system

Abstract

We provide the system. [Solution] A means for extracting evaluation information from the network, A means for analyzing extracted evaluation information using natural language processing to generate sentiment evaluation values, A means of detecting fraudulent evaluations using a machine learning algorithm and assigning a confidence score, A means of providing user terminals with priority-based and useful evaluation information based on reliability evaluation values, A means of summarizing the analyzed evaluation information and presenting it to the user, A means of collecting user feedback and using it to improve machine learning algorithms, A system that includes this.

Description

Technical Field

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

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor, the method including: receiving a user utterance; adding the user utterance to a prompt including an instruction sentence related to an explanation of a character of the chatbot; 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

Summary of the Invention

Problems to be Solved by the Invention

[0004] Product reviews provided on the Internet play an important role in consumers' purchasing decisions, but often lack reliability. Consumers may bear the risk of purchasing products with qualities different from their expectations. In particular, there is a possibility that purchasing decisions are made based on false information due to fake reviews. Therefore, there is a need for a system that automatically evaluates the reliability of review information so that consumers can make accurate judgments based on reliable reviews.

Means for Solving the Problems

[0005] This invention provides a means for collecting review information from a network, analyzing the sentiment of those reviews using natural language processing technology, and generating a numerical sentiment score. It also provides means for detecting fake reviews based on a machine learning model and assigning a reliability score to each review. Furthermore, the system includes means for prioritizing and providing users with highly reliable review information based on the reliability score. In addition, it aims for continuous accuracy improvement by collecting user feedback and using it to improve the machine learning model. This makes it possible to provide an environment in which consumers can make purchasing decisions with greater confidence.

[0006] "Online review information" refers to a collection of user ratings and comments about products and services that are publicly available on the internet.

[0007] "Natural language processing" is a technology that enables computers to understand and process human language appropriately, including the analysis of text data and the identification of emotions.

[0008] An "emotion score" is a numerical representation of the emotional tendencies within a review, indicating whether it is positive, negative, or neutral.

[0009] A "machine learning model" is an algorithm or system that learns from large amounts of data and performs pattern recognition and prediction, and is used to detect fake reviews.

[0010] A "fake review" is a review that contains false information and is usually created with the intention of unfairly manipulating purchasing decisions.

[0011] A "reliability score" is a numerical value that evaluates the reliability of review information, with higher reliability resulting in a higher score.

[0012] A "user terminal" is a device used by a user to receive, display, and manipulate information, and includes smartphones and computers.

[0013] "Feedback" refers to the opinions and impressions that users give to the system or reviews, and serves as data for improving the system. [Brief explanation of the drawing]

[0014] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11] This is a sequence diagram showing the processing flow of the data processing system in Example 1. [Figure 12] This is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13]It is a sequence diagram showing the processing flow of the data processing system in Embodiment 2 when the emotion engine is combined. [Figure 14] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 when the emotion engine is combined.

Modes for Carrying Out the Invention

[0015] Hereinafter, an example of an embodiment of the system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0016] First, the terms used in the following description will be explained.

[0017] In the following embodiments, the numbered processor (hereinafter simply referred to as "processor") may be one arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be one type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), and the like.

[0018] In the following embodiments, the numbered RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0019] In the following embodiments, the numbered storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0020] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0021] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0022] [First Embodiment]

[0023] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0024] As shown in Figure 1, the 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.

[0025] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0026] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0027] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and 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.

[0028] 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 perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0029] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0030] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0031] As shown in Figure 2, in the data processing device 12, a specific processing 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" related to the technology of this 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 according to the specific processing program 56 executed on the RAM 30.

[0032] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0033] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0034] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0035] The system according to the present invention consists of a server, a terminal, and a user, and aims to automatically evaluate the reliability of product reviews over a network and provide useful information to consumers.

[0036] Server Role

[0037] The server first collects review information from various sales sites via the network. The collected data is stored and managed on the server. The server processes this data using natural language processing algorithms to analyze the text of each review. Through this analysis, it generates a sentiment score for each review, identifying whether the review is positive, neutral, or negative. Furthermore, it uses machine learning models to detect fake reviews and evaluate the reliability of each review.

[0038] Terminal role

[0039] The device receives analyzed data from the server and displays it in a format suitable for the user's actions. Specifically, it prioritizes displaying highly reliable reviews based on their reliability scores, and generates and displays review summaries so that the user can quickly grasp the main points of the reviews.

[0040] User Roles

[0041] Users view review information provided by the system through their own devices and make purchasing decisions. If they have any opinions or comments regarding the reliability of the reviews, they can provide this information to the system through the feedback function. This feedback is sent to the server and used for continuous improvement of the model and to enhance the quality of reviews.

[0042] Specific example

[0043] For example, if a consumer is looking to buy a new smartphone, they can search for product reviews for the "XYZ smartphone" on their device. The user can then view highly reliable and summarized reviews displayed on their device. Summary information such as, "This smartphone has a long battery life and a great camera. However, many users find it heavy," can help them make a quicker and more informed purchasing decision.

[0044] Thus, the system of the present invention aims to improve consumer satisfaction through a process that combines reliability and convenience.

[0045] The following describes the processing flow.

[0046] Step 1:

[0047] The server uses web crawling technology to collect product reviews from multiple online sales platforms. During this process, the server searches based on specified keywords and extracts the URLs of the relevant reviews.

[0048] Step 2:

[0049] The server cleanses the collected review information. This cleansing process includes removing HTML tags and unnecessary strings, organizing the data into a text format for the reviews. This process formats the data in a way that facilitates subsequent analysis.

[0050] Step 3:

[0051] The server inputs clean data into a natural language processing model, analyzes the sentiment of each review, and generates a sentiment score. The models used here, such as BERT or logistic regression models, quantify and return whether the text is positive, negative, or neutral.

[0052] Step 4:

[0053] The server applies a machine learning model to detect fake reviews. This model is trained on previously collected data and identifies reviews that deviate from typical patterns. As a result, fake reviews are assigned a low confidence score.

[0054] Step 5:

[0055] The device receives analysis results from the server and prioritizes displaying reviews with high reliability scores to the user. When displaying reviews, it shows not only the details but also a summary, allowing the user to grasp important information quickly.

[0056] Step 6:

[0057] Users make purchasing decisions based on the displayed reviews and send feedback to the system regarding the information provided. This feedback is offered as an option in the UI and is stored on the server as raw feedback.

[0058] Step 7:

[0059] The server analyzes user feedback and uses the collected data to retrain the machine learning model. This improves the model's accuracy and helps in subsequent analyses.

[0060] (Example 1)

[0061] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0062] Traditionally, online reviews have significantly influenced buyer decisions, but this information is not always reliable. In particular, fake reviews and exaggerated self-promotional reviews can mislead consumers. To address this challenge, there is a need for a system that automatically evaluates the reliability of reviews and presents truly useful information to consumers.

[0063] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0064] In this invention, the server includes means for collecting purchase-related evaluation information on a computer network, means for analyzing the collected evaluation information using natural language processing techniques to generate sentiment evaluation values, and means for detecting non-regular evaluations using an inference model and assigning reliability evaluation values. This enables users to make purchasing decisions based on reliable review information.

[0065] A "computer network" is a system in which multiple computer devices communicate with each other to share and process data.

[0066] "Purchase-related evaluation information" refers to digital data that includes user opinions and evaluations of products and services.

[0067] "Natural language processing technology" refers to technical methods aimed at analyzing and understanding human language using computers.

[0068] An "emotional rating" is an index that quantifies the tendency of emotions and opinions within text or written data.

[0069] An "inference model" is a model that derives new inferences and decisions from data learned through machine learning algorithms.

[0070] "Irregular evaluation" refers to evaluation information that is inaccurate or intentionally created to be misleading.

[0071] A "reliability rating" is a numerical indicator assigned to show whether a particular piece of information is trustworthy.

[0072] "User terminal" refers to an electronic device used directly by the user, and typically includes computers, tablets, or smartphones.

[0073] This invention is a system that automatically evaluates the reliability of reviews based on purchase-related evaluation information collected via a computer network. This system consists of three entities: a server, a terminal, and a user, each playing a specific role to support consumers' purchasing decisions.

[0074] Server configuration and operation

[0075] The server collects evaluation information from various e-commerce platforms on the network using scraping techniques. For collection, it utilizes commonly used web scraping libraries as computer programs. The server stores this information in logical data management systems such as MySQL® or PostgreSQL. The evaluation information stored in the database undergoes sentiment analysis using natural language processing techniques. This analysis utilizes machine learning frameworks on the computer platform.

[0076] The information after sentiment analysis is input into an inference model to detect non-normalized ratings. This inference model is trained using historical data and machine learning algorithms. As a result of the processing, a reliability rating is assigned to each review.

[0077] Terminal configuration and operation

[0078] The device receives analyzed information delivered from the server. It then displays the received reviews in a format easily understood by the user. Specifically, it prioritizes listing reliable review information on the screen based on reliability ratings. Furthermore, it uses natural language generation technology to generate comprehensive summary information, prompting the user to make a quick purchase decision.

[0079] User roles

[0080] Users can operate the terminal and view product reviews for the items they need. Based on the summary information displayed on the screen, users make purchasing decisions. If they have any opinions about the reviews, they can send them from the terminal to the server through the feedback function. This feedback helps to further improve the model and enhance system accuracy.

[0081] Specific examples and prompt statements

[0082] For example, if a user is looking to buy a new electronic device, they might search for "reviews of the latest model electronic device" on their device. Information with a high reliability rating from the server will be prioritized and displayed, and a summary such as "This electronic device has excellent performance and a well-received design, but it is on the expensive side" will be provided. This allows the user to make a quicker and more effective purchase decision.

[0083] An example of a prompt message would be, "I'm considering purchasing the latest model of electronic device. Please provide reliable reviews and summaries of them."

[0084] This configuration allows servers, terminals, and users to work together to provide highly accurate and useful information, thereby improving the consumer purchasing experience.

[0085] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0086] Step 1:

[0087] The server collects evaluation information from various e-commerce platforms via a computer network. It requires platform URLs and page structure information as input. Using scraping libraries such as Beautiful Soup or Scrapy, it parses the HTML and extracts review text, evaluation scores, reviewer IDs, and other relevant data. The resulting output is structured data of evaluation information.

[0088] Step 2:

[0089] The server stores the collected evaluation information in a database. The input consists of structured evaluation information, which is stored in the database using MySQL or PostgreSQL. During the storage process, each review is assigned a unique identifier to facilitate access in subsequent processing. The output is the review information stored in the database.

[0090] Step 3:

[0091] The server analyzes reviews in the database using natural language processing (NLP) techniques and generates sentiment ratings. The input is the text of the reviews retrieved from the database, which is then converted into positive, neutral, or negative scores using an NLP model. The output of this process is data with a sentiment rating attached to each review.

[0092] Step 4:

[0093] The server evaluates the reliability of reviews using an inference model. The input is review information with sentiment ratings attached. A machine learning algorithm is executed to determine if the reviews are non-normalized and generate reliability ratings. The output is the review information with the reliability ratings added.

[0094] Step 5:

[0095] The server selects and delivers review information that should be displayed with priority based on the analyzed reliability rating. The input consists of review information with reliability ratings attached, which is sent to the terminal via a RESTful API. The output is evaluation information organized according to its usefulness.

[0096] Step 6:

[0097] The terminal receives review information sent from the server and displays it to the user. The input includes analyzed evaluation information from the server, and the received data is formatted into a user-friendly format before being displayed on the screen. The output is an interface that displays highly reliable review information in an organized manner.

[0098] Step 7:

[0099] Users view the provided review information and make purchasing decisions. Input includes reviews displayed on the device; users make decisions based on these reviews and, if necessary, send feedback to the system. Output is specific actions such as purchasing decisions or submitting feedback.

[0100] (Application Example 1)

[0101] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0102] In online marketplaces such as internet shopping, it is crucial to quickly and accurately assess the reliability of product and service reviews. Currently, consumers need to read through many reviews, and there is a risk of being misled by unreliable information. Furthermore, there is a lack of means to quickly grasp the key points, leading to delays in making purchasing decisions. In addition, methods for improving the system through feedback are not being effectively utilized.

[0103] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0104] In this invention, the server includes means for extracting evaluation information from the network, means for analyzing the extracted evaluation information using natural language processing to generate sentiment evaluation values, and means for detecting fraudulent evaluations using machine learning algorithms and assigning confidence evaluation values. This enables consumers to quickly obtain highly reliable evaluation information and supports their purchasing decisions. In addition, it makes it possible to continuously improve the system based on user feedback.

[0105] A "network" is an electronic connection method for exchanging information with each other.

[0106] "Evaluation information" refers to data that includes users' opinions and impressions of products and services.

[0107] "Natural language processing" is a technology that uses computers to analyze, understand, and generate human language.

[0108] An "emotional rating score" is an index that shows emotional tendencies such as positive and negative, extracted from evaluation information.

[0109] A "machine learning algorithm" is a set of computational procedures used to learn patterns from data and perform predictions and classifications.

[0110] "Fraudulent evaluation" refers to evaluation information that intentionally contains false information.

[0111] A "reliability rating" is a numerical value or indicator that shows the accuracy and reliability of evaluation information.

[0112] "Feedback" refers to opinions and reactions from system users and serves as a source of information for system improvement.

[0113] An "algorithm" is a series of computational steps that a computer performs to solve a problem.

[0114] In the system implementing the present invention, a server, a terminal, and a user collaborate to analyze the reliability of product reviews and provide consumers with useful information.

[0115] The server first collects evaluation information about products and services from multiple online sales platforms. This data collection can be done efficiently and securely by utilizing cloud infrastructure, such as AWS (Amazon Web Services). The collected evaluation information is analyzed on the server using natural language processing tools (e.g., NLTK and Spacy) to generate sentiment evaluation scores. This analysis objectively quantifies the emotional tendencies of the evaluation information, such as positive or negative. Furthermore, machine learning models using Scikit-learn and TENSORFLOW (registered trademark) detect fraudulent evaluations and assign confidence scores.

[0116] The device receives analyzed evaluation information provided by the server. When a user views product information, the device prioritizes displaying highly reliable evaluation information. This allows users to quickly grasp important information and make quick purchasing decisions. Mobile frameworks such as React Native are used for display, and summarization and feedback functions are implemented.

[0117] Users receive and view evaluation information from the system via smartphones and other mobile devices. If they have any opinions on the system's evaluation results, they can provide this information to the server through the feedback function. This feedback is used to continuously improve the machine learning model and helps to improve the accuracy of the evaluation information.

[0118] As a concrete example, consider a scenario where a user is considering purchasing a new household appliance. When the user searches for a product on their device, highly reliable reviews are prioritized, and a short overall evaluation, such as "This household appliance is energy-efficient and quiet," is presented. In this way, the present invention aims to improve the quality of reviews and the user experience.

[0119] An example of a prompt for a generative AI model would be: "New product review: Aggregate positive, negative, and neutral opinions and calculate a reliability score."

[0120] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0121] Step 1:

[0122] The server collects review information from various sales platforms via the network. This input data includes reviews in multiple text formats. The server stores the collected review information in a database and organizes it for subsequent analysis.

[0123] Step 2:

[0124] The server processes the stored evaluation information using a natural language processing algorithm. Text data and natural language processing tools (such as NLTK or Spacy) are used as input. The analysis results in the generation of sentiment evaluation values ​​for each review, with positive, negative, and neutral sentiment tendencies output as numerical values.

[0125] Step 3:

[0126] The server detects fraudulent reviews using a machine learning model (Scikit-learn or TensorFlow). The input consists of the sentiment rating data and review text data obtained in the previous step. Based on the patterns learned by the model, it identifies fraudulent reviews and assigns a confidence rating to each review. The output is the confidence rating for each review.

[0127] Step 4:

[0128] The terminal receives analyzed evaluation information provided by the server. The input is evaluation information with a confidence score assigned to it. Based on the confidence score, the terminal prioritizes displaying the evaluation information with the highest confidence score to the user. This allows the user to obtain important information in a short amount of time.

[0129] Step 5:

[0130] Users view highly reliable rating information displayed on their devices and make purchasing decisions based on that information. Input is the rating information from the device. Users can quickly make decisions about products they are interested in and send their opinions and suggestions for improvement regarding the rating information to the server through a feedback function.

[0131] Step 6:

[0132] The server collects user feedback and uses it to improve machine learning models. The input is user feedback data. This feedback data is added to the model's training data, contributing to improved accuracy in confidence assessments.

[0133] Through this process, the system of the present invention provides a mechanism for consumers to quickly and efficiently obtain reliable evaluation information. The prompt sentence used for the generating AI model is "New product review: Aggregate positive, negative, and neutral opinions and calculate a reliability score."

[0134] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0135] The system according to the present invention consists of a server, a terminal, a user, and an emotion engine. This creates a complete system that automatically evaluates the reliability of product reviews and analyzes user sentiment, aiming to improve the consumer experience.

[0136] Server Role

[0137] The server performs web crawling to extract and collect product review information from various platforms on the network. This data is analyzed using natural language processing algorithms to generate sentiment scores for the reviews. Machine learning models are also applied to identify fake reviews and assign reliability scores.

[0138] Terminal role

[0139] The device receives analyzed data from the server and displays the information in a user-friendly format. It prioritizes reviews with high reliability scores and simultaneously displays summary information to support users in quickly acquiring information. Furthermore, it uses an emotion engine to recognize the user's emotions in real time and dynamically adjusts the displayed content based on the results.

[0140] User Roles

[0141] Users provide various emotional data through wearable devices and built-in sensors designed to recognize emotions. This data is analyzed by the device and used to understand the user's emotional state. This allows for the provision of more appropriate review information tailored to the user's current emotional state.

[0142] The role of the emotional engine

[0143] The emotion engine analyzes data obtained from the user's voice, facial expressions, or vital signs to quantify their emotional state. Based on this state data, the system adaptively adjusts the selection and display of review information. This data is also sent to the server as feedback to help improve the entire system.

[0144] Specific example

[0145] For example, when a user tries to view reviews for an "ABC laptop" on an online store, the device recognizes the user's current mood along with a reliability score, and then selects and displays the most relevant reviews. For users feeling down, uplifting, positive reviews are highlighted, while users in a hurry are shown short, summarized information. This provides a personalized service tailored to each user's situation.

[0146] Thus, the present invention provides information adaptively, taking into account the user's emotional state along with highly reliable review information. This makes it possible to create an environment in which consumers can make more confident and satisfying purchasing decisions.

[0147] The following describes the processing flow.

[0148] Step 1:

[0149] The server uses web crawling technology to collect review information from multiple network platforms on the internet. This involves specifying product names and related keywords to extract data from relevant pages. The extracted data is temporarily stored as raw, unprocessed data.

[0150] Step 2:

[0151] The server performs data cleansing on the collected review information. This process uses regular expressions to remove unnecessary HTML tags and special characters, formatting the text data. By organizing the data into a clean state, it becomes suitable for analysis by natural language processing.

[0152] Step 3:

[0153] The server sends the cleansed data to a natural language processing (NLP) system to evaluate the sentiment of each review. This system uses the BERT model to calculate contextual word sentiment vectors and generates positive, negative, or neutral sentiment scores.

[0154] Step 4:

[0155] The server uses a machine learning model to detect fake reviews. This model employs a pre-trained anomaly detection algorithm to extract reviews that exhibit patterns deviating from typical reviews. A reliability score is assigned to each review and stored in a database.

[0156] Step 5:

[0157] The terminal receives analysis results sent from the server and displays the information to the user via an interface. It prioritizes presenting reviews with high reliability scores and further uses a summarization algorithm to present the key points of the reviews to the user in abbreviated form.

[0158] Step 6:

[0159] The emotion engine is activated, detecting the user's real-time emotional state from their voice and facial expressions. This data is sent to the device and used to determine which reviews are most relevant. If the user is anxious, dynamic adjustments are made, such as highlighting positive reviews.

[0160] Step 7:

[0161] Users can view product reviews through their devices and make purchasing decisions based on information tailored to their emotions. They are also presented with options to provide feedback on the accuracy of the reviews and the system. This feedback data is sent to the server and used to improve the system.

[0162] Step 8:

[0163] The server analyzes user feedback and uses it to retrain machine learning models. This enables continuous improvement to enhance system reliability and user satisfaction.

[0164] (Example 2)

[0165] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0166] Currently, online content, including reviews and ratings, may contain fraudulent or false information, making it difficult for users to make decisions based on accurate information. Furthermore, the provision of information tailored to each user's individual emotional state is not always optimized, resulting in an unbalanced user experience.

[0167] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0168] In this invention, the server includes means for collecting evaluation texts on a network, means for analyzing the collected evaluation texts using natural language processing to generate sentiment values, and means for detecting fraudulent evaluations using machine learning methods and assigning reliability values. This makes it possible to provide users with highly reliable information while also providing personalized information tailored to the user's emotional state.

[0169] A "network" is an infrastructure that connects multiple devices and systems to each other and allows them to exchange data.

[0170] "Evaluation text" refers to text data that includes user opinions and evaluations about a product or service.

[0171] "Collecting" refers to the process of selecting and accumulating specific data.

[0172] "Natural language processing" is a technology that allows computers to understand and process human language and extract information from it.

[0173] "Emotional value" is a numerical representation of the degree of emotion derived from text.

[0174] "Machine learning methods" are techniques that enable computers to learn from data and automate specific tasks.

[0175] "Fraudulent evaluation" refers to an evaluation document that is not based on actual experience or that intentionally contains false information.

[0176] A "reliability metric" is a numerical indicator that shows the accuracy and integrity of an evaluation document.

[0177] A "human-operated device" is a device that allows users to obtain information or perform control through user interaction.

[0178] "Dynamic adjustment" means changing or adapting in real time according to the user's situation and environment.

[0179] "Emotional data" refers to information related to a user's psychological state and emotions.

[0180] "Feedback" refers to information used to improve or adjust a system or device based on the data it has obtained.

[0181] "Analyzing" is the process of deriving a detailed understanding or conclusions based on the information obtained.

[0182] "Summary" means making detailed information concise and extracting only the essential information.

[0183] The system according to the present invention consists of a server, a terminal, a user, and an emotion analysis engine.

[0184] Server configuration and processing:

[0185] The server automatically collects review texts from various sources on the internet using web crawlers. This uses libraries such as Python's Beautiful Soup and Scrapy. This data is stored in a database, and sentiment scores are calculated using natural language processing algorithms. Python's NLTK and spaCy are used for this. The server also runs machine learning models (using Scikit-learn and TensorFlow) to identify fraudulent reviews and assign a reliability score to each review.

[0186] Terminal configuration and processing:

[0187] The device uses the analyzed data received from the server to display information in a format easily understandable to the user. Using a web interface based on HTML, CSS, and JavaScript (registered trademark), it prioritizes displaying reviews with high reliability scores and summarizes reviews using natural language processing technology. Furthermore, the device works in conjunction with a sentiment analysis engine to dynamically adjust the displayed content based on the user's real-time sentiment data.

[0188] User configuration and processing:

[0189] Users collect emotional data such as heart rate and voice tone via wearable devices like smartwatches and provide it to the device. This allows the device to analyze the user's current emotional state and select appropriate display information.

[0190] Structure and processing of the emotion analysis engine:

[0191] The emotion analysis engine quantifies emotions from the user's voice signals and facial expression data. This analysis utilizes OpenCV and voice analysis libraries. The analysis results are then used to optimize the overall information display of the system, providing content tailored to the user's individual needs. This data is returned to the server as feedback, contributing to the improvement of the machine learning model.

[0192] Examples of specific cases and prompt statements:

[0193] For example, if a user wants to view reviews of a particular product on an online platform, the device will provide information that takes the user's emotional state into consideration. In particular, if the user is feeling stressed, it will display short, easy-to-understand positive reviews.

[0194] An example of a prompt message would be: "Get reviews for ABC laptops and sort them based on sentiment and trustworthiness. If users are stressed, highlight one short, positive review."

[0195] This system aims to improve the consumer experience by providing each user with appropriately personalized evaluation information.

[0196] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0197] Step 1:

[0198] The server collects evaluation texts from information sources on the network. Given specific URLs or keywords as input, the server uses Python's Beautiful Soup or Scrapy to scrape matching evaluation texts. As output, the collected evaluation texts are stored in a database for subsequent analysis.

[0199] Step 2:

[0200] The server performs natural language processing on the collected evaluation texts to generate sentiment scores. It receives collected text data as input and analyzes the text's polarity using Python's NLTK and spaCy. Specifically, it scores the positive, negative, and neutral sentiment of each text. The output provides sentiment scores corresponding to each evaluation text, which serve as material for reliability evaluation.

[0201] Step 3:

[0202] The server executes machine learning methods to assign a reliability score to each evaluated text. As input, evaluated texts with sentiment scores attached are used, and these are passed through a model that identifies fraudulent evaluations using Scikit-learn and TensorFlow. Specifically, a model built using supervised learning determines the reliability of the evaluated text. As output, evaluations judged to be fraudulent are given a low reliability score, and evaluations recognized as legitimate are given a high reliability score.

[0203] Step 4:

[0204] The terminal receives analyzed data provided by the server and displays it to the user through a visual user interface. It receives evaluation texts sorted by reliability and sentiment scores as input. Specifically, view components created with HTML, CSS, and JavaScript are used to visualize high-priority information. The output is in a format that allows the user to easily identify the most useful information.

[0205] Step 5:

[0206] Users provide emotional data to the device using a wearable device. Inputs include heart rate and voice tone, transmitted via Bluetooth or Wi-Fi. The device then passes this data to an emotion analysis engine, which quantifies the emotional state. The analyzed emotional values ​​are then used as input for individual content adjustments.

[0207] Step 6:

[0208] The device dynamically adjusts its display content based on emotion values ​​obtained from an emotion analysis engine. The input is the user's real-time emotion values. In response, the UI content changes appropriately, providing information tailored to the user's emotional state. The output is a customized information display optimized for the user experience.

[0209] (Application Example 2)

[0210] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0211] It is difficult to efficiently provide users with reliable information from the countless evaluation data available on the network. Furthermore, appropriate information is not provided according to the user's emotional state, lacking the individual adaptability necessary for purchase decision-making. Therefore, support for users in making optimal purchasing choices from a diverse range of options is insufficient.

[0212] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0213] In this invention, the server includes means for extracting evaluation information from the network, means for analyzing the extracted evaluation information using natural language processing to generate an emotion index, means for detecting false evaluations using a machine learning algorithm and assigning a reliability index, means for detecting the user's emotional state and dynamically adjusting the displayed content based on the emotional state, and means for preferentially providing useful evaluation information to the user terminal based on the reliability index. This enables the provision of individually adapted information according to the user's emotional state, and makes it possible to support optimal purchasing choices based on highly reliable information.

[0214] A "network" is a communication system in which multiple digital devices are connected to each other and can share information.

[0215] "Evaluation information" refers to user opinions, impressions, and feedback regarding a particular product or service.

[0216] "Natural language processing" is a technology that uses computers to process human language, understand its meaning, and analyze it.

[0217] "Emotional indicators" are a way of expressing a user's emotions and psychological state using numbers or categories.

[0218] A "machine learning algorithm" is a computational method that uses data to allow a computer to automatically learn and perform a specific task.

[0219] A "false review" is a review or opinion that contains false information not based on actual usage experience or facts.

[0220] A "reliability index" is a standard or measure used to evaluate the reliability and accuracy of the information provided.

[0221] "User terminal" refers to a computer device used by a user for operations and information retrieval.

[0222] "Individually adapted information provision" is a method of providing information customized to the user's specific situation and needs.

[0223] The server extracts a wide variety of evaluation information from the network. This process uses web crawling technology to collect information from various websites and platforms. The collected data is analyzed by natural language processing (NLP) algorithms to generate sentiment metrics for each review. Machine learning algorithms are also applied to detect whether the evaluation information is false and to assign reliability metrics. This prepares the system to prioritize and deliver reliable information to the user's device.

[0224] The device receives analyzed data from the server and displays the information in a format that is easy for the user to see and use. The device uses sensors such as a camera and microphone to analyze the user's emotional state in real time and adjusts the information displayed on the screen based on that analysis. This adjustment provides information optimized for the user's current emotional state.

[0225] Users access the system using smartphones or other digital devices. This allows users to obtain the most relevant information that matches their emotional state when checking reviews of desired products. For example, if a user is looking for reviews of an "ABC laptop," the device will display detailed and reliable information while summarizing it according to the user's emotional state.

[0226] As a concrete example, consider a scenario where a user is browsing reviews of an "ABC laptop" while shopping on a holiday. If the user's facial expression is detected as calm, a review including technical details will be displayed. However, if the user is detected as being in a hurry, a summarized, shorter review will be presented.

[0227] A generative AI model is used to support this process, and an example of its prompt is: "Write a Python script that personalizes and displays reviews, taking into account the user's emotional state. Analyze the user's facial expressions and voice to select relevant product reviews."

[0228] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0229] Step 1:

[0230] The server uses web crawling technology to extract evaluation information from various platforms on the network. The input is keywords for a specific product, and the output is raw data of related review information. This data is stored in a database for later processing. This process is automated using web scraping tools.

[0231] Step 2:

[0232] The server analyzes the extracted raw data using natural language processing (NLP) algorithms and assigns a sentiment index to each review. Using the extracted review information as input, it generates a sentiment score for each review as output. This process utilizes morphological analysis and sentiment dictionaries to calculate the frequency of positive and negative words contained in the reviews.

[0233] Step 3:

[0234] The server uses a machine learning algorithm to evaluate the reliability of reviews. The input is review information with sentiment scores, and the output assigns high scores to highly reliable reviews. This process utilizes a learning model based on historical data and applies an algorithm to identify false reviews.

[0235] Step 4:

[0236] The terminal receives analyzed review information along with a reliability score from the server. The input is the analyzed review information, and the output is data suitable for display. This information is displayed in the user interface and organized for easy access by the user.

[0237] Step 5:

[0238] The device uses an emotional state analysis sensor to determine the user's emotional state. Inputs include user facial expression data and voice data, and output is a real-time emotion assessment. Based on this assessment, a dynamic display algorithm is executed, which uses an NLP model to adjust the review display to suit the user's situation.

[0239] Step 6:

[0240] Users review the displayed review information on their devices and use it as a reference for their purchasing decisions. The input is the user's desired product information, and the output is a list of reliable reviews for the selected product. This allows users to efficiently obtain information that suits their emotional state.

[0241] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating 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.

[0242] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0243] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0244] [Second Embodiment]

[0245] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0246] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0247] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0248] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0249] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0250] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0251] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0252] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0253] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0254] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0255] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0256] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. 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".

[0257] The system according to the present invention consists of a server, a terminal, and a user, and aims to automatically evaluate the reliability of product reviews over a network and provide useful information to consumers.

[0258] Server Role

[0259] The server first collects review information from various sales sites via the network. The collected data is stored and managed on the server. The server processes this data using natural language processing algorithms to analyze the text of each review. Through this analysis, it generates a sentiment score for each review, identifying whether the review is positive, neutral, or negative. Furthermore, it uses machine learning models to detect fake reviews and evaluate the reliability of each review.

[0260] Terminal role

[0261] The device receives analyzed data from the server and displays it in a format suitable for the user's actions. Specifically, it prioritizes displaying highly reliable reviews based on their reliability scores, and generates and displays review summaries so that the user can quickly grasp the main points of the reviews.

[0262] User Roles

[0263] Users view review information provided by the system through their own devices and make purchasing decisions. If they have any opinions or comments regarding the reliability of the reviews, they can provide this information to the system through the feedback function. This feedback is sent to the server and used for continuous improvement of the model and to enhance the quality of reviews.

[0264] Specific example

[0265] For example, if a consumer is looking to buy a new smartphone, they can search for product reviews for "XYZ smartphone" on their device. The user can then view highly reliable and summarized reviews displayed on their device. Summary information such as, "This smartphone has a long battery life and a good camera. However, many users find it heavy," can help them make a quicker and more informed purchasing decision.

[0266] Thus, the system of the present invention aims to improve consumer satisfaction through a process that combines reliability and convenience.

[0267] The following describes the processing flow.

[0268] Step 1:

[0269] The server uses web crawling technology to collect product reviews from multiple online sales platforms. During this process, the server searches based on specified keywords and extracts the URLs of the relevant reviews.

[0270] Step 2:

[0271] The server cleanses the collected review information. This cleansing process includes removing HTML tags and unnecessary strings, organizing the data into a text format for the reviews. This process formats the data in a way that facilitates subsequent analysis.

[0272] Step 3:

[0273] The server inputs clean data into a natural language processing model, analyzes the sentiment of each review, and generates a sentiment score. The models used here, such as BERT or logistic regression models, quantify and return whether the text is positive, negative, or neutral.

[0274] Step 4:

[0275] The server applies a machine learning model to detect fake reviews. This model is trained on previously collected data and identifies reviews that deviate from typical patterns. As a result, fake reviews are assigned a low confidence score.

[0276] Step 5:

[0277] The terminal receives the analysis results from the server and preferentially displays reviews with a high reliability score to the user. When displaying, not only the details of the review but also the summarized content is presented so that the user can grasp important information in a short time.

[0278] Step 6:

[0279] Based on the displayed reviews, the user makes a purchase decision and sends feedback on the provided information to the system. The feedback is provided as an option on the UI and is saved as raw opinions on the server.

[0280] Step 7:

[0281] The server analyzes the feedback from the user and retrains the machine learning model using the collected data. This improves the accuracy of the model and is useful for the next analysis.

[0282] (Example 1)

[0283] Next, Example 1 will be described. In the following description, the data processing device 12 is referred to as the "server", and the smart glasses 214 are referred to as the "terminal".

[0284] Conventionally, review information in the online market greatly affects the judgment of purchasers, but the information is not always reliable. In particular, false reviews and exaggerated self-promotional reviews may mislead consumers' judgments. To solve this problem, there is a need for a system that automatically evaluates the reliability of reviews and presents information that is truly useful to consumers.

[0285] The specific processing by the specific processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0286] In this invention, the server includes means for collecting purchase-related evaluation information on a computer network, means for analyzing the collected evaluation information by natural language analysis technology to generate an emotional evaluation value, and means for using an inference model to detect informal evaluations and assign a reliability evaluation value. Thereby, users can perform consumption behaviors based on reliable review information.

[0287] A "computer network" is a mechanism in which a plurality of computer devices communicate to share and process data.

[0288] "Purchase-related evaluation information" refers to digital data including users' opinions and evaluations on products and services.

[0289] "Natural language analysis technology" is a technical method aimed at analyzing and understanding human language by a computer.

[0290] An "emotional evaluation value" is an index that quantifies the tendency of emotions and opinions in a text or text data.

[0291] An "inference model" is a model that derives new inferences and decisions from data learned by a machine learning algorithm.

[0292] "Informal evaluation" refers to evaluation information that is inaccurate or may cause misunderstandings created intentionally.

[0293] A "reliability evaluation value" is a numerical index assigned to indicate whether specific information is reliable.

[0294] A "user terminal" refers to an electronic device for direct use by a user, usually including a computer, a tablet, or a smartphone.

[0295] This invention is a system that automatically evaluates the reliability of reviews based on purchase-related evaluation information collected via a computer network. This system consists of three entities: a server, a terminal, and a user, each playing a specific role to support consumers' purchasing decisions.

[0296] Server configuration and operation

[0297] The server collects evaluation information from various e-commerce platforms on the network using scraping techniques. For collection, it utilizes commonly used web scraping libraries as computer programs. The server stores this information in a logical data management system such as MySQL or PostgreSQL. The evaluation information stored in the database undergoes sentiment analysis using natural language processing techniques. This analysis utilizes machine learning frameworks on the computer platform.

[0298] The information after sentiment analysis is input into an inference model to detect non-normalized ratings. This inference model is trained using historical data and machine learning algorithms. As a result of the processing, a reliability rating is assigned to each review.

[0299] Terminal configuration and operation

[0300] The device receives analyzed information delivered from the server. It then displays the received reviews in a format easily understood by the user. Specifically, it prioritizes listing reliable review information on the screen based on reliability ratings. Furthermore, it uses natural language generation technology to generate comprehensive summary information, prompting the user to make a quick purchase decision.

[0301] User roles

[0302] The user can operate the terminal and view the evaluation information of the required products. Based on the summary information displayed on the screen, the user makes a decision to purchase. If there are opinions about the evaluation information, they are sent from the terminal to the server through the feedback function. This feedback helps to improve further models and system accuracy.

[0303] Specific examples and prompt sentences

[0304] For example, when a user intends to purchase a new electronic device, they search for "Reviews of the latest model electronic devices" on the terminal. Information with a high reliability evaluation value from the server is preferentially displayed, and as a summary, information such as "This electronic device has excellent performance and a well-received design. However, the price is on the high side" is provided. This enables the user to make a quicker and more effective purchase decision.

[0305] Examples of prompt sentences include "I am considering purchasing the latest model of an electronic device. Please provide reliable reviews and a summary."

[0306] With such a configuration, the server, terminal, and user can cooperate to provide highly accurate and useful information, making it possible to improve the consumer's purchasing experience.

[0307] The flow of the specific process in Example 1 will be described using FIG. 11.

[0308] Step 1:

[0309] The server collects evaluation information from various business transaction platforms via a computer network. As input, the URL of the platform and page configuration information are required. HTML is parsed using scraping libraries such as Beautiful Soup or Scrapy to extract review text, evaluation scores, reviewer IDs, etc. The output obtained thereby is structured data of the evaluation information.

[0310] Step 2:

[0311] The server stores the collected evaluation information in a database. The input consists of structured evaluation information, which is stored in the database using MySQL or PostgreSQL. During the storage process, each review is assigned a unique identifier to facilitate access in subsequent processing. The output is the review information stored in the database.

[0312] Step 3:

[0313] The server analyzes reviews in the database using natural language processing (NLP) techniques and generates sentiment ratings. The input is the text of the reviews retrieved from the database, which is then converted into positive, neutral, or negative scores using an NLP model. The output of this process is data with a sentiment rating attached to each review.

[0314] Step 4:

[0315] The server evaluates the reliability of reviews using an inference model. The input is review information with sentiment ratings attached. A machine learning algorithm is executed to determine if the reviews are non-normalized and generate reliability ratings. The output is the review information with the reliability ratings added.

[0316] Step 5:

[0317] The server selects and delivers review information that should be displayed with priority based on the analyzed reliability rating. The input consists of review information with reliability ratings attached, which is sent to the terminal via a RESTful API. The output is evaluation information organized according to its usefulness.

[0318] Step 6:

[0319] The terminal receives review information sent from the server and displays it to the user. The input includes analyzed evaluation information from the server, and the received data is formatted into a user-friendly format before being displayed on the screen. The output is an interface that displays highly reliable review information in an organized manner.

[0320] Step 7:

[0321] Users view the provided review information and make purchasing decisions. Input includes reviews displayed on the device; users make decisions based on these reviews and, if necessary, send feedback to the system. Output is specific actions such as purchasing decisions or submitting feedback.

[0322] (Application Example 1)

[0323] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0324] In online marketplaces such as internet shopping, it is crucial to quickly and accurately assess the reliability of product and service reviews. Currently, consumers need to read through many reviews, and there is a risk of being misled by unreliable information. Furthermore, there is a lack of means to quickly grasp the key points, leading to delays in making purchasing decisions. In addition, methods for improving the system through feedback are not being effectively utilized.

[0325] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0326] In this invention, the server includes means for extracting evaluation information from the network, means for analyzing the extracted evaluation information using natural language processing to generate sentiment evaluation values, and means for detecting fraudulent evaluations using machine learning algorithms and assigning confidence evaluation values. This enables consumers to quickly obtain highly reliable evaluation information and supports their purchasing decisions. In addition, it makes it possible to continuously improve the system based on user feedback.

[0327] A "network" is an electronic connection method for exchanging information with each other.

[0328] "Evaluation information" refers to data that includes users' opinions and impressions of products and services.

[0329] "Natural language processing" is a technology that uses computers to analyze, understand, and generate human language.

[0330] An "emotional rating score" is an index that shows emotional tendencies such as positive and negative, extracted from evaluation information.

[0331] A "machine learning algorithm" is a set of computational procedures used to learn patterns from data and perform predictions and classifications.

[0332] "Fraudulent evaluation" refers to evaluation information that intentionally contains false information.

[0333] A "reliability rating" is a numerical value or indicator that shows the accuracy and reliability of evaluation information.

[0334] "Feedback" refers to opinions and reactions from system users and serves as a source of information for system improvement.

[0335] An "algorithm" is a series of computational steps that a computer performs to solve a problem.

[0336] In the system implementing the present invention, a server, a terminal, and a user collaborate to analyze the reliability of product reviews and provide consumers with useful information.

[0337] The server first collects evaluation information about products and services from multiple online sales platforms. This data collection can be done efficiently and securely by utilizing cloud infrastructure, such as AWS (Amazon Web Services). The collected evaluation information is analyzed on the server using natural language processing tools (e.g., NLTK and Spacy) to generate sentiment evaluation scores. This analysis objectively quantifies the emotional tendencies of the evaluation information, such as positive or negative. Furthermore, machine learning models using Scikit-learn and TensorFlow detect fraudulent evaluations and assign confidence scores.

[0338] The device receives analyzed evaluation information provided by the server. When a user views product information, the device prioritizes displaying highly reliable evaluation information. This allows users to quickly grasp important information and make quick purchasing decisions. Mobile frameworks such as React Native are used for display, and summarization and feedback functions are implemented.

[0339] Users receive and view evaluation information from the system via smartphones and other mobile devices. If they have any opinions on the system's evaluation results, they can provide this information to the server through the feedback function. This feedback is used to continuously improve the machine learning model and helps to improve the accuracy of the evaluation information.

[0340] As a concrete example, consider a scenario where a user is considering purchasing a new household appliance. When the user searches for a product on their device, highly reliable reviews are prioritized, and a short overall evaluation, such as "This household appliance is energy-efficient and quiet," is presented. In this way, the present invention aims to improve the quality of reviews and the user experience.

[0341] An example of a prompt for a generative AI model would be: "New product review: Aggregate positive, negative, and neutral opinions and calculate a reliability score."

[0342] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0343] Step 1:

[0344] The server collects review information from various sales platforms via the network. This input data includes reviews in multiple text formats. The server stores the collected review information in a database and organizes it for subsequent analysis.

[0345] Step 2:

[0346] The server processes the stored evaluation information using a natural language processing algorithm. Text data and natural language processing tools (such as NLTK or Spacy) are used as input. The analysis results in the generation of sentiment evaluation values ​​for each review, with positive, negative, and neutral sentiment tendencies output as numerical values.

[0347] Step 3:

[0348] The server detects fraudulent reviews using a machine learning model (Scikit-learn or TensorFlow). The input consists of the sentiment rating data and review text data obtained in the previous step. Based on the patterns learned by the model, it identifies fraudulent reviews and assigns a confidence rating to each review. The output is the confidence rating for each review.

[0349] Step 4:

[0350] The terminal receives analyzed evaluation information provided by the server. The input is evaluation information with a confidence score assigned to it. Based on the confidence score, the terminal prioritizes displaying the evaluation information with the highest confidence score to the user. This allows the user to obtain important information in a short amount of time.

[0351] Step 5:

[0352] Users view highly reliable rating information displayed on their devices and make purchasing decisions based on that information. Input is the rating information from the device. Users can quickly make decisions about products they are interested in and send their opinions and suggestions for improvement regarding the rating information to the server through a feedback function.

[0353] Step 6:

[0354] The server collects user feedback and uses it to improve machine learning models. The input is user feedback data. This feedback data is added to the model's training data, contributing to improved accuracy in confidence assessments.

[0355] Through this process, the system of the present invention provides a mechanism for consumers to quickly and efficiently obtain reliable evaluation information. The prompt sentence used for the generating AI model is "New product review: Aggregate positive, negative, and neutral opinions and calculate a reliability score."

[0356] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0357] The system according to the present invention consists of a server, a terminal, a user, and an emotion engine. This creates a complete system that automatically evaluates the reliability of product reviews and analyzes user sentiment, aiming to improve the consumer experience.

[0358] Server Role

[0359] The server performs web crawling to extract and collect product review information from various platforms on the network. This data is analyzed using natural language processing algorithms to generate sentiment scores for the reviews. Machine learning models are also applied to identify fake reviews and assign reliability scores.

[0360] Terminal role

[0361] The device receives analyzed data from the server and displays the information in a user-friendly format. It prioritizes reviews with high reliability scores and simultaneously displays summary information to support users in quickly acquiring information. Furthermore, it uses an emotion engine to recognize the user's emotions in real time and dynamically adjusts the displayed content based on the results.

[0362] User Roles

[0363] Users provide various emotional data through wearable devices and built-in sensors designed to recognize emotions. This data is analyzed by the device and used to understand the user's emotional state. This allows for the provision of more appropriate review information tailored to the user's current emotional state.

[0364] The role of the emotional engine

[0365] The emotion engine analyzes data obtained from the user's voice, facial expressions, or vital signs to quantify their emotional state. Based on this state data, the system adaptively adjusts the selection and display of review information. This data is also sent to the server as feedback to help improve the entire system.

[0366] Specific example

[0367] For example, when a user tries to view reviews for an "ABC laptop" on an online store, the device recognizes the user's current mood along with a reliability score, and then selects and displays the most relevant reviews. For users feeling down, uplifting, positive reviews are highlighted, while users in a hurry are shown short, summarized information. This provides a personalized service tailored to each user's situation.

[0368] Thus, the present invention provides information adaptively, taking into account the user's emotional state along with highly reliable review information. This makes it possible to create an environment in which consumers can make more confident and satisfying purchasing decisions.

[0369] The following describes the processing flow.

[0370] Step 1:

[0371] The server uses web crawling technology to collect review information from multiple network platforms on the internet. This involves specifying product names and related keywords to extract data from relevant pages. The extracted data is temporarily stored as raw, unprocessed data.

[0372] Step 2:

[0373] The server performs data cleansing on the collected review information. This process uses regular expressions to remove unnecessary HTML tags and special characters, formatting the text data. By organizing the data into a clean state, it becomes suitable for analysis by natural language processing.

[0374] Step 3:

[0375] The server sends the cleansed data to a natural language processing (NLP) system to evaluate the sentiment of each review. This system uses the BERT model to calculate contextual word sentiment vectors and generates positive, negative, or neutral sentiment scores.

[0376] Step 4:

[0377] The server uses a machine learning model to detect fake reviews. This model employs a pre-trained anomaly detection algorithm to extract reviews that exhibit patterns deviating from typical reviews. A reliability score is assigned to each review and stored in a database.

[0378] Step 5:

[0379] The terminal receives analysis results sent from the server and displays the information to the user via an interface. It prioritizes presenting reviews with high reliability scores and further uses a summarization algorithm to present the key points of the reviews to the user in abbreviated form.

[0380] Step 6:

[0381] The emotion engine is activated, detecting the user's real-time emotional state from their voice and facial expressions. This data is sent to the device and used to determine which reviews are most relevant. If the user is anxious, dynamic adjustments are made, such as highlighting positive reviews.

[0382] Step 7:

[0383] Users can view product reviews through their devices and make purchasing decisions based on information tailored to their emotions. They are also presented with options to provide feedback on the accuracy of the reviews and the system. This feedback data is sent to the server and used to improve the system.

[0384] Step 8:

[0385] The server analyzes user feedback and uses it to retrain machine learning models. This enables continuous improvement to enhance system reliability and user satisfaction.

[0386] (Example 2)

[0387] Next, we will describe Example 2. 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".

[0388] Currently, online content, including reviews and ratings, may contain fraudulent or false information, making it difficult for users to make decisions based on accurate information. Furthermore, the provision of information tailored to each user's individual emotional state is not always optimized, resulting in an unbalanced user experience.

[0389] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0390] In this invention, the server includes means for collecting evaluation texts on a network, means for analyzing the collected evaluation texts using natural language processing to generate sentiment values, and means for detecting fraudulent evaluations using machine learning methods and assigning reliability values. This makes it possible to provide users with highly reliable information while also providing personalized information tailored to the user's emotional state.

[0391] A "network" is an infrastructure that connects multiple devices and systems to each other and allows them to exchange data.

[0392] "Evaluation text" refers to text data that includes user opinions and evaluations about a product or service.

[0393] "Collecting" refers to the process of selecting and accumulating specific data.

[0394] "Natural language processing" is a technology that allows computers to understand and process human language and extract information from it.

[0395] "Emotional value" is a numerical representation of the degree of emotion derived from text.

[0396] "Machine learning methods" are techniques that enable computers to learn from data and automate specific tasks.

[0397] "Fraudulent evaluation" refers to an evaluation document that is not based on actual experience or that intentionally contains false information.

[0398] A "reliability metric" is a numerical indicator that shows the accuracy and integrity of an evaluation document.

[0399] A "human-operated device" is a device that allows users to obtain information or perform control through user interaction.

[0400] "Dynamic adjustment" means changing or adapting in real time according to the user's situation and environment.

[0401] "Emotional data" refers to information related to a user's psychological state and emotions.

[0402] "Feedback" refers to information used to improve or adjust a system or device based on the data it has obtained.

[0403] "Analyzing" is the process of deriving a detailed understanding or conclusions based on the information obtained.

[0404] "Summary" means making detailed information concise and extracting only the essential information.

[0405] The system according to the present invention consists of a server, a terminal, a user, and an emotion analysis engine.

[0406] Server configuration and processing:

[0407] The server automatically collects review texts from various sources on the internet using web crawlers. This uses libraries such as Python's Beautiful Soup and Scrapy. This data is stored in a database, and sentiment scores are calculated using natural language processing algorithms. Python's NLTK and spaCy are used for this. The server also runs machine learning models (using Scikit-learn and TensorFlow) to identify fraudulent reviews and assign a reliability score to each review.

[0408] Terminal configuration and processing:

[0409] The device uses the analyzed data received from the server to display information in a format easily understandable to the user. A web interface using HTML, CSS, and JavaScript prioritizes the display of reviews with high reliability scores and summarizes reviews using natural language processing technology. Furthermore, the device works in conjunction with a sentiment analysis engine to dynamically adjust the displayed content based on the user's real-time sentiment data.

[0410] User configuration and processing:

[0411] Users collect emotional data such as heart rate and voice tone via wearable devices like smartwatches and provide it to the device. This allows the device to analyze the user's current emotional state and select appropriate display information.

[0412] Structure and processing of the emotion analysis engine:

[0413] The emotion analysis engine quantifies emotions from the user's voice signals and facial expression data. This analysis utilizes OpenCV and voice analysis libraries. The analysis results are then used to optimize the overall information display of the system, providing content tailored to the user's individual needs. This data is returned to the server as feedback, contributing to the improvement of the machine learning model.

[0414] Examples of specific cases and prompt statements:

[0415] For example, if a user wants to view reviews of a particular product on an online platform, the device will provide information that takes the user's emotional state into consideration. In particular, if the user is feeling stressed, it will display short, easy-to-understand positive reviews.

[0416] An example of a prompt message would be: "Get reviews for ABC laptops and sort them based on sentiment and trustworthiness. If users are stressed, highlight one short, positive review."

[0417] This system aims to improve the consumer experience by providing each user with appropriately personalized evaluation information.

[0418] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0419] Step 1:

[0420] The server collects evaluation texts from information sources on the network. Given specific URLs or keywords as input, the server uses Python's Beautiful Soup or Scrapy to scrape matching evaluation texts. As output, the collected evaluation texts are stored in a database for subsequent analysis.

[0421] Step 2:

[0422] The server performs natural language processing on the collected evaluation texts to generate sentiment scores. It receives collected text data as input and analyzes the text's polarity using Python's NLTK and spaCy. Specifically, it scores the positive, negative, and neutral sentiment of each text. The output provides sentiment scores corresponding to each evaluation text, which serve as material for reliability evaluation.

[0423] Step 3:

[0424] The server executes machine learning methods to assign a reliability score to each evaluated text. As input, evaluated texts with sentiment scores attached are used, and these are passed through a model that identifies fraudulent evaluations using Scikit-learn and TensorFlow. Specifically, a model built using supervised learning determines the reliability of the evaluated text. As output, evaluations judged to be fraudulent are given a low reliability score, and evaluations recognized as legitimate are given a high reliability score.

[0425] Step 4:

[0426] The terminal receives analyzed data provided by the server and displays it to the user through a visual user interface. It receives evaluation texts sorted by reliability and sentiment scores as input. Specifically, view components created with HTML, CSS, and JavaScript are used to visualize high-priority information. The output is in a format that allows the user to easily identify the most useful information.

[0427] Step 5:

[0428] Users provide emotional data to the device using a wearable device. Inputs include heart rate and voice tone, transmitted via Bluetooth or Wi-Fi. The device then passes this data to an emotion analysis engine, which quantifies the emotional state. The analyzed emotional values ​​are then used as input for individual content adjustments.

[0429] Step 6:

[0430] The device dynamically adjusts its display content based on emotion values ​​obtained from an emotion analysis engine. The input is the user's real-time emotion values. In response, the UI content changes appropriately, providing information tailored to the user's emotional state. The output is a customized information display optimized for the user experience.

[0431] (Application Example 2)

[0432] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0433] It is difficult to efficiently provide users with reliable information from the countless evaluation data available on the network. Furthermore, appropriate information is not provided according to the user's emotional state, lacking the individual adaptability necessary for purchase decision-making. Therefore, support for users in making optimal purchasing choices from a diverse range of options is insufficient.

[0434] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0435] In this invention, the server includes means for extracting evaluation information from the network, means for analyzing the extracted evaluation information using natural language processing to generate an emotion index, means for detecting false evaluations using a machine learning algorithm and assigning a reliability index, means for detecting the user's emotional state and dynamically adjusting the displayed content based on the emotional state, and means for preferentially providing useful evaluation information to the user terminal based on the reliability index. This enables the provision of individually adapted information according to the user's emotional state, and makes it possible to support optimal purchasing choices based on highly reliable information.

[0436] A "network" is a communication system in which multiple digital devices are connected to each other and can share information.

[0437] "Evaluation information" refers to user opinions, impressions, and feedback regarding a particular product or service.

[0438] "Natural language processing" is a technology that uses computers to process human language, understand its meaning, and analyze it.

[0439] "Emotional indicators" are a way of expressing a user's emotions and psychological state using numbers or categories.

[0440] A "machine learning algorithm" is a computational method that uses data to allow a computer to automatically learn and perform a specific task.

[0441] A "false review" is a review or opinion that contains false information not based on actual usage experience or facts.

[0442] A "reliability index" is a standard or measure used to evaluate the reliability and accuracy of the information provided.

[0443] "User terminal" refers to a computer device used by a user for operations and information retrieval.

[0444] "Individually adapted information provision" is a method of providing information customized to the user's specific situation and needs.

[0445] The server extracts a wide variety of evaluation information from the network. This process uses web crawling technology to collect information from various websites and platforms. The collected data is analyzed by natural language processing (NLP) algorithms to generate sentiment metrics for each review. Machine learning algorithms are also applied to detect whether the evaluation information is false and to assign reliability metrics. This prepares the system to prioritize and deliver reliable information to the user's device.

[0446] The device receives analyzed data from the server and displays the information in a format that is easy for the user to see and use. The device uses sensors such as a camera and microphone to analyze the user's emotional state in real time and adjusts the information displayed on the screen based on that analysis. This adjustment provides information optimized for the user's current emotional state.

[0447] Users access the system using smartphones or other digital devices. This allows users to obtain the most relevant information that matches their emotional state when checking reviews of desired products. For example, if a user is looking for reviews of an "ABC laptop," the device will display detailed and reliable information while summarizing it according to the user's emotional state.

[0448] As a concrete example, consider a scenario where a user is browsing reviews of an "ABC laptop" while shopping on a holiday. If the user's facial expression is detected as calm, a review including technical details will be displayed. However, if the user is detected as being in a hurry, a summarized, shorter review will be presented.

[0449] A generative AI model is used to support this process, and an example of its prompt is: "Write a Python script that personalizes and displays reviews, taking into account the user's emotional state. Analyze the user's facial expressions and voice to select relevant product reviews."

[0450] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0451] Step 1:

[0452] The server uses web crawling technology to extract evaluation information from various platforms on the network. The input is keywords for a specific product, and the output is raw data of related review information. This data is stored in a database for later processing. This process is automated using web scraping tools.

[0453] Step 2:

[0454] The server analyzes the extracted raw data using natural language processing (NLP) algorithms and assigns a sentiment index to each review. Using the extracted review information as input, it generates a sentiment score for each review as output. This process utilizes morphological analysis and sentiment dictionaries to calculate the frequency of positive and negative words contained in the reviews.

[0455] Step 3:

[0456] The server uses a machine learning algorithm to evaluate the reliability of reviews. The input is review information with sentiment scores, and the output assigns high scores to highly reliable reviews. This process utilizes a learning model based on historical data and applies an algorithm to identify false reviews.

[0457] Step 4:

[0458] The terminal receives analyzed review information along with a reliability score from the server. The input is the analyzed review information, and the output is data suitable for display. This information is displayed in the user interface and organized for easy access by the user.

[0459] Step 5:

[0460] The device uses an emotional state analysis sensor to determine the user's emotional state. Inputs include user facial expression data and voice data, and output is a real-time emotion assessment. Based on this assessment, a dynamic display algorithm is executed, which uses an NLP model to adjust the review display to suit the user's situation.

[0461] Step 6:

[0462] Users review the displayed review information on their devices and use it as a reference for their purchasing decisions. The input is the user's desired product information, and the output is a list of reliable reviews for the selected product. This allows users to efficiently obtain information that suits their emotional state.

[0463] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0464] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0465] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0466] [Third Embodiment]

[0467] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0468] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0469] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0470] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0471] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0472] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0473] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0474] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0475] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0476] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0477] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0478] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0479] The system according to the present invention consists of a server, a terminal, and a user, and aims to automatically evaluate the reliability of product reviews over a network and provide useful information to consumers.

[0480] Server Role

[0481] The server first collects review information from various sales sites via the network. The collected data is stored and managed on the server. The server processes this data using natural language processing algorithms to analyze the text of each review. Through this analysis, it generates a sentiment score for each review, identifying whether the review is positive, neutral, or negative. Furthermore, it uses machine learning models to detect fake reviews and evaluate the reliability of each review.

[0482] Terminal role

[0483] The device receives analyzed data from the server and displays it in a format suitable for the user's actions. Specifically, it prioritizes displaying highly reliable reviews based on their reliability scores, and generates and displays review summaries so that the user can quickly grasp the main points of the reviews.

[0484] User Roles

[0485] Users view review information provided by the system through their own devices and make purchasing decisions. If they have any opinions or comments regarding the reliability of the reviews, they can provide this information to the system through the feedback function. This feedback is sent to the server and used for continuous improvement of the model and to enhance the quality of reviews.

[0486] Specific example

[0487] For example, if a consumer is looking to buy a new smartphone, they can search for product reviews for "XYZ smartphone" on their device. The user can then view highly reliable and summarized reviews displayed on their device. Summary information such as, "This smartphone has a long battery life and a good camera. However, many users find it heavy," can help them make a quicker and more informed purchasing decision.

[0488] Thus, the system of the present invention aims to improve consumer satisfaction through a process that combines reliability and convenience.

[0489] The following describes the processing flow.

[0490] Step 1:

[0491] The server uses web crawling technology to collect product reviews from multiple online sales platforms. During this process, the server searches based on specified keywords and extracts the URLs of the relevant reviews.

[0492] Step 2:

[0493] The server cleanses the collected review information. This cleansing process includes removing HTML tags and unnecessary strings, organizing the data into a text format for the reviews. This process formats the data in a way that facilitates subsequent analysis.

[0494] Step 3:

[0495] The server inputs clean data into a natural language processing model, analyzes the sentiment of each review, and generates a sentiment score. The models used here, such as BERT or logistic regression models, quantify and return whether the text is positive, negative, or neutral.

[0496] Step 4:

[0497] The server applies a machine learning model to detect fake reviews. This model is trained on previously collected data and identifies reviews that deviate from typical patterns. As a result, fake reviews are assigned a low confidence score.

[0498] Step 5:

[0499] The device receives analysis results from the server and prioritizes displaying reviews with high reliability scores to the user. When displaying reviews, it shows not only the details but also a summary, allowing the user to grasp important information quickly.

[0500] Step 6:

[0501] Users make purchasing decisions based on the displayed reviews and send feedback to the system regarding the information provided. This feedback is offered as an option in the UI and is stored on the server as raw feedback.

[0502] Step 7:

[0503] The server analyzes user feedback and uses the collected data to retrain the machine learning model. This improves the model's accuracy and helps in subsequent analyses.

[0504] (Example 1)

[0505] Next, we will describe Example 1. 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."

[0506] Traditionally, online reviews have significantly influenced buyer decisions, but this information is not always reliable. In particular, fake reviews and exaggerated self-promotional reviews can mislead consumers. To address this challenge, there is a need for a system that automatically evaluates the reliability of reviews and presents truly useful information to consumers.

[0507] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0508] In this invention, the server includes means for collecting purchase-related evaluation information on a computer network, means for analyzing the collected evaluation information using natural language processing techniques to generate sentiment evaluation values, and means for detecting non-regular evaluations using an inference model and assigning reliability evaluation values. This enables users to make purchasing decisions based on reliable review information.

[0509] A "computer network" is a system in which multiple computer devices communicate with each other to share and process data.

[0510] "Purchase-related evaluation information" refers to digital data that includes user opinions and evaluations of products and services.

[0511] "Natural language processing technology" refers to technical methods aimed at analyzing and understanding human language using computers.

[0512] An "emotional rating" is an index that quantifies the tendency of emotions and opinions within text or written data.

[0513] An "inference model" is a model that derives new inferences and decisions from data learned through machine learning algorithms.

[0514] "Irregular evaluation" refers to evaluation information that is inaccurate or intentionally created to be misleading.

[0515] A "reliability rating" is a numerical indicator assigned to show whether a particular piece of information is trustworthy.

[0516] "User terminal" refers to an electronic device used directly by the user, and typically includes computers, tablets, or smartphones.

[0517] This invention is a system that automatically evaluates the reliability of reviews based on purchase-related evaluation information collected via a computer network. This system consists of three entities: a server, a terminal, and a user, each playing a specific role to support consumers' purchasing decisions.

[0518] Server configuration and operation

[0519] The server collects evaluation information from various e-commerce platforms on the network using scraping techniques. For collection, it utilizes commonly used web scraping libraries as computer programs. The server stores this information in a logical data management system such as MySQL or PostgreSQL. The evaluation information stored in the database undergoes sentiment analysis using natural language processing techniques. This analysis utilizes machine learning frameworks on the computer platform.

[0520] The information after sentiment analysis is input into an inference model to detect non-normalized ratings. This inference model is trained using historical data and machine learning algorithms. As a result of the processing, a reliability rating is assigned to each review.

[0521] Terminal configuration and operation

[0522] The device receives analyzed information delivered from the server. It then displays the received reviews in a format easily understood by the user. Specifically, it prioritizes listing reliable review information on the screen based on reliability ratings. Furthermore, it uses natural language generation technology to generate comprehensive summary information, prompting the user to make a quick purchase decision.

[0523] User roles

[0524] Users can operate the terminal and view product reviews for the items they need. Based on the summary information displayed on the screen, users make purchasing decisions. If they have any opinions about the reviews, they can send them from the terminal to the server through the feedback function. This feedback helps to further improve the model and enhance system accuracy.

[0525] Specific examples and prompt statements

[0526] For example, if a user is looking to buy a new electronic device, they might search for "reviews of the latest model electronic device" on their device. Information with a high reliability rating from the server will be prioritized and displayed, and a summary such as "This electronic device has excellent performance and a well-received design, but it is on the expensive side" will be provided. This allows the user to make a quicker and more effective purchase decision.

[0527] An example of a prompt message would be, "I'm considering purchasing the latest model of electronic device. Please provide reliable reviews and summaries of them."

[0528] This configuration allows servers, terminals, and users to work together to provide highly accurate and useful information, thereby improving the consumer purchasing experience.

[0529] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0530] Step 1:

[0531] The server collects evaluation information from various e-commerce platforms via a computer network. It requires platform URLs and page structure information as input. Using scraping libraries such as Beautiful Soup or Scrapy, it parses the HTML and extracts review text, evaluation scores, reviewer IDs, and other relevant data. The resulting output is structured data of evaluation information.

[0532] Step 2:

[0533] The server stores the collected evaluation information in a database. The input consists of structured evaluation information, which is stored in the database using MySQL or PostgreSQL. During the storage process, each review is assigned a unique identifier to facilitate access in subsequent processing. The output is the review information stored in the database.

[0534] Step 3:

[0535] The server analyzes reviews in the database using natural language processing (NLP) techniques and generates sentiment ratings. The input is the text of the reviews retrieved from the database, which is then converted into positive, neutral, or negative scores using an NLP model. The output of this process is data with a sentiment rating attached to each review.

[0536] Step 4:

[0537] The server evaluates the reliability of reviews using an inference model. The input is review information with sentiment ratings attached. A machine learning algorithm is executed to determine if the reviews are non-normalized and generate reliability ratings. The output is the review information with the reliability ratings added.

[0538] Step 5:

[0539] The server selects and delivers review information that should be displayed with priority based on the analyzed reliability rating. The input consists of review information with reliability ratings attached, which is sent to the terminal via a RESTful API. The output is evaluation information organized according to its usefulness.

[0540] Step 6:

[0541] The terminal receives review information sent from the server and displays it to the user. The input includes analyzed evaluation information from the server, and the received data is formatted into a user-friendly format before being displayed on the screen. The output is an interface that displays highly reliable review information in an organized manner.

[0542] Step 7:

[0543] Users view the provided review information and make purchasing decisions. Input includes reviews displayed on the device; users make decisions based on these reviews and, if necessary, send feedback to the system. Output is specific actions such as purchasing decisions or submitting feedback.

[0544] (Application Example 1)

[0545] Next, we will explain Application Example 1. In the following explanation, 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."

[0546] In online marketplaces such as internet shopping, it is crucial to quickly and accurately assess the reliability of product and service reviews. Currently, consumers need to read through many reviews, and there is a risk of being misled by unreliable information. Furthermore, there is a lack of means to quickly grasp the key points, leading to delays in making purchasing decisions. In addition, methods for improving the system through feedback are not being effectively utilized.

[0547] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0548] In this invention, the server includes means for extracting evaluation information from the network, means for analyzing the extracted evaluation information using natural language processing to generate sentiment evaluation values, and means for detecting fraudulent evaluations using machine learning algorithms and assigning confidence evaluation values. This enables consumers to quickly obtain highly reliable evaluation information and supports their purchasing decisions. In addition, it makes it possible to continuously improve the system based on user feedback.

[0549] A "network" is an electronic connection method for exchanging information with each other.

[0550] "Evaluation information" refers to data that includes users' opinions and impressions of products and services.

[0551] "Natural language processing" is a technology that uses computers to analyze, understand, and generate human language.

[0552] An "emotional rating score" is an index that shows emotional tendencies such as positive and negative, extracted from evaluation information.

[0553] A "machine learning algorithm" is a set of computational procedures used to learn patterns from data and perform predictions and classifications.

[0554] "Fraudulent evaluation" refers to evaluation information that intentionally contains false information.

[0555] A "reliability rating" is a numerical value or indicator that shows the accuracy and reliability of evaluation information.

[0556] "Feedback" refers to opinions and reactions from system users and serves as a source of information for system improvement.

[0557] An "algorithm" is a series of computational steps that a computer performs to solve a problem.

[0558] In the system implementing the present invention, a server, a terminal, and a user collaborate to analyze the reliability of product reviews and provide consumers with useful information.

[0559] The server first collects evaluation information about products and services from multiple online sales platforms. This data collection can be done efficiently and securely by utilizing cloud infrastructure, such as AWS (Amazon Web Services). The collected evaluation information is analyzed on the server using natural language processing tools (e.g., NLTK and Spacy) to generate sentiment evaluation scores. This analysis objectively quantifies the emotional tendencies of the evaluation information, such as positive or negative. Furthermore, machine learning models using Scikit-learn and TensorFlow detect fraudulent evaluations and assign confidence scores.

[0560] The device receives analyzed evaluation information provided by the server. When a user views product information, the device prioritizes displaying highly reliable evaluation information. This allows users to quickly grasp important information and make quick purchasing decisions. Mobile frameworks such as React Native are used for display, and summarization and feedback functions are implemented.

[0561] Users receive and view evaluation information from the system via smartphones and other mobile devices. If they have any opinions on the system's evaluation results, they can provide this information to the server through the feedback function. This feedback is used to continuously improve the machine learning model and helps to improve the accuracy of the evaluation information.

[0562] As a concrete example, consider a scenario where a user is considering purchasing a new household appliance. When the user searches for a product on their device, highly reliable reviews are prioritized, and a short overall evaluation, such as "This household appliance is energy-efficient and quiet," is presented. In this way, the present invention aims to improve the quality of reviews and the user experience.

[0563] An example of a prompt for a generative AI model would be: "New product review: Aggregate positive, negative, and neutral opinions and calculate a reliability score."

[0564] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0565] Step 1:

[0566] The server collects review information from various sales platforms via the network. This input data includes reviews in multiple text formats. The server stores the collected review information in a database and organizes it for subsequent analysis.

[0567] Step 2:

[0568] The server processes the stored evaluation information using a natural language processing algorithm. Text data and natural language processing tools (such as NLTK or Spacy) are used as input. The analysis results in the generation of sentiment evaluation values ​​for each review, with positive, negative, and neutral sentiment tendencies output as numerical values.

[0569] Step 3:

[0570] The server detects fraudulent reviews using a machine learning model (Scikit-learn or TensorFlow). The input consists of the sentiment rating data and review text data obtained in the previous step. Based on the patterns learned by the model, it identifies fraudulent reviews and assigns a confidence rating to each review. The output is the confidence rating for each review.

[0571] Step 4:

[0572] The terminal receives analyzed evaluation information provided by the server. The input is evaluation information with a confidence score assigned to it. Based on the confidence score, the terminal prioritizes displaying the evaluation information with the highest confidence score to the user. This allows the user to obtain important information in a short amount of time.

[0573] Step 5:

[0574] Users view highly reliable rating information displayed on their devices and make purchasing decisions based on that information. Input is the rating information from the device. Users can quickly make decisions about products they are interested in and send their opinions and suggestions for improvement regarding the rating information to the server through a feedback function.

[0575] Step 6:

[0576] The server collects user feedback and uses it to improve machine learning models. The input is user feedback data. This feedback data is added to the model's training data, contributing to improved accuracy in confidence assessments.

[0577] Through this process, the system of the present invention provides a mechanism for consumers to quickly and efficiently obtain reliable evaluation information. The prompt sentence used for the generating AI model is "New product review: Aggregate positive, negative, and neutral opinions and calculate a reliability score."

[0578] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0579] The system according to the present invention consists of a server, a terminal, a user, and an emotion engine. This creates a complete system that automatically evaluates the reliability of product reviews and analyzes user sentiment, aiming to improve the consumer experience.

[0580] Server Role

[0581] The server performs web crawling to extract and collect product review information from various platforms on the network. This data is analyzed using natural language processing algorithms to generate sentiment scores for the reviews. Machine learning models are also applied to identify fake reviews and assign reliability scores.

[0582] Terminal role

[0583] The device receives analyzed data from the server and displays the information in a user-friendly format. It prioritizes reviews with high reliability scores and simultaneously displays summary information to support users in quickly acquiring information. Furthermore, it uses an emotion engine to recognize the user's emotions in real time and dynamically adjusts the displayed content based on the results.

[0584] User Roles

[0585] Users provide various emotional data through wearable devices and built-in sensors designed to recognize emotions. This data is analyzed by the device and used to understand the user's emotional state. This allows for the provision of more appropriate review information tailored to the user's current emotional state.

[0586] The role of the emotional engine

[0587] The emotion engine analyzes data obtained from the user's voice, facial expressions, or vital signs to quantify their emotional state. Based on this state data, the system adaptively adjusts the selection and display of review information. This data is also sent to the server as feedback to help improve the entire system.

[0588] Specific example

[0589] For example, when a user tries to view reviews for an "ABC laptop" on an online store, the device recognizes the user's current mood along with a reliability score, and then selects and displays the most relevant reviews. For users feeling down, uplifting, positive reviews are highlighted, while users in a hurry are shown short, summarized information. This provides a personalized service tailored to each user's situation.

[0590] Thus, the present invention provides information adaptively, taking into account the user's emotional state along with highly reliable review information. This makes it possible to create an environment in which consumers can make more confident and satisfying purchasing decisions.

[0591] The following describes the processing flow.

[0592] Step 1:

[0593] The server uses web crawling technology to collect review information from multiple network platforms on the internet. This involves specifying product names and related keywords to extract data from relevant pages. The extracted data is temporarily stored as raw, unprocessed data.

[0594] Step 2:

[0595] The server performs data cleansing on the collected review information. This process uses regular expressions to remove unnecessary HTML tags and special characters, formatting the text data. By organizing the data into a clean state, it becomes suitable for analysis by natural language processing.

[0596] Step 3:

[0597] The server sends the cleansed data to a natural language processing (NLP) system to evaluate the sentiment of each review. This system uses the BERT model to calculate contextual word sentiment vectors and generates positive, negative, or neutral sentiment scores.

[0598] Step 4:

[0599] The server uses a machine learning model to detect fake reviews. This model employs a pre-trained anomaly detection algorithm to extract reviews that exhibit patterns deviating from typical reviews. A reliability score is assigned to each review and stored in a database.

[0600] Step 5:

[0601] The terminal receives analysis results sent from the server and displays the information to the user via an interface. It prioritizes presenting reviews with high reliability scores and further uses a summarization algorithm to present the key points of the reviews to the user in abbreviated form.

[0602] Step 6:

[0603] The emotion engine is activated, detecting the user's real-time emotional state from their voice and facial expressions. This data is sent to the device and used to determine which reviews are most relevant. If the user is anxious, dynamic adjustments are made, such as highlighting positive reviews.

[0604] Step 7:

[0605] Users can view product reviews through their devices and make purchasing decisions based on information tailored to their emotions. They are also presented with options to provide feedback on the accuracy of the reviews and the system. This feedback data is sent to the server and used to improve the system.

[0606] Step 8:

[0607] The server analyzes user feedback and uses it to retrain machine learning models. This enables continuous improvement to enhance system reliability and user satisfaction.

[0608] (Example 2)

[0609] Next, we will describe Example 2. 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."

[0610] Currently, online content, including reviews and ratings, may contain fraudulent or false information, making it difficult for users to make decisions based on accurate information. Furthermore, the provision of information tailored to each user's individual emotional state is not always optimized, resulting in an unbalanced user experience.

[0611] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0612] In this invention, the server includes means for collecting evaluation texts on a network, means for analyzing the collected evaluation texts using natural language processing to generate sentiment values, and means for detecting fraudulent evaluations using machine learning methods and assigning reliability values. This makes it possible to provide users with highly reliable information while also providing personalized information tailored to the user's emotional state.

[0613] A "network" is an infrastructure that connects multiple devices and systems to each other and allows them to exchange data.

[0614] "Evaluation text" refers to text data that includes user opinions and evaluations about a product or service.

[0615] "Collecting" refers to the process of selecting and accumulating specific data.

[0616] "Natural language processing" is a technology that allows computers to understand and process human language and extract information from it.

[0617] "Emotional value" is a numerical representation of the degree of emotion derived from text.

[0618] "Machine learning methods" are techniques that enable computers to learn from data and automate specific tasks.

[0619] "Fraudulent evaluation" refers to an evaluation document that is not based on actual experience or that intentionally contains false information.

[0620] A "reliability metric" is a numerical indicator that shows the accuracy and integrity of an evaluation document.

[0621] A "human-operated device" is a device that allows users to obtain information or perform control through user interaction.

[0622] "Dynamic adjustment" means changing or adapting in real time according to the user's situation and environment.

[0623] "Emotional data" refers to information related to a user's psychological state and emotions.

[0624] "Feedback" refers to information used to improve or adjust a system or device based on the data it has obtained.

[0625] "Analyzing" is the process of deriving a detailed understanding or conclusions based on the information obtained.

[0626] "Summary" means making detailed information concise and extracting only the essential information.

[0627] The system according to the present invention consists of a server, a terminal, a user, and an emotion analysis engine.

[0628] Server configuration and processing:

[0629] The server automatically collects review texts from various sources on the internet using web crawlers. This uses libraries such as Python's Beautiful Soup and Scrapy. This data is stored in a database, and sentiment scores are calculated using natural language processing algorithms. Python's NLTK and spaCy are used for this. The server also runs machine learning models (using Scikit-learn and TensorFlow) to identify fraudulent reviews and assign a reliability score to each review.

[0630] Terminal configuration and processing:

[0631] The device uses the analyzed data received from the server to display information in a format easily understandable to the user. A web interface using HTML, CSS, and JavaScript prioritizes the display of reviews with high reliability scores and summarizes reviews using natural language processing technology. Furthermore, the device works in conjunction with a sentiment analysis engine to dynamically adjust the displayed content based on the user's real-time sentiment data.

[0632] User configuration and processing:

[0633] Users collect emotional data such as heart rate and voice tone via wearable devices like smartwatches and provide it to the device. This allows the device to analyze the user's current emotional state and select appropriate display information.

[0634] Structure and processing of the emotion analysis engine:

[0635] The emotion analysis engine quantifies emotions from the user's voice signals and facial expression data. This analysis utilizes OpenCV and voice analysis libraries. The analysis results are then used to optimize the overall information display of the system, providing content tailored to the user's individual needs. This data is returned to the server as feedback, contributing to the improvement of the machine learning model.

[0636] Examples of specific cases and prompt statements:

[0637] For example, if a user wants to view reviews of a particular product on an online platform, the device will provide information that takes the user's emotional state into consideration. In particular, if the user is feeling stressed, it will display short, easy-to-understand positive reviews.

[0638] An example of a prompt message would be: "Get reviews for ABC laptops and sort them based on sentiment and trustworthiness. If users are stressed, highlight one short, positive review."

[0639] This system aims to improve the consumer experience by providing each user with appropriately personalized evaluation information.

[0640] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0641] Step 1:

[0642] The server collects evaluation texts from information sources on the network. Given specific URLs or keywords as input, the server uses Python's Beautiful Soup or Scrapy to scrape matching evaluation texts. As output, the collected evaluation texts are stored in a database for subsequent analysis.

[0643] Step 2:

[0644] The server performs natural language processing on the collected evaluation texts to generate sentiment scores. It receives collected text data as input and analyzes the text's polarity using Python's NLTK and spaCy. Specifically, it scores the positive, negative, and neutral sentiment of each text. The output provides sentiment scores corresponding to each evaluation text, which serve as material for reliability evaluation.

[0645] Step 3:

[0646] The server executes machine learning methods to assign a reliability score to each evaluated text. As input, evaluated texts with sentiment scores attached are used, and these are passed through a model that identifies fraudulent evaluations using Scikit-learn and TensorFlow. Specifically, a model built using supervised learning determines the reliability of the evaluated text. As output, evaluations judged to be fraudulent are given a low reliability score, and evaluations recognized as legitimate are given a high reliability score.

[0647] Step 4:

[0648] The terminal receives analyzed data provided by the server and displays it to the user through a visual user interface. It receives evaluation texts sorted by reliability and sentiment scores as input. Specifically, view components created with HTML, CSS, and JavaScript are used to visualize high-priority information. The output is in a format that allows the user to easily identify the most useful information.

[0649] Step 5:

[0650] Users provide emotional data to the device using a wearable device. Inputs include heart rate and voice tone, transmitted via Bluetooth or Wi-Fi. The device then passes this data to an emotion analysis engine, which quantifies the emotional state. The analyzed emotional values ​​are then used as input for individual content adjustments.

[0651] Step 6:

[0652] The device dynamically adjusts its display content based on emotion values ​​obtained from an emotion analysis engine. The input is the user's real-time emotion values. In response, the UI content changes appropriately, providing information tailored to the user's emotional state. The output is a customized information display optimized for the user experience.

[0653] (Application Example 2)

[0654] Next, we will explain application example 2. In the following explanation, 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."

[0655] It is difficult to efficiently provide users with reliable information from the countless evaluation data available on the network. Furthermore, appropriate information is not provided according to the user's emotional state, lacking the individual adaptability necessary for purchase decision-making. Therefore, support for users in making optimal purchasing choices from a diverse range of options is insufficient.

[0656] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0657] In this invention, the server includes means for extracting evaluation information from the network, means for analyzing the extracted evaluation information using natural language processing to generate an emotion index, means for detecting false evaluations using a machine learning algorithm and assigning a reliability index, means for detecting the user's emotional state and dynamically adjusting the displayed content based on the emotional state, and means for preferentially providing useful evaluation information to the user terminal based on the reliability index. This enables the provision of individually adapted information according to the user's emotional state, and makes it possible to support optimal purchasing choices based on highly reliable information.

[0658] A "network" is a communication system in which multiple digital devices are connected to each other and can share information.

[0659] "Evaluation information" refers to user opinions, impressions, and feedback regarding a particular product or service.

[0660] "Natural language processing" is a technology that uses computers to process human language, understand its meaning, and analyze it.

[0661] "Emotional indicators" are a way of expressing a user's emotions and psychological state using numbers or categories.

[0662] A "machine learning algorithm" is a computational method that uses data to allow a computer to automatically learn and perform a specific task.

[0663] A "false review" is a review or opinion that contains false information not based on actual usage experience or facts.

[0664] A "reliability index" is a standard or measure used to evaluate the reliability and accuracy of the information provided.

[0665] "User terminal" refers to a computer device used by a user for operations and information retrieval.

[0666] "Individually adapted information provision" is a method of providing information customized to the user's specific situation and needs.

[0667] The server extracts a wide variety of evaluation information from the network. This process uses web crawling technology to collect information from various websites and platforms. The collected data is analyzed by natural language processing (NLP) algorithms to generate sentiment metrics for each review. Machine learning algorithms are also applied to detect whether the evaluation information is false and to assign reliability metrics. This prepares the system to prioritize and deliver reliable information to the user's device.

[0668] The device receives analyzed data from the server and displays the information in a format that is easy for the user to see and use. The device uses sensors such as a camera and microphone to analyze the user's emotional state in real time and adjusts the information displayed on the screen based on that analysis. This adjustment provides information optimized for the user's current emotional state.

[0669] Users access the system using smartphones or other digital devices. This allows users to obtain the most relevant information that matches their emotional state when checking reviews of desired products. For example, if a user is looking for reviews of an "ABC laptop," the device will display detailed and reliable information while summarizing it according to the user's emotional state.

[0670] As a concrete example, consider a scenario where a user is browsing reviews of an "ABC laptop" while shopping on a holiday. If the user's facial expression is detected as calm, a review including technical details will be displayed. However, if the user is detected as being in a hurry, a summarized, shorter review will be presented.

[0671] A generative AI model is used to support this process, and an example of its prompt is: "Write a Python script that personalizes and displays reviews, taking into account the user's emotional state. Analyze the user's facial expressions and voice to select relevant product reviews."

[0672] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0673] Step 1:

[0674] The server uses web crawling technology to extract evaluation information from various platforms on the network. The input is keywords for a specific product, and the output is raw data of related review information. This data is stored in a database for later processing. This process is automated using web scraping tools.

[0675] Step 2:

[0676] The server analyzes the extracted raw data using natural language processing (NLP) algorithms and assigns a sentiment index to each review. Using the extracted review information as input, it generates a sentiment score for each review as output. This process utilizes morphological analysis and sentiment dictionaries to calculate the frequency of positive and negative words contained in the reviews.

[0677] Step 3:

[0678] The server uses a machine learning algorithm to evaluate the reliability of reviews. The input is review information with sentiment scores, and the output assigns high scores to highly reliable reviews. This process utilizes a learning model based on historical data and applies an algorithm to identify false reviews.

[0679] Step 4:

[0680] The terminal receives analyzed review information along with a reliability score from the server. The input is the analyzed review information, and the output is data suitable for display. This information is displayed in the user interface and organized for easy access by the user.

[0681] Step 5:

[0682] The device uses an emotional state analysis sensor to determine the user's emotional state. Inputs include user facial expression data and voice data, and output is a real-time emotion assessment. Based on this assessment, a dynamic display algorithm is executed, which uses an NLP model to adjust the review display to suit the user's situation.

[0683] Step 6:

[0684] Users review the displayed review information on their devices and use it as a reference for their purchasing decisions. The input is the user's desired product information, and the output is a list of reliable reviews for the selected product. This allows users to efficiently obtain information that suits their emotional state.

[0685] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0686] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0687] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[0688] [Fourth Embodiment]

[0689] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0690] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0691] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0692] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[0693] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0694] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0695] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0696] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[0697] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0698] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0699] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0700] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0701] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0702] The system according to the present invention consists of a server, a terminal, and a user, and aims to automatically evaluate the reliability of product reviews over a network and provide useful information to consumers.

[0703] Server Role

[0704] The server first collects review information from various sales sites via the network. The collected data is stored and managed on the server. The server processes this data using natural language processing algorithms to analyze the text of each review. Through this analysis, it generates a sentiment score for each review, identifying whether the review is positive, neutral, or negative. Furthermore, it uses machine learning models to detect fake reviews and evaluate the reliability of each review.

[0705] Terminal role

[0706] The device receives analyzed data from the server and displays it in a format suitable for the user's actions. Specifically, it prioritizes displaying highly reliable reviews based on their reliability scores, and generates and displays review summaries so that the user can quickly grasp the main points of the reviews.

[0707] User Roles

[0708] Users view review information provided by the system through their own devices and make purchasing decisions. If they have any opinions or comments regarding the reliability of the reviews, they can provide this information to the system through the feedback function. This feedback is sent to the server and used for continuous improvement of the model and to enhance the quality of reviews.

[0709] Specific example

[0710] For example, if a consumer is looking to buy a new smartphone, they can search for product reviews for "XYZ smartphone" on their device. The user can then view highly reliable and summarized reviews displayed on their device. Summary information such as, "This smartphone has a long battery life and a good camera. However, many users find it heavy," can help them make a quicker and more informed purchasing decision.

[0711] Thus, the system of the present invention aims to improve consumer satisfaction through a process that combines reliability and convenience.

[0712] The following describes the processing flow.

[0713] Step 1:

[0714] The server uses web crawling technology to collect product reviews from multiple online sales platforms. During this process, the server searches based on specified keywords and extracts the URLs of the relevant reviews.

[0715] Step 2:

[0716] The server cleanses the collected review information. This cleansing process includes removing HTML tags and unnecessary strings, organizing the data into a text format for the reviews. This process formats the data in a way that facilitates subsequent analysis.

[0717] Step 3:

[0718] The server inputs clean data into a natural language processing model, analyzes the sentiment of each review, and generates a sentiment score. The models used here, such as BERT or logistic regression models, quantify and return whether the text is positive, negative, or neutral.

[0719] Step 4:

[0720] The server applies a machine learning model to detect fake reviews. This model is trained on previously collected data and identifies reviews that deviate from typical patterns. As a result, fake reviews are assigned a low confidence score.

[0721] Step 5:

[0722] The device receives analysis results from the server and prioritizes displaying reviews with high reliability scores to the user. When displaying reviews, it shows not only the details but also a summary, allowing the user to grasp important information quickly.

[0723] Step 6:

[0724] Users make purchasing decisions based on the displayed reviews and send feedback to the system regarding the information provided. This feedback is offered as an option in the UI and is stored on the server as raw feedback.

[0725] Step 7:

[0726] The server analyzes user feedback and uses the collected data to retrain the machine learning model. This improves the model's accuracy and helps in subsequent analyses.

[0727] (Example 1)

[0728] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0729] Traditionally, online reviews have significantly influenced buyer decisions, but this information is not always reliable. In particular, fake reviews and exaggerated self-promotional reviews can mislead consumers. To address this challenge, there is a need for a system that automatically evaluates the reliability of reviews and presents truly useful information to consumers.

[0730] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0731] In this invention, the server includes means for collecting purchase-related evaluation information on a computer network, means for analyzing the collected evaluation information using natural language processing techniques to generate sentiment evaluation values, and means for detecting non-regular evaluations using an inference model and assigning reliability evaluation values. This enables users to make purchasing decisions based on reliable review information.

[0732] A "computer network" is a system in which multiple computer devices communicate with each other to share and process data.

[0733] "Purchase-related evaluation information" refers to digital data that includes user opinions and evaluations of products and services.

[0734] "Natural language processing technology" refers to technical methods aimed at analyzing and understanding human language using computers.

[0735] An "emotional rating" is an index that quantifies the tendency of emotions and opinions within text or written data.

[0736] An "inference model" is a model that derives new inferences and decisions from data learned through machine learning algorithms.

[0737] "Irregular evaluation" refers to evaluation information that is inaccurate or intentionally created to be misleading.

[0738] A "reliability rating" is a numerical indicator assigned to show whether a particular piece of information is trustworthy.

[0739] "User terminal" refers to an electronic device used directly by the user, and typically includes computers, tablets, or smartphones.

[0740] This invention is a system that automatically evaluates the reliability of reviews based on purchase-related evaluation information collected via a computer network. This system consists of three entities: a server, a terminal, and a user, each playing a specific role to support consumers' purchasing decisions.

[0741] Server configuration and operation

[0742] The server collects evaluation information from various e-commerce platforms on the network using scraping techniques. For collection, it utilizes commonly used web scraping libraries as computer programs. The server stores this information in a logical data management system such as MySQL or PostgreSQL. The evaluation information stored in the database undergoes sentiment analysis using natural language processing techniques. This analysis utilizes machine learning frameworks on the computer platform.

[0743] The information after sentiment analysis is input into an inference model to detect non-normalized ratings. This inference model is trained using historical data and machine learning algorithms. As a result of the processing, a reliability rating is assigned to each review.

[0744] Terminal configuration and operation

[0745] The device receives analyzed information delivered from the server. It then displays the received reviews in a format easily understood by the user. Specifically, it prioritizes listing reliable review information on the screen based on reliability ratings. Furthermore, it uses natural language generation technology to generate comprehensive summary information, prompting the user to make a quick purchase decision.

[0746] User roles

[0747] Users can operate the terminal and view product reviews for the items they need. Based on the summary information displayed on the screen, users make purchasing decisions. If they have any opinions about the reviews, they can send them from the terminal to the server through the feedback function. This feedback helps to further improve the model and enhance system accuracy.

[0748] Specific examples and prompt statements

[0749] For example, if a user is looking to buy a new electronic device, they might search for "reviews of the latest model electronic device" on their device. Information with a high reliability rating from the server will be prioritized and displayed, and a summary such as "This electronic device has excellent performance and a well-received design, but it is on the expensive side" will be provided. This allows the user to make a quicker and more effective purchase decision.

[0750] An example of a prompt message would be, "I'm considering purchasing the latest model of electronic device. Please provide reliable reviews and summaries of them."

[0751] This configuration allows servers, terminals, and users to work together to provide highly accurate and useful information, thereby improving the consumer purchasing experience.

[0752] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0753] Step 1:

[0754] The server collects evaluation information from various e-commerce platforms via a computer network. It requires platform URLs and page structure information as input. Using scraping libraries such as Beautiful Soup or Scrapy, it parses the HTML and extracts review text, evaluation scores, reviewer IDs, and other relevant data. The resulting output is structured data of evaluation information.

[0755] Step 2:

[0756] The server stores the collected evaluation information in a database. The input consists of structured evaluation information, which is stored in the database using MySQL or PostgreSQL. During the storage process, each review is assigned a unique identifier to facilitate access in subsequent processing. The output is the review information stored in the database.

[0757] Step 3:

[0758] The server analyzes reviews in the database using natural language processing (NLP) techniques and generates sentiment ratings. The input is the text of the reviews retrieved from the database, which is then converted into positive, neutral, or negative scores using an NLP model. The output of this process is data with a sentiment rating attached to each review.

[0759] Step 4:

[0760] The server evaluates the reliability of reviews using an inference model. The input is review information with sentiment ratings attached. A machine learning algorithm is executed to determine if the reviews are non-normalized and generate reliability ratings. The output is the review information with the reliability ratings added.

[0761] Step 5:

[0762] The server selects and delivers review information that should be displayed with priority based on the analyzed reliability rating. The input consists of review information with reliability ratings attached, which is sent to the terminal via a RESTful API. The output is evaluation information organized according to its usefulness.

[0763] Step 6:

[0764] The terminal receives review information sent from the server and displays it to the user. The input includes analyzed evaluation information from the server, and the received data is formatted into a user-friendly format before being displayed on the screen. The output is an interface that displays highly reliable review information in an organized manner.

[0765] Step 7:

[0766] Users view the provided review information and make purchasing decisions. Input includes reviews displayed on the device; users make decisions based on these reviews and, if necessary, send feedback to the system. Output is specific actions such as purchasing decisions or submitting feedback.

[0767] (Application Example 1)

[0768] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0769] In online marketplaces such as internet shopping, it is crucial to quickly and accurately assess the reliability of product and service reviews. Currently, consumers need to read through many reviews, and there is a risk of being misled by unreliable information. Furthermore, there is a lack of means to quickly grasp the key points, leading to delays in making purchasing decisions. In addition, methods for improving the system through feedback are not being effectively utilized.

[0770] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0771] In this invention, the server includes means for extracting evaluation information from the network, means for analyzing the extracted evaluation information using natural language processing to generate sentiment evaluation values, and means for detecting fraudulent evaluations using machine learning algorithms and assigning confidence evaluation values. This enables consumers to quickly obtain highly reliable evaluation information and supports their purchasing decisions. In addition, it makes it possible to continuously improve the system based on user feedback.

[0772] A "network" is an electronic connection method for exchanging information with each other.

[0773] "Evaluation information" refers to data that includes users' opinions and impressions of products and services.

[0774] "Natural language processing" is a technology that uses computers to analyze, understand, and generate human language.

[0775] An "emotional rating score" is an index that shows emotional tendencies such as positive and negative, extracted from evaluation information.

[0776] A "machine learning algorithm" is a set of computational procedures used to learn patterns from data and perform predictions and classifications.

[0777] "Fraudulent evaluation" refers to evaluation information that intentionally contains false information.

[0778] A "reliability rating" is a numerical value or indicator that shows the accuracy and reliability of evaluation information.

[0779] "Feedback" refers to opinions and reactions from system users and serves as a source of information for system improvement.

[0780] An "algorithm" is a series of computational steps that a computer performs to solve a problem.

[0781] In the system implementing the present invention, a server, a terminal, and a user collaborate to analyze the reliability of product reviews and provide consumers with useful information.

[0782] The server first collects evaluation information about products and services from multiple online sales platforms. This data collection can be done efficiently and securely by utilizing cloud infrastructure, such as AWS (Amazon Web Services). The collected evaluation information is analyzed on the server using natural language processing tools (e.g., NLTK and Spacy) to generate sentiment evaluation scores. This analysis objectively quantifies the emotional tendencies of the evaluation information, such as positive or negative. Furthermore, machine learning models using Scikit-learn and TensorFlow detect fraudulent evaluations and assign confidence scores.

[0783] The device receives analyzed evaluation information provided by the server. When a user views product information, the device prioritizes displaying highly reliable evaluation information. This allows users to quickly grasp important information and make quick purchasing decisions. Mobile frameworks such as React Native are used for display, and summarization and feedback functions are implemented.

[0784] Users receive and view evaluation information from the system via smartphones and other mobile devices. If they have any opinions on the system's evaluation results, they can provide this information to the server through the feedback function. This feedback is used to continuously improve the machine learning model and helps to improve the accuracy of the evaluation information.

[0785] As a concrete example, consider a scenario where a user is considering purchasing a new household appliance. When the user searches for a product on their device, highly reliable reviews are prioritized, and a short overall evaluation, such as "This household appliance is energy-efficient and quiet," is presented. In this way, the present invention aims to improve the quality of reviews and the user experience.

[0786] An example of a prompt for a generative AI model would be: "New product review: Aggregate positive, negative, and neutral opinions and calculate a reliability score."

[0787] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0788] Step 1:

[0789] The server collects review information from various sales platforms via the network. This input data includes reviews in multiple text formats. The server stores the collected review information in a database and organizes it for subsequent analysis.

[0790] Step 2:

[0791] The server processes the stored evaluation information using a natural language processing algorithm. Text data and natural language processing tools (such as NLTK or Spacy) are used as input. The analysis results in the generation of sentiment evaluation values ​​for each review, with positive, negative, and neutral sentiment tendencies output as numerical values.

[0792] Step 3:

[0793] The server detects fraudulent reviews using a machine learning model (Scikit-learn or TensorFlow). The input consists of the sentiment rating data and review text data obtained in the previous step. Based on the patterns learned by the model, it identifies fraudulent reviews and assigns a confidence rating to each review. The output is the confidence rating for each review.

[0794] Step 4:

[0795] The terminal receives analyzed evaluation information provided by the server. The input is evaluation information with a confidence score assigned to it. Based on the confidence score, the terminal prioritizes displaying the evaluation information with the highest confidence score to the user. This allows the user to obtain important information in a short amount of time.

[0796] Step 5:

[0797] Users view highly reliable rating information displayed on their devices and make purchasing decisions based on that information. Input is the rating information from the device. Users can quickly make decisions about products they are interested in and send their opinions and suggestions for improvement regarding the rating information to the server through a feedback function.

[0798] Step 6:

[0799] The server collects user feedback and uses it to improve machine learning models. The input is user feedback data. This feedback data is added to the model's training data, contributing to improved accuracy in confidence assessments.

[0800] Through this process, the system of the present invention provides a mechanism for consumers to quickly and efficiently obtain reliable evaluation information. The prompt sentence used for the generating AI model is "New product review: Aggregate positive, negative, and neutral opinions and calculate a reliability score."

[0801] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0802] The system according to the present invention consists of a server, a terminal, a user, and an emotion engine. This creates a complete system that automatically evaluates the reliability of product reviews and analyzes user sentiment, aiming to improve the consumer experience.

[0803] Server Role

[0804] The server performs web crawling to extract and collect product review information from various platforms on the network. This data is analyzed using natural language processing algorithms to generate sentiment scores for the reviews. Machine learning models are also applied to identify fake reviews and assign reliability scores.

[0805] Terminal role

[0806] The device receives analyzed data from the server and displays the information in a user-friendly format. It prioritizes reviews with high reliability scores and simultaneously displays summary information to support users in quickly acquiring information. Furthermore, it uses an emotion engine to recognize the user's emotions in real time and dynamically adjusts the displayed content based on the results.

[0807] User Roles

[0808] Users provide various emotional data through wearable devices and built-in sensors designed to recognize emotions. This data is analyzed by the device and used to understand the user's emotional state. This allows for the provision of more appropriate review information tailored to the user's current emotional state.

[0809] The role of the emotional engine

[0810] The emotion engine analyzes data obtained from the user's voice, facial expressions, or vital signs to quantify their emotional state. Based on this state data, the system adaptively adjusts the selection and display of review information. This data is also sent to the server as feedback to help improve the entire system.

[0811] Specific example

[0812] For example, when a user tries to view reviews for an "ABC laptop" on an online store, the device recognizes the user's current mood along with a reliability score, and then selects and displays the most relevant reviews. For users feeling down, uplifting, positive reviews are highlighted, while users in a hurry are shown short, summarized information. This provides a personalized service tailored to each user's situation.

[0813] Thus, the present invention provides information adaptively, taking into account the user's emotional state along with highly reliable review information. This makes it possible to create an environment in which consumers can make more confident and satisfying purchasing decisions.

[0814] The following describes the processing flow.

[0815] Step 1:

[0816] The server uses web crawling technology to collect review information from multiple network platforms on the internet. This involves specifying product names and related keywords to extract data from relevant pages. The extracted data is temporarily stored as raw, unprocessed data.

[0817] Step 2:

[0818] The server performs data cleansing on the collected review information. This process uses regular expressions to remove unnecessary HTML tags and special characters, formatting the text data. By organizing the data into a clean state, it becomes suitable for analysis by natural language processing.

[0819] Step 3:

[0820] The server sends the cleansed data to a natural language processing (NLP) system to evaluate the sentiment of each review. This system uses the BERT model to calculate contextual word sentiment vectors and generates positive, negative, or neutral sentiment scores.

[0821] Step 4:

[0822] The server uses a machine learning model to detect fake reviews. This model employs a pre-trained anomaly detection algorithm to extract reviews that exhibit patterns deviating from typical reviews. A reliability score is assigned to each review and stored in a database.

[0823] Step 5:

[0824] The terminal receives analysis results sent from the server and displays the information to the user via an interface. It prioritizes presenting reviews with high reliability scores and further uses a summarization algorithm to present the key points of the reviews to the user in abbreviated form.

[0825] Step 6:

[0826] The emotion engine is activated, detecting the user's real-time emotional state from their voice and facial expressions. This data is sent to the device and used to determine which reviews are most relevant. If the user is anxious, dynamic adjustments are made, such as highlighting positive reviews.

[0827] Step 7:

[0828] Users can view product reviews through their devices and make purchasing decisions based on information tailored to their emotions. They are also presented with options to provide feedback on the accuracy of the reviews and the system. This feedback data is sent to the server and used to improve the system.

[0829] Step 8:

[0830] The server analyzes user feedback and uses it to retrain machine learning models. This enables continuous improvement to enhance system reliability and user satisfaction.

[0831] (Example 2)

[0832] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0833] Currently, online content, including reviews and ratings, may contain fraudulent or false information, making it difficult for users to make decisions based on accurate information. Furthermore, the provision of information tailored to each user's individual emotional state is not always optimized, resulting in an unbalanced user experience.

[0834] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0835] In this invention, the server includes means for collecting evaluation texts on a network, means for analyzing the collected evaluation texts using natural language processing to generate sentiment values, and means for detecting fraudulent evaluations using machine learning methods and assigning reliability values. This makes it possible to provide users with highly reliable information while also providing personalized information tailored to the user's emotional state.

[0836] A "network" is an infrastructure that connects multiple devices and systems to each other and allows them to exchange data.

[0837] "Evaluation text" refers to text data that includes user opinions and evaluations about a product or service.

[0838] "Collecting" refers to the process of selecting and accumulating specific data.

[0839] "Natural language processing" is a technology that allows computers to understand and process human language and extract information from it.

[0840] "Emotional value" is a numerical representation of the degree of emotion derived from text.

[0841] "Machine learning methods" are techniques that enable computers to learn from data and automate specific tasks.

[0842] "Fraudulent evaluation" refers to an evaluation document that is not based on actual experience or that intentionally contains false information.

[0843] A "reliability metric" is a numerical indicator that shows the accuracy and integrity of an evaluation document.

[0844] A "human-operated device" is a device that allows users to obtain information or perform control through user interaction.

[0845] "Dynamic adjustment" means changing or adapting in real time according to the user's situation and environment.

[0846] "Emotional data" refers to information related to a user's psychological state and emotions.

[0847] "Feedback" refers to information used to improve or adjust a system or device based on the data it has obtained.

[0848] "Analyzing" is the process of deriving a detailed understanding or conclusions based on the information obtained.

[0849] "Summary" means making detailed information concise and extracting only the essential information.

[0850] The system according to the present invention consists of a server, a terminal, a user, and an emotion analysis engine.

[0851] Server configuration and processing:

[0852] The server automatically collects review texts from various sources on the internet using web crawlers. This uses libraries such as Python's Beautiful Soup and Scrapy. This data is stored in a database, and sentiment scores are calculated using natural language processing algorithms. Python's NLTK and spaCy are used for this. The server also runs machine learning models (using Scikit-learn and TensorFlow) to identify fraudulent reviews and assign a reliability score to each review.

[0853] Terminal configuration and processing:

[0854] The device uses the analyzed data received from the server to display information in a format easily understandable to the user. A web interface using HTML, CSS, and JavaScript prioritizes the display of reviews with high reliability scores and summarizes reviews using natural language processing technology. Furthermore, the device works in conjunction with a sentiment analysis engine to dynamically adjust the displayed content based on the user's real-time sentiment data.

[0855] User configuration and processing:

[0856] Users collect emotional data such as heart rate and voice tone via wearable devices like smartwatches and provide it to the device. This allows the device to analyze the user's current emotional state and select appropriate display information.

[0857] Structure and processing of the emotion analysis engine:

[0858] The emotion analysis engine quantifies emotions from the user's voice signals and facial expression data. This analysis utilizes OpenCV and voice analysis libraries. The analysis results are then used to optimize the overall information display of the system, providing content tailored to the user's individual needs. This data is returned to the server as feedback, contributing to the improvement of the machine learning model.

[0859] Examples of specific cases and prompt statements:

[0860] For example, if a user wants to view reviews of a particular product on an online platform, the device will provide information that takes the user's emotional state into consideration. In particular, if the user is feeling stressed, it will display short, easy-to-understand positive reviews.

[0861] An example of a prompt message would be: "Get reviews for ABC laptops and sort them based on sentiment and trustworthiness. If users are stressed, highlight one short, positive review."

[0862] This system aims to improve the consumer experience by providing each user with appropriately personalized evaluation information.

[0863] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0864] Step 1:

[0865] The server collects evaluation texts from information sources on the network. Given specific URLs or keywords as input, the server uses Python's Beautiful Soup or Scrapy to scrape matching evaluation texts. As output, the collected evaluation texts are stored in a database for subsequent analysis.

[0866] Step 2:

[0867] The server performs natural language processing on the collected evaluation texts to generate sentiment scores. It receives collected text data as input and analyzes the text's polarity using Python's NLTK and spaCy. Specifically, it scores the positive, negative, and neutral sentiment of each text. The output provides sentiment scores corresponding to each evaluation text, which serve as material for reliability evaluation.

[0868] Step 3:

[0869] The server executes machine learning methods to assign a reliability score to each evaluated text. As input, evaluated texts with sentiment scores attached are used, and these are passed through a model that identifies fraudulent evaluations using Scikit-learn and TensorFlow. Specifically, a model built using supervised learning determines the reliability of the evaluated text. As output, evaluations judged to be fraudulent are given a low reliability score, and evaluations recognized as legitimate are given a high reliability score.

[0870] Step 4:

[0871] The terminal receives analyzed data provided by the server and displays it to the user through a visual user interface. It receives evaluation texts sorted by reliability and sentiment scores as input. Specifically, view components created with HTML, CSS, and JavaScript are used to visualize high-priority information. The output is in a format that allows the user to easily identify the most useful information.

[0872] Step 5:

[0873] Users provide emotional data to the device using a wearable device. Inputs include heart rate and voice tone, transmitted via Bluetooth or Wi-Fi. The device then passes this data to an emotion analysis engine, which quantifies the emotional state. The analyzed emotional values ​​are then used as input for individual content adjustments.

[0874] Step 6:

[0875] The device dynamically adjusts its display content based on emotion values ​​obtained from an emotion analysis engine. The input is the user's real-time emotion values. In response, the UI content changes appropriately, providing information tailored to the user's emotional state. The output is a customized information display optimized for the user experience.

[0876] (Application Example 2)

[0877] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[0878] It is difficult to efficiently provide users with reliable information from the countless evaluation data available on the network. Furthermore, appropriate information is not provided according to the user's emotional state, lacking the individual adaptability necessary for purchase decision-making. Therefore, support for users in making optimal purchasing choices from a diverse range of options is insufficient.

[0879] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0880] In this invention, the server includes means for extracting evaluation information from the network, means for analyzing the extracted evaluation information using natural language processing to generate an emotion index, means for detecting false evaluations using a machine learning algorithm and assigning a reliability index, means for detecting the user's emotional state and dynamically adjusting the displayed content based on the emotional state, and means for preferentially providing useful evaluation information to the user terminal based on the reliability index. This enables the provision of individually adapted information according to the user's emotional state, and makes it possible to support optimal purchasing choices based on highly reliable information.

[0881] A "network" is a communication system in which multiple digital devices are connected to each other and can share information.

[0882] "Evaluation information" refers to user opinions, impressions, and feedback regarding a particular product or service.

[0883] "Natural language processing" is a technology that uses computers to process human language, understand its meaning, and analyze it.

[0884] "Emotional indicators" are a way of expressing a user's emotions and psychological state using numbers or categories.

[0885] A "machine learning algorithm" is a computational method that uses data to allow a computer to automatically learn and perform a specific task.

[0886] A "false review" is a review or opinion that contains false information not based on actual usage experience or facts.

[0887] A "reliability index" is a standard or measure used to evaluate the reliability and accuracy of the information provided.

[0888] "User terminal" refers to a computer device used by a user for operations and information retrieval.

[0889] "Individually adapted information provision" is a method of providing information customized to the user's specific situation and needs.

[0890] The server extracts a wide variety of evaluation information from the network. This process uses web crawling technology to collect information from various websites and platforms. The collected data is analyzed by natural language processing (NLP) algorithms to generate sentiment metrics for each review. Machine learning algorithms are also applied to detect whether the evaluation information is false and to assign reliability metrics. This prepares the system to prioritize and deliver reliable information to the user's device.

[0891] The device receives analyzed data from the server and displays the information in a format that is easy for the user to see and use. The device uses sensors such as a camera and microphone to analyze the user's emotional state in real time and adjusts the information displayed on the screen based on that analysis. This adjustment provides information optimized for the user's current emotional state.

[0892] Users access the system using smartphones or other digital devices. This allows users to obtain the most relevant information that matches their emotional state when checking reviews of desired products. For example, if a user is looking for reviews of an "ABC laptop," the device will display detailed and reliable information while summarizing it according to the user's emotional state.

[0893] As a concrete example, consider a scenario where a user is browsing reviews of an "ABC laptop" while shopping on a holiday. If the user's facial expression is detected as calm, a review including technical details will be displayed. However, if the user is detected as being in a hurry, a summarized, shorter review will be presented.

[0894] A generative AI model is used to support this process, and an example of its prompt is: "Write a Python script that personalizes and displays reviews, taking into account the user's emotional state. Analyze the user's facial expressions and voice to select relevant product reviews."

[0895] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0896] Step 1:

[0897] The server uses web crawling technology to extract evaluation information from various platforms on the network. The input is keywords for a specific product, and the output is raw data of related review information. This data is stored in a database for later processing. This process is automated using web scraping tools.

[0898] Step 2:

[0899] The server analyzes the extracted raw data using natural language processing (NLP) algorithms and assigns a sentiment index to each review. Using the extracted review information as input, it generates a sentiment score for each review as output. This process utilizes morphological analysis and sentiment dictionaries to calculate the frequency of positive and negative words contained in the reviews.

[0900] Step 3:

[0901] The server uses a machine learning algorithm to evaluate the reliability of reviews. The input is review information with sentiment scores, and the output assigns high scores to highly reliable reviews. This process utilizes a learning model based on historical data and applies an algorithm to identify false reviews.

[0902] Step 4:

[0903] The terminal receives analyzed review information along with a reliability score from the server. The input is the analyzed review information, and the output is data suitable for display. This information is displayed in the user interface and organized for easy access by the user.

[0904] Step 5:

[0905] The device uses an emotional state analysis sensor to determine the user's emotional state. Inputs include user facial expression data and voice data, and output is a real-time emotion assessment. Based on this assessment, a dynamic display algorithm is executed, which uses an NLP model to adjust the review display to suit the user's situation.

[0906] Step 6:

[0907] Users review the displayed review information on their devices and use it as a reference for their purchasing decisions. The input is the user's desired product information, and the output is a list of reliable reviews for the selected product. This allows users to efficiently obtain information that suits their emotional state.

[0908] 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 controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0909] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0910] 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 this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[0911] Furthermore, the emotion identification model 59, acting 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 a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[0912] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[0913] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[0914] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[0915] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[0916] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is 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 the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[0917] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[0918] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[0919] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[0920] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[0921] 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.

[0922] Furthermore, it is not necessary to store the entirety 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 the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[0923] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[0924] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of 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). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[0925] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[0926] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[0927] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[0928] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted as being incorporated by reference.

[0929] The following is further disclosed regarding the embodiments described above.

[0930] (Claim 1)

[0931] A means of extracting review information from the network,

[0932] A method for analyzing extracted review information using natural language processing and generating sentiment scores,

[0933] A method for detecting fake reviews and assigning reliability scores using machine learning models,

[0934] A means of providing user terminals with priority access to useful review information based on a reliability score,

[0935] A system that includes this.

[0936] (Claim 2)

[0937] The system according to claim 1, further comprising means for collecting user feedback and using it to improve a machine learning model.

[0938] (Claim 3)

[0939] The system according to claim 1, further comprising means for summarizing the analyzed review information and displaying it to the user.

[0940] "Example 1"

[0941] (Claim 1)

[0942] A means for collecting purchase-related evaluation information on a computer network,

[0943] A means for analyzing collected evaluation information using natural language processing technology to generate sentiment evaluation values,

[0944] A means for detecting non-normalized evaluations using an inference model and assigning reliability evaluation values,

[0945] A means for providing user terminals with priority-based and useful evaluation information based on reliability evaluation values,

[0946] A means of summarizing the analyzed evaluation information and displaying it to the user,

[0947] A system that includes this.

[0948] (Claim 2)

[0949] The system according to claim 1, further comprising means for collecting user response information and using it to improve an inference model.

[0950] (Claim 3)

[0951] The system according to claim 1, further comprising means for summarizing the analyzed evaluation information and displaying it to the user.

[0952] "Application Example 1"

[0953] (Claim 1)

[0954] A means for extracting evaluation information from the network,

[0955] A means for analyzing extracted evaluation information using natural language processing to generate sentiment evaluation values,

[0956] A means of detecting fraudulent evaluations using a machine learning algorithm and assigning a confidence score,

[0957] A means of providing user terminals with priority-based and useful evaluation information based on reliability evaluation values,

[0958] A means of summarizing the analyzed evaluation information and presenting it to the user,

[0959] A means of collecting user feedback and using it to improve machine learning algorithms,

[0960] A system that includes this.

[0961] (Claim 2)

[0962] The system according to claim 1, which presents evaluation information in real time using a display means.

[0963] (Claim 3)

[0964] The system according to claim 1, further comprising means for analyzing user evaluations and automatically generating key points.

[0965] "Example 2 of combining an emotion engine"

[0966] (Claim 1)

[0967] A means of collecting evaluation documents on the network,

[0968] A means for analyzing collected evaluation texts using natural language processing to generate sentiment values,

[0969] A means of detecting fraudulent evaluations using machine learning methods and assigning reliability scores,

[0970] A means for providing a human-operated device with preferentially useful evaluation texts based on reliability figures,

[0971] A means for dynamically adjusting the displayed content based on emotional data provided via a human-operated device,

[0972] A system that includes this.

[0973] (Claim 2)

[0974] The system according to claim 1, further comprising means for collecting emotional data from a human-operated device and understanding the user's emotional state through analysis.

[0975] (Claim 3)

[0976] The system according to claim 1, further comprising means for summarizing the analyzed evaluation text and displaying it on a human-operated device.

[0977] "Application example 2 when combining with an emotional engine"

[0978] (Claim 1)

[0979] A means for extracting evaluation information from the network,

[0980] A means for analyzing extracted evaluation information using natural language processing to generate sentiment indicators,

[0981] A means for detecting false evaluations using machine learning algorithms and assigning reliability indicators,

[0982] A means for detecting the user's emotional state and dynamically adjusting the displayed content based on that emotional state,

[0983] A means of providing user terminals with priority-based and useful evaluation information based on reliability indicators,

[0984] A system that includes this.

[0985] (Claim 2)

[0986] The system according to claim 1, further comprising means for collecting user feedback information and using it to improve machine learning algorithms.

[0987] (Claim 3)

[0988] The system according to claim 1, further comprising means for summarizing the analyzed evaluation information and displaying it to the user. [Explanation of Symbols]

[0989] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. A means for extracting evaluation information from the network, A means for analyzing extracted evaluation information using natural language processing to generate sentiment evaluation values, A means of detecting fraudulent evaluations using a machine learning algorithm and assigning a confidence score, A means of providing user terminals with priority-based and useful evaluation information based on reliability evaluation values, A means of summarizing the analyzed evaluation information and presenting it to the user, A means of collecting user feedback and using it to improve machine learning algorithms, A system that includes this.

2. The system according to claim 1, wherein evaluation information is presented in real time by a display means.

3. The system according to claim 1, further comprising means for analyzing user evaluations and automatically generating key points.

Images