analyzing listing data from the Yahoo! Auctions database to detect and respond to fraud. This system is primarily composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, and a warning and blocking module.

[0933] Data Collection Module

[0934] The server periodically connects to the Yahoo! Auctions database to collect listing data. An API key or authentication token is used for the connection. The collected data includes the product title, description, image link, and seller ID, and this data is temporarily stored in storage. Specifically, a structured database such as MySQL or PostgreSQL is generally used.

[0935] AI Learning Module

[0936] The server extracts labeled data that is deemed to be fraudulent from the data stored in the storage and inputs it into a machine learning model to learn fraudulent patterns. The machine learning model used can be the Gemini AI model, which is widely used in industry, or another appropriate model. Python and the Scikit-learn library are often used for the learning process.

[0937] Fraud detection module

[0938] The server continuously monitors new listings in real time and detects new listing data. The detected data is immediately input into a machine learning model, which uses that data to predict the possibility of fraud. If the output exceeds a certain threshold, it is deemed fraudulent. Specifically, a Python script is used to input data into the model and obtain prediction results.

[0939] Warning and Blocking Module

[0940] Any listings that are deemed fraudulent are automatically blocked by the server and are set to not be published on the auction site. At the same time, a warning message is sent to the seller stating, "This item is not permitted for listing." A detailed report is also sent to the administrator, including the listing ID, seller ID, and reason for the fraud. Notification systems typically use email or the administrator dashboard.

[0941] Specific examples

[0942] Case 1: Detecting counterfeit branded products

[0943] User: Sells fake designer handbags.

[0944] Server: Obtains new listing data and saves it to storage.

[0945] Server: Trains machine learning models based on the stored data.

[0946] Specifically, the system extracts features such as "if the brand name is not listed on the official website, it is considered a fraudulent listing."

[0947] Server: Data on newly listed handbags is instantly fed into the model and analyzed.

[0948] Specific operations include techniques such as "detecting logo mismatches using image analysis."

[0949] Result: If the model determines that the product is a counterfeit, the server blocks the listing and sends a warning to the seller. A detailed report is sent to the administrator.

[0950] Case 2: Large number of identical products listed

[0951] User: Listing a large number of identical items in a short period of time.

[0952] Server: Collects listing data and stores it in storage.

[0953] Server: Trains the machine learning model based on the collected data.

[0954] Specifically, the system extracts features such as "detecting when the same title or image link appears frequently in a short period of time."

[0955] Server: Analyzes data on a large number of identical products listed.

[0956] Specifically, the model determines that the appearance of a certain number of identical products is fraudulent.

[0957] Result: The server will block all affected listings and send a warning message to the seller. A detailed report will be sent to the administrator.

[0958] Prompt Sentence Examples

[0959] "Use listing data collected from Yahoo! Auctions to train a machine learning model to detect fraudulent listings. Then, when new listing data is entered, analyze it for potential fraud."

[0960] This invention efficiently detects and prevents fraudulent listings, providing a safe and secure auction environment for users. It also enables administrators to take prompt and appropriate action, contributing to improving the reliability of the entire site.

[0961] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0962] Step 1:

[0963] Input: Yahoo! Auctions database connection information (API key, authentication token, etc.)

[0964] Server: The server periodically connects to the Yahoo! Auctions database to collect new listing data. Specifically, it uses the API to obtain information such as product titles, descriptions, image links, and seller IDs.

[0965] Output: Listing data (product title, description, image link, seller ID)

[0966] Step 2:

[0967] Input: Listing data collected in Step 1

[0968] Server: The server stores the acquired listing data in storage. Specifically, it stores the data in a database such as MySQL or PostgreSQL.

[0969] Output: Listing data saved in storage

[0970] Step 3:

[0971] Input: Past listing data stored in storage

[0972] Server: The server extracts labeled data that is deemed to be fraudulent listings and inputs it into a machine learning model to train it. Specifically, it uses Python and Scikit-learn to preprocess the data, extract features, and train the model.

[0973] Output: A trained machine learning model (generative AI model)

[0974] Step 4:

[0975] Input: New listing data retrieved in real time

[0976] Server: The server monitors the Yahoo! Auctions database in real time, detecting new listings and immediately feeding that data into the machine learning model.

[0977] Output: New listing data that is fed into the machine learning model

[0978] Step 5:

[0979] Input: New listing data fed into the machine learning model in Step 4

[0980] Server: The server uses machine learning models to predict the likelihood of fraud in new listings. If the prediction exceeds a certain threshold, it is deemed fraudulent. A Python script is used for the specific processing.

[0981] Output: Identification information (flag) of listing data that is determined to be fraudulent

[0982] Step 6:

[0983] Input: Listing data identified as fraudulent in Step 5

[0984] Server: The server automatically blocks listings that are deemed fraudulent, preventing them from being published on the auction site.

[0985] Output: Blocked listing data

[0986] Step 7:

[0987] Input: Identification information of the listing data that was determined to be fraudulent in Step 5

[0988] Server: Using the transmission system, send a warning message to the seller stating "This item is not an authorized listing." A detailed report is also sent to the administrator, including the listing ID, seller ID, and reason for the fraud.

[0989] Output: Warning message to seller and report to administrator

[0990] By implementing the above steps, the system can quickly and accurately detect and respond to fraudulent listings, providing a safe and reliable auction environment. Furthermore, by incorporating specific actions, the processing content of each step can be more clearly understood.

[0991] (Application example 1)

[0992] 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."

[0993] The listing of counterfeit goods, fake branded goods, and large quantities of identical goods on Internet auction and mail-order sites has become a problem. Such fraudulent listings not only undermine the trust of buyers, but also hinder sound market transactions. However, there is still no effective system for efficiently detecting and quickly dealing with these fraudulent listings. The present invention aims to provide a new method for solving these problems.

[0994] 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.

[0995] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for blocking listings in response to detected fraud, means for notifying the seller and the administrator when fraud is detected, and means for monitoring the listing data in real time via a terminal equipped with a fraudulent product listing detection application and analyzing the data using a generative AI model to evaluate the authenticity of the product, thereby enabling the listing of counterfeit products and fake brand products to be detected and prevented quickly and efficiently.

[0996] "Listing data" refers to information including the title, description, image link, seller ID, etc. of a product that a seller has listed on an online shopping site or auction site.

[0997] "Means for analyzing" refers to means for processing the collected data and conducting analysis to identify counterfeit or fake branded products.

[0998] The "means for detecting fraud" is a means for determining whether an item is a fraudulent listing based on the analyzed data.

[0999] "Methods for blocking listings" refers to removing listings that are detected as fraudulent from auction sites and mail-order sites, preventing them from being made public.

[1000] "Means of notification" refers to the means used to notify sellers and administrators when fraud is detected.

[1001] The "fraudulent product listing detection application" is a software application for detecting fraudulent product listings in real time.

[1002] A "generative AI model" is a model trained by artificial intelligence that is used to predict fraudulent product listings with high accuracy.

[1003] A "prompt" is text data to be input into the generative AI model, and includes the product title, description, image link, seller ID, etc.

[1004] This invention is a system that efficiently detects and prevents fraudulent product listings on mail order and auction sites by using a terminal and a server equipped with a fraudulent product listing detection application. Specific methods for implementing the invention are described in detail below.

[1005] Data Collection Module

[1006] The server periodically connects to the database of the mail order or auction site to retrieve new listing data. This data includes product titles, descriptions, image links, seller IDs, etc., and the retrieved data is temporarily stored in storage. Communication is via the Internet and REST API.

[1007] AI Learning Module

[1008] The server extracts listing data that is deemed fraudulent from the large amount of listing data stored in storage, and trains a generative AI model, a machine learning model. This generative AI model is used to identify fraudulent listing patterns from large amounts of data. High-performance servers and GPUs are used to train the AI ​​model.

[1009] Fraud detection module

[1010] The server continuously monitors new listings in real time and acquires data as they occur. Each time new listing data is entered, the server inputs this data into a generative AI model and analyzes the possibility of listing fraud. If the model output exceeds a certain threshold, it is deemed to be fraudulent. An example of a prompt used at this time is as follows:

[1011] Example prompt sentence:

[1012] Product Title: "Luxury brand bag"

[1013] Product Description: "Brand new, unused luxury brand bag."

[1014] Image link: "https: / / example.com / image.jpg"

[1015] Seller ID: "seller1234"

[1016] Warning and Blocking Module

[1017] The server automatically blocks listings that are deemed fraudulent. This blocking prevents the product from being displayed on online shopping or auction sites. If fraud is detected, a warning message is automatically sent to the seller. At the same time, a detailed report is sent to the administrator. This report includes the listing ID, seller ID, and the reason why the fraud was detected.

[1018] Overall system description

[1019] This system helps maintain the integrity of online shopping and auction sites. Users can purchase products with peace of mind, and administrators can quickly address fraudulent listings. The specific process involves analyzing collected listing data, predicting fraud using machine learning models, and blocking listings in real time. All of these processes are automated, increasing the efficiency and reliability of the entire system.

[1020] By utilizing the technical features of the present invention and using the above-described method to effectively prevent fraudulent product listings on mail-order sites and auction sites, it is possible to realize a highly reliable online marketplace.

[1021] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1022] Step 1:

[1023] The server periodically connects to the online shopping site's database to collect new listing data. The input is listing information received from the online shopping site's API, and the output is raw data stored in the server's storage. This data includes product titles, descriptions, image links, and seller IDs.

[1024] Step 2:

[1025] The server inputs the stored listing data into an analysis module, which converts the data into a format suitable for the generative AI model. The input here is raw data, and the output is data converted into a format that the AI ​​model can understand.

[1026] Step 3:

[1027] The server uses a generative AI model to train the converted data. It uses the fraudulent listing data that was originally accumulated and learns fraudulent patterns from newly collected data. The input of this step is the reformatted listing data, and the output is the trained AI model.

[1028] Step 4:

[1029] The server monitors new listing data in real time and inputs the collected data into the generative AI model. The input data is in the form of a prompt, and the output is a score indicating the likelihood of fraud. The prompt can be in the following form:

[1030] Product Title: "Luxury brand bag"

[1031] Product Description: "Brand new, unused luxury brand bag."

[1032] Image link: "https: / / example.com / image.jpg"

[1033] Seller ID: "seller1234"

[1034] Step 5:

[1035] The server determines the listing to be fraudulent and flags it if the score obtained from the generated AI model exceeds a certain threshold. The input is the AI ​​model score, and the output is the flagged listing data.

[1036] Step 6:

[1037] The server automatically blocks flagged listings and removes them from the online store's database. The input is the flagged listing data, and the output is the blocked listing data.

[1038] Step 7:

[1039] The server sends a warning message to the seller of any listings that are determined to be fraudulent. The message includes the content "This item has been determined to be a fraudulent listing," and at the same time, a detailed report is sent to the administrator. The input is the flagged listing data and seller information, and the output is a warning message to the seller and a report to the administrator.

[1040] 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.

[1041] This invention is a system that detects and prevents fraudulent listings in internet auctions, and by combining it with an emotion engine that recognizes user emotions, it achieves highly accurate fraudulent listing detection. This system is mainly composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, a warning / blocking module, and an emotion engine.

[1042] Data Collection Module

[1043] The server periodically connects to the internet auction database to retrieve new listing data, including product titles, descriptions, image links, and seller IDs, and temporarily stores the retrieved data in storage.

[1044] AI Learning Module

[1045] The server extracts listing data that is deemed fraudulent from the data stored in the storage and trains it into a machine learning model (e.g., an AI model). The training process is carried out so that the AI ​​can understand past fraudulent listing patterns and predict future fraudulent listings.

[1046] Fraud detection module

[1047] The server continuously monitors and captures new listings in real time. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. If the model's output exceeds a certain threshold, it is flagged as fraudulent.

[1048] Emotion Engine

[1049] The emotion engine is designed to recognize the user's emotions. It analyzes the seller's emotional state from their behavior and input and generates emotion data. Specifically, it can detect emotional states such as "anger" and "excitement."

[1050] Emotional Data Integration

[1051] The server combines the emotion data provided by the emotion engine with the results of analyzing the listing data to perform more accurate fraud detection, making it easier to identify sellers who are likely to be engaging in fraudulent activity.

[1052] Warning and Blocking Module

[1053] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and a warning message is sent to the seller. For example, the seller may receive a notification that "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, reason for fraud, and product data.

[1054] Natural language explanation of the process

[1055] 1. Data Collection

[1056] Server: The server periodically collects new listing data from the auction database and stores it in storage.

[1057] Users: When users list items, their information is collected automatically.

[1058] 2. AI Learning

[1059] Server: The server inputs the collected data into a machine learning model to train the model on fraudulent patterns. This process identifies counterfeit branded products and bulk listing data.

[1060] 3. Real-time analysis

[1061] Server: When a new listing is made, the server immediately feeds the data into a machine learning model to analyze it for fraud.

[1062] Device: The device checks the status of the listing based on the analysis results sent from the server.

[1063] 4. Acquiring Emotion Data

[1064] Server: Analyzes the seller's input and behavior, and uses the emotion engine to obtain emotion data.

[1065] Emotion engine: Detects the seller's emotional state (e.g., anger or excitement) and sends the emotional data to the server.

[1066] 5. Emotional Data Integration

[1067] Server: Integrates the analysis results of emotion data and listing data to determine whether there is any possibility of fraud.

[1068] 6. Warning Block

[1069] Server: Blocks listings that are deemed fraudulent, sends a warning to the seller, and also sends a detailed report to the administrator.

[1070] Users: Sellers receive a warning message and are informed that their problematic listings have been blocked.

[1071] Administrator: The administrator will receive a detailed report from the server and take further action if necessary.

[1072] Specific examples

[1073] Case 1: Detecting counterfeit branded products

[1074] User: Sells fake designer handbags.

[1075] Server: Collects new listings and stores them in storage.

[1076] Server: Trains the machine learning model on the collected data.

[1077] Server: The machine learning model analyzes data on newly listed handbags.

[1078] Emotion Engine: Analyzes the seller's emotional state and detects anger and excitement.

[1079] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[1080] Server: Blocks the listing and sends a warning notice to the seller. Sends a detailed report to the administrator.

[1081] Case 2: Large number of identical products listed

[1082] User: Listing a large number of identical items in a short period of time.

[1083] Server: Collects listing data from the auction database and stores it in storage.

[1084] Server: Trains the machine learning model on the collected data.

[1085] Server: The machine learning model analyzes data on a large number of identical product listings.

[1086] Emotion Engine: Analyzes the seller's emotional state and detects abnormal excitement.

[1087] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[1088] Server: Block all listings and send a warning message to the seller. Send a detailed report to the administrator.

[1089] In this way, the system can effectively prevent illegal activities and fraudulent use, providing users with a safe and secure auction service. Furthermore, the integration of emotion data can further improve the accuracy of fraud detection.

[1090] The processing flow will be explained below.

[1091] Step 1:

[1092] The server connects to the Internet auction database every hour to retrieve all new listing data. The listing data includes the product title, description, image link, and seller ID. The retrieved data is temporarily stored in storage.

[1093] Step 2:

[1094] The server extracts the product title, description, image link, and seller ID from the listing data stored in the storage. It checks the data format and logs any errors.

[1095] Step 3:

[1096] The server then inputs the extracted data into a machine learning model (e.g., an AI model) and the model begins the learning process, which allows the AI ​​to understand past fraudulent listing patterns and predict future fraudulent listings.

[1097] Step 4:

[1098] A user lists a new item on an auction site, and the listing is automatically added to the auction site's database.

[1099] Step 5:

[1100] When a new item is listed, the server immediately detects and collects the listing data, which is then stored in the storage.

[1101] Step 6:

[1102] The server inputs collected new listing data into a machine learning model in real time, and the AI ​​analyzes the listing for possible fraud. If the analysis result exceeds a threshold, it is flagged as fraudulent.

[1103] Step 7:

[1104] The server analyzes the seller's behavior and input content and acquires emotional data using an emotion engine, which detects the seller's emotional state (e.g., anger or excitement).

[1105] Step 8:

[1106] The server combines the emotion data with the results of the machine learning model analysis to perform more accurate fraud detection. If a fraud flag is raised, the server immediately blocks the listing. It changes the listing status in the database to "blocked" and prohibits access to the product page from front-end devices.

[1107] Step 9:

[1108] The server will send a warning message to the fraudulent seller, either by email or via the notification function, informing them that "This item is not an authorized listing."

[1109] Step 10:

[1110] The server will send a detailed report to the administrator, including the listing ID, seller ID, reason for the fraud, and product data. The report will be displayed on the administrator dashboard.

[1111] Through this series of processes, the system detects and prevents fraudulent listings with high accuracy and efficiency, providing an environment where users can use the auction service with peace of mind.The integration of emotional data makes it easier to identify sellers who are likely to have fraudulent intentions.

[1112] Example 2

[1113] 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."

[1114] The increase in fraudulent listings in internet auctions has created a demand for a safe and reliable trading environment. However, conventional fraud detection systems only detect fraud based on the analysis of listing data and do not take into account the emotional state of the seller, resulting in insufficient accuracy. Furthermore, there are problems with delays in appropriate response and overlooking fraud. Therefore, there is a need for a system that can more accurately detect fraud and enable prompt response by integrating not only listing data but also seller emotional data.

[1115] 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.

[1116] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for acquiring emotional data of the seller, means for integrating the acquired emotional data with the analysis means, means for blocking the listing in response to detected fraud, and means for notifying the seller and the administrator when fraud is detected. This makes it possible to detect fraudulent listings with higher accuracy and to respond quickly and appropriately.

[1117] "Listing data" refers to information such as the product title, description, image link, and seller ID provided by a seller in an Internet auction.

[1118] The "analysis means" is a function for processing collected listing data and determining whether or not there is any fraud.

[1119] The "fraud detection means" is a function for identifying fraud based on the analyzed auction data.

[1120] "Emotional data" is information indicating the emotional state of the seller obtained from their behavior and input content.

[1121] The "emotion acquisition means" is a function for acquiring emotional data of the seller.

[1122] The "data integration means" is a function for combining the analysis results of the auction data by the analysis means with emotion data.

[1123] "Blocking measures" are functions that respond to detected fraud and remove listings from the auction site.

[1124] "Notification means" is a function for notifying the seller and administrator of the relevant information when fraud is detected.

[1125] This invention is a system for detecting and preventing fraudulent listings in internet auctions. This invention is centered around a server and consists of a data collection module, an AI learning module, a fraud detection module, a warning / blocking module, and an emotion engine.

[1126] Data Collection Module

[1127] The server periodically connects to the internet auction database to retrieve new listing data. This data includes the product title, description, image link, and seller ID, and the retrieved data is temporarily stored in storage. For example, the operation is "The server sends an API request to the auction database to retrieve new listing information."

[1128] AI Learning Module

[1129] The server extracts listing data deemed fraudulent from the data stored in storage and trains it on a machine learning model (e.g., a generative AI model). During the training process, the AI ​​learns past fraudulent listing patterns and becomes capable of predicting future fraudulent listings. A high-performance server equipped with a GPU is recommended as the hardware to be used.

[1130] Fraud detection module

[1131] The server continuously monitors new listings in real time and captures that data. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. For example, if the AI ​​model exceeds a certain threshold, it flags the listing as fraudulent.

[1132] Emotion Engine

[1133] The emotion engine is designed to recognize the user's emotions. It analyzes the seller's emotional state from their behavior and input content and generates emotional data. Specifically, it can detect emotional states such as "anger" and "excitement." For example, the emotion engine works by reading emotions based on the user's input content and generating data.

[1134] Emotional Data Integration

[1135] The server combines the emotion data provided by the emotion engine with the results of analyzing the listing data to perform more accurate fraud detection, making it easier to identify sellers who are likely to be engaging in fraudulent activity.

[1136] Warning and Blocking Module

[1137] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and the seller receives a warning message, such as "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, reason for fraud, and product data.

[1138] Specific examples

[1139] Case 1: Detecting counterfeit branded products

[1140] User: Sells fake designer handbags.

[1141] Server: Collects new listing data and stores it in storage.

[1142] Server: Trains the machine learning model on the collected data.

[1143] Server: The machine learning model analyzes data on newly listed handbags.

[1144] Emotion Engine: Analyzes the seller's emotional state and detects anger and excitement.

[1145] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[1146] Server: Blocks the listing and sends a warning notice to the seller. Sends a detailed report to the administrator.

[1147] Case 2: Large number of identical products listed

[1148] User: Listing a large number of identical items in a short period of time.

[1149] Server: Collects listing data from the auction database and stores it in storage.

[1150] Server: Trains the machine learning model on the collected data.

[1151] Server: The machine learning model analyzes data on a large number of identical product listings.

[1152] Emotion Engine: Analyzes the seller's emotional state and detects abnormal excitement.

[1153] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[1154] Server: Block all listings and send a warning message to the seller. Send a detailed report to the administrator.

[1155] This allows the system to effectively prevent illegal activities and fraudulent use, providing users with a safe and secure auction service. Furthermore, the integration of emotion data can further improve the accuracy of fraud detection.

[1156] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1157] Step 1:

[1158] Data collection

[1159] Server: The server periodically connects to the auction database and collects new listing data.

[1160] Input: New listing data (item title, description, image link, seller ID) to be retrieved from the auction database.

[1161] Output: The retrieved listing data saved to storage.

[1162] Specific operation: The server sends a request to the API endpoint and saves the obtained listing data as various fields (product title, description, image link, seller ID) in the internal storage database.

[1163] Step 2:

[1164] AI Learning

[1165] Server: Extracts fraudulent listing data from the data stored in the storage and trains it using a machine learning model.

[1166] Input: Listing data saved in storage.

[1167] Output: An updated and further trained machine learning model.

[1168] What it does: The server pre-processes the stored listing data, performs feature extraction, and feeds it into a machine learning model, which learns to recognize new fraudulent listings based on data patterns.

[1169] Step 3:

[1170] Real-time analytics

[1171] Server: Monitors and validates new listing data in real time.

[1172] Input: New listing data.

[1173] Output: Judgment result of whether or not there is suspicion of fraud. Fraud flag.

[1174] What it does: The server retrieves new listing data as it arrives, feeds it into a machine learning model for real-time analysis, and flags it as fraudulent if it exceeds a certain threshold.

[1175] Step 4:

[1176] Acquiring emotion data

[1177] Server: Uses an emotion engine to obtain emotion data based on the seller's behavior and input.

[1178] Input: Seller's behavior log and input content.

[1179] Output: Emotion data (e.g., anger, excitement, etc.).

[1180] Specific operation: The server sends the seller's behavioral data to the emotion engine, which then obtains emotion data as the analysis result. At that time, natural language processing technology is used to detect signs of emotion from the text data.

[1181] Step 5:

[1182] Emotional Data Integration

[1183] Server: The results of the analysis of the listing data are combined with the emotional data to determine the likelihood of fraud.

[1184] Input: Fraud detection data and emotion data.

[1185] Output: Integrated analysis results, invalid flag.

[1186] Specific operation: The server integrates the fraud judgment results from the machine learning model with the emotion data from the emotion engine, and performs specific scoring to further increase the likelihood of fraud.

[1187] Step 6:

[1188] Warning Block

[1189] Server: Automatically blocks listings that are deemed fraudulent and sends warning messages to sellers.

[1190] Input: Combined fraud flags and listing data.

[1191] Output: Blocked listing status, warning notice to seller, detailed report to administrator.

[1192] Specific operation: The server executes the command to block the fraudulent listing that matches the conditions, sends a warning message to the seller saying "This item is not permitted for listing," and also sends a detailed report to the administrator, including the listing ID, seller ID, reason for the fraud, etc.

[1193] (Application example 2)

[1194] 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."

[1195] Conventional auction systems have difficulty detecting fraudulent listings, allowing sellers to intentionally manipulate listing information to deceive the system. Furthermore, fraud detection does not take into account the emotional state of the seller, resulting in a lack of accuracy. Furthermore, it is difficult to detect and notify fraudulent activity in real time, requiring rapid response.

[1196] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1197] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for blocking listings in response to detected fraud, means for notifying sellers and administrators when fraud is detected, means for collecting user emotion data in real time and analyzing the emotion data, and means for integrating the analysis means and the emotion data to improve fraud detection accuracy. This improves accuracy compared to conventional fraud detection methods, enabling rapid detection of fraudulent activity and countermeasures.

[1198] "Listing data" refers to information entered by a user when listing an item on an auction site, and specifically includes the item title, description, image link, seller ID, and the like.

[1199] The "collection means" refers to a device or program that has the function of acquiring the auction data from the auction site database and storing it on a server.

[1200] "Analysis means" refers to a device or program that executes a machine learning model or algorithm to assess the likelihood of fraud using collected listing data.

[1201] The "fraud detection means" is a device or program that has the function of identifying items that are likely to be fraudulent based on data analyzed using the analysis means.

[1202] "Blocking means" refers to a device or program that has the function of preventing items that are determined to be fraudulent from being published on the auction site.

[1203] The "notification means" is a device or program that has the function of sending a message to notify the seller and administrator when fraud is detected.

[1204] "Emotion data" is data that quantifies or categorizes the user's emotional state, and specifically represents emotions such as "anger," "excitement," and "fear."

[1205] The "emotion data collection means" is a device or program that has the function of collecting emotion data from the user's facial expressions, voice, etc. in real time.

[1206] An "emotion analysis engine" is an algorithm or program that analyzes collected emotional data and classifies the user's emotional state.

[1207] The "integration means" is a device or program that has the function of combining the results of the analysis of the auction data obtained from the analysis means with the emotion data to improve the accuracy of fraud detection.

[1208] This invention is a system that detects and prevents fraudulent listings based on listing data and user emotion data. This system is mainly composed of a server and has the function of detecting fraudulent listings in real time by using data from devices such as browsers and smart glasses.

[1209] Hardware and software used

[1210] Hardware

[1211] 1. Server: The central unit that performs analysis and data integration.

[1212] 2. Smart glasses: A wearable device that collects the user's emotional data (facial expressions and behavior).

[1213] 3. Storage device: A device for temporarily storing collected data.

[1214] software

[1215] 1. Data collection module: A program that periodically retrieves listing data from the auction site database.

[1216] 2. Analysis module: A machine learning model (e.g., TensorFlow) that uses the acquired data to detect fraudulent patterns.

[1217] 3. Sentiment analysis engine: A program that analyzes user emotional data (e.g., Affectiva SDK).

[1218] 4. Integration module: A program that integrates listing data and sentiment data to improve the accuracy of fraud detection.

[1219] 5. Warning and Notification Module: A program that notifies sellers and administrators when fraud is detected.

[1220] Overall system flow

[1221] 1. Data Collection

[1222] The server uses a data collection module to periodically retrieve listing data from the auction site database, as well as user emotion data obtained from the smart glasses.

[1223] 2. Data Analysis

[1224] The collected listing data is analyzed by the analysis module, which uses machine learning models to detect fraudulent patterns. At the same time, the sentiment analysis engine analyzes users' emotional data and quantifies or categorizes their emotional states.

[1225] 3. Data Integration and Fraud Detection

[1226] The integration module integrates the analysis results of the listing data with the emotion data. Using this integrated data, the server makes a decision to further increase the likelihood of fraud.

[1227] 4. Warning and Blocking

[1228] If fraud is detected, the warning and notification module will be activated and the fraudulent listing will be blocked immediately, a warning message will be sent to the seller, and a detailed report will be sent to the administrator.

[1229] Examples of concrete examples and prompts

[1230] Specific examples

[1231] Consider a case where a user is wearing smart glasses and working in a warehouse. The camera in the smart glasses monitors the user's behavior and facial expressions in real time and sends the data to a server. If the user is detected repeatedly picking up items from a specific shelf and the emotional data from that event indicates a high level of "excitement," the analysis module and emotion analysis engine will work together to assess the possibility of fraud.

[1232] Prompt Sentence Examples

[1233] By inputting the following text prompts into the generative AI model, sentiment analysis and fraud detection can be performed more accurately.

[1234] We are currently analyzing the behavior of employees wearing smart glasses. Based on the following data, we would like to infer the employee's emotional state and determine whether they are suspected of misconduct.

[1235] data:

[1236] Facial Expression Capture: Image Data

[1237] Audio Tone: Audio Data

[1238] Behavioral pattern: The behavior of repeatedly picking up products from the shelf

[1239] Rate the emotional state as "anger," "excitement," "fear," etc., and estimate the likelihood of cheating.

[1240] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1241] Step 1:

[1242] Data collection

[1243] The server uses a data collection module to connect to the auction site's database and periodically retrieve new listing data. Specifically, data such as product titles, descriptions, image links, and seller IDs are collected. At the same time, the server obtains the user's emotional data (facial expressions and voice tone) in real time from the smart glasses. The input is the auction site data and the smart glasses' sensor data, and the output is the listing data and emotional data stored in the server's storage.

[1244] Step 2:

[1245] Data storage

[1246] The server stores the collected listing data and emotion data in a storage device. Specifically, the listing data obtained from the auction site is saved in CSV format or database format, and the emotion data is recorded over time. This ensures that all data used in the subsequent analysis process is retained. The input is data from the data collection module, and the output is the data saved in the storage device.

[1247] Step 3:

[1248] Data analysis

[1249] The server analyzes the stored listing data and emotion data. The analysis module uses a machine learning model (e.g., TensorFlow) to analyze the listing data and detect fraudulent listing patterns. At the same time, the emotion analysis engine (e.g., Affectiva SDK) analyzes the user's emotion data and quantifies emotional states such as "anger," "excitement," and "fear." The input is the stored listing data and emotion data, and the output is the analysis results for fraud detection.

[1250] Step 4:

[1251] Data Integration

[1252] The server integrates the results of the analysis of the listing data and the analysis of the emotion data. The integration module combines the output of the machine learning model with the emotion data to make a comprehensive judgment to increase the likelihood of fraud. As a specific example, if a listing of a counterfeit branded product is detected and the seller is abnormally excited, the integration result will further increase the likelihood of fraud. The input is the analysis result and emotion data, and the output is the integrated fraud detection result.

[1253] Step 5:

[1254] Warn and block

[1255] If fraud is detected, the server notifies the seller and administrator via the warning and notification module. Specifically, listings determined to be fraudulent are immediately not published on the auction site, and a warning message is sent to the seller. A detailed report is sent to the administrator, including the reason for the fraud and specific listing data. The input is the consolidated fraud detection result, and the output is the warning message sent and the report to the administrator.

[1256] 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.

[1257] 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.

[1258] 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.

[1259] [Fourth embodiment]

[1260] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1261] 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.

[1262] 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).

[1263] 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.

[1264] 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.

[1265] 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).

[1266] 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.

[1267] 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.

[1268] 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.

[1269] 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.

[1270] 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.

[1271] 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.

[1272] 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."

[1273] This invention is a system for detecting and preventing fraudulent listings in internet auctions, and implements a series of processes for collecting and analyzing listing data and detecting fraud. This system is mainly composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, and a warning / blocking module.

[1274] Data Collection Module

[1275] The server periodically connects to the Yahoo! Auctions database to retrieve new listing data, including product titles, descriptions, image links, seller IDs, etc. The retrieved data is temporarily stored in storage.

[1276] AI Learning Module

[1277] The server extracts listing data that is deemed fraudulent from the data stored in the storage and trains it into a machine learning model (e.g., the Gemini AI model). This training is based on a specific algorithm and is used to identify fraudulent listing patterns. Once training is complete, the model will be able to predict the likelihood of fraud for new listing data with high accuracy.

[1278] Fraud detection module

[1279] The server continuously monitors and captures new listings in real time. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. If the model's output exceeds a certain threshold, the listing is deemed fraudulent and flagged.

[1280] Warning and Blocking Module

[1281] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and the seller receives a warning message, such as "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, and the reason for the fraud.

[1282] Natural language explanation of the process

[1283] 1. Data Collection

[1284] Server: The server periodically collects new listing data from the auction database and stores it in storage.

[1285] Users: When users list items, their information is collected automatically.

[1286] 2. AI Learning

[1287] Server: The server inputs the collected data into a machine learning model to train the model on fraudulent patterns. This process identifies counterfeit branded products and bulk listing data.

[1288] 3. Real-time analysis

[1289] Server: When a new listing is made, the server immediately feeds the data into a machine learning model to analyze it for fraud.

[1290] Device: The device checks the status of the listing based on the analysis results sent from the server.

[1291] 4. Warning / Block

[1292] Server: If a listing is deemed fraudulent, the server will automatically block the listing and send a warning to the seller.

[1293] Users: Sellers receive a warning message and are informed that their problematic listings have been blocked.

[1294] Administrator: The administrator will receive a detailed report from the server and take further action if necessary.

[1295] Specific examples

[1296] Case 1: Detecting counterfeit branded products

[1297] User: Sells fake designer handbags.

[1298] Server: Collects new listings and stores them in storage.

[1299] Server: Trains the machine learning model on the collected data.

[1300] Server: The machine learning model analyzes data on newly listed handbags.

[1301] Result: If the model determines that the product is a counterfeit, the server blocks the listing and sends a warning to the seller. A detailed report is sent to the administrator.

[1302] Case 2: Large number of identical products listed

[1303] User: Listing a large number of identical items in a short period of time.

[1304] Server: Collects listing data from the auction database and stores it in storage.

[1305] Server: Trains the machine learning model on the collected data.

[1306] Server: The machine learning model analyzes data on a large number of identical product listings.

[1307] Result: If it is determined to be a bulk listing, the server will block all of the listings, send a warning message to the seller, and send a detailed report to the administrator.

[1308] This allows the system to efficiently detect and prevent fraudulent listings, providing an environment where users can use the service with peace of mind, while also enabling administrators to respond quickly.

[1309] The processing flow will be explained below.

[1310] Step 1:

[1311] The server connects to the Yahoo! Auctions database every hour to retrieve all new listing data. The retrieved data includes the product title, description, image link, and seller ID. The retrieved data is temporarily stored in storage.

[1312] Step 2:

[1313] The server extracts the product title, description, image link, and seller ID from the listing data stored in storage. It checks the data format and logs any errors.

[1314] Step 3:

[1315] The server feeds the extracted data into a machine learning model (e.g., the Gemini AI model) and the model begins the learning process, which allows the AI ​​to understand past fraudulent listing patterns and predict future fraudulent listings.

[1316] Step 4:

[1317] A user lists a new item on an auction site, and the listing is automatically added to the auction site's database.

[1318] Step 5:

[1319] When a new item is listed, the server immediately detects and collects the listing data, which is then stored in the storage.

[1320] Step 6:

[1321] The server inputs collected new listing data into a machine learning model in real time, and the AI ​​analyzes the listing for possible fraud. If the analysis result exceeds a threshold, it is flagged as fraudulent.

[1322] Step 7:

[1323] The server immediately blocks any listings that are flagged as fraudulent, changing the listing status in the database to "blocked" and prohibiting access to the product page from front-end devices.

[1324] Step 8:

[1325] The server will send a warning message to the fraudulent seller, either by email or via the notification function, informing them that "This item is not an authorized listing."

[1326] Step 9:

[1327] The server will send a detailed report to the administrator, which will include the listing ID, seller ID, reason for the fraud, and product data. The report will be displayed directly in the administrator dashboard.

[1328] This process enables the system to quickly and accurately detect fraudulent listings, allowing users to use the auction service with peace of mind.

[1329] Example 1

[1330] 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."

[1331] Fraudulent listings in internet auctions can cause a decline in credibility and harm users, so it is important to detect and deal with them quickly and accurately. However, current systems have difficulty effectively monitoring and analyzing large amounts of listing data, and fraudulent listings can be overlooked. Furthermore, it takes time to deal with fraud after detecting it, which can ultimately increase the damage. There is a need to solve these problems and provide a safer and more reliable auction environment.

[1332] 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.

[1333] In this invention, the server includes means for periodically connecting to the database to collect listing data, means for saving the collected listing data in storage, and means for inputting the saved listing data into a machine learning model to learn fraudulent patterns. This makes it possible to quickly and accurately detect and deal with fraudulent listings. In addition, new listing data can be monitored in real time and immediately input into the machine learning model to predict the possibility of fraud, enabling timely response.

[1334] "Means for periodically connecting to the database to collect listing data" refers to a technical element for accessing the auction site's database at regular intervals to obtain new listing information.

[1335] The "means for storing collected listing data in storage" refers to a data storage system or storage media for temporarily or long-term storage of acquired listing information.

[1336] The "means of inputting stored listing data into a machine learning model to learn fraudulent patterns" refers to a technical element that uses an algorithm to train a machine learning model based on accumulated listing data, thereby learning fraudulent listing patterns.

[1337] "Means for monitoring new listing data in real time" refers to the technology that instantly detects and collects newly registered listing information on the auction site.

[1338] The "means of inputting monitored listing data into a machine learning model to predict the possibility of fraud" refers to a technical element that inputs collected listing data into a trained machine learning model to predict whether a listing is fraudulent.

[1339] "Means for blocking listings determined to be fraudulent" refers to a technical element that, based on the output of a machine learning model, blocks listings that are deemed fraudulent from being published and makes them invisible on the auction site.

[1340] "Means for notifying sellers and administrators when fraud is detected" refers to a notification system that sends warnings and detailed information to sellers and site administrators when fraudulent listings are detected.

[1341] "Product Title" is a data item that indicates the name or title of the product being put up for sale.

[1342] The "description" is text data that describes detailed information about the product being sold, and explains the product's characteristics, condition, usage, etc.

[1343] "Image link" is a data item that indicates the URL or link to access the image or photo of the item being sold.

[1344] The "seller ID" is an identifier or account ID for identifying the user who put up an item for sale.

[1345] This invention is a system for detecting and preventing fraudulent listings in internet auctions, specifically for collecting and analyzing listing data from the Yahoo! Auctions database to detect and respond to fraud. This system is primarily composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, and a warning and blocking module.

[1346] Data Collection Module

[1347] The server periodically connects to the Yahoo! Auctions database to collect listing data. An API key or authentication token is used for the connection. The collected data includes the product title, description, image link, and seller ID, and this data is temporarily stored in storage. Specifically, a structured database such as MySQL or PostgreSQL is generally used.

[1348] AI Learning Module

[1349] The server extracts labeled data that is deemed to be fraudulent from the data stored in the storage and inputs it into a machine learning model to learn fraudulent patterns. The machine learning model used can be the Gemini AI model, which is widely used in industry, or another appropriate model. Python and the Scikit-learn library are often used for the learning process.

[1350] Fraud detection module

[1351] The server continuously monitors new listings in real time and detects new listing data. The detected data is immediately input into a machine learning model, which uses that data to predict the possibility of fraud. If the output exceeds a certain threshold, it is deemed fraudulent. Specifically, a Python script is used to input data into the model and obtain prediction results.

[1352] Warning and Blocking Module

[1353] Any listings that are deemed fraudulent are automatically blocked by the server and are set to not be published on the auction site. At the same time, a warning message is sent to the seller stating, "This item is not permitted for listing." A detailed report is also sent to the administrator, including the listing ID, seller ID, and reason for the fraud. Notification systems typically use email or the administrator dashboard.

[1354] Specific examples

[1355] Case 1: Detecting counterfeit branded products

[1356] User: Sells fake designer handbags.

[1357] Server: Obtains new listing data and saves it to storage.

[1358] Server: Trains machine learning models based on the stored data.

[1359] Specifically, the system extracts features such as "if the brand name is not listed on the official website, it is considered a fraudulent listing."

[1360] Server: Data on newly listed handbags is instantly fed into the model and analyzed.

[1361] Specific operations include techniques such as "detecting logo mismatches using image analysis."

[1362] Result: If the model determines that the product is a counterfeit, the server blocks the listing and sends a warning to the seller. A detailed report is sent to the administrator.

[1363] Case 2: Large number of identical products listed

[1364] User: Listing a large number of identical items in a short period of time.

[1365] Server: Collects listing data and stores it in storage.

[1366] Server: Trains the machine learning model based on the collected data.

[1367] Specifically, the system extracts features such as "detecting when the same title or image link appears frequently in a short period of time."

[1368] Server: Analyzes data on a large number of identical products listed.

[1369] Specifically, the model determines that the appearance of a certain number of identical products is fraudulent.

[1370] Result: The server will block all affected listings and send a warning message to the seller. A detailed report will be sent to the administrator.

[1371] Prompt Sentence Examples

[1372] "Use listing data collected from Yahoo! Auctions to train a machine learning model to detect fraudulent listings. Then, when new listing data is entered, analyze it for potential fraud."

[1373] This invention efficiently detects and prevents fraudulent listings, providing a safe and secure auction environment for users. It also enables administrators to take prompt and appropriate action, contributing to improving the reliability of the entire site.

[1374] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1375] Step 1:

[1376] Input: Yahoo! Auctions database connection information (API key, authentication token, etc.)

[1377] Server: The server periodically connects to the Yahoo! Auctions database to collect new listing data. Specifically, it uses the API to obtain information such as product titles, descriptions, image links, and seller IDs.

[1378] Output: Listing data (product title, description, image link, seller ID)

[1379] Step 2:

[1380] Input: Listing data collected in Step 1

[1381] Server: The server stores the acquired listing data in storage. Specifically, it stores the data in a database such as MySQL or PostgreSQL.

[1382] Output: Listing data saved in storage

[1383] Step 3:

[1384] Input: Past listing data stored in storage

[1385] Server: The server extracts labeled data that is deemed to be fraudulent listings and inputs it into a machine learning model to train it. Specifically, it uses Python and Scikit-learn to preprocess the data, extract features, and train the model.

[1386] Output: A trained machine learning model (generative AI model)

[1387] Step 4:

[1388] Input: New listing data retrieved in real time

[1389] Server: The server monitors the Yahoo! Auctions database in real time, detecting new listings and immediately feeding that data into the machine learning model.

[1390] Output: New listing data that is fed into the machine learning model

[1391] Step 5:

[1392] Input: New listing data fed into the machine learning model in Step 4

[1393] Server: The server uses machine learning models to predict the likelihood of fraud in new listings. If the prediction exceeds a certain threshold, it is deemed fraudulent. A Python script is used for the specific processing.

[1394] Output: Identification information (flag) of listing data that is determined to be fraudulent

[1395] Step 6:

[1396] Input: Listing data identified as fraudulent in Step 5

[1397] Server: The server automatically blocks listings that are deemed fraudulent, preventing them from being published on the auction site.

[1398] Output: Blocked listing data

[1399] Step 7:

[1400] Input: Identification information of the listing data that was determined to be fraudulent in Step 5

[1401] Server: Using the transmission system, send a warning message to the seller stating "This item is not an authorized listing." A detailed report is also sent to the administrator, including the listing ID, seller ID, and reason for the fraud.

[1402] Output: Warning message to seller and report to administrator

[1403] By implementing the above steps, the system can quickly and accurately detect and respond to fraudulent listings, providing a safe and reliable auction environment. Furthermore, by incorporating specific actions, the processing content of each step can be more clearly understood.

[1404] (Application example 1)

[1405] 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."

[1406] The listing of counterfeit goods, fake branded goods, and large quantities of identical goods on Internet auction and mail-order sites has become a problem. Such fraudulent listings not only undermine the trust of buyers, but also hinder sound market transactions. However, there is still no effective system for efficiently detecting and quickly dealing with these fraudulent listings. The present invention aims to provide a new method for solving these problems.

[1407] 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.

[1408] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for blocking listings in response to detected fraud, means for notifying the seller and the administrator when fraud is detected, and means for monitoring the listing data in real time via a terminal equipped with a fraudulent product listing detection application and analyzing the data using a generative AI model to evaluate the authenticity of the product, thereby enabling the listing of counterfeit products and fake brand products to be detected and prevented quickly and efficiently.

[1409] "Listing data" refers to information including the title, description, image link, seller ID, etc. of a product that a seller has listed on an online shopping site or auction site.

[1410] "Means for analyzing" refers to means for processing the collected data and conducting analysis to identify counterfeit or fake branded products.

[1411] The "means for detecting fraud" is a means for determining whether an item is a fraudulent listing based on the analyzed data.

[1412] "Methods for blocking listings" refers to removing listings that are detected as fraudulent from auction sites and mail-order sites, preventing them from being made public.

[1413] "Means of notification" refers to the means used to notify sellers and administrators when fraud is detected.

[1414] The "fraudulent product listing detection application" is a software application for detecting fraudulent product listings in real time.

[1415] A "generative AI model" is a model trained by artificial intelligence that is used to predict fraudulent product listings with high accuracy.

[1416] A "prompt" is text data to be input into the generative AI model, and includes the product title, description, image link, seller ID, etc.

[1417] This invention is a system that efficiently detects and prevents fraudulent product listings on mail order and auction sites by using a terminal and a server equipped with a fraudulent product listing detection application. Specific methods for implementing the invention are described in detail below.

[1418] Data Collection Module

[1419] The server periodically connects to the database of the mail order or auction site to retrieve new listing data. This data includes product titles, descriptions, image links, seller IDs, etc., and the retrieved data is temporarily stored in storage. Communication is via the Internet and REST API.

[1420] AI Learning Module

[1421] The server extracts listing data that is deemed fraudulent from the large amount of listing data stored in storage, and trains a generative AI model, a machine learning model. This generative AI model is used to identify fraudulent listing patterns from large amounts of data. High-performance servers and GPUs are used to train the AI ​​model.

[1422] Fraud detection module

[1423] The server continuously monitors new listings in real time and acquires data as they occur. Each time new listing data is entered, the server inputs this data into a generative AI model and analyzes the possibility of listing fraud. If the model output exceeds a certain threshold, it is deemed to be fraudulent. An example of a prompt used at this time is as follows:

[1424] Example prompt sentence:

[1425] Product Title: "Luxury brand bag"

[1426] Product Description: "Brand new, unused luxury brand bag."

[1427] Image link: "https: / / example.com / image.jpg"

[1428] Seller ID: "seller1234"

[1429] Warning and Blocking Module

[1430] The server automatically blocks listings that are deemed fraudulent. This blocking prevents the product from being displayed on online shopping or auction sites. If fraud is detected, a warning message is automatically sent to the seller. At the same time, a detailed report is sent to the administrator. This report includes the listing ID, seller ID, and the reason why the fraud was detected.

[1431] Overall system description

[1432] This system helps maintain the integrity of online shopping and auction sites. Users can purchase products with peace of mind, and administrators can quickly address fraudulent listings. The specific process involves analyzing collected listing data, predicting fraud using machine learning models, and blocking listings in real time. All of these processes are automated, increasing the efficiency and reliability of the entire system.

[1433] By utilizing the technical features of the present invention and using the above-described method to effectively prevent fraudulent product listings on mail-order sites and auction sites, it is possible to realize a highly reliable online marketplace.

[1434] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1435] Step 1:

[1436] The server periodically connects to the online shopping site's database to collect new listing data. The input is listing information received from the online shopping site's API, and the output is raw data stored in the server's storage. This data includes product titles, descriptions, image links, and seller IDs.

[1437] Step 2:

[1438] The server inputs the stored listing data into an analysis module, which converts the data into a format suitable for the generative AI model. The input here is raw data, and the output is data converted into a format that the AI ​​model can understand.

[1439] Step 3:

[1440] The server uses a generative AI model to train the converted data. It uses the fraudulent listing data that was originally accumulated and learns fraudulent patterns from newly collected data. The input of this step is the reformatted listing data, and the output is the trained AI model.

[1441] Step 4:

[1442] The server monitors new listing data in real time and inputs the collected data into the generative AI model. The input data is in the form of a prompt, and the output is a score indicating the likelihood of fraud. The prompt can be in the following form:

[1443] Product Title: "Luxury brand bag"

[1444] Product Description: "Brand new, unused luxury brand bag."

[1445] Image link: "https: / / example.com / image.jpg"

[1446] Seller ID: "seller1234"

[1447] Step 5:

[1448] The server determines the listing to be fraudulent and flags it if the score obtained from the generated AI model exceeds a certain threshold. The input is the AI ​​model score, and the output is the flagged listing data.

[1449] Step 6:

[1450] The server automatically blocks flagged listings and removes them from the online store's database. The input is the flagged listing data, and the output is the blocked listing data.

[1451] Step 7:

[1452] The server sends a warning message to the seller of any listings that are determined to be fraudulent. The message includes the content "This item has been determined to be a fraudulent listing," and at the same time, a detailed report is sent to the administrator. The input is the flagged listing data and seller information, and the output is a warning message to the seller and a report to the administrator.

[1453] 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.

[1454] This invention is a system that detects and prevents fraudulent listings in internet auctions, and by combining it with an emotion engine that recognizes user emotions, it achieves highly accurate fraudulent listing detection. This system is mainly composed of a server, and consists of a data collection module, an AI learning module, a fraud detection module, a warning / blocking module, and an emotion engine.

[1455] Data Collection Module

[1456] The server periodically connects to the internet auction database to retrieve new listing data, including product titles, descriptions, image links, and seller IDs, and temporarily stores the retrieved data in storage.

[1457] AI Learning Module

[1458] The server extracts listing data that is deemed fraudulent from the data stored in the storage and trains it into a machine learning model (e.g., an AI model). The training process is carried out so that the AI ​​can understand past fraudulent listing patterns and predict future fraudulent listings.

[1459] Fraud detection module

[1460] The server continuously monitors and captures new listings in real time. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. If the model's output exceeds a certain threshold, it is flagged as fraudulent.

[1461] Emotion Engine

[1462] The emotion engine is designed to recognize the user's emotions. It analyzes the seller's emotional state from their behavior and input and generates emotion data. Specifically, it can detect emotional states such as "anger" and "excitement."

[1463] Emotional Data Integration

[1464] The server combines the emotion data provided by the emotion engine with the results of analyzing the listing data to perform more accurate fraud detection, making it easier to identify sellers who are likely to be engaging in fraudulent activity.

[1465] Warning and Blocking Module

[1466] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and a warning message is sent to the seller. For example, the seller may receive a notification that "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, reason for fraud, and product data.

[1467] Natural language explanation of the process

[1468] 1. Data Collection

[1469] Server: The server periodically collects new listing data from the auction database and stores it in storage.

[1470] Users: When users list items, their information is collected automatically.

[1471] 2. AI Learning

[1472] Server: The server inputs the collected data into a machine learning model to train the model on fraudulent patterns. This process identifies counterfeit branded products and bulk listing data.

[1473] 3. Real-time analysis

[1474] Server: When a new listing is made, the server immediately feeds the data into a machine learning model to analyze it for fraud.

[1475] Device: The device checks the status of the listing based on the analysis results sent from the server.

[1476] 4. Acquiring Emotion Data

[1477] Server: Analyzes the seller's input and behavior, and uses the emotion engine to obtain emotion data.

[1478] Emotion engine: Detects the seller's emotional state (e.g., anger or excitement) and sends the emotional data to the server.

[1479] 5. Emotional Data Integration

[1480] Server: Integrates the analysis results of emotion data and listing data to determine whether there is any possibility of fraud.

[1481] 6. Warning Block

[1482] Server: Blocks listings that are deemed fraudulent, sends a warning to the seller, and also sends a detailed report to the administrator.

[1483] Users: Sellers receive a warning message and are informed that their problematic listings have been blocked.

[1484] Administrator: The administrator will receive a detailed report from the server and take further action if necessary.

[1485] Specific examples

[1486] Case 1: Detecting counterfeit branded products

[1487] User: Sells fake designer handbags.

[1488] Server: Collects new listings and stores them in storage.

[1489] Server: Trains the machine learning model on the collected data.

[1490] Server: The machine learning model analyzes data on newly listed handbags.

[1491] Emotion Engine: Analyzes the seller's emotional state and detects anger and excitement.

[1492] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[1493] Server: Blocks the listing and sends a warning notice to the seller. Sends a detailed report to the administrator.

[1494] Case 2: Large number of identical products listed

[1495] User: Listing a large number of identical items in a short period of time.

[1496] Server: Collects listing data from the auction database and stores it in storage.

[1497] Server: Trains the machine learning model on the collected data.

[1498] Server: The machine learning model analyzes data on a large number of identical product listings.

[1499] Emotion Engine: Analyzes the seller's emotional state and detects abnormal excitement.

[1500] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[1501] Server: Block all listings and send a warning message to the seller. Send a detailed report to the administrator.

[1502] In this way, the system can effectively prevent illegal activities and fraudulent use, providing users with a safe and secure auction service. Furthermore, the integration of emotion data can further improve the accuracy of fraud detection.

[1503] The processing flow will be explained below.

[1504] Step 1:

[1505] The server connects to the Internet auction database every hour to retrieve all new listing data. The listing data includes the product title, description, image link, and seller ID. The retrieved data is temporarily stored in storage.

[1506] Step 2:

[1507] The server extracts the product title, description, image link, and seller ID from the listing data stored in the storage. It checks the data format and logs any errors.

[1508] Step 3:

[1509] The server then inputs the extracted data into a machine learning model (e.g., an AI model) and the model begins the learning process, which allows the AI ​​to understand past fraudulent listing patterns and predict future fraudulent listings.

[1510] Step 4:

[1511] A user lists a new item on an auction site, and the listing is automatically added to the auction site's database.

[1512] Step 5:

[1513] When a new item is listed, the server immediately detects and collects the listing data, which is then stored in the storage.

[1514] Step 6:

[1515] The server inputs collected new listing data into a machine learning model in real time, and the AI ​​analyzes the listing for possible fraud. If the analysis result exceeds a threshold, it is flagged as fraudulent.

[1516] Step 7:

[1517] The server analyzes the seller's behavior and input content and acquires emotional data using an emotion engine, which detects the seller's emotional state (e.g., anger or excitement).

[1518] Step 8:

[1519] The server combines the emotion data with the results of the machine learning model analysis to perform more accurate fraud detection. If a fraud flag is raised, the server immediately blocks the listing. It changes the listing status in the database to "blocked" and prohibits access to the product page from front-end devices.

[1520] Step 9:

[1521] The server will send a warning message to the fraudulent seller, either by email or via the notification function, informing them that "This item is not an authorized listing."

[1522] Step 10:

[1523] The server will send a detailed report to the administrator, including the listing ID, seller ID, reason for the fraud, and product data. The report will be displayed on the administrator dashboard.

[1524] Through this series of processes, the system detects and prevents fraudulent listings with high accuracy and efficiency, providing an environment where users can use the auction service with peace of mind.The integration of emotional data makes it easier to identify sellers who are likely to have fraudulent intentions.

[1525] Example 2

[1526] 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."

[1527] The increase in fraudulent listings in internet auctions has created a demand for a safe and reliable trading environment. However, conventional fraud detection systems only detect fraud based on the analysis of listing data and do not take into account the emotional state of the seller, resulting in insufficient accuracy. Furthermore, there are problems with delays in appropriate response and overlooking fraud. Therefore, there is a need for a system that can more accurately detect fraud and enable prompt response by integrating not only listing data but also seller emotional data.

[1528] 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.

[1529] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for acquiring emotional data of the seller, means for integrating the acquired emotional data with the analysis means, means for blocking the listing in response to detected fraud, and means for notifying the seller and the administrator when fraud is detected. This makes it possible to detect fraudulent listings with higher accuracy and to respond quickly and appropriately.

[1530] "Listing data" refers to information such as the product title, description, image link, and seller ID provided by a seller in an Internet auction.

[1531] The "analysis means" is a function for processing collected listing data and determining whether or not there is any fraud.

[1532] The "fraud detection means" is a function for identifying fraud based on the analyzed auction data.

[1533] "Emotional data" is information indicating the emotional state of the seller obtained from their behavior and input content.

[1534] The "emotion acquisition means" is a function for acquiring emotional data of the seller.

[1535] The "data integration means" is a function for combining the analysis results of the auction data by the analysis means with emotion data.

[1536] "Blocking measures" are functions that respond to detected fraud and remove listings from the auction site.

[1537] "Notification means" is a function for notifying the seller and administrator of the relevant information when fraud is detected.

[1538] This invention is a system for detecting and preventing fraudulent listings in internet auctions. This invention is centered around a server and consists of a data collection module, an AI learning module, a fraud detection module, a warning / blocking module, and an emotion engine.

[1539] Data Collection Module

[1540] The server periodically connects to the internet auction database to retrieve new listing data. This data includes the product title, description, image link, and seller ID, and the retrieved data is temporarily stored in storage. For example, the operation is "The server sends an API request to the auction database to retrieve new listing information."

[1541] AI Learning Module

[1542] The server extracts listing data deemed fraudulent from the data stored in storage and trains it on a machine learning model (e.g., a generative AI model). During the training process, the AI ​​learns past fraudulent listing patterns and becomes capable of predicting future fraudulent listings. A high-performance server equipped with a GPU is recommended as the hardware to be used.

[1543] Fraud detection module

[1544] The server continuously monitors new listings in real time and captures that data. Each time new listing data is entered, the server feeds it into a machine learning model, which analyzes the listing for possible fraud. For example, if the AI ​​model exceeds a certain threshold, it flags the listing as fraudulent.

[1545] Emotion Engine

[1546] The emotion engine is designed to recognize the user's emotions. It analyzes the seller's emotional state from their behavior and input content and generates emotional data. Specifically, it can detect emotional states such as "anger" and "excitement." For example, the emotion engine works by reading emotions based on the user's input content and generating data.

[1547] Emotional Data Integration

[1548] The server combines the emotion data provided by the emotion engine with the results of analyzing the listing data to perform more accurate fraud detection, making it easier to identify sellers who are likely to be engaging in fraudulent activity.

[1549] Warning and Blocking Module

[1550] The server automatically blocks listings that are deemed fraudulent. Blocked listings are not published on the auction site, and the seller receives a warning message, such as "This item is not permitted for sale." At the same time, a detailed report is sent to the administrator, including the listing ID, seller ID, reason for fraud, and product data.

[1551] Specific examples

[1552] Case 1: Detecting counterfeit branded products

[1553] User: Sells fake designer handbags.

[1554] Server: Collects new listing data and stores it in storage.

[1555] Server: Trains the machine learning model on the collected data.

[1556] Server: The machine learning model analyzes data on newly listed handbags.

[1557] Emotion Engine: Analyzes the seller's emotional state and detects anger and excitement.

[1558] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[1559] Server: Blocks the listing and sends a warning notice to the seller. Sends a detailed report to the administrator.

[1560] Case 2: Large number of identical products listed

[1561] User: Listing a large number of identical items in a short period of time.

[1562] Server: Collects listing data from the auction database and stores it in storage.

[1563] Server: Trains the machine learning model on the collected data.

[1564] Server: The machine learning model analyzes data on a large number of identical product listings.

[1565] Emotion Engine: Analyzes the seller's emotional state and detects abnormal excitement.

[1566] Server: Integrates the results of the machine learning model and sentiment engine to flag fraud.

[1567] Server: Block all listings and send a warning message to the seller. Send a detailed report to the administrator.

[1568] This allows the system to effectively prevent illegal activities and fraudulent use, providing users with a safe and secure auction service. Furthermore, the integration of emotion data can further improve the accuracy of fraud detection.

[1569] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1570] Step 1:

[1571] Data collection

[1572] Server: The server periodically connects to the auction database and collects new listing data.

[1573] Input: New listing data (item title, description, image link, seller ID) to be retrieved from the auction database.

[1574] Output: The retrieved listing data saved to storage.

[1575] Specific operation: The server sends a request to the API endpoint and saves the obtained listing data as various fields (product title, description, image link, seller ID) in the internal storage database.

[1576] Step 2:

[1577] AI Learning

[1578] Server: Extracts fraudulent listing data from the data stored in the storage and trains it using a machine learning model.

[1579] Input: Listing data saved in storage.

[1580] Output: An updated and further trained machine learning model.

[1581] What it does: The server pre-processes the stored listing data, performs feature extraction, and feeds it into a machine learning model, which learns to recognize new fraudulent listings based on data patterns.

[1582] Step 3:

[1583] Real-time analytics

[1584] Server: Monitors and validates new listing data in real time.

[1585] Input: New listing data.

[1586] Output: Judgment result of whether or not there is suspicion of fraud. Fraud flag.

[1587] What it does: The server retrieves new listing data as it arrives, feeds it into a machine learning model for real-time analysis, and flags it as fraudulent if it exceeds a certain threshold.

[1588] Step 4:

[1589] Acquiring emotion data

[1590] Server: Uses an emotion engine to obtain emotion data based on the seller's behavior and input.

[1591] Input: Seller's behavior log and input content.

[1592] Output: Emotion data (e.g., anger, excitement, etc.).

[1593] Specific operation: The server sends the seller's behavioral data to the emotion engine, which then obtains emotion data as the analysis result. At that time, natural language processing technology is used to detect signs of emotion from the text data.

[1594] Step 5:

[1595] Emotional Data Integration

[1596] Server: The results of the analysis of the listing data are combined with the emotional data to determine the likelihood of fraud.

[1597] Input: Fraud detection data and emotion data.

[1598] Output: Integrated analysis results, invalid flag.

[1599] Specific operation: The server integrates the fraud judgment results from the machine learning model with the emotion data from the emotion engine, and performs specific scoring to further increase the likelihood of fraud.

[1600] Step 6:

[1601] Warning Block

[1602] Server: Automatically blocks listings that are deemed fraudulent and sends warning messages to sellers.

[1603] Input: Combined fraud flags and listing data.

[1604] Output: Blocked listing status, warning notice to seller, detailed report to administrator.

[1605] Specific operation: The server executes the command to block the fraudulent listing that matches the conditions, sends a warning message to the seller saying "This item is not permitted for listing," and also sends a detailed report to the administrator, including the listing ID, seller ID, reason for the fraud, etc.

[1606] (Application example 2)

[1607] 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."

[1608] Conventional auction systems have difficulty detecting fraudulent listings, allowing sellers to intentionally manipulate listing information to deceive the system. Furthermore, fraud detection does not take into account the emotional state of the seller, resulting in a lack of accuracy. Furthermore, it is difficult to detect and notify fraudulent activity in real time, requiring rapid response.

[1609] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1610] In this invention, the server includes means for collecting listing data, means for analyzing the collected listing data, means for detecting fraud based on the analyzed listing data, means for blocking listings in response to detected fraud, means for notifying sellers and administrators when fraud is detected, means for collecting user emotion data in real time and analyzing the emotion data, and means for integrating the analysis means and the emotion data to improve fraud detection accuracy. This improves accuracy compared to conventional fraud detection methods, enabling rapid detection of fraudulent activity and countermeasures.

[1611] "Listing data" refers to information entered by a user when listing an item on an auction site, and specifically includes the item title, description, image link, seller ID, and the like.

[1612] The "collection means" refers to a device or program that has the function of acquiring the auction data from the auction site database and storing it on a server.

[1613] "Analysis means" refers to a device or program that executes a machine learning model or algorithm to assess the likelihood of fraud using collected listing data.

[1614] The "fraud detection means" is a device or program that has the function of identifying items that are likely to be fraudulent based on data analyzed using the analysis means.

[1615] "Blocking means" refers to a device or program that has the function of preventing items that are determined to be fraudulent from being published on the auction site.

[1616] The "notification means" is a device or program that has the function of sending a message to notify the seller and administrator when fraud is detected.

[1617] "Emotion data" is data that quantifies or categorizes the user's emotional state, and specifically represents emotions such as "anger," "excitement," and "fear."

[1618] The "emotion data collection means" is a device or program that has the function of collecting emotion data from the user's facial expressions, voice, etc. in real time.

[1619] An "emotion analysis engine" is an algorithm or program that analyzes collected emotional data and classifies the user's emotional state.

[1620] The "integration means" is a device or program that has the function of combining the results of the analysis of the auction data obtained from the analysis means with the emotion data to improve the accuracy of fraud detection.

[1621] This invention is a system that detects and prevents fraudulent listings based on listing data and user emotion data. This system is mainly composed of a server and has the function of detecting fraudulent listings in real time by using data from devices such as browsers and smart glasses.

[1622] Hardware and software used

[1623] Hardware

[1624] 1. Server: The central unit that performs analysis and data integration.

[1625] 2. Smart glasses: A wearable device that collects the user's emotional data (facial expressions and behavior).

[1626] 3. Storage device: A device for temporarily storing collected data.

[1627] software

[1628] 1. Data collection module: A program that periodically retrieves listing data from the auction site database.

[1629] 2. Analysis module: A machine learning model (e.g., TensorFlow) that uses the acquired data to detect fraudulent patterns.

[1630] 3. Sentiment analysis engine: A program that analyzes user emotional data (e.g., Affectiva SDK).

[1631] 4. Integration module: A program that integrates listing data and sentiment data to improve the accuracy of fraud detection.

[1632] 5. Warning and Notification Module: A program that notifies sellers and administrators when fraud is detected.

[1633] Overall system flow

[1634] 1. Data Collection

[1635] The server uses a data collection module to periodically retrieve listing data from the auction site database, as well as user emotion data obtained from the smart glasses.

[1636] 2. Data Analysis

[1637] The collected listing data is analyzed by the analysis module, which uses machine learning models to detect fraudulent patterns. At the same time, the sentiment analysis engine analyzes users' emotional data and quantifies or categorizes their emotional states.

[1638] 3. Data Integration and Fraud Detection

[1639] The integration module integrates the analysis results of the listing data with the emotion data. Using this integrated data, the server makes a decision to further increase the likelihood of fraud.

[1640] 4. Warning and Blocking

[1641] If fraud is detected, the warning and notification module will be activated and the fraudulent listing will be blocked immediately, a warning message will be sent to the seller, and a detailed report will be sent to the administrator.

[1642] Examples of concrete examples and prompts

[1643] Specific examples

[1644] Consider a case where a user is wearing smart glasses and working in a warehouse. The camera in the smart glasses monitors the user's behavior and facial expressions in real time and sends the data to a server. If the user is detected repeatedly picking up items from a specific shelf and the emotional data from that event indicates a high level of "excitement," the analysis module and emotion analysis engine will work together to assess the possibility of fraud.

[1645] Prompt Sentence Examples

[1646] By inputting the following text prompts into the generative AI model, sentiment analysis and fraud detection can be performed more accurately.

[1647] We are currently analyzing the behavior of employees wearing smart glasses. Based on the following data, we would like to infer the employee's emotional state and determine whether they are suspected of misconduct.

[1648] data:

[1649] Facial Expression Capture: Image Data

[1650] Audio Tone: Audio Data

[1651] Behavioral pattern: The behavior of repeatedly picking up products from the shelf

[1652] Rate the emotional state as "anger," "excitement," "fear," etc., and estimate the likelihood of cheating.

[1653] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1654] Step 1:

[1655] Data collection

[1656] The server uses a data collection module to connect to the auction site's database and periodically retrieve new listing data. Specifically, data such as product titles, descriptions, image links, and seller IDs are collected. At the same time, the server obtains the user's emotional data (facial expressions and voice tone) in real time from the smart glasses. The input is the auction site data and the smart glasses' sensor data, and the output is the listing data and emotional data stored in the server's storage.

[1657] Step 2:

[1658] Data storage

[1659] The server stores the collected listing data and emotion data in a storage device. Specifically, the listing data obtained from the auction site is saved in CSV format or database format, and the emotion data is recorded over time. This ensures that all data used in the subsequent analysis process is retained. The input is data from the data collection module, and the output is the data saved in the storage device.

[1660] Step 3:

[1661] Data analysis

[1662] The server analyzes the stored listing data and emotion data. The analysis module uses a machine learning model (e.g., TensorFlow) to analyze the listing data and detect fraudulent listing patterns. At the same time, the emotion analysis engine (e.g., Affectiva SDK) analyzes the user's emotion data and quantifies emotional states such as "anger," "excitement," and "fear." The input is the stored listing data and emotion data, and the output is the analysis results for fraud detection.

[1663] Step 4:

[1664] Data Integration

[1665] The server integrates the results of the analysis of the listing data and the analysis of the emotion data. The integration module combines the output of the machine learning model with the emotion data to make a comprehensive judgment to increase the likelihood of fraud. As a specific example, if a listing of a counterfeit branded product is detected and the seller is abnormally excited, the integration result will further increase the likelihood of fraud. The input is the analysis result and emotion data, and the output is the integrated fraud detection result.

[1666] Step 5:

[1667] Warn and block

[1668] If fraud is detected, the server notifies the seller and administrator via the warning and notification module. Specifically, listings determined to be fraudulent are immediately not published on the auction site, and a warning message is sent to the seller. A detailed report is sent to the administrator, including the reason for the fraud and specific listing data. The input is the consolidated fraud detection result, and the output is the warning message sent and the report to the administrator.

[1669] 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.

[1670] 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.

[1671] 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.

[1672] 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.

[1673] 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.

[1674] 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.

[1675] 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).

[1676] 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.

[1677] 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."

[1678] 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.

[1679] 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).

[1680] 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.

[1681] 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.

[1682] 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.

[1683] 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.

[1684] 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.

[1685] 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.

[1686] 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.

[1687] 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.

[1688] 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.

[1689] 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.

[1690] The following is further disclosed regarding the above embodiment.

[1691] (Claim 1)

[1692] a means for collecting listing data;

[1693] a means for analyzing the collected listing data;

[1694] A means for detecting fraud based on the analyzed listing data;

[1695] A means to block listings in response to detected fraud; and

[1696] A means of notifying the seller and administrator when fraud is detected;

[1697] A system including:

[1698] (Claim 2)

[1699] The system of claim 1 , further comprising: means for extracting a product title, description, image link, and seller ID from the collected listing data.

[1700] (Claim 3)

[1701] The system of claim 1, wherein the analysis means is a means for inputting the collected listing data into a machine learning model and detecting fraud based on the output of the machine learning model.

[1702] "Example 1"

[1703] (Claim 1)

[1704] a means of periodically connecting to the database to collect listing data;

[1705] A means for storing the collected listing data in storage;

[1706] A method for inputting the saved listing data into a machine learning model to learn fraudulent patterns;

[1707] A means of monitoring new listing data in real time;

[1708] A means of inputting monitored listing data into a machine learning model to predict potential fraud; and

[1709] A means to block listings that are deemed fraudulent,

[1710] A means of notifying sellers and administrators when fraud is detected;

[1711] A system including:

[1712] (Claim 2)

[1713] 10. The system of claim 1, further comprising means for extracting product titles, descriptions, image links, and seller IDs from the collected listing data.

[1714] (Claim 3)

[1715] The system of claim 1, which is a means for inputting collected listing data into a machine learning model and detecting fraud based on the output of the machine learning model.

[1716] "Application Example 1"

[1717] (Claim 1)

[1718] a means for collecting listing data;

[1719] a means for analyzing the collected listing data;

[1720] A means for detecting fraud based on the analyzed listing data;

[1721] A means to block listings in response to detected fraud; and

[1722] A means of notifying the seller and administrator when fraud is detected;

[1723] We monitor listing data in real time through devices equipped with a fraudulent listing detection application.

[1724] A system that includes analytical means using a generative AI model to evaluate the reliability of a product.

[1725] (Claim 2)

[1726] The method further includes means for extracting a product title, a description, an image link, and a seller ID from the collected listing data.

[1727] further comprising means for sending a warning message to the seller of the detected counterfeit product;

[1728] 10. The system of claim 1.

[1729] (Claim 3)

[1730] The analysis means is a means for inputting the collected listing data into a machine learning model and detecting fraud based on the output of the machine learning model.

[1731] further comprising means for generating a prompt sentence to be input to the generative AI model;

[1732] 10. The system of claim 1.

[1733] "Example 2: Combining Emotion Engines"

[1734] (Claim 1)

[1735] a means for collecting listing data;

[1736] a means for analyzing the collected listing data;

[1737] A means for detecting fraud based on the analyzed listing data;

[1738] A means for acquiring seller emotion data;

[1739] means for integrating the acquired emotion data with said analysis means;

[1740] A means to block listings in response to detected fraud; and

[1741] A means of notifying the seller and administrator when fraud is detected;

[1742] A system including:

[1743] (Claim 2)

[1744] The system of claim 1 , further comprising: means for extracting a product title, description, image link, and seller ID from the collected listing data.

[1745] (Claim 3)

[1746] The system of claim 1, wherein the analysis means is a means for inputting the collected listing data into a machine learning model and detecting fraud based on the output of the machine learning model.

[1747] "Application example 2 when combining emotion engines"

[1748] (Claim 1)

[1749] a means for collecting listing data;

[1750] a means for analyzing the collected listing data;

[1751] A means for detecting fraud based on the analyzed listing data;

[1752] A means to block listings in response to detected fraud; and

[1753] A means of notifying the seller and administrator when fraud is detected;

[1754] a means for collecting user emotion data in real time and analyzing the emotion data;

[1755] a means for integrating the analysis means and the emotion data to improve fraud detection accuracy;

[1756] A system including:

[1757] (Claim 2)

[1758] The system of claim 1 , further comprising: means for extracting a product title, description, image link, and seller ID from the collected listing data.

[1759] (Claim 3)

[1760] 2. The system of claim 1, wherein the analysis means is a means for inputting the collected listing data into a machine learning model and detecting fraud based on the output of the machine learning model, and the emotion data collection means is a means for analyzing the user's emotional state using an emotion analysis engine and integrating the results into the fraud detection process. [Explanation of symbols]

[1761] 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 collecting listing data; a means for analyzing the collected listing data; A means for detecting fraud based on the analyzed listing data; A means to block listings in response to detected fraud; and A means of notifying the seller and administrator when fraud is detected; A system including:

2. The system of claim 1 , further comprising means for extracting a product title, description, image link, and seller ID from the collected listing data.

3. The system according to claim 1 , wherein the analysis means is a means for inputting the collected listing data into a machine learning model and detecting fraud based on the output of the machine learning model.

Images