System

Abstract

To provide a system.SOLUTION: A system comprises: means for receiving a comment from a user; an emotion engine which recognizes emotion of the user providing the comment; determination means for calculating a score indicating whether the user has positive or negative emotion based on the recognized emotion and determines whether the comment is aggressive based on the score; conversion means for converting the aggressive comment to a specific comment by inputting a predetermined prompt text into a generative AI model when the comment is determined to be aggressive; and means for displaying the converted comment on a terminal of the user.SELECTED DRAWING: Figure 1

Description

[Technical field]

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

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including a description and related instruction sentence regarding the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2022-180282 A Summary of the Invention [Problem to be solved by the invention]

[0004] There is a problem that aggressive comments from users can hurt the mental health of service providers, which can lead to them being less likely to respond to user feedback. [Means for solving the problem]

[0005] The system of the present disclosure includes a means for receiving comments from a user, an emotion engine for recognizing the emotions of the user who made the comment, a determination means for calculating a score indicating whether the user has positive or negative emotions based on the emotion and determining whether the comment is aggressive or not based on the score, a conversion means for converting the aggressive comment into a specific comment by inputting a specific prompt sentence into a generative AI model if the comment is aggressive, and a means for displaying the converted comment on the user's terminal. [Brief description of the drawings]

[0006] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Diagram 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. FIG. [Diagram 3] FIG. 11 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Diagram 5] FIG. 13 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 13 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 13 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 2 is a sequence diagram showing a process flow of the data processing system in the first embodiment of the first form example. [Figure 12] FIG. 11 is a sequence diagram showing a process flow of the data processing system in the application example 1 of the first embodiment. [Figure 13] FIG. 11 is a sequence diagram showing a process flow of the data processing system in the second embodiment of the second embodiment. [Figure 14] FIG. 13 is a sequence diagram showing a process flow of the data processing system in application example 2 of embodiment 2. [Figure 15] FIG. 13 is a sequence diagram showing a process flow of a data processing system in a third embodiment of the present invention. [Figure 16] FIG. 13 is a sequence diagram showing a process flow of the data processing system in the third application example of the third embodiment. [Figure 17] FIG. 13 is a sequence diagram showing the flow of processing in the data processing system in the first embodiment of the first form example when an emotion engine is combined. [Figure 18] FIG. 13 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1 of Form Example 1 when combined with an emotion engine. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0007] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0008] First, the terms used in the following description will be explained.

[0009] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be one arithmetic device or a combination of multiple arithmetic devices. The processor may be one type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit),

[0010] Examples of such processors include a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), or a TPU (TENSOR PROCESSING UNIT (registered trademark)).

[0011] In the following embodiments, a signed RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by the processor.

[0012] In the following embodiments, the storage with the code is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g.,

[0013] Examples include hard disks, or magnetic tapes.

[0014] In the following embodiments, a communication I / F (Interface) with a code is an interface including a communication processor and an antenna. The communication I / F controls communication between multiple computers. An example of a communication standard applied to the communication I / F is a wireless communication standard including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0015] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. In addition, in this specification, the same idea as "A and / or B" is also applied when three or more things are expressed by connecting them with "and / or."

[0016] [First embodiment]

[0017] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0018] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0019] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a wide area network (WAN) and / or a local area network (LAN).

[0020] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0021] The reception device 38 includes a touch panel 38A and a microphone 38B, and receives user input. The touch panel 38A detects contact with a pointer (e.g., a pen or a finger) to receive user input by the touch of the pointer. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0022] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (e.g., voice 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 voice according to instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a Complementary Metal-Oxide-Semiconductor (CMOS) image sensor or a Charge Coupled Device (CCD) image sensor.

[0023] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for transmitting and receiving various types of information between the processor 46 and the processor 28 via the network 54.

[0024] FIG. 2 shows an example of main functions of the data processing device 12 and the smart device 14.

[0025] As shown in Fig. 2, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32. The specific process program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific process program 56 from the storage 32, and executes the read specific process program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific process program 56 executed on the RAM 30.

[0026] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0027] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores a reception output program 60. The reception output program 60 is used together with the specific processing program 56 by the data processing system 10. The processor 46 reads out 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 the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0028] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0029] "Example 1"

[0030] As one embodiment of the present invention, a feedback form on a website or application, a comment section on a social networking site, or the like is used as a means for receiving comments from users. As a means for determining whether a received comment is aggressive, natural language processing technology or machine learning is used to analyze the content of the comment and determine its tone and emotion. Then, as a means for converting aggressive comments into positive and kind words, a generation API is used to convert the expression of the comment while keeping its content. Specifically, "Your service is the worst" is converted to "Your service has room for improvement."

[0031] "Example 2"

[0032] As one embodiment of the present invention, a feedback form on a website or application, a comment section on a social networking site, or the like is used as a means for receiving comments from users. As a means for determining whether a received comment is aggressive, natural language processing technology or machine learning is used to analyze the content of the comment and determine its tone and emotion. Then, as a means for converting aggressive comments into positive and kind words, a generation API is used to convert the expression of the comment while keeping its content. Specifically, "Your service is the worst" is converted to "Your service has room for improvement."

[0033] The process flow of each embodiment will be described below.

[0034] "Example 1"

[0035] Step 1: Receive comments from users. In this step, you collect comments from users through feedback forms on your website or application, or through comments sections on social media.

[0036] Step 2: Determine whether the comment received is offensive. In this step, natural language processing technology and machine learning are used to analyze the content of the comment and determine its tone and sentiment. For example, a comment like "Your service is the worst" would be deemed offensive.

[0037] Step 3: Convert aggressive comments to positive and kind words. In this step, we use the generation API to convert the expression of the comment while preserving its content. For example, we convert a comment such as "Your service is the worst" to "Your service has room for improvement."

[0038] Example 1

[0039] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal".

[0040] In communication on the Internet, if a user's comment is aggressive, it may cause discomfort to other users and service providers. There is a demand for providing a healthy communication environment by converting such aggressive comments into positive and kind words. With conventional systems, it was difficult to properly detect aggressive comments and convert them into positive expressions while preserving their content.

[0041] The identification process by the identification processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means. The system of the present disclosure includes a means for receiving a comment from a user, an emotion engine for recognizing the emotion of the user who made the comment, a determination means for calculating a score indicating whether the user has a positive emotion or a negative emotion based on the emotion and determining whether the comment is aggressive or not based on the score, a conversion means for converting the aggressive comment into a specific comment by inputting a specific prompt sentence into a generative AI model when the comment is aggressive, and a means for displaying the converted comment on the user's terminal. The conversion means may convert the expression of the comment while maintaining the content of the comment from the user. The conversion means may convert an aggressive comment into a comment with a positive and gentle expression.

[0042] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for displaying the converted comments to the user. This makes it possible to appropriately detect offensive comments and convert them into positive expressions while preserving their contents.

[0043] "User" refers to an individual or organization that uses the system to enter comments.

[0044] "Comment" refers to any text information that a User enters into a feedback form on a Website or Application, or into a comment section on a social networking site.

[0045] "Server" refers to a computer system that receives, analyzes, and converts comments from users.

[0046] "Means for receiving" refers to a function that enables the server to receive comments sent by users.

[0047] "Means of determining" refers to the function for analyzing and evaluating whether a received comment is offensive.

[0048] "Means of transformation" refers to the function of transforming aggressive comments into positive, kind words.

[0049] "Means for displaying" refers to a function for displaying the converted comment to the user.

[0050] "Offensive comments" refers to negative comments that may cause offense to other users or service providers.

[0051] "Positive and kind words" refers to words that preserve the content of aggressive comments but transform them into more constructive and friendly language.

[0052] "Generative AI Model" refers to an artificial intelligence model that generates and transforms text using natural language processing techniques.

[0053] A "prompt" refers to an instruction entered into a generative AI model to cause it to perform a specific task.

[0054] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0055] The present invention relates to a system that receives comments from users, determines whether the comments are offensive, and converts the offensive comments into positive and kind words. A specific embodiment of this system will be described below.

[0056] Hardware and Software Used

[0057] The servers include a web server and an application server for receiving comments from users. These servers receive comment data through HTTP requests. The servers analyze the comments using Python natural language processing libraries NLTK and spaCy. They also use a generative AI model (e.g., GPT-3 (registered trademark) from OpenAI (registered trademark)) to convert aggressive comments into positive expressions.

[0058] Program Processing

[0059] A user enters a comment into a feedback form on a website or application, or into a comment section on a social networking site. For example, a user might enter "your service is the worst." When a user submits a comment, the server receives the comment. The server receives the comment data through an HTTP request.

[0060] The server then uses a natural language processing library, such as Python's NLTK or spaCy, to parse the comments it receives. The server analyzes the tone and sentiment of the comment to determine whether it is offensive or not. For example, it might recognize the word "terrible" as offensive.

[0061] The server will determine whether the comment is offensive based on the analysis results. If it is determined to be offensive, it will proceed to the next step. For example, a comment like "Your service is the worst" will be detected as offensive.

[0062] The server uses a generative AI model (e.g., OpenAI's GPT-3) to convert aggressive comments into positive ones. The server sends a prompt to the generative AI model, which converts the comment's expression while preserving its content. For example, the comment "Your service is terrible" is converted to "Your service has room for improvement."

[0063] Specific prompt examples:

[0064] Original comment: "Your service is terrible"

[0065] Prompt: "Transform the following offensive comment into a positive, kind statement: 'Your service is awful.'"

[0066] Finally, the server displays the converted comments received from the generative AI model to the user. The user can see the converted comments in the feedback form or in the comments section of a social networking site. For example, a comment such as "Your service has room for improvement" can be displayed.

[0067] In this way, a system is realized that converts aggressive comments from users into positive expressions.

[0068] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0069] Step 1:

[0070] The user enters a comment

[0071] A user enters a comment into a feedback form on a website or application, or into a comment section on a social networking site. For example, a user enters "Your service is the worst." The comment is sent to the server by pressing the send button.

[0072] Step 2:

[0073] The server receives the comment.

[0074] The server receives comments sent by users through HTTP requests. The received comments are stored in a database or temporarily held in memory. The input is the user's comments, and the output is the comment data stored on the server.

[0075] Step 3:

[0076] The server parses the comments

[0077] The server uses Python natural language processing libraries NLTK and spaCy to analyze the received comments. Specifically, it performs tokenization of comments, part-of-speech tagging, sentiment analysis, etc. The input is the comment data stored on the server, and the output is the analysis results (e.g., the tone and sentiment of the comments).

[0078] Step 4:

[0079] The server detects offensive comments

[0080] The server determines whether the comment is offensive based on the analysis result. If it is determined to be offensive, it proceeds to the next step. For example, if the comment contains the word "worst", it is determined to be offensive. The input is the analysis result, and the output is the detection result of the offensive comment.

[0081] Step 5:

[0082] The server converts the comments

[0083] The server uses a generative AI model (e.g., OpenAI's GPT-3) to convert offensive comments into positive expressions. The server sends a prompt to the generative AI model, which converts the expression while preserving the content of the comment. For example, the comment "Your service is terrible" is converted to "Your service could be improved." The input is the offensive comment and the prompt, and the output is the converted comment.

[0084] Specific prompt examples:

[0085] Original comment: "Your service is terrible"

[0086] Prompt: "Transform the following offensive comment into a positive, kind statement: 'Your service is awful.'"

[0087] Step 6:

[0088] The server displays the converted comment to the user.

[0089] The server displays the converted comments received from the generative AI model to the user. The user can see the converted comments in a feedback form or in a comment section on a social networking site. For example, a comment such as "Your service has room for improvement" is displayed. The input is the converted comment, and the output is the comment displayed to the user.

[0090] (Application example 1)

[0091] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0092] In conventional customer support systems, aggressive comments from users are sent directly to the support staff, which often causes stress for the support staff. In addition, it is difficult to respond appropriately to aggressive comments, which makes communication with users less smooth.

[0093] The identification process by the identification processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for providing the converted comments to a customer support representative. This enables the customer support representative to receive constructive feedback without feeling stressed.

[0094] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[0095] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[0096] A "means of transforming aggressive comments into positive and kind words" is a technique for transforming aggressive comments into positive and kind words while preserving their content.

[0097] The "means for providing the converted comments to the customer support representative" refers to an interface for displaying or notifying the converted comments to the customer support representative.

[0098] "Generation API" means an application programming interface for generating text using a generative AI model.

[0099] A "customer support representative" is a person who handles inquiries and feedback from users.

[0100] A system for implementing the present invention includes means for receiving comments from a user, means for determining whether the received comments are offensive, means for converting the offensive comments into positive, kind words, and means for providing the converted comments to a customer support representative.

[0101] The server provides an interface, such as a feedback form on a website or application or a comment section on a social networking site, to receive comments from users. When a user enters a comment, the comment is sent to the server.

[0102] The server uses natural language processing techniques to analyze the received comments. Specifically, it uses the TextBlob library to analyze the sentiment of the comment and determine whether the comment is offensive. If the sentiment score is negative, the comment is deemed offensive.

[0103] The server then uses a generative AI model to convert offensive comments into positive, kind words. Specifically, it uses OpenAI's generative API to convert the wording of the offensive comment while preserving its content. For example, it converts a comment like "Your service is terrible" to "Your service could use some improvement."

[0104] The converted comments are provided to a customer support representative. The server provides an interface for displaying the converted comments on the customer support representative's terminal. This allows the customer support representative to receive constructive feedback without feeling stressed about offensive comments.

[0105] As a concrete example, if a user comments, "Your service is terrible," the system will translate that comment into, "Your service could be improved," and provide that to a customer support representative.

[0106] Example of a prompt to be input to a generative AI model:

[0107] Please convert the following comments into positive and kind words: Your service is awful.

[0108] In this way, the system converts offensive comments from users into positive feedback, enabling customer support agents to respond more effectively.

[0109] The flow of the specific process in the application example 1 will be described with reference to FIG.

[0110] Step 1:

[0111] A user enters a comment. A user enters a comment using a feedback form on a website or application, or a comment section on a social networking site. The comment is sent to the server. The input is the user's comment, and the output is the comment sent to the server.

[0112] Step 2:

[0113] The server receives the comments. The server receives the comments sent by the user and stores them in a database. The input is the comment sent by the user, and the output is the stored comment data.

[0114] Step 3:

[0115] The server analyzes the sentiment of the comment. The server uses the TextBlob library to analyze the sentiment of the received comment. Specifically, it calculates the positive / negative score of the comment and judges whether the comment is aggressive based on the score. The input is the stored comment data, and the output is the sentiment score and the judgment result of whether the comment is aggressive or not. For example, using a learning model, the positive / negative score of the comment is obtained by machine learning using teacher data in which a label (sentiment score; for example, a multiple-level numerical value in the range of +10 to -10) is added to a large amount of data (a combination of vital data and face images). To generate the learning model, any algorithm such as a support vector machine, boosting, or neural network can be used. Note that the algorithm such as the support vector machine can be a publicly known technology, so the explanation is omitted.

[0116] Step 4:

[0117] The server converts aggressive comments. The server uses OpenAI's generation API to convert aggressive comments into positive and kind words. Specifically, it inputs a prompt sentence to the generative AI model and obtains the converted comment. The input is the comment judged to be aggressive and the prompt sentence, and the output is the converted comment.

[0118] Step 5:

[0119] The server provides the converted comments to the customer support representative. The server provides an interface for displaying the converted comments on the customer support representative's terminal. The input is the converted comments and the output is the comments displayed on the customer support representative's terminal.

[0120] Step 6:

[0121] The customer support representative checks the converted comments and responds. The customer support representative checks the positive comments displayed on the terminal and takes appropriate action. The input is the converted comments displayed on the terminal, and the output is the customer support representative's response action.

[0122] Example 2

[0123] Next, a second embodiment of the second embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0124] Comments from users on the Internet often contain offensive content. Such offensive comments cause discomfort to other users and service providers, and cause a decline in the quality of communication. Conventional systems have had difficulty automatically detecting and appropriately processing offensive comments. Therefore, the present invention aims to provide a system that automatically detects aggressive comments from users and converts them into positive, kind words.

[0125] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for storing the received comments in a database, a means for analyzing the stored comments using natural language processing technology and a machine learning model, a means for determining whether the comments are aggressive based on the analysis results, a means for converting aggressive comments into positive and kind words using a generative AI model, and a means for storing the converted comments in a database. This makes it possible to automatically detect aggressive comments from users and convert them into positive expressions.

[0126] A "User" is an individual or entity that submits a Comment to the System.

[0127] A "comment" is an opinion, feedback, or message provided by a user to the system.

[0128] A "database" is a system for systematically storing and managing data such as comments and analysis results.

[0129] "Natural language processing technology" is a technology that enables computers to understand, interpret, and generate human language.

[0130] A "machine learning model" is an algorithm or system that learns from data to perform a specific task.

[0131] "Analysis" is the process of tokenizing the content of comments and evaluating their sentiment and tone.

[0132] A "generative AI model" is an artificial intelligence model that generates new text based on input text.

[0133] "Aggression" refers to words or expressions that are offensive or hostile to others.

[0134] "Positive and kind words" are words and expressions that give a favorable and friendly impression to others.

[0135] "Transformation" is the process of replacing aggressive comments with positive, kind words.

[0136] The present invention relates to a system for automatically detecting offensive comments from users and converting them into positive and kind words. This system is implemented using the following hardware and software.

[0137] 1. Receiving comments

[0138] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site, and the device sends the comment to the server as an HTTP POST request.

[0139] 2. Saving comments

[0140] The server stores the received comments in a database. Specifically, a database management system such as MySQL (registered trademark) or PostgreSQL is used.

[0141] 3. Comment Analysis

[0142] The server analyzes the stored comments using natural language processing (NLP) and machine learning models, tokenizing the comments and performing sentiment analysis using the Python libraries NLTK and spaCy.

[0143] 4. Judging offensive comments

[0144] The server uses the analysis results to determine whether a comment is offensive. For example, if it contains negative words such as "the worst" or "horrible," it will be deemed offensive.

[0145] 5. Comment Conversion

[0146] The server inputs the comments that are judged to be offensive into a generative AI model (e.g., OpenAI's GPT-3) and converts them into positive and kind expressions. Using a generative API, the expression is changed while preserving the content of the comment. An example of a specific prompt sentence is "Please convert the following offensive comment into a positive and kind expression: 'Your service is the worst.'"

[0147] 6. Saving comments after conversion

[0148] The server saves the converted comment back into the database, so that both the original comment and the converted comment are stored.

[0149] Examples

[0150] For example, if a user sends a comment saying "Your service is the worst," the server receives this comment and analyzes it using natural language processing technology. Since it is determined to be an offensive comment, "Your service is the worst" is input into the generative AI model. The generative AI model converts this into a more positive expression, "Your service has room for improvement." The server saves the converted comment in a database and provides feedback to the user.

[0151] In this way, the present invention is able to automatically detect offensive comments from users and convert them into positive expressions.

[0152] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0153] Step 1: Receiving comments

[0154] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site. The device sends this comment to the server as an HTTP POST request. The input is the user's comment, and the output is the comment data sent to the server.

[0155] Step 2: Save your comments

[0156] The server stores the received comments in a database, typically using a database management system such as MySQL or PostgreSQL. The input is the comment data sent to the server, and the output is the comment records stored in the database.

[0157] Step 3: Analyzing comments

[0158] The server analyzes the stored comments using natural language processing (NLP) technology and machine learning models. It uses Python libraries NLTK and spaCy to tokenize comments and perform sentiment analysis. The input is the comment data obtained from the database, and the output is the analysis results.

[0159] Step 4: Identifying offensive comments

[0160] The server uses the analysis results to determine whether a comment is offensive. For example, if a comment contains negative words such as "the worst" or "horrible," it is deemed offensive. The input is the analysis results, and the output is the judgment result of whether the comment is offensive or not.

[0161] Step 5: Converting comments

[0162] The server inputs the comments judged as offensive into a generative AI model (e.g., OpenAI's GPT-3) and converts them into positive and kind expressions. Using a generative API, the expression is changed while preserving the content of the comment. An example of a specific prompt sentence is "Please convert the following offensive comment into a positive and kind expression: 'Your service is the worst.'" The input is the comment judged as offensive, and the output is the converted positive comment.

[0163] Step 6: Save comments after conversion

[0164] The server stores the converted comment back in the database, thus storing both the original and the converted comment. The input is the converted positive comment, and the output is the converted comment record stored in the database.

[0165] (Application example 2)

[0166] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0167] On conventional online shopping sites, feedback and reviews from users may contain offensive content, which may have a negative impact on other users. Such offensive comments can negatively impact the overall atmosphere of the site and reduce other users' desire to purchase. Therefore, there is a need to convert offensive comments into positive expressions and provide a healthy communication environment.

[0168] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0169] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for displaying the converted comments to the user. This makes it possible to convert the offensive comments into positive expressions and turn the review section of the online shopping site into a place for healthy communication.

[0170] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[0171] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[0172] The "means for converting aggressive comments into positive and kind words" refers to a technique for converting aggressive comments into positive and kind words while retaining their content. Positive and kind words may be construed as specific comments in the present disclosure.

[0173] The "means for displaying the converted comment to the user" refers to an interface for visually presenting the converted comment to the user.

[0174] "Generation API" means an application programming interface for performing natural language generation for a particular input.

[0175] A system for implementing the present invention includes a series of processes for receiving comments from users, determining whether the comments are offensive, converting the offensive comments into positive, kind words, and displaying the converted comments to the user.

[0176] Hardware and Software Used

[0177] Hardware: Servers, user devices (PCs, smartphones, etc.)

[0178] Software: Python, OpenAI API, Natural Language Processing libraries (e.g. NLTK, spaCy)

[0179] Data processing and calculation

[0180] 1. Receiving comments:

[0181] When a user enters a comment in the review section of an online shopping site and presses the send button, the comment is sent to the server.

[0182] The server receives the comments from the user using a means for receiving the comments.

[0183] 2. Judging offensive comments:

[0184] The server analyzes the content of the received comments using a method to determine whether the comments are offensive.

[0185] Specifically, it uses natural language processing technology and machine learning models to analyze the tone and sentiment of comments and determine whether they contain offensive keywords.

[0186] 3. Comment conversion:

[0187] The server uses a method to convert offensive comments into positive, kind words while preserving the content of the comments.

[0188] Specifically, it uses OpenAI's generative API to transform offensive comments into positive ones.

[0189] 4. Viewing comments after conversion:

[0190] The server transmits the converted comment to the user terminal and displays it using a means for displaying the converted comment to the user.

[0191] Examples

[0192] Original comment: "This product is awful and completely unusable."

[0193] Post-conversion comment: "This product could be improved, but depending on how you use it, it may be useful."

[0194] Examples of prompt statements

[0195] Please convert the following comment into a positive one: This product is terrible, I can't use it at all.

[0196] In this way, by transforming aggressive comments into positive expressions, it is possible to turn the review section of an online shopping site into a place for healthy communication.

[0197] The flow of the specific process in the application example 2 will be described with reference to FIG.

[0198] Step 1:

[0199] A user inputs a comment in the review section of an online shopping site and presses the send button. The input comment is sent from the user terminal to the server. The input data is the comment text input by the user. The server receives this comment text.

[0200] Step 2:

[0201] The server analyzes the received comment text and determines whether it is offensive. Specifically, it uses a natural language processing library (e.g., NLTK, spaCy) to analyze the tone and sentiment of the comment. The input data is the received comment text, and the output data is the judgment result of whether the comment is offensive or not.

[0202] Step 3:

[0203] The server converts comments that are judged as offensive into positive and kind words. Specifically, it uses OpenAI's generation API to convert aggressive comments into positive expressions. The input data is the comment text that is judged as offensive, and the output data is the comment text that has been converted into positive expressions.

[0204] Step 4:

[0205] The server transmits the converted comment text to the user terminal and displays it. The input data is the comment text converted into a positive comment, and the output data is the converted comment text displayed on the user terminal. The user terminal displays the received converted comment in the review section.

[0206] Through this series of processes, aggressive comments are transformed into positive expressions, which can then be displayed in the review section of the online shopping site.

[0207] Furthermore, an emotion engine that estimates the emotion of the user may be combined. That is, the identification processing unit 290 may estimate the emotion of the user using the emotion identification model 59, and perform identification processing using the emotion of the user.

[0208] "Example 1"

[0209] In the second embodiment of the present invention, in addition to the means for receiving comments from users, the means for determining whether the received comments are aggressive, and the means for converting aggressive comments into positive and kind words, an emotion engine for recognizing the emotions of users is built in. This emotion engine uses natural language processing technology and machine learning to read emotions from user comments. For example, emotions such as "anger" and "dissatisfaction" are recognized from a comment such as "your service is the worst." Then, based on this emotion information, it is determined with higher accuracy whether the comment is aggressive. Furthermore, the emotion engine is also utilized when converting aggressive comments. Specifically, a comment including the emotion of "anger" such as "your service is the worst" is converted into an expression that softens the emotional element, such as "your service has room for improvement." This makes it possible to properly understand the emotions of the user and take appropriate measures based on them.

[0210] "Example 2"

[0211] In the second embodiment of the present invention, in addition to the means for receiving comments from users, the means for determining whether the received comments are aggressive, and the means for converting aggressive comments into positive and kind words, an emotion engine for recognizing the emotions of users is built in. This emotion engine uses natural language processing technology and machine learning to read emotions from user comments. For example, emotions such as "anger" and "dissatisfaction" are recognized from a comment such as "your service is the worst." Then, based on this emotion information, it is determined with higher accuracy whether the comment is aggressive. Furthermore, the emotion engine is also utilized when converting aggressive comments. Specifically, a comment including the emotion of "anger" such as "your service is the worst" is converted into an expression that softens the emotional element, such as "your service has room for improvement." This makes it possible to properly understand the emotions of the user and take appropriate measures based on them.

[0212] The process flow of each embodiment will be described below.

[0213] "Example 1"

[0214] Step 1: Receive user comments. These comments can come in a variety of forms, such as a feedback form on your website or a comment section on social media.

[0215] Step 2: The received comments are input into the emotion engine, which uses natural language processing technology and machine learning to read the user's emotions from the comments.

[0216] Step 3: Based on the emotion information obtained from the emotion engine, determine whether the comment is offensive. This determination is made based on the type and intensity of the emotion recognized by the emotion engine.

[0217] Step 4: If a comment is deemed offensive, we transform it into positive, kinder language, taking into account the emotions recognized by the sentiment engine, and converting it into a more toned-down version of the comment.

[0218] Step 5: The converted comment is output. This comment retains the content of the original comment but removes any offensive elements. This allows the service provider to reduce the mental burden when receiving feedback from users.

[0219] Example 1

[0220] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal".

[0221] In communication on the Internet, comments from users often contain offensive content, which can cause discomfort and stress to other users and service providers. It is also difficult to properly process aggressive comments and convert them into positive, kind words. Another challenge is to accurately recognize the sentiment of the comments and respond appropriately based on that sentiment.

[0222] The identification process by the identification processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are aggressive, a means for converting aggressive comments into positive and kind words, a means for analyzing the tone and emotions of the comments using natural language processing technology, and a means for converting the expression while maintaining the content of the comments using a generation API. This makes it possible to appropriately analyze the user's comments and convert aggressive content into positive expressions. In addition, by using an emotion engine, it is possible to accurately recognize the emotions of the comments and to make more accurate judgments and conversions based on the emotions.

[0223] "User" means an individual or entity that submits comments using the System.

[0224] "Comment" means a text message that a User sends via a feedback form on the Website or Application or through a comment section on a social networking site.

[0225] A "server" is a computer system for receiving, parsing, and optionally transforming comments from users.

[0226] "Natural language processing technology" is a technology that enables computers to understand, analyze, and generate human language.

[0227] A "machine learning model" is an algorithm that learns from data and makes predictions and classifications.

[0228] A "generation API" is an application programming interface for generating natural language text for particular input.

[0229] An "emotion engine" is a system that recognizes emotions from user comments and analyzes and converts the comments based on that information.

[0230] "Tone" refers to the overall mood or feeling of a comment.

[0231] An "offensive comment" is a comment that includes content that causes discomfort or stress to other users or service providers.

[0232] "Positive and kind words" are those that soften aggressive comments and turn them into constructive, friendly language.

[0233] The present invention relates to a system for receiving comments from users and converting aggressive comments into positive and kind words. A specific embodiment of this system will be described below.

[0234] Hardware and Software Configuration

[0235] The server is a computer system for receiving, analyzing, and optionally converting comments from users. The server uses the following software and libraries:

[0236] Natural Language Processing (NLP): Analyze the tone and sentiment of comments using Python libraries NLTK and spaCy.

[0237] Machine learning model: As an emotion engine, we use the Python libraries TextBlob and VADER to recognize emotions from user comments.

[0238] Generative API: Uses generative AI models such as OpenAI’s GPT-3 to transform comments while preserving their content.

[0239] System Operation

[0240] 1. Receiving comments

[0241] Users can submit comments via feedback forms on websites and applications, or through comment sections on social media.

[0242] The server receives comments from users through these means. Specifically, the server receives an HTTP request and extracts the comment data.

[0243] 2. Comment Analysis

[0244] The server analyzes the received comments using natural language processing (NLP) and machine learning models. Specifically, the server tokenizes the comment text and inputs it into a sentiment analysis model.

[0245] An emotion engine is incorporated to recognize emotions such as "anger" or "dissatisfaction" from user comments.

[0246] 3. Judging offensive comments

[0247] The server uses the emotion engine to more accurately determine whether a comment is offensive. Specifically, if the emotion engine detects emotions such as "anger" or "dissatisfaction," the server determines the comment is offensive.

[0248] 4. Converting offensive comments

[0249] When the server detects an offensive comment, it uses the generation API to convert the comment's expression while preserving its content. Specifically, the server sends the offensive comment to the generation API as a prompt text and receives the converted comment.

[0250] Examples

[0251] Example of an embodiment

[0252] User comment: "Your service is the worst."

[0253] Server action:

[0254] Receive comments.

[0255] Comments are analyzed using NLTK and spaCy and deemed offensive.

[0256] Send the prompt text "Please convert the following comment into positive and kind words. Comment: 'Your service is the worst'" to the generation API.

[0257] Receive the converted comment "Your service has room for improvement" from the generation API.

[0258] Example of the second form

[0259] User comment: "Your service is the worst."

[0260] Server action:

[0261] Receive comments.

[0262] Comments are analyzed using TextBlob and VADER to recognize emotions such as "anger" and "dissatisfaction."

[0263] The sentiment engine determines that the comment is offensive.

[0264] Send the prompt text "Please convert the following comment into positive and kind words. Comment: 'Your service is the worst'" to the generation API.

[0265] Receive the converted comment "Your service has room for improvement" from the generation API.

[0266] In this way, we have created a system that properly analyzes user comments and converts offensive content into positive expressions. By using an emotion engine, we can accurately recognize the emotions in comments and make more accurate judgments and conversions based on those emotions.

[0267] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0268] Step 1: Receiving comments

[0269] Users can submit comments via feedback forms on websites and applications, or through comment sections on social media.

[0270] The server receives the HTTP request and extracts the comment data.

[0271] Input: User submitted comment (e.g. "Your service is terrible")

[0272] Output: Extracted comment data

[0273] Step 2: Analyzing comments

[0274] The server analyzes the received comments using natural language processing technology (NLP) and machine learning models.

[0275] How it works: The server uses Python libraries NLTK and spaCy to tokenize the comment text and input it into a sentiment analysis model.

[0276] Input: Extracted comment data

[0277] Output: An analysis of the comments' tone and sentiment.

[0278] Step 3: Recognize emotions

[0279] The server uses an emotion engine to recognize emotions such as "anger" or "dissatisfaction" from user comments.

[0280] Specific operation: The server analyzes sentiment using Python libraries TextBlob and VADER.

[0281] Input: Analysis of the comment's tone and sentiment

[0282] Output: Recognized emotion information (e.g. anger, frustration)

[0283] Step 4: Identifying offensive comments

[0284] The server uses a sentiment engine to determine whether a comment is offensive.

[0285] What it does: If the emotion engine detects emotions like "anger" or "frustration," it marks the comment as offensive.

[0286] Input: Recognized emotion information

[0287] Output: Judgment result of whether the comment is offensive (e.g., offensive)

[0288] Step 5: Convert offensive comments

[0289] When the server detects an offensive comment, it uses a generation API to convert the expression while preserving the content of the comment.

[0290] Specific behavior: The server sends the offensive comment as a prompt to the generation API and receives the converted comment.

[0291] Input: Offensive comment (e.g. "Your service is terrible")

[0292] Output: The converted comment (e.g. "Your service could be improved")

[0293] In this way, we have created a system that properly analyzes user comments and converts offensive content into positive expressions. By using an emotion engine, we can accurately recognize the emotions in comments and make more accurate judgments and conversions based on those emotions.

[0294] (Application example 1)

[0295] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0296] Conventional customer support systems had problems with poor user experience due to difficulties in responding appropriately to aggressive comments from users. In addition, there was a lack of technology to accurately recognize users' emotions and convert comments based on them, so users' dissatisfaction could not be effectively alleviated.

[0297] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0298] In this invention, the server includes a means for receiving comments from a user, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, a means for recognizing the user's emotions, a means for converting the comments based on the recognized emotions, and a means for being installed in a smartphone application. This makes it possible to convert the offensive comments from the user into positive expressions and to appropriately understand and respond to the user's emotions.

[0299] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[0300] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[0301] "Methods for transforming aggressive comments into positive and kind words" are techniques for transforming aggressive comments into more positive and kind expressions while preserving their content.

[0302] "Means for recognizing user emotions" refers to technology for reading emotions from user comments and identifying those emotions.

[0303] The "means for converting a comment based on a recognized emotion" is a technique for converting a comment into an appropriate expression based on the recognized emotion of the user.

[0304] "Means for being installed in a smartphone application" refers to technology for being incorporated into an application running on a smartphone.

[0305] As a form for implementing this invention, a system is constructed for a customer support chatbot on an online shopping site that receives comments from users and converts aggressive comments into positive expressions.

[0306] System Programming

[0307] The system consists of the following main components:

[0308] 1. Comment Receiving Module: An interface for receiving comments from users. This is implemented as a chatbot function in a smartphone application.

[0309] 2. Sentiment Analysis Module: Natural language processing techniques to determine whether the comments received are offensive or not. Specifically, it uses natural language processing libraries such as spaCy and NLTK.

[0310] 3. Comment Transformation Module: A generative AI model for transforming offensive comments into positive ones. Specifically, we use generative APIs such as OpenAI's GPT-3.

[0311] 4. Emotion Recognition Module: Technology to read and identify emotions from user comments. This also uses natural language processing libraries.

[0312] 5. Smartphone application: A platform to integrate the above modules and make them accessible to users.

[0313] Processing Description

[0314] When the server receives a comment from a user, it first analyzes the content of the comment using the sentiment analysis module. If the comment is determined to be offensive based on the analysis results, the comment conversion module converts the comment into a positive expression using a generative AI model. At this time, the emotion recognition module identifies the user's emotion and performs the conversion based on that emotion. Finally, the converted comment is returned to the user through the smartphone application.

[0315] Hardware and Software Used

[0316] Hardware: Servers, smartphones

[0317] Software: spaCy, NLTK, OpenAI GPT-3, smartphone applications

[0318] Examples

[0319] For example, if a user submits a comment saying, "Your service is terrible," the sentiment analysis module determines the comment as "aggressive." The emotion recognition module then identifies that the comment contains the emotion of "anger." The comment transformation module uses a generative AI model to transform the comment into a positive expression, such as, "Your service could use some improvement."

[0320] Examples of prompt statements

[0321] Please convert the following offensive comments into positive ones: Your service is awful.

[0322] In this way, it is possible to convert aggressive comments from users into positive expressions and appropriately understand and respond to the user's feelings.

[0323] The flow of the specific process in the application example 1 will be described with reference to FIG.

[0324] Step 1:

[0325] A user inputs and submits a comment through a smartphone application. The input comment is sent to the server. The input data is the user's comment text.

[0326] Step 2:

[0327] The server passes the received comments to a sentiment analysis module, which uses a natural language processing library (e.g., spaCy or NLTK) to analyze the content of the comment and determine whether it is offensive. The input data is the user's comment text, and the output data is the judgment result of whether the comment is offensive or not.

[0328] Step 3:

[0329] The server receives the output of the sentiment analysis module, and if the comment is determined to be offensive, passes the comment to the emotion recognition module, which uses natural language processing techniques to identify the user's emotion from the comment. The input data is the user's comment text, and the output data is the identified emotion information.

[0330] Step 4:

[0331] The server receives the output of the emotion recognition module and passes the comment and emotion information to the comment conversion module, which uses a generative AI model (e.g., OpenAI's GPT-3) to convert aggressive comments into positive expressions. The input data is the user's comment text and emotion information, and the output data is the converted positive comment text.

[0332] Step 5:

[0333] The server returns the converted comment to the smartphone application, which displays the converted comment to the user. The input data is the converted positive comment text, and the output data is the comment displayed on the user's screen.

[0334] In this way, it is possible to convert aggressive comments from users into positive expressions and appropriately understand and respond to the user's feelings.

[0335] Example 2

[0336] Next, a second embodiment of the second embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0337] When a user's comment on the Internet contains offensive content, it can cause discomfort and stress to other users and service providers. There is also the issue that it is difficult to properly process offensive comments and convert them into positive expressions. Furthermore, there is a need to accurately recognize the sentiment of the comments and respond appropriately based on that.

[0338] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0339] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, a means for recognizing emotions from the user's comments, a means for determining whether the comments are offensive based on the recognized emotional information, and a means for converting the offensive comments based on the recognized emotional information. This makes it possible to appropriately understand the emotions of the user and convert the offensive comments into positive expressions.

[0340] "User" means an individual or entity that submits comments using the System.

[0341] A "comment" is text data such as an opinion, impression, or feedback that a user sends through the system.

[0342] A "server" is a computer system that receives comments from users, analyzes them, and performs the necessary processing.

[0343] The "means for receiving" is a function that enables the server to receive comments sent by users.

[0344] The "means of judgment" is a function for analyzing and determining whether a received comment is offensive or not.

[0345] "Means of transformation" is a function for transforming aggressive comments into positive and kind words.

[0346] "Means for recognizing emotions" is a function for analyzing and recognizing emotions from user comments.

[0347] “Generation API” means an application programming interface for utilizing external generative AI models to generate or transform text.

[0348] A "database" is a data management system for storing data such as comments and analysis results.

[0349] This invention is a system that receives comments from users, judges whether the comments are offensive, and converts the offensive comments into positive and kind words. It also includes a function to recognize emotions from the user's comments, judge whether the comments are offensive based on the emotional information, and convert them appropriately.

[0350] Hardware and Software Used

[0351] The system uses the following hardware and software:

[0352] Server: A computer system for receiving, parsing, and converting comments. Specifically, it includes a web server (e.g., Apache (registered trademark), Nginx) for processing HTTP requests, and a database (e.g., MySQL, PostgreSQL).

[0353] Natural language processing technology: Libraries and APIs for analyzing the content of comments and determining their tone and sentiment. Specifically, we use the Python NLTK library and Google Cloud Natural Language API.

[0354] Generative API: A generative AI model for converting aggressive comments into positive and kind words. Specifically, it uses OpenAI's GPT-3 and GPT-4 (registered trademarks).

[0355] Explanation of program processing

[0356] 1. Receiving comments

[0357] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site.

[0358] The server receives the HTTP request, extracts the comment data, and stores it in a database.

[0359] 2. Judging offensive comments

[0360] The server retrieves the comments from the database and analyzes the content of the comments using natural language processing techniques.

[0361] The server uses Python's NLTK library and the Google Cloud Natural Language API to determine the tone and sentiment of the comments.

[0362] The server determines whether a comment is offensive and stores the result as a flag in a database.

[0363] 3. Emotion Recognition

[0364] The server retrieves comments from the database and analyzes their sentiment using natural language processing technology and machine learning.

[0365] The server uses Python's NLTK library and the Google Cloud Natural Language API to recognize emotions such as "anger" or "dissatisfaction" from the comments.

[0366] The server stores the recognized emotion information in a database.

[0367] 4. Converting offensive comments

[0368] The server sends comments determined to be offensive to the generation API.

[0369] The server uses OpenAI's GPT-3 and GPT-4 to convert offensive comments into positive, kind words.

[0370] The server stores the converted comment in its database, replacing the original comment.

[0371] Examples of concrete examples and prompts

[0372] Examples

[0373] A user enters the comment "Your service is terrible" into a feedback form on a website and clicks the submit button.

[0374] The server receives the HTTP POST request, extracts the comment data, and stores it in a database.

[0375] The server uses Python's NLTK library to analyze the tone and sentiment of comments and determine which ones are offensive.

[0376] The server uses OpenAI's GPT-3 to translate the comment into "Your service could use some improvement."

[0377] The server stores the converted comment in its database, replacing the original comment.

[0378] Examples of prompt statements

[0379] "Please translate the following comment into something positive and kind: 'Your service is awful.'"

[0380] "Recognize the emotion in the following comment and tone down the offensive language: 'Your service is terrible.'"

[0381] In this way, it is possible to properly understand the user's emotions and convert aggressive comments into positive expressions.

[0382] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0383] Step 1:

[0384] Receiving comments

[0385] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site.

[0386] Input: Comment entered by the user (e.g. "Your service is terrible").

[0387] The server receives the HTTP POST request and extracts the comment data.

[0388] Output: Extracted comment data (e.g. "Your service is the worst").

[0389] Specific operation: The server stores the received comments in a database.

[0390] Step 2:

[0391] Judging offensive comments

[0392] The server retrieves the comments from the database.

[0393] Input: Comment data stored in database (e.g. "Your service is terrible").

[0394] The server analyzes the content of the comments using natural language processing technology.

[0395] How it works: The server uses Python's NLTK library and the Google Cloud Natural Language API to determine the tone and sentiment of the comments.

[0396] Output: A judgement as to whether the comment is offensive (e.g. "Offensive").

[0397] The server stores the result of the determination in a database as a flag.

[0398] Step 3:

[0399] Emotion recognition

[0400] The server retrieves the comments from the database.

[0401] Input: Comment data stored in database (e.g. "Your service is terrible").

[0402] The server uses natural language processing technology and machine learning to analyze the sentiment of the comments.

[0403] How it works: The server uses Python's NLTK library and the Google Cloud Natural Language API to recognize emotions such as "anger" or "dissatisfaction" from comments.

[0404] Output: Recognized emotion information (e.g., "anger").

[0405] The server stores the recognized emotion information in a database.

[0406] Step 4:

[0407] Converting offensive comments

[0408] The server sends comments determined to be offensive to the generation API.

[0409] Input: aggressive comments (e.g., “your service is the worst”) and perceived emotion information (e.g., “anger”).

[0410] The server uses generative AI models (e.g., OpenAI's GPT-3 and GPT-4) to transform offensive comments into positive, kind words.

[0411] Specific operation: The server sends a prompt text to the generation API and receives the converted comment.

[0412] Output: The transformed comment (e.g. "Your service could be improved").

[0413] The server stores the converted comment in its database, replacing the original comment.

[0414] (Application example 2)

[0415] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0416] Conventional systems have difficulty in responding appropriately to aggressive comments from users, and responses to comments that contain emotional elements are often insufficient. In addition, when converting aggressive comments into positive expressions, it is necessary to convert them into appropriate expressions while retaining the content of the comment, but this is difficult to achieve. Furthermore, there is a demand to implement these functions in smartphone applications.

[0417] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0418] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are aggressive, a means for converting aggressive comments into positive and kind words, a means for recognizing the user's emotions, a means for determining the aggressiveness of the comments based on the emotion information, a means for converting aggressive comments based on the emotion information, and a means for being installed in a smartphone application. This makes it possible to appropriately respond to aggressive comments from users, accurately determine comments that include emotional elements, and convert them into positive expressions.

[0419] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[0420] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[0421] "Means of converting aggressive comments into positive and kind words" refers to a technique for converting aggressive comments into positive and kind expressions while retaining their content.

[0422] "Means for recognizing user emotions" refers to technology for reading emotions from users' comments and analyzing those emotions.

[0423] The "means for determining the aggressiveness of a comment based on emotional information" is a technique for determining with higher accuracy whether a comment is aggressive or not based on the emotional information of a user.

[0424] The "means for converting aggressive comments based on emotional information" is a technique for converting aggressive comments into positive expressions by taking into account the emotional information of the user.

[0425] "Means to be installed in a smartphone application" refers to technology for incorporating this system into an application running on a smartphone.

[0426] As an embodiment of the present invention, a method for installing a system in a smartphone application that receives comments from users, determines whether the comments are offensive, and converts the offensive comments into positive, kind words will be described.

[0427] First, a user inputs a comment through a smartphone application. The comment is then sent to the server through an interface within the application. The server then analyzes the received comment using natural language processing technology (e.g., spaCy) to understand the content of the comment.

[0428] The server then uses a machine learning model (e.g., the sentiment analysis pipeline in the transformers library) to determine whether the comment is offensive or not. At the same time, the user's sentiment is also analyzed to obtain sentiment information, which is then used to more accurately determine the offensiveness of the comment.

[0429] If a comment is deemed offensive, the server uses a generative AI model (e.g., GPT-2) to convert the offensive comment into positive, kind words. Specifically, the server inputs the following prompt sentences into the generative AI model:

[0430] Please convert the following offensive comments into positive ones: Your service is awful.

[0431] Positive comments:

[0432] The generative AI model uses this prompt to convert aggressive comments into positive ones, which are then fed back to the user.

[0433] This system responds appropriately to aggressive comments from users, and can accurately identify comments that contain emotional elements and convert them into positive expressions. This is expected to improve the user experience and reduce problems caused by aggressive comments.

[0434] For example, if a user enters a comment like "Your service is the worst," the system will convert this comment into a positive expression like "Your service could be improved." In this way, the system can properly understand the user's sentiment and take appropriate action based on it.

[0435] The flow of the specific process in the application example 2 will be described with reference to FIG.

[0436] Step 1:

[0437] A user inputs a comment through a smartphone application. The input comment is sent to a server through an interface within the application. The input data is the user's comment, and the output data is the comment sent to the server.

[0438] Step 2:

[0439] The server analyzes the received comments using natural language processing technology (e.g., spaCy). Specifically, it performs tokenization of the comments, part-of-speech tagging, dependency analysis, etc. The input data are the received comments, and the output data is the structural information of the analyzed comments.

[0440] Step 3:

[0441] The server uses a machine learning model (e.g., a sentiment analysis pipeline from the transformers library) to determine whether a comment is offensive. The input data is the parsed comment structure information, and the output data is a judgment result about the offensiveness of the comment.

[0442] Step 4:

[0443] The server simultaneously analyzes the user's emotions and obtains the emotion information. Using the emotion analysis pipeline, emotions such as "anger" and "dissatisfaction" are extracted from the comments. The input data is the structure information of the analyzed comments, and the output data is the user's emotion information.

[0444] Step 5:

[0445] The server judges the offensiveness of the comment more accurately based on the emotion information. It integrates the emotion information and the offensiveness judgment result to make a final offensiveness judgment. The input data are the emotion information and the offensiveness judgment result, and the output data is the final offensiveness judgment result.

[0446] Step 6:

[0447] If a comment is deemed offensive, the server uses a generative AI model (e.g., GPT-2) to convert the offensive comment into positive, kind words. Specifically, the server inputs the following prompt sentences into the generative AI model:

[0448] Please convert the following offensive comments into positive ones: Your service is awful.

[0449] Positive comments:

[0450] The input data are offensive comments, and the output data are the comments transformed into positive expressions.

[0451] Step 7:

[0452] The converted comments are fed back from the server to the smartphone application. The user can check the comments that have been converted into positive expressions. The input data are the comments that have been converted into positive expressions, and the output data are the comments that have been fed back to the user.

[0453] 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 a voice indicating a user input for the result of the specific processing. The control unit 46A transmits the voice data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0454] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search<URL: https: / / openai.com / blog / chatgpt> ) and other generation AIs. The data generation model 58 is obtained by making a neural network perform deep learning. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating a voice, text data indicating a text, and image data indicating an image is input. The data generation model 58 performs inference on the input inference data according to the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0455] Another example of generative AI is Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) are mentioned.

[0456] In the above embodiment, an example was given in which the specific process was performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0457] [Second embodiment]

[0458] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0459] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0460] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a wide area network (WAN) and / or a local area network (LAN).

[0461] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0462] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs the voice according to instructions from the processor 46.

[0463] Camera 42 is a small digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (e.g., an imaging range defined by an angle of view equivalent to the width of the field of vision of an average healthy person).

[0464] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for transmitting and receiving various types of information between the processor 46 and the processor 28 via the network 54. The transmission and reception of various types of information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is performed in a secure state.

[0465] Fig. 4 shows an example of main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0466] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32, and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0467] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0468] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50, and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0469] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0470] "Example 1"

[0471] As one embodiment of the present invention, a feedback form on a website or application, a comment section on a social networking site, or the like is used as a means for receiving comments from users. As a means for determining whether a received comment is aggressive, natural language processing technology or machine learning is used to analyze the content of the comment and determine its tone and emotion. Then, as a means for converting aggressive comments into positive and kind words, a generation API is used to convert the expression of the comment while keeping its content. Specifically, "Your service is the worst" is converted to "Your service has room for improvement."

[0472] "Example 2"

[0473] As one embodiment of the present invention, a feedback form on a website or application, a comment section on a social networking site, or the like is used as a means for receiving comments from users. As a means for determining whether a received comment is aggressive, natural language processing technology or machine learning is used to analyze the content of the comment and determine its tone and emotion. Then, as a means for converting aggressive comments into positive and kind words, a generation API is used to convert the expression of the comment while keeping its content. Specifically, "Your service is the worst" is converted to "Your service has room for improvement."

[0474] The process flow of each embodiment will be described below.

[0475] "Example 1"

[0476] Step 1: Receive comments from users. In this step, you collect comments from users through feedback forms on your website or application, or through comments sections on social media.

[0477] Step 2: Determine whether the comment received is offensive. In this step, natural language processing technology and machine learning are used to analyze the content of the comment and determine its tone and sentiment. For example, a comment like "Your service is the worst" would be deemed offensive.

[0478] Step 3: Convert aggressive comments to positive and kind words. In this step, we use the generation API to convert the expression of the comment while preserving its content. Specifically, we convert a comment such as "Your service is the worst" to "Your service has room for improvement."

[0479] Example 1

[0480] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[0481] In online communication, if a user makes an aggressive comment, that comment may cause discomfort to other users and service providers. There is a need to provide a healthy communication environment by converting such aggressive comments into positive and kind words. With conventional systems, it was difficult to properly detect aggressive comments and convert them into positive expressions while preserving their content.

[0482] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0483] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for displaying the converted comments to the user. This makes it possible to appropriately detect offensive comments and convert them into positive expressions while preserving their contents.

[0484] "User" refers to an individual or organization that uses the system to enter comments.

[0485] "Comment" refers to any text information that a User enters into a feedback form on a Website or Application, or into a comment section on a social networking site.

[0486] "Server" refers to a computer system that receives, analyzes, and converts comments from users.

[0487] "Means for receiving" refers to a function that enables the server to receive comments sent by users.

[0488] "Means of determining" refers to the function for analyzing and evaluating whether a received comment is offensive.

[0489] "Means of transformation" refers to the function of transforming aggressive comments into positive, kind words.

[0490] "Means for displaying" refers to a function for displaying the converted comment to the user.

[0491] "Offensive comments" refers to negative comments that may cause offense to other users or service providers.

[0492] "Positive and kind words" refers to words that preserve the content of aggressive comments but transform them into more constructive and friendly language.

[0493] "Generative AI Model" refers to an artificial intelligence model that generates and transforms text using natural language processing techniques.

[0494] A "prompt" refers to an instruction entered into a generative AI model to cause it to perform a specific task.

[0495] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0496] The present invention relates to a system that receives comments from users, determines whether the comments are offensive, and converts the offensive comments into positive and kind words. A specific embodiment of this system will be described below.

[0497] Hardware and Software Used

[0498] The servers include a web server and an application server for receiving comments from users. These servers receive comment data through HTTP requests. The servers analyze the comments using Python natural language processing libraries NLTK and spaCy. They also use generative AI models (e.g., OpenAI's GPT-3) to convert offensive comments into positive expressions.

[0499] Program Processing

[0500] A user enters a comment into a feedback form on a website or application, or into a comment section on a social networking site. For example, a user might enter "your service is the worst." When a user submits a comment, the server receives the comment. The server receives the comment data through an HTTP request.

[0501] The server then uses a natural language processing library, such as Python's NLTK or spaCy, to parse the comments it receives. The server analyzes the tone and sentiment of the comment to determine whether it is offensive or not. For example, it might recognize the word "terrible" as offensive.

[0502] The server will determine whether the comment is offensive based on the analysis results. If it is determined to be offensive, it will proceed to the next step. For example, a comment like "Your service is the worst" will be detected as offensive.

[0503] The server uses a generative AI model (e.g., OpenAI's GPT-3) to convert aggressive comments into positive ones. The server sends a prompt to the generative AI model, which converts the comment's expression while preserving its content. For example, the comment "Your service is terrible" is converted to "Your service has room for improvement."

[0504] Specific prompt examples:

[0505] Original comment: "Your service is terrible"

[0506] Prompt: "Transform the following offensive comment into a positive, kind statement: 'Your service is awful.'"

[0507] Finally, the server displays the converted comments received from the generative AI model to the user. The user can see the converted comments in the feedback form or in the comments section of a social networking site. For example, a comment such as "Your service has room for improvement" can be displayed.

[0508] In this way, a system is realized that converts aggressive comments from users into positive expressions.

[0509] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0510] Step 1:

[0511] The user enters a comment

[0512] A user enters a comment into a feedback form on a website or application, or into a comment section on a social networking site. For example, a user enters "Your service is the worst." The comment is sent to the server by pressing the send button.

[0513] Step 2:

[0514] The server receives the comment.

[0515] The server receives comments sent by users through HTTP requests. The received comments are stored in a database or temporarily held in memory. The input is the user's comments, and the output is the comment data stored on the server.

[0516] Step 3:

[0517] The server parses the comments

[0518] The server uses Python natural language processing libraries NLTK and spaCy to analyze the received comments. Specifically, it performs tokenization of comments, part-of-speech tagging, sentiment analysis, etc. The input is the comment data stored on the server, and the output is the analysis results (e.g., the tone and sentiment of the comments).

[0519] Step 4:

[0520] The server detects offensive comments

[0521] The server determines whether the comment is offensive based on the analysis result. If it is determined to be offensive, it proceeds to the next step. For example, if the comment contains the word "worst", it is determined to be offensive. The input is the analysis result, and the output is the detection result of the offensive comment.

[0522] Step 5:

[0523] The server converts the comments

[0524] The server uses a generative AI model (e.g., OpenAI's GPT-3) to convert offensive comments into positive expressions. The server sends a prompt to the generative AI model, which converts the expression while preserving the content of the comment. For example, the comment "Your service is terrible" is converted to "Your service could be improved." The input is the offensive comment and the prompt, and the output is the converted comment.

[0525] Specific prompt examples:

[0526] Original comment: "Your service is terrible"

[0527] Prompt: "Transform the following offensive comment into a positive, kind statement: 'Your service is awful.'"

[0528] Step 6:

[0529] The server displays the converted comment to the user.

[0530] The server displays the converted comments received from the generative AI model to the user. The user can see the converted comments in a feedback form or in a comment section on a social networking site. For example, a comment such as "Your service has room for improvement" is displayed. The input is the converted comment, and the output is the comment displayed to the user.

[0531] (Application example 1)

[0532] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[0533] In conventional customer support systems, aggressive comments from users are sent directly to the support staff, which often causes stress for the support staff. In addition, it is difficult to respond appropriately to aggressive comments, which makes communication with users less smooth.

[0534] The identification process by the identification processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for providing the converted comments to a customer support representative. This enables the customer support representative to receive constructive feedback without feeling stressed.

[0535] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[0536] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[0537] A "means of transforming aggressive comments into positive and kind words" is a technique for transforming aggressive comments into positive and kind words while preserving their content.

[0538] The "means for providing the converted comments to the customer support representative" refers to an interface for displaying or notifying the converted comments to the customer support representative.

[0539] "Generation API" means an application programming interface for generating text using a generative AI model.

[0540] A "customer support representative" is a person who handles inquiries and feedback from users.

[0541] A system for implementing the present invention includes means for receiving comments from a user, means for determining whether the received comments are offensive, means for converting the offensive comments into positive, kind words, and means for providing the converted comments to a customer support representative.

[0542] The server provides an interface, such as a feedback form on a website or application or a comment section on a social networking site, to receive comments from users. When a user enters a comment, the comment is sent to the server.

[0543] The server uses natural language processing techniques to analyze the received comments. Specifically, it uses the TextBlob library to analyze the sentiment of the comment and determine whether the comment is offensive. If the sentiment score is negative, the comment is deemed offensive.

[0544] The server then uses a generative AI model to convert offensive comments into positive, kind words. Specifically, it uses OpenAI's generative API to convert the wording of the offensive comment while preserving its content. For example, it converts a comment like "Your service is terrible" to "Your service could use some improvement."

[0545] The converted comments are provided to a customer support representative. The server provides an interface for displaying the converted comments on the customer support representative's terminal. This allows the customer support representative to receive constructive feedback without feeling stressed about offensive comments.

[0546] As a concrete example, if a user comments, "Your service is terrible," the system will translate that comment into, "Your service could be improved," and provide that to a customer support representative.

[0547] Example of a prompt to be input to a generative AI model:

[0548] Please convert the following comments into positive and kind words: Your service is awful.

[0549] In this way, the system converts offensive comments from users into positive feedback, enabling customer support agents to respond more effectively.

[0550] The flow of the specific process in the application example 1 will be described with reference to FIG.

[0551] Step 1:

[0552] A user enters a comment. A user enters a comment using a feedback form on a website or application, or a comment section on a social networking site. The comment is sent to the server. The input is the user's comment, and the output is the comment sent to the server.

[0553] Step 2:

[0554] The server receives the comments. The server receives the comments sent by the user and stores them in a database. The input is the comment sent by the user, and the output is the stored comment data.

[0555] Step 3:

[0556] The server analyzes the sentiment of the comments. The server uses the TextBlob library to analyze the sentiment of the received comments. Specifically, it calculates the positive / negative score of the comment and judges whether the comment is offensive based on the score. The input is the stored comment data, and the output is the sentiment score and the judgment result of whether the comment is offensive or not.

[0557] Step 4:

[0558] The server converts aggressive comments. The server uses OpenAI's generation API to convert aggressive comments into positive and kind words. Specifically, it inputs a prompt sentence to the generative AI model and obtains the converted comment. The input is the comment judged to be aggressive and the prompt sentence, and the output is the converted comment.

[0559] Step 5:

[0560] The server provides the converted comments to the customer support representative. The server provides an interface for displaying the converted comments on the customer support representative's terminal. The input is the converted comments and the output is the comments displayed on the customer support representative's terminal.

[0561] Step 6:

[0562] The customer support representative checks the converted comments and responds. The customer support representative checks the positive comments displayed on the terminal and takes appropriate action. The input is the converted comments displayed on the terminal, and the output is the customer support representative's response action.

[0563] Example 2

[0564] Next, a description will be given of Example 2 of the embodiment 2. In the following description, the data processing device 12 is referred to as a "server" and the smart glasses 214 are referred to as a "terminal".

[0565] Comments from users on the Internet often contain offensive content. Such offensive comments cause discomfort to other users and service providers, and cause a decline in the quality of communication. Conventional systems have had difficulty automatically detecting and appropriately processing offensive comments. Therefore, the present invention aims to provide a system that automatically detects aggressive comments from users and converts them into positive, kind words.

[0566] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for storing the received comments in a database, a means for analyzing the stored comments using natural language processing technology and a machine learning model, a means for determining whether the comments are aggressive based on the analysis results, a means for converting aggressive comments into positive and kind words using a generative AI model, and a means for storing the converted comments in a database. This makes it possible to automatically detect aggressive comments from users and convert them into positive expressions.

[0567] A "User" is an individual or entity that submits a Comment to the System.

[0568] A "comment" is an opinion, feedback, or message provided by a user to the system.

[0569] A "database" is a system for systematically storing and managing data such as comments and analysis results.

[0570] "Natural language processing technology" is a technology that enables computers to understand, interpret, and generate human language.

[0571] A "machine learning model" is an algorithm or system that learns from data to perform a specific task.

[0572] "Analysis" is the process of tokenizing the content of comments and evaluating their sentiment and tone.

[0573] A "generative AI model" is an artificial intelligence model that generates new text based on input text.

[0574] "Aggression" refers to words or expressions that are offensive or hostile to others.

[0575] "Positive and kind words" are words and expressions that give a favorable and friendly impression to others.

[0576] "Transformation" is the process of replacing aggressive comments with positive, kind words.

[0577] The present invention relates to a system for automatically detecting offensive comments from users and converting them into positive and kind words. This system is implemented using the following hardware and software.

[0578] 1. Receiving comments

[0579] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site, and the device sends the comment to the server as an HTTP POST request.

[0580] 2. Saving comments

[0581] The server stores the received comments in a database, typically using a database management system such as MySQL or PostgreSQL.

[0582] 3. Comment Analysis

[0583] The server analyzes the stored comments using natural language processing (NLP) and machine learning models, tokenizing the comments and performing sentiment analysis using the Python libraries NLTK and spaCy.

[0584] 4. Judging offensive comments

[0585] The server uses the analysis results to determine whether a comment is offensive. For example, if it contains negative words such as "the worst" or "horrible," it will be deemed offensive.

[0586] 5. Comment Conversion

[0587] The server inputs the comments that are judged to be offensive into a generative AI model (e.g., OpenAI's GPT-3) and converts them into positive and kind expressions. Using a generative API, the expression is changed while preserving the content of the comment. An example of a specific prompt sentence is "Please convert the following offensive comment into a positive and kind expression: 'Your service is the worst.'"

[0588] 6. Saving comments after conversion

[0589] The server saves the converted comment back into the database, so that both the original comment and the converted comment are stored.

[0590] Examples

[0591] For example, if a user sends a comment saying "Your service is the worst," the server receives this comment and analyzes it using natural language processing technology. Since it is determined to be an offensive comment, "Your service is the worst" is input into the generative AI model. The generative AI model converts this into a more positive expression, "Your service has room for improvement." The server saves the converted comment in a database and provides feedback to the user.

[0592] In this way, the present invention is able to automatically detect offensive comments from users and convert them into positive expressions.

[0593] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0594] Step 1: Receiving comments

[0595] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site. The device sends this comment to the server as an HTTP POST request. The input is the user's comment, and the output is the comment data sent to the server.

[0596] Step 2: Save your comments

[0597] The server stores the received comments in a database, typically using a database management system such as MySQL or PostgreSQL. The input is the comment data sent to the server, and the output is the comment records stored in the database.

[0598] Step 3: Analyzing comments

[0599] The server analyzes the stored comments using natural language processing (NLP) technology and machine learning models. It uses Python libraries NLTK and spaCy to tokenize comments and perform sentiment analysis. The input is the comment data obtained from the database, and the output is the analysis results.

[0600] Step 4: Identifying offensive comments

[0601] The server uses the analysis results to determine whether a comment is offensive. For example, if a comment contains negative words such as "the worst" or "horrible," it is deemed offensive. The input is the analysis results, and the output is the judgment result of whether the comment is offensive or not.

[0602] Step 5: Converting comments

[0603] The server inputs the comments judged as offensive into a generative AI model (e.g., OpenAI's GPT-3) and converts them into positive and kind expressions. Using a generative API, the expression is changed while preserving the content of the comment. An example of a specific prompt sentence is "Please convert the following offensive comment into a positive and kind expression: 'Your service is the worst.'" The input is the comment judged as offensive, and the output is the converted positive comment.

[0604] Step 6: Save comments after conversion

[0605] The server stores the converted comment back in the database, thus storing both the original and the converted comment. The input is the converted positive comment, and the output is the converted comment record stored in the database.

[0606] (Application example 2)

[0607] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server", and the smart glasses 214 will be referred to as a "terminal".

[0608] On conventional online shopping sites, feedback and reviews from users may contain offensive content, which may have a negative impact on other users. Such offensive comments can negatively impact the overall atmosphere of the site and reduce other users' desire to purchase. Therefore, there is a need to convert offensive comments into positive expressions and provide a healthy communication environment.

[0609] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0610] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for displaying the converted comments to the user. This makes it possible to convert the offensive comments into positive expressions and turn the review section of the online shopping site into a place for healthy communication.

[0611] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[0612] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[0613] "Methods for transforming aggressive comments into positive and kind words" are techniques for transforming aggressive comments into positive and kind expressions while preserving their content.

[0614] The "means for displaying the converted comment to the user" refers to an interface for visually presenting the converted comment to the user.

[0615] "Generation API" means an application programming interface for performing natural language generation for a particular input.

[0616] A system for implementing the present invention includes a series of processes for receiving comments from users, determining whether the comments are offensive, converting the offensive comments into positive, kind words, and displaying the converted comments to the user.

[0617] Hardware and Software Used

[0618] Hardware: Servers, user devices (PCs, smartphones, etc.)

[0619] Software: Python, OpenAI API, Natural Language Processing libraries (e.g. NLTK, spaCy)

[0620] Data processing and calculation

[0621] 1. Receiving comments:

[0622] When a user enters a comment in the review section of an online shopping site and presses the send button, the comment is sent to the server.

[0623] The server receives the comments from the user using a means for receiving the comments.

[0624] 2. Judging offensive comments:

[0625] The server analyzes the content of the received comments using a method to determine whether the comments are offensive.

[0626] Specifically, it uses natural language processing technology and machine learning models to analyze the tone and sentiment of comments and determine whether they contain offensive keywords.

[0627] 3. Comment conversion:

[0628] The server uses a method to convert offensive comments into positive, kind words while preserving the content of the comments.

[0629] Specifically, it uses OpenAI's generative API to transform offensive comments into positive ones.

[0630] 4. Viewing comments after conversion:

[0631] The server transmits the converted comment to the user terminal and displays it using a means for displaying the converted comment to the user.

[0632] Examples

[0633] Original comment: "This product is awful and completely unusable."

[0634] Post-conversion comment: "This product could be improved, but depending on how you use it, it may be useful."

[0635] Examples of prompt statements

[0636] Please convert the following comment into a positive one: This product is terrible, I can't use it at all.

[0637] In this way, by transforming aggressive comments into positive expressions, it is possible to turn the review section of an online shopping site into a place for healthy communication.

[0638] The flow of the specific process in the application example 2 will be described with reference to FIG.

[0639] Step 1:

[0640] A user inputs a comment in the review section of an online shopping site and presses the send button. The input comment is sent from the user terminal to the server. The input data is the comment text input by the user. The server receives this comment text.

[0641] Step 2:

[0642] The server analyzes the received comment text and determines whether it is offensive. Specifically, it uses a natural language processing library (e.g., NLTK, spaCy) to analyze the tone and sentiment of the comment. The input data is the received comment text, and the output data is the judgment result of whether the comment is offensive or not.

[0643] Step 3:

[0644] The server converts comments that are judged as offensive into positive and kind words. Specifically, it uses OpenAI's generation API to convert aggressive comments into positive expressions. The input data is the comment text that is judged as offensive, and the output data is the comment text that has been converted into positive expressions.

[0645] Step 4:

[0646] The server transmits the converted comment text to the user terminal and displays it. The input data is the comment text converted into a positive comment, and the output data is the converted comment text displayed on the user terminal. The user terminal displays the received converted comment in the review section.

[0647] Through this series of processes, aggressive comments are transformed into positive expressions, which can then be displayed in the review section of the online shopping site.

[0648] In addition, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0649] "Example 1"

[0650] In the second embodiment of the present invention, in addition to the means for receiving comments from users, the means for determining whether the received comments are aggressive, and the means for converting aggressive comments into positive and kind words, an emotion engine for recognizing the emotions of users is built in. This emotion engine uses natural language processing technology and machine learning to read emotions from user comments. For example, emotions such as "anger" and "dissatisfaction" are recognized from a comment such as "your service is the worst." Then, based on this emotion information, it is determined with higher accuracy whether the comment is aggressive. Furthermore, the emotion engine is also utilized when converting aggressive comments. Specifically, a comment including the emotion of "anger" such as "your service is the worst" is converted into an expression that softens the emotional element, such as "your service has room for improvement." This makes it possible to properly understand the emotions of the user and take appropriate measures based on them.

[0651] "Example 2"

[0652] In the second embodiment of the present invention, in addition to the means for receiving comments from users, the means for determining whether the received comments are aggressive, and the means for converting aggressive comments into positive and kind words, an emotion engine for recognizing the emotions of users is built in. This emotion engine uses natural language processing technology and machine learning to read emotions from user comments. For example, emotions such as "anger" and "dissatisfaction" are recognized from a comment such as "your service is the worst." Then, based on this emotion information, it is determined with higher accuracy whether the comment is aggressive. Furthermore, the emotion engine is also utilized when converting aggressive comments. Specifically, a comment including the emotion of "anger" such as "your service is the worst" is converted into an expression that softens the emotional element, such as "your service has room for improvement." This makes it possible to properly understand the emotions of the user and take appropriate measures based on them.

[0653] The process flow of each embodiment will be described below.

[0654] "Example 1"

[0655] Step 1: Receive user comments. These comments can come in a variety of forms, such as a feedback form on your website or a comment section on social media.

[0656] Step 2: The received comments are input into the emotion engine, which uses natural language processing technology and machine learning to read the user's emotions from the comments.

[0657] Step 3: Based on the emotion information obtained from the emotion engine, determine whether the comment is offensive. This determination is made based on the type and intensity of the emotion recognized by the emotion engine.

[0658] Step 4: If a comment is deemed offensive, we transform it into positive, kinder language, taking into account the emotions recognized by the sentiment engine, and converting it into a more toned-down version of the comment.

[0659] Step 5: The converted comment is output. This comment retains the content of the original comment but removes any offensive elements. This allows the service provider to reduce the mental burden when receiving feedback from users.

[0660] Example 1

[0661] Next, a description will be given of Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[0662] In communication on the Internet, comments from users often contain offensive content, which can cause discomfort and stress to other users and service providers. It is also difficult to properly process aggressive comments and convert them into positive, kind words. Another challenge is to accurately recognize the sentiment of the comments and respond appropriately based on that sentiment.

[0663] The identification process by the identification processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are aggressive, a means for converting aggressive comments into positive and kind words, a means for analyzing the tone and emotions of the comments using natural language processing technology, and a means for converting the expression while maintaining the content of the comments using a generation API. This makes it possible to appropriately analyze the user's comments and convert aggressive content into positive expressions. In addition, by using an emotion engine, it is possible to accurately recognize the emotions of the comments and to make more accurate judgments and conversions based on the emotions.

[0664] "User" means an individual or entity that submits comments using the System.

[0665] "Comment" means a text message that a User sends via a feedback form on the Website or Application or through a comment section on a social networking site.

[0666] A "server" is a computer system for receiving, parsing, and optionally transforming comments from users.

[0667] "Natural language processing technology" is a technology that enables computers to understand, analyze, and generate human language.

[0668] A "machine learning model" is an algorithm that learns from data and makes predictions and classifications.

[0669] A "generation API" is an application programming interface for generating natural language text for particular input.

[0670] An "emotion engine" is a system that recognizes emotions from user comments and analyzes and converts the comments based on that information.

[0671] "Tone" refers to the overall mood or feeling of a comment.

[0672] An "offensive comment" is a comment that includes content that causes discomfort or stress to other users or service providers.

[0673] "Positive and kind words" are those that soften aggressive comments and turn them into constructive, friendly language.

[0674] The present invention relates to a system for receiving comments from users and converting aggressive comments into positive and kind words. A specific embodiment of this system will be described below.

[0675] Hardware and Software Configuration

[0676] The server is a computer system for receiving, analyzing, and optionally converting comments from users. The server uses the following software and libraries:

[0677] Natural Language Processing (NLP): Analyze the tone and sentiment of comments using Python libraries NLTK and spaCy.

[0678] Machine learning model: As an emotion engine, we use the Python libraries TextBlob and VADER to recognize emotions from user comments.

[0679] Generative API: Uses generative AI models such as OpenAI’s GPT-3 to transform comments while preserving their content.

[0680] System Operation

[0681] 1. Receiving comments

[0682] Users can submit comments via feedback forms on websites and applications, or through comment sections on social media.

[0683] The server receives comments from users through these means. Specifically, the server receives an HTTP request and extracts the comment data.

[0684] 2. Comment Analysis

[0685] The server analyzes the received comments using natural language processing (NLP) and machine learning models. Specifically, the server tokenizes the comment text and inputs it into a sentiment analysis model.

[0686] An emotion engine is incorporated to recognize emotions such as "anger" or "dissatisfaction" from user comments.

[0687] 3. Judging offensive comments

[0688] The server uses the emotion engine to more accurately determine whether a comment is offensive. Specifically, if the emotion engine detects emotions such as "anger" or "dissatisfaction," the server determines the comment is offensive.

[0689] 4. Converting offensive comments

[0690] When the server detects an offensive comment, it uses the generation API to convert the comment's expression while preserving its content. Specifically, the server sends the offensive comment to the generation API as a prompt text and receives the converted comment.

[0691] Examples

[0692] Example of an embodiment

[0693] User comment: "Your service is the worst."

[0694] Server action:

[0695] Receive comments.

[0696] Comments are analyzed using NLTK and spaCy and deemed offensive.

[0697] Send the prompt text "Please convert the following comment into positive and kind words. Comment: 'Your service is the worst'" to the generation API.

[0698] Receive the converted comment "Your service has room for improvement" from the generation API.

[0699] Example of the second form

[0700] User comment: "Your service is the worst."

[0701] Server action:

[0702] Receive comments.

[0703] Comments are analyzed using TextBlob and VADER to recognize emotions such as "anger" and "dissatisfaction."

[0704] The sentiment engine determines that the comment is offensive.

[0705] Send the prompt text "Please convert the following comment into positive and kind words. Comment: 'Your service is the worst'" to the generation API.

[0706] Receive the converted comment "Your service has room for improvement" from the generation API.

[0707] In this way, we have created a system that properly analyzes user comments and converts offensive content into positive expressions. By using an emotion engine, we can accurately recognize the emotions in comments and make more accurate judgments and conversions based on those emotions.

[0708] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0709] Step 1: Receiving comments

[0710] Users can submit comments via feedback forms on websites and applications, or through comment sections on social media.

[0711] The server receives the HTTP request and extracts the comment data.

[0712] Input: User submitted comment (e.g. "Your service is terrible")

[0713] Output: Extracted comment data

[0714] Step 2: Analyzing comments

[0715] The server analyzes the received comments using natural language processing technology (NLP) and machine learning models.

[0716] How it works: The server uses Python libraries NLTK and spaCy to tokenize the comment text and input it into a sentiment analysis model.

[0717] Input: Extracted comment data

[0718] Output: An analysis of the comments' tone and sentiment.

[0719] Step 3: Recognize emotions

[0720] The server uses an emotion engine to recognize emotions such as "anger" or "dissatisfaction" from user comments.

[0721] Specific operation: The server analyzes sentiment using Python libraries TextBlob and VADER.

[0722] Input: Analysis of the comment's tone and sentiment

[0723] Output: Recognized emotion information (e.g. anger, frustration)

[0724] Step 4: Identifying offensive comments

[0725] The server uses a sentiment engine to determine whether a comment is offensive.

[0726] What it does: If the emotion engine detects emotions like "anger" or "frustration," it marks the comment as offensive.

[0727] Input: Recognized emotion information

[0728] Output: Judgment result of whether the comment is offensive (e.g., offensive)

[0729] Step 5: Convert offensive comments

[0730] When the server detects an offensive comment, it uses a generation API to convert the expression while preserving the content of the comment.

[0731] Specific behavior: The server sends the offensive comment as a prompt to the generation API and receives the converted comment.

[0732] Input: Offensive comment (e.g. "Your service is terrible")

[0733] Output: The converted comment (e.g. "Your service could be improved")

[0734] In this way, we have created a system that properly analyzes user comments and converts offensive content into positive expressions. By using an emotion engine, we can accurately recognize the emotions in comments and make more accurate judgments and conversions based on those emotions.

[0735] (Application example 1)

[0736] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal".

[0737] Conventional customer support systems had problems with poor user experience due to difficulties in responding appropriately to aggressive comments from users. In addition, there was a lack of technology to accurately recognize users' emotions and convert comments based on them, so users' dissatisfaction could not be effectively alleviated.

[0738] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0739] In this invention, the server includes a means for receiving comments from a user, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, a means for recognizing the user's emotions, a means for converting the comments based on the recognized emotions, and a means for being installed in a smartphone application. This makes it possible to convert the offensive comments from the user into positive expressions and to appropriately understand and respond to the user's emotions.

[0740] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[0741] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[0742] "Methods for transforming aggressive comments into positive and kind words" are techniques for transforming aggressive comments into more positive and kind expressions while preserving their content.

[0743] "Means for recognizing user emotions" refers to technology for reading emotions from user comments and identifying those emotions.

[0744] The "means for converting a comment based on a recognized emotion" is a technique for converting a comment into an appropriate expression based on the recognized emotion of the user.

[0745] "Means for being installed in a smartphone application" refers to technology for being incorporated into an application running on a smartphone.

[0746] As a form for implementing this invention, a system is constructed for a customer support chatbot on an online shopping site that receives comments from users and converts aggressive comments into positive expressions.

[0747] System Programming

[0748] The system consists of the following main components:

[0749] 1. Comment Receiving Module: An interface for receiving comments from users. This is implemented as a chatbot function in a smartphone application.

[0750] 2. Sentiment Analysis Module: Natural language processing techniques to determine whether the comments received are offensive or not. Specifically, it uses natural language processing libraries such as spaCy and NLTK.

[0751] 3. Comment Transformation Module: A generative AI model for transforming offensive comments into positive ones. Specifically, we use generative APIs such as OpenAI's GPT-3.

[0752] 4. Emotion Recognition Module: Technology to read and identify emotions from user comments. This also uses natural language processing libraries.

[0753] 5. Smartphone application: A platform to integrate the above modules and make them accessible to users.

[0754] Processing Description

[0755] When the server receives a comment from a user, it first analyzes the content of the comment using the sentiment analysis module. If the comment is determined to be offensive based on the analysis results, the comment conversion module converts the comment into a positive expression using a generative AI model. At this time, the emotion recognition module identifies the user's emotion and performs the conversion based on that emotion. Finally, the converted comment is returned to the user through the smartphone application.

[0756] Hardware and Software Used

[0757] Hardware: Servers, smartphones

[0758] Software: spaCy, NLTK, OpenAI GPT-3, smartphone applications

[0759] Examples

[0760] For example, if a user submits a comment saying, "Your service is terrible," the sentiment analysis module determines the comment as "aggressive." The emotion recognition module then identifies that the comment contains the emotion of "anger." The comment transformation module uses a generative AI model to transform the comment into a positive expression, such as, "Your service could use some improvement."

[0761] Examples of prompt statements

[0762] Please convert the following offensive comments into positive ones: Your service is awful.

[0763] In this way, it is possible to convert aggressive comments from users into positive expressions and appropriately understand and respond to the user's feelings.

[0764] The flow of the specific process in the application example 1 will be described with reference to FIG.

[0765] Step 1:

[0766] A user inputs and submits a comment through a smartphone application. The input comment is sent to the server. The input data is the user's comment text.

[0767] Step 2:

[0768] The server passes the received comments to a sentiment analysis module, which uses a natural language processing library (e.g., spaCy or NLTK) to analyze the content of the comment and determine whether it is offensive. The input data is the user's comment text, and the output data is the judgment result of whether the comment is offensive or not.

[0769] Step 3:

[0770] The server receives the output of the sentiment analysis module, and if the comment is determined to be offensive, passes the comment to the emotion recognition module, which uses natural language processing techniques to identify the user's emotion from the comment. The input data is the user's comment text, and the output data is the identified emotion information.

[0771] Step 4:

[0772] The server receives the output of the emotion recognition module and passes the comment and emotion information to the comment conversion module, which uses a generative AI model (e.g., OpenAI's GPT-3) to convert aggressive comments into positive expressions. The input data is the user's comment text and emotion information, and the output data is the converted positive comment text.

[0773] Step 5:

[0774] The server returns the converted comment to the smartphone application, which displays the converted comment to the user. The input data is the converted positive comment text, and the output data is the comment displayed on the user's screen.

[0775] In this way, it is possible to convert aggressive comments from users into positive expressions and appropriately understand and respond to the user's feelings.

[0776] Example 2

[0777] Next, a description will be given of Example 2 of the embodiment 2. In the following description, the data processing device 12 is referred to as a "server" and the smart glasses 214 are referred to as a "terminal".

[0778] When a user's comment on the Internet contains offensive content, it can cause discomfort and stress to other users and service providers. There is also the issue that it is difficult to properly process offensive comments and convert them into positive expressions. Furthermore, there is a need to accurately recognize the sentiment of the comments and respond appropriately based on that.

[0779] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0780] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, a means for recognizing emotions from the user's comments, a means for determining whether the comments are offensive based on the recognized emotional information, and a means for converting the offensive comments based on the recognized emotional information. This makes it possible to appropriately understand the emotions of the user and convert the offensive comments into positive expressions.

[0781] "User" means an individual or entity that submits comments using the System.

[0782] A "comment" is text data such as an opinion, impression, or feedback that a user sends through the system.

[0783] A "server" is a computer system that receives comments from users, analyzes them, and performs the necessary processing.

[0784] The "means for receiving" is a function that enables the server to receive comments sent by users.

[0785] The "means of judgment" is a function for analyzing and determining whether a received comment is offensive or not.

[0786] "Means of transformation" is a function for transforming aggressive comments into positive and kind words.

[0787] "Means for recognizing emotions" is a function for analyzing and recognizing emotions from user comments.

[0788] “Generation API” means an application programming interface for utilizing external generative AI models to generate or transform text.

[0789] A "database" is a data management system for storing data such as comments and analysis results.

[0790] This invention is a system that receives comments from users, judges whether the comments are offensive, and converts the offensive comments into positive and kind words. It also includes a function to recognize emotions from the user's comments, judge whether the comments are offensive based on the emotional information, and convert them appropriately.

[0791] Hardware and Software Used

[0792] The system uses the following hardware and software:

[0793] Server: A computer system used to receive, parse, and transform comments, including a web server (e.g., Apache, Nginx) to process HTTP requests, and a database (e.g., MySQL, PostgreSQL).

[0794] Natural language processing technology: Libraries and APIs for analyzing the content of comments and determining their tone and sentiment. Specifically, we use the Python NLTK library and the Google Cloud Natural Language API.

[0795] Generative API: A generative AI model for converting offensive comments into positive and kind words. Specifically, it uses OpenAI's GPT-3 and GPT-4.

[0796] Explanation of program processing

[0797] 1. Receiving comments

[0798] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site.

[0799] The server receives the HTTP request, extracts the comment data, and stores it in a database.

[0800] 2. Judging offensive comments

[0801] The server retrieves the comments from the database and analyzes the content of the comments using natural language processing techniques.

[0802] The server uses Python's NLTK library and the Google Cloud Natural Language API to determine the tone and sentiment of the comments.

[0803] The server determines whether a comment is offensive and stores the result as a flag in a database.

[0804] 3. Emotion Recognition

[0805] The server retrieves comments from the database and analyzes their sentiment using natural language processing technology and machine learning.

[0806] The server uses Python's NLTK library and the Google Cloud Natural Language API to recognize emotions such as "anger" or "dissatisfaction" from the comments.

[0807] The server stores the recognized emotion information in a database.

[0808] 4. Converting offensive comments

[0809] The server sends comments determined to be offensive to the generation API.

[0810] The server uses OpenAI's GPT-3 and GPT-4 to convert offensive comments into positive, kind words.

[0811] The server stores the converted comment in its database, replacing the original comment.

[0812] Examples of concrete examples and prompts

[0813] Examples

[0814] A user enters the comment "Your service is terrible" into a feedback form on a website and clicks the submit button.

[0815] The server receives the HTTP POST request, extracts the comment data, and stores it in a database.

[0816] The server uses Python's NLTK library to analyze the tone and sentiment of comments and determine which ones are offensive.

[0817] The server uses OpenAI's GPT-3 to translate the comment into "Your service could use some improvement."

[0818] The server stores the converted comment in its database, replacing the original comment.

[0819] Examples of prompt statements

[0820] "Please translate the following comment into something positive and kind: 'Your service is awful.'"

[0821] "Recognize the emotion in the following comment and tone down the offensive language: 'Your service is terrible.'"

[0822] In this way, it is possible to properly understand the user's emotions and convert aggressive comments into positive expressions.

[0823] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0824] Step 1:

[0825] Receiving comments

[0826] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site.

[0827] Input: Comment entered by the user (e.g. "Your service is terrible").

[0828] The server receives the HTTP POST request and extracts the comment data.

[0829] Output: Extracted comment data (e.g. "Your service is the worst").

[0830] Specific operation: The server stores the received comments in a database.

[0831] Step 2:

[0832] Judging offensive comments

[0833] The server retrieves the comments from the database.

[0834] Input: Comment data stored in database (e.g. "Your service is terrible").

[0835] The server analyzes the content of the comments using natural language processing technology.

[0836] How it works: The server uses Python's NLTK library and the Google Cloud Natural Language API to determine the tone and sentiment of the comments.

[0837] Output: A judgement as to whether the comment is offensive (e.g. "Offensive").

[0838] The server stores the result of the determination in a database as a flag.

[0839] Step 3:

[0840] Emotion recognition

[0841] The server retrieves the comments from the database.

[0842] Input: Comment data stored in database (e.g. "Your service is terrible").

[0843] The server uses natural language processing technology and machine learning to analyze the sentiment of the comments.

[0844] How it works: The server uses Python's NLTK library and the Google Cloud Natural Language API to recognize emotions such as "anger" or "dissatisfaction" from comments.

[0845] Output: Recognized emotion information (e.g., "anger").

[0846] The server stores the recognized emotion information in a database.

[0847] Step 4:

[0848] Converting offensive comments

[0849] The server sends comments determined to be offensive to the generation API.

[0850] Input: aggressive comments (e.g., “your service is the worst”) and perceived emotion information (e.g., “anger”).

[0851] The server uses generative AI models (e.g., OpenAI's GPT-3 and GPT-4) to transform offensive comments into positive, kind words.

[0852] Specific operation: The server sends a prompt text to the generation API and receives the converted comment.

[0853] Output: The transformed comment (e.g. "Your service could be improved").

[0854] The server stores the converted comment in its database, replacing the original comment.

[0855] (Application example 2)

[0856] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server", and the smart glasses 214 will be referred to as a "terminal".

[0857] Conventional systems have difficulty in responding appropriately to aggressive comments from users, and responses to comments that contain emotional elements are often insufficient. In addition, when converting aggressive comments into positive expressions, it is necessary to convert them into appropriate expressions while retaining the content of the comment, but this is difficult to achieve. Furthermore, there is a demand to implement these functions in smartphone applications.

[0858] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0859] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are aggressive, a means for converting aggressive comments into positive and kind words, a means for recognizing the user's emotions, a means for determining the aggressiveness of the comments based on the emotion information, a means for converting aggressive comments based on the emotion information, and a means for being installed in a smartphone application. This makes it possible to appropriately respond to aggressive comments from users, accurately determine comments that include emotional elements, and convert them into positive expressions.

[0860] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[0861] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[0862] "Means of converting aggressive comments into positive and kind words" refers to a technique for converting aggressive comments into positive and kind expressions while retaining their content.

[0863] "Means for recognizing user emotions" refers to technology for reading emotions from users' comments and analyzing those emotions.

[0864] The "means for determining the aggressiveness of a comment based on emotional information" is a technique for determining with higher accuracy whether a comment is aggressive or not based on the emotional information of a user.

[0865] The "means for converting aggressive comments based on emotional information" is a technique for converting aggressive comments into positive expressions by taking into account the emotional information of the user.

[0866] "Means to be installed in a smartphone application" refers to technology for incorporating this system into an application running on a smartphone.

[0867] As an embodiment of the present invention, a method for installing a system in a smartphone application that receives comments from users, determines whether the comments are offensive, and converts the offensive comments into positive, kind words will be described.

[0868] First, a user inputs a comment through a smartphone application. The comment is then sent to the server through an interface within the application. The server then analyzes the received comment using natural language processing technology (e.g., spaCy) to understand the content of the comment.

[0869] The server then uses a machine learning model (e.g., the sentiment analysis pipeline in the transformers library) to determine whether the comment is offensive or not. At the same time, the user's sentiment is also analyzed to obtain sentiment information, which is then used to more accurately determine the offensiveness of the comment.

[0870] If a comment is deemed offensive, the server uses a generative AI model (e.g., GPT-2) to convert the offensive comment into positive, kind words. Specifically, the server inputs the following prompt sentences into the generative AI model:

[0871] Please convert the following offensive comments into positive ones: Your service is awful.

[0872] Positive comments:

[0873] The generative AI model uses this prompt to convert aggressive comments into positive ones, which are then fed back to the user.

[0874] This system responds appropriately to aggressive comments from users, and can accurately identify comments that contain emotional elements and convert them into positive expressions. This is expected to improve the user experience and reduce problems caused by aggressive comments.

[0875] For example, if a user enters a comment like "Your service is the worst," the system will convert this comment into a positive expression like "Your service could be improved." In this way, the system can properly understand the user's sentiment and take appropriate action based on it.

[0876] The flow of the specific process in the application example 2 will be described with reference to FIG.

[0877] Step 1:

[0878] A user inputs a comment through a smartphone application. The input comment is sent to a server through an interface within the application. The input data is the user's comment, and the output data is the comment sent to the server.

[0879] Step 2:

[0880] The server analyzes the received comments using natural language processing technology (e.g., spaCy). Specifically, it performs tokenization of the comments, part-of-speech tagging, dependency analysis, etc. The input data are the received comments, and the output data is the structural information of the analyzed comments.

[0881] Step 3:

[0882] The server uses a machine learning model (e.g., a sentiment analysis pipeline from the transformers library) to determine whether a comment is offensive. The input data is the parsed comment structure information, and the output data is a judgment result about the offensiveness of the comment.

[0883] Step 4:

[0884] The server simultaneously analyzes the user's emotions and obtains the emotion information. Using the emotion analysis pipeline, emotions such as "anger" and "dissatisfaction" are extracted from the comments. The input data is the structure information of the analyzed comments, and the output data is the user's emotion information.

[0885] Step 5:

[0886] The server judges the offensiveness of the comment more accurately based on the emotion information. It integrates the emotion information and the offensiveness judgment result to make a final offensiveness judgment. The input data are the emotion information and the offensiveness judgment result, and the output data is the final offensiveness judgment result.

[0887] Step 6:

[0888] If a comment is deemed offensive, the server uses a generative AI model (e.g., GPT-2) to convert the offensive comment into positive, kind words. Specifically, the server inputs the following prompt sentences into the generative AI model:

[0889] Please convert the following offensive comments into positive ones: Your service is awful.

[0890] Positive comments:

[0891] The input data are offensive comments, and the output data are the comments transformed into positive expressions.

[0892] Step 7:

[0893] The converted comments are fed back from the server to the smartphone application. The user can check the comments that have been converted into positive expressions. The input data are the comments that have been converted into positive expressions, and the output data are the comments that have been fed back to the user.

[0894] 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 a voice indicating a user input for the result of the specific processing. The control unit 46A transmits the voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0895] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ) and other generation AIs. The data generation model 58 is obtained by making a neural network perform deep learning. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating a voice, text data indicating a text, and image data indicating an image is input. The data generation model 58 performs inference on the input inference data according to the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0896] Another example of generative AI is Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) are mentioned.

[0897] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0898] [Third embodiment]

[0899] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0900] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0901] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a wide area network (WAN) and / or a local area network (LAN).

[0902] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0903] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs the voice according to instructions from the processor 46.

[0904] Camera 42 is a small digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (e.g., an imaging range defined by an angle of view equivalent to the width of the field of vision of an average healthy person).

[0905] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for transmitting and receiving various types of information between the processor 46 and the processor 28 via the network 54. The transmission and reception of various types of information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is performed in a secure state.

[0906] Fig. 6 shows an example of main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0907] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32, and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0908] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0909] In the headset type terminal 314, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50, and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0910] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[0911] "Example 1"

[0912] As one embodiment of the present invention, a feedback form on a website or application, a comment section on a social networking site, or the like is used as a means for receiving comments from users. As a means for determining whether a received comment is aggressive, natural language processing technology or machine learning is used to analyze the content of the comment and determine its tone and emotion. Then, as a means for converting aggressive comments into positive and kind words, a generation API is used to convert the expression of the comment while keeping its content. Specifically, "Your service is the worst" is converted to "Your service has room for improvement."

[0913] "Example 2"

[0914] As one embodiment of the present invention, a feedback form on a website or application, a comment section on a social networking site, or the like is used as a means for receiving comments from users. As a means for determining whether a received comment is aggressive, natural language processing technology or machine learning is used to analyze the content of the comment and determine its tone and emotion. Then, as a means for converting aggressive comments into positive and kind words, a generation API is used to convert the expression of the comment while keeping its content. Specifically, "Your service is the worst" is converted to "Your service has room for improvement."

[0915] The process flow of each embodiment will be described below.

[0916] "Example 1"

[0917] Step 1: Receive comments from users. In this step, you collect comments from users through feedback forms on your website or application, or through comments sections on social media.

[0918] Step 2: Determine whether the comment received is offensive. In this step, natural language processing technology and machine learning are used to analyze the content of the comment and determine its tone and sentiment. For example, a comment like "Your service is the worst" would be deemed offensive.

[0919] Step 3: Convert aggressive comments to positive and kind words. In this step, we use the generation API to convert the expression of the comment while preserving its content. For example, we convert a comment such as "Your service is the worst" to "Your service has room for improvement."

[0920] Example 1

[0921] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal".

[0922] In online communication, if a user makes an aggressive comment, that comment may cause discomfort to other users and service providers. There is a need to provide a healthy communication environment by converting such aggressive comments into positive and kind words. With conventional systems, it was difficult to properly detect aggressive comments and convert them into positive expressions while preserving their content.

[0923] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0924] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for displaying the converted comments to the user. This makes it possible to appropriately detect offensive comments and convert them into positive expressions while preserving their contents.

[0925] "User" refers to an individual or organization that uses the system to enter comments.

[0926] "Comment" refers to any text information that a User enters into a feedback form on a Website or Application, or into a comment section on a social networking site.

[0927] "Server" refers to a computer system that receives, analyzes, and converts comments from users.

[0928] "Means for receiving" refers to a function that enables the server to receive comments sent by users.

[0929] "Means of determining" refers to the function for analyzing and evaluating whether a received comment is offensive.

[0930] "Means of transformation" refers to the function of transforming aggressive comments into positive, kind words.

[0931] "Means for displaying" refers to a function for displaying the converted comment to the user.

[0932] "Offensive comments" refers to negative comments that may cause offense to other users or service providers.

[0933] "Positive and kind words" refers to words that preserve the content of aggressive comments but transform them into more constructive and friendly language.

[0934] "Generative AI Model" refers to an artificial intelligence model that generates and transforms text using natural language processing techniques.

[0935] A "prompt" refers to an instruction entered into a generative AI model to cause it to perform a specific task.

[0936] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0937] The present invention relates to a system that receives comments from users, determines whether the comments are offensive, and converts the offensive comments into positive and kind words. A specific embodiment of this system will be described below.

[0938] Hardware and Software Used

[0939] The servers include a web server and an application server for receiving comments from users. These servers receive comment data through HTTP requests. The servers analyze the comments using Python natural language processing libraries NLTK and spaCy. They also use generative AI models (e.g., OpenAI's GPT-3) to convert offensive comments into positive expressions.

[0940] Program Processing

[0941] A user enters a comment into a feedback form on a website or application, or into a comment section on a social networking site. For example, a user might enter "your service is the worst." When a user submits a comment, the server receives the comment. The server receives the comment data through an HTTP request.

[0942] The server then uses a natural language processing library, such as Python's NLTK or spaCy, to parse the comments it receives. The server analyzes the tone and sentiment of the comment to determine whether it is offensive or not. For example, it might recognize the word "terrible" as offensive.

[0943] The server will determine whether the comment is offensive based on the analysis results. If it is determined to be offensive, it will proceed to the next step. For example, a comment like "Your service is the worst" will be detected as offensive.

[0944] The server uses a generative AI model (e.g., OpenAI's GPT-3) to convert aggressive comments into positive ones. The server sends a prompt to the generative AI model, which converts the comment's expression while preserving its content. For example, the comment "Your service is terrible" is converted to "Your service has room for improvement."

[0945] Specific prompt examples:

[0946] Original comment: "Your service is terrible"

[0947] Prompt: "Transform the following offensive comment into a positive, kind statement: 'Your service is awful.'"

[0948] Finally, the server displays the converted comments received from the generative AI model to the user. The user can see the converted comments in the feedback form or in the comments section of a social networking site. For example, a comment such as "Your service has room for improvement" may be displayed.

[0949] In this way, a system is realized that converts aggressive comments from users into positive expressions.

[0950] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0951] Step 1:

[0952] The user enters a comment

[0953] A user enters a comment into a feedback form on a website or application, or into a comment section on a social networking site. For example, a user enters "Your service is the worst." The comment is sent to the server by pressing the send button.

[0954] Step 2:

[0955] The server receives the comment.

[0956] The server receives comments sent by users through HTTP requests. The received comments are stored in a database or temporarily held in memory. The input is the user's comments, and the output is the comment data stored on the server.

[0957] Step 3:

[0958] The server parses the comments

[0959] The server uses Python natural language processing libraries NLTK and spaCy to analyze the received comments. Specifically, it performs tokenization of comments, part-of-speech tagging, sentiment analysis, etc. The input is the comment data stored on the server, and the output is the analysis results (e.g., the tone and sentiment of the comments).

[0960] Step 4:

[0961] The server detects offensive comments

[0962] The server determines whether the comment is offensive based on the analysis result. If it is determined to be offensive, it proceeds to the next step. For example, if the comment contains the word "worst", it is determined to be offensive. The input is the analysis result, and the output is the detection result of the offensive comment.

[0963] Step 5:

[0964] The server converts the comments

[0965] The server uses a generative AI model (e.g., OpenAI's GPT-3) to convert offensive comments into positive expressions. The server sends a prompt to the generative AI model, which converts the expression while preserving the content of the comment. For example, the comment "Your service is terrible" is converted to "Your service could be improved." The input is the offensive comment and the prompt, and the output is the converted comment.

[0966] Specific prompt examples:

[0967] Original comment: "Your service is terrible"

[0968] Prompt: "Transform the following offensive comment into a positive, kind statement: 'Your service is awful.'"

[0969] Step 6:

[0970] The server displays the converted comment to the user.

[0971] The server displays the converted comments received from the generative AI model to the user. The user can see the converted comments in a feedback form or in a comment section on a social networking site. For example, a comment such as "Your service has room for improvement" is displayed. The input is the converted comment, and the output is the comment displayed to the user.

[0972] (Application example 1)

[0973] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server", and the headset type terminal 314 will be referred to as a "terminal".

[0974] In conventional customer support systems, aggressive comments from users are sent directly to the support staff, which often causes stress for the support staff. In addition, it is difficult to respond appropriately to aggressive comments, which makes communication with users less smooth.

[0975] The identification process by the identification processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for providing the converted comments to a customer support representative. This enables the customer support representative to receive constructive feedback without feeling stressed.

[0976] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[0977] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[0978] A "means of transforming aggressive comments into positive and kind words" is a technique for transforming aggressive comments into positive and kind words while preserving their content.

[0979] The "means for providing the converted comments to the customer support representative" refers to an interface for displaying or notifying the converted comments to the customer support representative.

[0980] "Generation API" means an application programming interface for generating text using a generative AI model.

[0981] A "customer support representative" is a person who handles inquiries and feedback from users.

[0982] A system for implementing the present invention includes means for receiving comments from a user, means for determining whether the received comments are offensive, means for converting the offensive comments into positive, kind words, and means for providing the converted comments to a customer support representative.

[0983] The server provides an interface, such as a feedback form on a website or application or a comment section on a social networking site, to receive comments from users. When a user enters a comment, the comment is sent to the server.

[0984] The server uses natural language processing techniques to analyze the received comments. Specifically, it uses the TextBlob library to analyze the sentiment of the comment and determine whether the comment is offensive. If the sentiment score is negative, the comment is deemed offensive.

[0985] The server then uses a generative AI model to convert offensive comments into positive, kind words. Specifically, it uses OpenAI's generative API to convert the wording of the offensive comment while preserving its content. For example, it converts a comment like "Your service is terrible" to "Your service could use some improvement."

[0986] The converted comments are provided to a customer support representative. The server provides an interface for displaying the converted comments on the customer support representative's terminal. This allows the customer support representative to receive constructive feedback without feeling stressed about offensive comments.

[0987] As a concrete example, if a user comments, "Your service is terrible," the system will translate that comment into, "Your service could be improved," and provide that to a customer support representative.

[0988] Example of a prompt to be input to a generative AI model:

[0989] Please convert the following comments into positive and kind words: Your service is awful.

[0990] In this way, the system converts offensive comments from users into positive feedback, enabling customer support agents to respond more effectively.

[0991] The flow of the specific process in the application example 1 will be described with reference to FIG.

[0992] Step 1:

[0993] A user enters a comment. A user enters a comment using a feedback form on a website or application, or a comment section on a social networking site. The comment is sent to the server. The input is the user's comment, and the output is the comment sent to the server.

[0994] Step 2:

[0995] The server receives the comments. The server receives the comments sent by the user and stores them in a database. The input is the comment sent by the user, and the output is the stored comment data.

[0996] Step 3:

[0997] The server analyzes the sentiment of the comments. The server uses the TextBlob library to analyze the sentiment of the received comments. Specifically, it calculates the positive / negative score of the comment and judges whether the comment is offensive based on the score. The input is the stored comment data, and the output is the sentiment score and the judgment result of whether the comment is offensive or not.

[0998] Step 4:

[0999] The server converts aggressive comments. The server uses OpenAI's generation API to convert aggressive comments into positive and kind words. Specifically, it inputs a prompt sentence to the generative AI model and obtains the converted comment. The input is the comment judged to be aggressive and the prompt sentence, and the output is the converted comment.

[1000] Step 5:

[1001] The server provides the converted comments to the customer support representative. The server provides an interface for displaying the converted comments on the customer support representative's terminal. The input is the converted comments and the output is the comments displayed on the customer support representative's terminal.

[1002] Step 6:

[1003] The customer support representative checks the converted comments and responds. The customer support representative checks the positive comments displayed on the terminal and takes appropriate action. The input is the converted comments displayed on the terminal, and the output is the customer support representative's response action.

[1004] Example 2

[1005] Next, a second embodiment of the second form will be described. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal".

[1006] Comments from users on the Internet often contain offensive content. Such offensive comments cause discomfort to other users and service providers, and cause a decline in the quality of communication. Conventional systems have had difficulty automatically detecting and appropriately processing offensive comments. Therefore, the present invention aims to provide a system that automatically detects aggressive comments from users and converts them into positive, kind words.

[1007] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for storing the received comments in a database, a means for analyzing the stored comments using natural language processing technology and a machine learning model, a means for determining whether the comments are aggressive based on the analysis results, a means for converting aggressive comments into positive and kind words using a generative AI model, and a means for storing the converted comments in a database. This makes it possible to automatically detect aggressive comments from users and convert them into positive expressions.

[1008] A "User" is an individual or entity that submits a Comment to the System.

[1009] A "comment" is an opinion, feedback, or message provided by a user to the system.

[1010] A "database" is a system for systematically storing and managing data such as comments and analysis results.

[1011] "Natural language processing technology" is a technology that enables computers to understand, interpret, and generate human language.

[1012] A "machine learning model" is an algorithm or system that learns from data to perform a specific task.

[1013] "Analysis" is the process of tokenizing the content of comments and evaluating their sentiment and tone.

[1014] A "generative AI model" is an artificial intelligence model that generates new text based on input text.

[1015] "Aggression" refers to words or expressions that are offensive or hostile to others.

[1016] "Positive and kind words" are words and expressions that give a favorable and friendly impression to others.

[1017] "Transformation" is the process of replacing aggressive comments with positive, kind words.

[1018] The present invention relates to a system for automatically detecting offensive comments from users and converting them into positive and kind words. This system is implemented using the following hardware and software.

[1019] 1. Receiving comments

[1020] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site, and the device sends the comment to the server as an HTTP POST request.

[1021] 2. Saving comments

[1022] The server stores the received comments in a database, typically using a database management system such as MySQL or PostgreSQL.

[1023] 3. Comment Analysis

[1024] The server analyzes the stored comments using natural language processing (NLP) and machine learning models, tokenizing the comments and performing sentiment analysis using the Python libraries NLTK and spaCy.

[1025] 4. Judging offensive comments

[1026] The server uses the analysis results to determine whether a comment is offensive. For example, if it contains negative words such as "the worst" or "horrible," it will be deemed offensive.

[1027] 5. Comment Conversion

[1028] The server inputs the comments that are judged to be offensive into a generative AI model (e.g., OpenAI's GPT-3) and converts them into positive and kind expressions. Using a generative API, the expression is changed while preserving the content of the comment. An example of a specific prompt sentence is "Please convert the following offensive comment into a positive and kind expression: 'Your service is the worst.'"

[1029] 6. Saving comments after conversion

[1030] The server saves the converted comment back into the database, so that both the original comment and the converted comment are stored.

[1031] Examples

[1032] For example, if a user sends a comment saying "Your service is the worst," the server receives this comment and analyzes it using natural language processing technology. Since it is determined to be an offensive comment, "Your service is the worst" is input into the generative AI model. The generative AI model converts this into a more positive expression, "Your service has room for improvement." The server saves the converted comment in a database and provides feedback to the user.

[1033] In this way, the present invention is able to automatically detect offensive comments from users and convert them into positive expressions.

[1034] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1035] Step 1: Receiving comments

[1036] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site. The device sends this comment to the server as an HTTP POST request. The input is the user's comment, and the output is the comment data sent to the server.

[1037] Step 2: Saving comments

[1038] The server stores the received comments in a database, typically using a database management system such as MySQL or PostgreSQL. The input is the comment data sent to the server, and the output is the comment records stored in the database.

[1039] Step 3: Analyzing comments

[1040] The server analyzes the stored comments using natural language processing (NLP) technology and machine learning models. It uses Python libraries NLTK and spaCy to tokenize comments and perform sentiment analysis. The input is the comment data obtained from the database, and the output is the analysis results.

[1041] Step 4: Identifying offensive comments

[1042] The server uses the analysis results to determine whether a comment is offensive. For example, if a comment contains negative words such as "the worst" or "horrible," it is deemed offensive. The input is the analysis results, and the output is the judgment result of whether the comment is offensive or not.

[1043] Step 5: Converting comments

[1044] The server inputs the comments judged as offensive into a generative AI model (e.g., OpenAI's GPT-3) and converts them into positive and kind expressions. Using a generative API, the expression is changed while preserving the content of the comment. An example of a specific prompt sentence is "Please convert the following offensive comment into a positive and kind expression: 'Your service is the worst.'" The input is the comment judged as offensive, and the output is the converted positive comment.

[1045] Step 6: Save the converted comments

[1046] The server stores the converted comment back in the database, thus storing both the original and the converted comment. The input is the converted positive comment, and the output is the converted comment record stored in the database.

[1047] (Application example 2)

[1048] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server", and the headset type terminal 314 will be referred to as a "terminal".

[1049] On conventional online shopping sites, feedback and reviews from users may contain offensive content, which may have a negative impact on other users. Such offensive comments can negatively impact the overall atmosphere of the site and reduce other users' desire to purchase. Therefore, there is a need to convert offensive comments into positive expressions and provide a healthy communication environment.

[1050] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1051] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for displaying the converted comments to the user. This makes it possible to convert the offensive comments into positive expressions and turn the review section of the online shopping site into a place for healthy communication.

[1052] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[1053] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[1054] "Methods for transforming aggressive comments into positive and kind words" are techniques for transforming aggressive comments into positive and kind expressions while preserving their content.

[1055] The "means for displaying the converted comment to the user" refers to an interface for visually presenting the converted comment to the user.

[1056] "Generation API" means an application programming interface for performing natural language generation for a particular input.

[1057] A system for implementing the present invention includes a series of processes for receiving comments from a user, determining whether the comment is offensive, converting the offensive comment into positive, kind words, and displaying the converted comment to the user.

[1058] Hardware and Software Used

[1059] Hardware: Servers, user devices (PCs, smartphones, etc.)

[1060] Software: Python, OpenAI API, Natural Language Processing libraries (e.g. NLTK, spaCy)

[1061] Data processing and calculation

[1062] 1. Receiving comments:

[1063] When a user enters a comment in the review section of an online shopping site and presses the send button, the comment is sent to the server.

[1064] The server receives the comments from the user using a means for receiving the comments.

[1065] 2. Judging offensive comments:

[1066] The server analyzes the content of the received comments using a method to determine whether the comments are offensive.

[1067] Specifically, it uses natural language processing technology and machine learning models to analyze the tone and sentiment of comments and determine whether they contain offensive keywords.

[1068] 3. Comment conversion:

[1069] The server uses a method to convert offensive comments into positive, kind words while preserving the content of the comments.

[1070] Specifically, it uses OpenAI's generative API to transform offensive comments into positive ones.

[1071] 4. Viewing comments after conversion:

[1072] The server transmits the converted comment to the user terminal and displays it using a means for displaying the converted comment to the user.

[1073] Examples

[1074] Original comment: "This product is awful and completely unusable."

[1075] Post-conversion comment: "This product could use some improvement, but it may be useful depending on how you use it."

[1076] Examples of prompt statements

[1077] Please convert the following comment into a positive one: This product is terrible, I can't use it at all.

[1078] In this way, by transforming aggressive comments into positive expressions, it is possible to turn the review section of an online shopping site into a place for healthy communication.

[1079] The flow of the specific process in the application example 2 will be described with reference to FIG.

[1080] Step 1:

[1081] A user inputs a comment in the review section of an online shopping site and presses the send button. The input comment is sent from the user terminal to the server. The input data is the comment text input by the user. The server receives this comment text.

[1082] Step 2:

[1083] The server analyzes the received comment text and determines whether it is offensive. Specifically, it uses a natural language processing library (e.g., NLTK, spaCy) to analyze the tone and sentiment of the comment. The input data is the received comment text, and the output data is the judgment result of whether the comment is offensive or not.

[1084] Step 3:

[1085] The server converts comments that are judged as offensive into positive and kind words. Specifically, it uses OpenAI's generation API to convert aggressive comments into positive expressions. The input data is the comment text that is judged as offensive, and the output data is the comment text that has been converted into positive expressions.

[1086] Step 4:

[1087] The server transmits the converted comment text to the user terminal and displays it. The input data is the comment text converted into a positive comment, and the output data is the converted comment text displayed on the user terminal. The user terminal displays the received converted comment in the review section.

[1088] Through this series of processes, aggressive comments are transformed into positive expressions, which can then be displayed in the review section of the online shopping site.

[1089] In addition, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1090] "Example 1"

[1091] In the second embodiment of the present invention, in addition to the means for receiving comments from users, the means for determining whether the received comments are aggressive, and the means for converting aggressive comments into positive and kind words, an emotion engine for recognizing the emotions of users is built in. This emotion engine uses natural language processing technology and machine learning to read emotions from user comments. For example, emotions such as "anger" and "dissatisfaction" are recognized from a comment such as "your service is the worst." Then, based on this emotion information, it is determined with higher accuracy whether the comment is aggressive. Furthermore, the emotion engine is also utilized when converting aggressive comments. Specifically, a comment including the emotion of "anger" such as "your service is the worst" is converted into an expression that softens the emotional element, such as "your service has room for improvement." This makes it possible to properly understand the emotions of the user and take appropriate measures based on them.

[1092] "Example 2"

[1093] In the second embodiment of the present invention, in addition to the means for receiving comments from users, the means for determining whether the received comments are aggressive, and the means for converting aggressive comments into positive and kind words, an emotion engine for recognizing the emotions of users is built in. This emotion engine uses natural language processing technology and machine learning to read emotions from user comments. For example, emotions such as "anger" and "dissatisfaction" are recognized from a comment such as "your service is the worst." Then, based on this emotion information, it is determined with higher accuracy whether the comment is aggressive. Furthermore, the emotion engine is also utilized when converting aggressive comments. Specifically, a comment including the emotion of "anger" such as "your service is the worst" is converted into an expression that softens the emotional element, such as "your service has room for improvement." This makes it possible to properly understand the emotions of the user and take appropriate measures based on them.

[1094] The process flow of each embodiment will be described below.

[1095] "Example 1"

[1096] Step 1: Receive user comments. These comments can come in a variety of forms, such as a feedback form on your website or a comment section on social media.

[1097] Step 2: The received comments are input into the emotion engine, which uses natural language processing technology and machine learning to read the user's emotions from the comments.

[1098] Step 3: Based on the emotion information obtained from the emotion engine, determine whether the comment is offensive. This determination is made based on the type and intensity of the emotion recognized by the emotion engine.

[1099] Step 4: If a comment is deemed offensive, we transform it into positive, kinder language, taking into account the emotions recognized by the sentiment engine, and converting it into a more toned-down version of the comment.

[1100] Step 5: The converted comment is output. This comment retains the content of the original comment but removes any offensive elements. This allows the service provider to reduce the mental burden when receiving feedback from users.

[1101] Example 1

[1102] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal".

[1103] In communication on the Internet, comments from users often contain offensive content, which can cause discomfort and stress to other users and service providers. It is also difficult to properly process aggressive comments and convert them into positive, kind words. Another challenge is to accurately recognize the sentiment of the comments and respond appropriately based on that sentiment.

[1104] The identification process by the identification processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are aggressive, a means for converting aggressive comments into positive and kind words, a means for analyzing the tone and emotions of the comments using natural language processing technology, and a means for converting the expression while maintaining the content of the comments using a generation API. This makes it possible to appropriately analyze the user's comments and convert aggressive content into positive expressions. In addition, by using an emotion engine, it is possible to accurately recognize the emotions of the comments and to make more accurate judgments and conversions based on the emotions.

[1105] "User" means an individual or entity that submits comments using the System.

[1106] "Comment" means a text message that a User sends via a feedback form on the Website or Application or through a comment section on a social networking site.

[1107] A "server" is a computer system for receiving, parsing, and optionally transforming comments from users.

[1108] "Natural language processing technology" is a technology that enables computers to understand, analyze, and generate human language.

[1109] A "machine learning model" is an algorithm that learns from data and makes predictions and classifications.

[1110] A "generation API" is an application programming interface for generating natural language text for particular input.

[1111] An "emotion engine" is a system that recognizes emotions from user comments and analyzes and converts the comments based on that information.

[1112] "Tone" refers to the overall mood or feeling of a comment.

[1113] An "offensive comment" is a comment that includes content that causes discomfort or stress to other users or service providers.

[1114] "Positive and kind words" are those that soften aggressive comments and turn them into constructive, friendly language.

[1115] The present invention relates to a system for receiving comments from users and converting aggressive comments into positive and kind words. A specific embodiment of this system will be described below.

[1116] Hardware and Software Configuration

[1117] The server is a computer system for receiving, analyzing, and optionally converting comments from users. The server uses the following software and libraries:

[1118] Natural Language Processing (NLP): Analyze the tone and sentiment of comments using Python libraries NLTK and spaCy.

[1119] Machine learning model: As an emotion engine, we use the Python libraries TextBlob and VADER to recognize emotions from user comments.

[1120] Generative API: Uses generative AI models such as OpenAI’s GPT-3 to transform comments while preserving their content.

[1121] System Operation

[1122] 1. Receiving comments

[1123] Users can submit comments via feedback forms on websites and applications, or through comment sections on social media.

[1124] The server receives comments from users through these means. Specifically, the server receives an HTTP request and extracts the comment data.

[1125] 2. Comment Analysis

[1126] The server analyzes the received comments using natural language processing (NLP) and machine learning models. Specifically, the server tokenizes the comment text and inputs it into a sentiment analysis model.

[1127] An emotion engine is incorporated to recognize emotions such as "anger" or "dissatisfaction" from user comments.

[1128] 3. Judging offensive comments

[1129] The server uses the emotion engine to more accurately determine whether a comment is offensive. Specifically, if the emotion engine detects emotions such as "anger" or "dissatisfaction," the server determines the comment is offensive.

[1130] 4. Converting offensive comments

[1131] When the server detects an offensive comment, it uses the generation API to convert the comment's expression while preserving its content. Specifically, the server sends the offensive comment to the generation API as a prompt text and receives the converted comment.

[1132] Examples

[1133] Example of an embodiment

[1134] User comment: "Your service is the worst."

[1135] Server action:

[1136] Receive comments.

[1137] Comments are analyzed using NLTK and spaCy and deemed offensive.

[1138] Send the prompt text "Please convert the following comment into positive and kind words. Comment: 'Your service is the worst'" to the generation API.

[1139] Receive the converted comment "Your service has room for improvement" from the generation API.

[1140] Example of the second form

[1141] User comment: "Your service is the worst."

[1142] Server action:

[1143] Receive comments.

[1144] Comments are analyzed using TextBlob and VADER to recognize emotions such as "anger" and "dissatisfaction."

[1145] The sentiment engine determines that the comment is offensive.

[1146] Send the prompt text "Please convert the following comment into positive and kind words. Comment: 'Your service is the worst'" to the generation API.

[1147] Receive the converted comment "Your service has room for improvement" from the generation API.

[1148] In this way, we have created a system that properly analyzes user comments and converts offensive content into positive expressions. By using an emotion engine, we can accurately recognize the emotions in comments and make more accurate judgments and conversions based on those emotions.

[1149] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1150] Step 1: Receiving comments

[1151] Users can submit comments via feedback forms on websites and applications, or through comment sections on social media.

[1152] The server receives the HTTP request and extracts the comment data.

[1153] Input: User submitted comment (e.g. "Your service is terrible")

[1154] Output: Extracted comment data

[1155] Step 2: Analyzing comments

[1156] The server analyzes the received comments using natural language processing technology (NLP) and machine learning models.

[1157] How it works: The server uses Python libraries NLTK and spaCy to tokenize the comment text and input it into a sentiment analysis model.

[1158] Input: Extracted comment data

[1159] Output: An analysis of the comments' tone and sentiment.

[1160] Step 3: Recognize emotions

[1161] The server uses an emotion engine to recognize emotions such as "anger" or "dissatisfaction" from user comments.

[1162] Specific operation: The server analyzes sentiment using Python libraries TextBlob and VADER.

[1163] Input: Analysis of the comment's tone and sentiment

[1164] Output: Recognized emotion information (e.g. anger, frustration)

[1165] Step 4: Identifying offensive comments

[1166] The server uses a sentiment engine to determine whether a comment is offensive.

[1167] What it does: If the emotion engine detects emotions like "anger" or "frustration," it marks the comment as offensive.

[1168] Input: Recognized emotion information

[1169] Output: Judgment result of whether the comment is offensive (e.g., offensive)

[1170] Step 5: Convert offensive comments

[1171] When the server detects an offensive comment, it uses a generation API to convert the expression while preserving the content of the comment.

[1172] Specific behavior: The server sends the offensive comment as a prompt to the generation API and receives the converted comment.

[1173] Input: Offensive comment (e.g. "Your service is terrible")

[1174] Output: The converted comment (e.g. "Your service could be improved")

[1175] In this way, we have created a system that properly analyzes user comments and converts offensive content into positive expressions. By using an emotion engine, we can accurately recognize the emotions in comments and make more accurate judgments and conversions based on those emotions.

[1176] (Application example 1)

[1177] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server", and the headset type terminal 314 will be referred to as a "terminal".

[1178] Conventional customer support systems had problems with poor user experience due to difficulties in responding appropriately to aggressive comments from users. In addition, there was a lack of technology to accurately recognize users' emotions and convert comments based on them, so users' dissatisfaction could not be effectively alleviated.

[1179] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1180] In this invention, the server includes a means for receiving comments from a user, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, a means for recognizing the user's emotions, a means for converting the comments based on the recognized emotions, and a means for being installed in a smartphone application. This makes it possible to convert the offensive comments from the user into positive expressions and to appropriately understand and respond to the user's emotions.

[1181] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[1182] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[1183] "Methods for transforming aggressive comments into positive and kind words" are techniques for transforming aggressive comments into more positive and kind expressions while preserving their content.

[1184] "Means for recognizing user emotions" refers to technology for reading emotions from user comments and identifying those emotions.

[1185] The "means for converting a comment based on a recognized emotion" is a technique for converting a comment into an appropriate expression based on the recognized emotion of the user.

[1186] "Means for being installed in a smartphone application" refers to technology for being incorporated into an application running on a smartphone.

[1187] As a form for implementing this invention, a system is constructed for a customer support chatbot on an online shopping site that receives comments from users and converts aggressive comments into positive expressions.

[1188] System Programming

[1189] The system consists of the following main components:

[1190] 1. Comment Receiving Module: An interface for receiving comments from users. This is implemented as a chatbot function in a smartphone application.

[1191] 2. Sentiment Analysis Module: Natural language processing techniques to determine whether the comments received are offensive or not. Specifically, it uses natural language processing libraries such as spaCy and NLTK.

[1192] 3. Comment Transformation Module: A generative AI model for transforming offensive comments into positive ones. Specifically, we use generative APIs such as OpenAI's GPT-3.

[1193] 4. Emotion Recognition Module: Technology to read and identify emotions from user comments. This also uses natural language processing libraries.

[1194] 5. Smartphone application: A platform to integrate the above modules and make them accessible to users.

[1195] Processing Description

[1196] When the server receives a comment from a user, it first analyzes the content of the comment using the sentiment analysis module. If the comment is determined to be offensive based on the analysis results, the comment conversion module converts the comment into a positive expression using a generative AI model. At this time, the emotion recognition module identifies the user's emotion and performs the conversion based on that emotion. Finally, the converted comment is returned to the user through the smartphone application.

[1197] Hardware and Software Used

[1198] Hardware: Servers, smartphones

[1199] Software: spaCy, NLTK, OpenAI GPT-3, smartphone applications

[1200] Examples

[1201] For example, if a user submits a comment saying, "Your service is terrible," the sentiment analysis module determines the comment as "aggressive." The emotion recognition module then identifies that the comment contains the emotion of "anger." The comment transformation module uses a generative AI model to transform the comment into a positive expression, such as, "Your service could use some improvement."

[1202] Examples of prompt statements

[1203] Please convert the following offensive comments into positive ones: Your service is awful.

[1204] In this way, it is possible to convert aggressive comments from users into positive expressions and appropriately understand and respond to the user's feelings.

[1205] The flow of the specific process in the application example 1 will be described with reference to FIG.

[1206] Step 1:

[1207] A user inputs and submits a comment through a smartphone application. The input comment is sent to the server. The input data is the user's comment text.

[1208] Step 2:

[1209] The server passes the received comments to a sentiment analysis module, which uses a natural language processing library (e.g., spaCy or NLTK) to analyze the content of the comment and determine whether it is offensive. The input data is the user's comment text, and the output data is the judgment result of whether the comment is offensive or not.

[1210] Step 3:

[1211] The server receives the output of the sentiment analysis module, and if the comment is determined to be offensive, passes the comment to the emotion recognition module, which uses natural language processing techniques to identify the user's emotion from the comment. The input data is the user's comment text, and the output data is the identified emotion information.

[1212] Step 4:

[1213] The server receives the output of the emotion recognition module and passes the comment and emotion information to the comment conversion module, which uses a generative AI model (e.g., OpenAI's GPT-3) to convert aggressive comments into positive expressions. The input data is the user's comment text and emotion information, and the output data is the converted positive comment text.

[1214] Step 5:

[1215] The server returns the converted comment to the smartphone application, which displays the converted comment to the user. The input data is the converted positive comment text, and the output data is the comment displayed on the user's screen.

[1216] In this way, it is possible to convert aggressive comments from users into positive expressions and appropriately understand and respond to the user's feelings.

[1217] Example 2

[1218] Next, a second embodiment of the second form will be described. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal".

[1219] When a user's comment on the Internet contains offensive content, it can cause discomfort and stress to other users and service providers. There is also the issue that it is difficult to properly process offensive comments and convert them into positive expressions. Furthermore, there is a need to accurately recognize the sentiment of the comments and respond appropriately based on that.

[1220] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1221] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, a means for recognizing emotions from the user's comments, a means for determining whether the comments are offensive based on the recognized emotional information, and a means for converting the offensive comments based on the recognized emotional information. This makes it possible to appropriately understand the emotions of the user and convert the offensive comments into positive expressions.

[1222] "User" means an individual or entity that submits comments using the System.

[1223] A "comment" is text data such as an opinion, impression, or feedback that a user sends through the system.

[1224] A "server" is a computer system that receives comments from users, analyzes them, and performs the necessary processing.

[1225] The "means for receiving" is a function that enables the server to receive comments sent by users.

[1226] The "means of judgment" is a function for analyzing and determining whether a received comment is offensive or not.

[1227] "Means of transformation" is a function for transforming aggressive comments into positive and kind words.

[1228] "Means for recognizing emotions" is a function for analyzing and recognizing emotions from user comments.

[1229] “Generation API” means an application programming interface for utilizing external generative AI models to generate or transform text.

[1230] A "database" is a data management system for storing data such as comments and analysis results.

[1231] This invention is a system that receives comments from users, judges whether the comments are offensive, and converts the offensive comments into positive and kind words. It also includes a function to recognize emotions from the user's comments, judge whether the comments are offensive based on the emotional information, and convert them appropriately.

[1232] Hardware and Software Used

[1233] The system uses the following hardware and software:

[1234] Server: A computer system used to receive, parse, and transform comments, including a web server (e.g., Apache, Nginx) to process HTTP requests, and a database (e.g., MySQL, PostgreSQL).

[1235] Natural language processing technology: Libraries and APIs for analyzing the content of comments and determining their tone and sentiment. Specifically, we use the Python NLTK library and the Google Cloud Natural Language API.

[1236] Generative API: A generative AI model for converting offensive comments into positive and kind words. Specifically, it uses OpenAI's GPT-3 and GPT-4.

[1237] Explanation of program processing

[1238] 1. Receiving comments

[1239] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site.

[1240] The server receives the HTTP request, extracts the comment data, and stores it in a database.

[1241] 2. Judging offensive comments

[1242] The server retrieves the comments from the database and analyzes the content of the comments using natural language processing techniques.

[1243] The server uses Python's NLTK library and the Google Cloud Natural Language API to determine the tone and sentiment of the comments.

[1244] The server determines whether a comment is offensive and stores the result as a flag in a database.

[1245] 3. Emotion Recognition

[1246] The server retrieves comments from the database and analyzes their sentiment using natural language processing technology and machine learning.

[1247] The server uses Python's NLTK library and the Google Cloud Natural Language API to recognize emotions such as "anger" or "dissatisfaction" from the comments.

[1248] The server stores the recognized emotion information in a database.

[1249] 4. Converting offensive comments

[1250] The server sends comments determined to be offensive to the generation API.

[1251] The server uses OpenAI's GPT-3 and GPT-4 to convert offensive comments into positive, kind words.

[1252] The server stores the converted comment in its database, replacing the original comment.

[1253] Examples of concrete examples and prompts

[1254] Examples

[1255] A user enters the comment "Your service is terrible" into a feedback form on a website and clicks the submit button.

[1256] The server receives the HTTP POST request, extracts the comment data, and stores it in a database.

[1257] The server uses Python's NLTK library to analyze the tone and sentiment of comments and determine which ones are offensive.

[1258] The server uses OpenAI's GPT-3 to translate the comment into "Your service could use some improvement."

[1259] The server stores the converted comment in its database, replacing the original comment.

[1260] Examples of prompt statements

[1261] "Please translate the following comment into something positive and kind: 'Your service is awful.'"

[1262] "Recognize the emotion in the following comment and tone down the offensive language: 'Your service is terrible.'"

[1263] In this way, it is possible to properly understand the user's emotions and convert aggressive comments into positive expressions.

[1264] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1265] Step 1:

[1266] Receiving comments

[1267] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site.

[1268] Input: Comment entered by the user (e.g. "Your service is terrible").

[1269] The server receives the HTTP POST request and extracts the comment data.

[1270] Output: Extracted comment data (e.g. "Your service is the worst").

[1271] Specific operation: The server stores the received comments in a database.

[1272] Step 2:

[1273] Judging offensive comments

[1274] The server retrieves the comments from the database.

[1275] Input: Comment data stored in database (e.g. "Your service is terrible").

[1276] The server analyzes the content of the comments using natural language processing technology.

[1277] How it works: The server uses Python's NLTK library and the Google Cloud Natural Language API to determine the tone and sentiment of the comments.

[1278] Output: A judgement as to whether the comment is offensive (e.g. "Offensive").

[1279] The server stores the result of the determination in a database as a flag.

[1280] Step 3:

[1281] Emotion recognition

[1282] The server retrieves the comments from the database.

[1283] Input: Comment data stored in database (e.g. "Your service is terrible").

[1284] The server uses natural language processing technology and machine learning to analyze the sentiment of the comments.

[1285] How it works: The server uses Python's NLTK library and the Google Cloud Natural Language API to recognize emotions such as "anger" or "dissatisfaction" from comments.

[1286] Output: Recognized emotion information (e.g., "anger").

[1287] The server stores the recognized emotion information in a database.

[1288] Step 4:

[1289] Converting offensive comments

[1290] The server sends comments determined to be offensive to the generation API.

[1291] Input: aggressive comments (e.g., “your service is the worst”) and perceived emotion information (e.g., “anger”).

[1292] The server uses generative AI models (e.g., OpenAI's GPT-3 and GPT-4) to transform offensive comments into positive, kind words.

[1293] Specific operation: The server sends a prompt text to the generation API and receives the converted comment.

[1294] Output: The transformed comment (e.g. "Your service could be improved").

[1295] The server stores the converted comment in its database, replacing the original comment.

[1296] (Application example 2)

[1297] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server", and the headset type terminal 314 will be referred to as a "terminal".

[1298] Conventional systems have difficulty in responding appropriately to aggressive comments from users, and responses to comments that contain emotional elements are often insufficient. In addition, when converting aggressive comments into positive expressions, it is necessary to convert them into appropriate expressions while retaining the content of the comment, but this is difficult to achieve. Furthermore, there is a demand to implement these functions in smartphone applications.

[1299] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1300] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are aggressive, a means for converting aggressive comments into positive and kind words, a means for recognizing the user's emotions, a means for determining the aggressiveness of the comments based on the emotion information, a means for converting aggressive comments based on the emotion information, and a means for being installed in a smartphone application. This makes it possible to appropriately respond to aggressive comments from users, accurately determine comments that include emotional elements, and convert them into positive expressions.

[1301] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[1302] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[1303] "Means of converting aggressive comments into positive and kind words" refers to a technique for converting aggressive comments into positive and kind expressions while retaining their content.

[1304] "Means for recognizing user emotions" refers to technology for reading emotions from users' comments and analyzing those emotions.

[1305] The "means for determining the aggressiveness of a comment based on emotional information" is a technique for determining with higher accuracy whether a comment is aggressive or not based on the emotional information of a user.

[1306] The "means for converting aggressive comments based on emotional information" is a technique for converting aggressive comments into positive expressions by taking into account the emotional information of the user.

[1307] "Means to be installed in a smartphone application" refers to technology for incorporating this system into an application running on a smartphone.

[1308] As an embodiment of the present invention, a method for installing a system in a smartphone application that receives comments from users, determines whether the comments are offensive, and converts the offensive comments into positive, kind words will be described.

[1309] First, a user inputs a comment through a smartphone application. The comment is then sent to the server through an interface within the application. The server then analyzes the received comment using natural language processing technology (e.g., spaCy) to understand the content of the comment.

[1310] The server then uses a machine learning model (e.g., the sentiment analysis pipeline in the transformers library) to determine whether the comment is offensive or not. At the same time, the user's sentiment is also analyzed to obtain sentiment information, which is then used to more accurately determine the offensiveness of the comment.

[1311] If a comment is deemed offensive, the server uses a generative AI model (e.g., GPT-2) to convert the offensive comment into positive, kind words. Specifically, the server inputs the following prompt sentences into the generative AI model:

[1312] Please convert the following offensive comments into positive ones: Your service is awful.

[1313] Positive comments:

[1314] The generative AI model uses this prompt to convert aggressive comments into positive ones, which are then fed back to the user.

[1315] This system responds appropriately to aggressive comments from users, and can accurately identify comments that contain emotional elements and convert them into positive expressions. This is expected to improve the user experience and reduce problems caused by aggressive comments.

[1316] For example, if a user enters a comment like "Your service is the worst," the system will convert this comment into a positive expression like "Your service could be improved." In this way, the system can properly understand the user's sentiment and take appropriate action based on it.

[1317] The flow of the specific process in the application example 2 will be described with reference to FIG.

[1318] Step 1:

[1319] A user inputs a comment through a smartphone application. The input comment is sent to a server through an interface within the application. The input data is the user's comment, and the output data is the comment sent to the server.

[1320] Step 2:

[1321] The server analyzes the received comments using natural language processing technology (e.g., spaCy). Specifically, it performs tokenization of the comments, part-of-speech tagging, dependency analysis, etc. The input data are the received comments, and the output data is the structural information of the analyzed comments.

[1322] Step 3:

[1323] The server uses a machine learning model (e.g., a sentiment analysis pipeline from the transformers library) to determine whether a comment is offensive. The input data is the parsed comment structure information, and the output data is a judgment result about the offensiveness of the comment.

[1324] Step 4:

[1325] The server simultaneously analyzes the user's emotions and obtains the emotion information. Using the emotion analysis pipeline, emotions such as "anger" and "dissatisfaction" are extracted from the comments. The input data is the structure information of the analyzed comments, and the output data is the user's emotion information.

[1326] Step 5:

[1327] The server judges the offensiveness of the comment more accurately based on the emotion information. It integrates the emotion information and the offensiveness judgment result to make a final offensiveness judgment. The input data are the emotion information and the offensiveness judgment result, and the output data is the final offensiveness judgment result.

[1328] Step 6:

[1329] If a comment is deemed offensive, the server uses a generative AI model (e.g., GPT-2) to convert the offensive comment into positive, kind words. Specifically, the server inputs the following prompt sentences into the generative AI model:

[1330] Please convert the following offensive comments into positive ones: Your service is awful.

[1331] Positive comments:

[1332] The input data are offensive comments, and the output data are the comments transformed into positive expressions.

[1333] Step 7:

[1334] The converted comments are fed back from the server to the smartphone application. The user can check the comments that have been converted into positive expressions. The input data are the comments that have been converted into positive expressions, and the output data are the comments that have been fed back to the user.

[1335] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input for the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1336] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ) and other generation AIs. The data generation model 58 is obtained by making a neural network perform deep learning. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating a voice, text data indicating a text, and image data indicating an image is input. The data generation model 58 performs inference on the input inference data according to the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1337] Another example of generative AI is Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) are mentioned.

[1338] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1339] [Fourth embodiment]

[1340] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1341] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1342] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a wide area network (WAN) and / or a local area network (LAN).

[1343] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. In addition, the microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1344] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs the voice according to instructions from the processor 46.

[1345] Camera 42 is a small digital camera equipped with an optical system including a lens, an aperture, and a shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (e.g., an imaging range defined by an angle of view equivalent to the width of the field of vision of an average healthy person).

[1346] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for transmitting and receiving various types of information between the processor 46 and the processor 28 via the network 54. The transmission and reception of various types of information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is performed in a secure state.

[1347] The control target 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, legs, etc. The posture and behavior of the robot 414 are controlled by controlling the motors of the arms, hands, legs, etc. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1348] Fig. 8 shows an example of main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1349] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32, and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1350] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1351] In the robot 414, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50, and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1352] Next, the specific processing by the specific processing unit 290 of the data processing device 12 will be described.

[1353] "Example 1"

[1354] As one embodiment of the present invention, a feedback form on a website or application, a comment section on a social networking site, or the like is used as a means for receiving comments from users. As a means for determining whether a received comment is aggressive, natural language processing technology or machine learning is used to analyze the content of the comment and determine its tone and emotion. Then, as a means for converting aggressive comments into positive and kind words, a generation API is used to convert the expression of the comment while keeping its content. Specifically, "Your service is the worst" is converted to "Your service has room for improvement."

[1355] "Example 2"

[1356] As one embodiment of the present invention, a feedback form on a website or application, a comment section on a social networking site, or the like is used as a means for receiving comments from users. As a means for determining whether a received comment is aggressive, natural language processing technology or machine learning is used to analyze the content of the comment and determine its tone and emotion. Then, as a means for converting aggressive comments into positive and kind words, a generation API is used to convert the expression of the comment while keeping its content. Specifically, "Your service is the worst" is converted to "Your service has room for improvement."

[1357] The process flow of each embodiment will be described below.

[1358] "Example 1"

[1359] Step 1: Receive comments from users. In this step, you collect comments from users through feedback forms on your website or application, or through comments sections on social media.

[1360] Step 2: Determine whether the comment received is offensive. In this step, natural language processing technology and machine learning are used to analyze the content of the comment and determine its tone and sentiment. For example, a comment like "Your service is the worst" would be deemed offensive.

[1361] Step 3: Convert aggressive comments to positive and kind words. In this step, we use the generation API to convert the expression of the comment while preserving its content. For example, we convert a comment such as "Your service is the worst" to "Your service has room for improvement."

[1362] Example 1

[1363] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal".

[1364] In online communication, if a user makes an aggressive comment, that comment may cause discomfort to other users and service providers. There is a need to provide a healthy communication environment by converting such aggressive comments into positive and kind words. With conventional systems, it was difficult to properly detect aggressive comments and convert them into positive expressions while preserving their content.

[1365] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1366] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for displaying the converted comments to the user. This makes it possible to appropriately detect offensive comments and convert them into positive expressions while preserving their contents.

[1367] "User" refers to an individual or organization that uses the system to enter comments.

[1368] "Comment" refers to any text information that a User enters into a feedback form on a Website or Application, or into a comment section on a social networking site.

[1369] "Server" refers to a computer system that receives, analyzes, and converts comments from users.

[1370] "Means for receiving" refers to a function that enables the server to receive comments sent by users.

[1371] "Means of determining" refers to the function for analyzing and evaluating whether a received comment is offensive.

[1372] "Means of transformation" refers to the function of transforming aggressive comments into positive, kind words.

[1373] "Means for displaying" refers to a function for displaying the converted comment to the user.

[1374] "Offensive comments" refers to negative comments that may cause offense to other users or service providers.

[1375] "Positive and kind words" refers to words that preserve the content of aggressive comments but transform them into more constructive and friendly language.

[1376] "Generative AI Model" refers to an artificial intelligence model that generates and transforms text using natural language processing techniques.

[1377] A "prompt" refers to an instruction entered into a generative AI model to cause it to perform a specific task.

[1378] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[1379] The present invention relates to a system that receives comments from users, determines whether the comments are offensive, and converts the offensive comments into positive and kind words. A specific embodiment of this system will be described below.

[1380] Hardware and Software Used

[1381] The servers include a web server and an application server for receiving comments from users. These servers receive comment data through HTTP requests. The servers analyze the comments using Python natural language processing libraries NLTK and spaCy. They also use generative AI models (e.g., OpenAI's GPT-3) to convert offensive comments into positive expressions.

[1382] Program Processing

[1383] A user enters a comment into a feedback form on a website or application, or into a comment section on a social networking site. For example, a user might enter "your service is the worst." When a user submits a comment, the server receives the comment. The server receives the comment data through an HTTP request.

[1384] The server then uses a natural language processing library, such as Python's NLTK or spaCy, to parse the comments it receives. The server analyzes the tone and sentiment of the comment to determine whether it is offensive or not. For example, it might recognize the word "terrible" as offensive.

[1385] The server will determine whether the comment is offensive based on the analysis results. If it is determined to be offensive, it will proceed to the next step. For example, a comment like "Your service is the worst" will be detected as offensive.

[1386] The server uses a generative AI model (e.g., OpenAI's GPT-3) to convert aggressive comments into positive ones. The server sends a prompt to the generative AI model, which converts the comment's expression while preserving its content. For example, the comment "Your service is terrible" is converted to "Your service has room for improvement."

[1387] Specific prompt examples:

[1388] Original comment: "Your service is terrible"

[1389] Prompt: "Transform the following offensive comment into a positive, kind statement: 'Your service is awful.'"

[1390] Finally, the server displays the converted comments received from the generative AI model to the user. The user can see the converted comments in the feedback form or in the comments section of a social networking site. For example, a comment such as "Your service has room for improvement" may be displayed.

[1391] In this way, a system is realized that converts aggressive comments from users into positive expressions.

[1392] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1393] Step 1:

[1394] The user enters a comment

[1395] A user enters a comment into a feedback form on a website or application, or into a comment section on a social networking site. For example, a user enters "Your service is the worst." The comment is sent to the server by pressing the send button.

[1396] Step 2:

[1397] The server receives the comment.

[1398] The server receives comments sent by users through HTTP requests. The received comments are stored in a database or temporarily held in memory. The input is the user's comments, and the output is the comment data stored on the server.

[1399] Step 3:

[1400] The server parses the comments

[1401] The server uses Python natural language processing libraries NLTK and spaCy to analyze the received comments. Specifically, it performs tokenization of comments, part-of-speech tagging, sentiment analysis, etc. The input is the comment data stored on the server, and the output is the analysis results (e.g., the tone and sentiment of the comments).

[1402] Step 4:

[1403] The server detects offensive comments

[1404] The server determines whether the comment is offensive based on the analysis result. If it is determined to be offensive, it proceeds to the next step. For example, if the comment contains the word "worst", it is determined to be offensive. The input is the analysis result, and the output is the detection result of the offensive comment.

[1405] Step 5:

[1406] The server converts the comments

[1407] The server uses a generative AI model (e.g., OpenAI's GPT-3) to convert offensive comments into positive expressions. The server sends a prompt to the generative AI model, which converts the expression while preserving the content of the comment. For example, the comment "Your service is terrible" is converted to "Your service could be improved." The input is the offensive comment and the prompt, and the output is the converted comment.

[1408] Specific prompt examples:

[1409] Original comment: "Your service is terrible"

[1410] Prompt: "Transform the following offensive comment into a positive, kind statement: 'Your service is awful.'"

[1411] Step 6:

[1412] The server displays the converted comment to the user.

[1413] The server displays the converted comments received from the generative AI model to the user. The user can see the converted comments in a feedback form or in a comment section on a social networking site. For example, a comment such as "Your service has room for improvement" is displayed. The input is the converted comment, and the output is the comment displayed to the user.

[1414] (Application example 1)

[1415] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal".

[1416] In conventional customer support systems, aggressive comments from users are sent directly to the support staff, which often causes stress for the support staff. In addition, it is difficult to respond appropriately to aggressive comments, which makes communication with users less smooth.

[1417] The identification process by the identification processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for providing the converted comments to a customer support representative. This enables the customer support representative to receive constructive feedback without feeling stressed.

[1418] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[1419] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[1420] A "means of transforming aggressive comments into positive and kind words" is a technique for transforming aggressive comments into positive and kind words while preserving their content.

[1421] The "means for providing the converted comments to the customer support representative" refers to an interface for displaying or notifying the converted comments to the customer support representative.

[1422] "Generation API" means an application programming interface for generating text using a generative AI model.

[1423] A "customer support representative" is a person who handles inquiries and feedback from users.

[1424] A system for implementing the present invention includes means for receiving comments from a user, means for determining whether the received comments are offensive, means for converting the offensive comments into positive, kind words, and means for providing the converted comments to a customer support representative.

[1425] The server provides an interface, such as a feedback form on a website or application or a comment section on a social networking site, to receive comments from users. When a user enters a comment, the comment is sent to the server.

[1426] The server uses natural language processing techniques to analyze the received comments. Specifically, it uses the TextBlob library to analyze the sentiment of the comment and determine whether the comment is offensive. If the sentiment score is negative, the comment is deemed offensive.

[1427] The server then uses a generative AI model to convert offensive comments into positive, kind words. Specifically, it uses OpenAI's generative API to convert the wording of the offensive comment while preserving its content. For example, it converts a comment like "Your service is terrible" to "Your service could use some improvement."

[1428] The converted comments are provided to a customer support representative. The server provides an interface for displaying the converted comments on the customer support representative's terminal. This allows the customer support representative to receive constructive feedback without feeling stressed about offensive comments.

[1429] As a concrete example, if a user comments, "Your service is terrible," the system will translate that comment into, "Your service could be improved," and provide that to a customer support representative.

[1430] Example of a prompt to be input to a generative AI model:

[1431] Please convert the following comments into positive and kind words: Your service is awful.

[1432] In this way, the system converts offensive comments from users into positive feedback, enabling customer support agents to respond more effectively.

[1433] The flow of the specific process in the application example 1 will be described with reference to FIG.

[1434] Step 1:

[1435] A user enters a comment. A user enters a comment using a feedback form on a website or application, or a comment section on a social networking site. The comment is sent to the server. The input is the user's comment, and the output is the comment sent to the server.

[1436] Step 2:

[1437] The server receives the comments. The server receives the comments sent by the user and stores them in a database. The input is the comment sent by the user, and the output is the stored comment data.

[1438] Step 3:

[1439] The server analyzes the sentiment of the comments. The server uses the TextBlob library to analyze the sentiment of the received comments. Specifically, it calculates the positive / negative score of the comment and judges whether the comment is offensive based on the score. The input is the stored comment data, and the output is the sentiment score and the judgment result of whether the comment is offensive or not.

[1440] Step 4:

[1441] The server converts aggressive comments. The server uses OpenAI's generation API to convert aggressive comments into positive and kind words. Specifically, it inputs a prompt sentence to the generative AI model and obtains the converted comment. The input is the comment judged to be aggressive and the prompt sentence, and the output is the converted comment.

[1442] Step 5:

[1443] The server provides the converted comments to the customer support representative. The server provides an interface for displaying the converted comments on the customer support representative's terminal. The input is the converted comments and the output is the comments displayed on the customer support representative's terminal.

[1444] Step 6:

[1445] The customer support representative checks the converted comments and responds. The customer support representative checks the positive comments displayed on the terminal and takes appropriate action. The input is the converted comments displayed on the terminal, and the output is the customer support representative's response action.

[1446] Example 2

[1447] Next, a second embodiment of the second embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal".

[1448] Comments from users on the Internet often contain offensive content. Such offensive comments cause discomfort to other users and service providers, and cause a decline in the quality of communication. Conventional systems have had difficulty automatically detecting and appropriately processing offensive comments. Therefore, the present invention aims to provide a system that automatically detects aggressive comments from users and converts them into positive, kind words.

[1449] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for storing the received comments in a database, a means for analyzing the stored comments using natural language processing technology and a machine learning model, a means for determining whether the comments are aggressive based on the analysis results, a means for converting aggressive comments into positive and kind words using a generative AI model, and a means for storing the converted comments in a database. This makes it possible to automatically detect aggressive comments from users and convert them into positive expressions.

[1450] A "User" is an individual or entity that submits a Comment to the System.

[1451] A "comment" is an opinion, feedback, or message provided by a user to the system.

[1452] A "database" is a system for systematically storing and managing data such as comments and analysis results.

[1453] "Natural language processing technology" is a technology that enables computers to understand, interpret, and generate human language.

[1454] A "machine learning model" is an algorithm or system that learns from data to perform a specific task.

[1455] "Analysis" is the process of tokenizing the content of comments and evaluating their sentiment and tone.

[1456] A "generative AI model" is an artificial intelligence model that generates new text based on input text.

[1457] "Aggression" refers to words or expressions that are offensive or hostile to others.

[1458] "Positive and kind words" are words and expressions that give a favorable and friendly impression to others.

[1459] "Transformation" is the process of replacing aggressive comments with positive, kind words.

[1460] The present invention relates to a system for automatically detecting offensive comments from users and converting them into positive and kind words. This system is implemented using the following hardware and software.

[1461] 1. Receiving comments

[1462] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site, and the device sends the comment to the server as an HTTP POST request.

[1463] 2. Saving comments

[1464] The server stores the received comments in a database, typically using a database management system such as MySQL or PostgreSQL.

[1465] 3. Comment Analysis

[1466] The server analyzes the stored comments using natural language processing (NLP) and machine learning models, tokenizing the comments and performing sentiment analysis using the Python libraries NLTK and spaCy.

[1467] 4. Judging offensive comments

[1468] The server uses the analysis results to determine whether a comment is offensive. For example, if it contains negative words such as "the worst" or "horrible," it will be deemed offensive.

[1469] 5. Comment Conversion

[1470] The server inputs the comments that are judged to be offensive into a generative AI model (e.g., OpenAI's GPT-3) and converts them into positive and kind expressions. Using a generative API, the expression is changed while preserving the content of the comment. An example of a specific prompt sentence is "Please convert the following offensive comment into a positive and kind expression: 'Your service is the worst.'"

[1471] 6. Saving comments after conversion

[1472] The server saves the converted comment back into the database, so that both the original comment and the converted comment are stored.

[1473] Examples

[1474] For example, if a user sends a comment saying "Your service is the worst," the server receives this comment and analyzes it using natural language processing technology. Since it is determined to be an offensive comment, "Your service is the worst" is input into the generative AI model. The generative AI model converts this into a more positive expression, "Your service has room for improvement." The server saves the converted comment in a database and provides feedback to the user.

[1475] In this way, the present invention is able to automatically detect offensive comments from users and convert them into positive expressions.

[1476] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1477] Step 1: Receiving comments

[1478] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site. The device sends this comment to the server as an HTTP POST request. The input is the user's comment, and the output is the comment data sent to the server.

[1479] Step 2: Saving comments

[1480] The server stores the received comments in a database, typically using a database management system such as MySQL or PostgreSQL. The input is the comment data sent to the server, and the output is the comment records stored in the database.

[1481] Step 3: Analyzing comments

[1482] The server analyzes the stored comments using natural language processing (NLP) technology and machine learning models. It uses Python libraries NLTK and spaCy to tokenize comments and perform sentiment analysis. The input is the comment data obtained from the database, and the output is the analysis results.

[1483] Step 4: Identifying offensive comments

[1484] The server uses the analysis results to determine whether a comment is offensive. For example, if a comment contains negative words such as "the worst" or "horrible," it is deemed offensive. The input is the analysis results, and the output is the judgment result of whether the comment is offensive or not.

[1485] Step 5: Converting comments

[1486] The server inputs the comments judged as offensive into a generative AI model (e.g., OpenAI's GPT-3) and converts them into positive and kind expressions. Using a generative API, the expression is changed while preserving the content of the comment. An example of a specific prompt sentence is "Please convert the following offensive comment into a positive and kind expression: 'Your service is the worst.'" The input is the comment judged as offensive, and the output is the converted positive comment.

[1487] Step 6: Save the converted comments

[1488] The server stores the converted comment back in the database, thus storing both the original and the converted comment. The input is the converted positive comment, and the output is the converted comment record stored in the database.

[1489] (Application example 2)

[1490] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal".

[1491] On conventional online shopping sites, feedback and reviews from users may contain offensive content, which may have a negative impact on other users. Such offensive comments can negatively impact the overall atmosphere of the site and reduce other users' desire to purchase. Therefore, there is a need to convert offensive comments into positive expressions and provide a healthy communication environment.

[1492] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1493] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, and a means for displaying the converted comments to the user. This makes it possible to convert the offensive comments into positive expressions and turn the review section of the online shopping site into a place for healthy communication.

[1494] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[1495] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[1496] "Methods for transforming aggressive comments into positive and kind words" are techniques for transforming aggressive comments into positive and kind expressions while preserving their content.

[1497] The "means for displaying the converted comment to the user" refers to an interface for visually presenting the converted comment to the user.

[1498] "Generation API" means an application programming interface for performing natural language generation for a particular input.

[1499] A system for implementing the present invention includes a series of processes for receiving comments from a user, determining whether the comment is offensive, converting the offensive comment into positive, kind words, and displaying the converted comment to the user.

[1500] Hardware and Software Used

[1501] Hardware: Servers, user devices (PCs, smartphones, etc.)

[1502] Software: Python, OpenAI API, Natural Language Processing libraries (e.g. NLTK, spaCy)

[1503] Data processing and calculation

[1504] 1. Receiving comments:

[1505] When a user enters a comment in the review section of an online shopping site and presses the send button, the comment is sent to the server.

[1506] The server receives the comments from the user using a means for receiving the comments.

[1507] 2. Judging offensive comments:

[1508] The server analyzes the content of the received comments using a method to determine whether the comments are offensive.

[1509] Specifically, it uses natural language processing technology and machine learning models to analyze the tone and sentiment of comments and determine whether they contain offensive keywords.

[1510] 3. Comment conversion:

[1511] The server uses a method to convert offensive comments into positive, kind words while preserving the content of the comments.

[1512] Specifically, it uses OpenAI's generative API to transform offensive comments into positive ones.

[1513] 4. Viewing comments after conversion:

[1514] The server transmits the converted comment to the user terminal and displays it using a means for displaying the converted comment to the user.

[1515] Examples

[1516] Original comment: "This product is awful and completely unusable."

[1517] Post-conversion comment: "This product could use some improvement, but it may be useful depending on how you use it."

[1518] Examples of prompt statements

[1519] Please convert the following comment into a positive one: This product is terrible, I can't use it at all.

[1520] In this way, by transforming aggressive comments into positive expressions, it is possible to turn the review section of an online shopping site into a place for healthy communication.

[1521] The flow of the specific process in the application example 2 will be described with reference to FIG.

[1522] Step 1:

[1523] A user inputs a comment in the review section of an online shopping site and presses the send button. The input comment is sent from the user terminal to the server. The input data is the comment text input by the user. The server receives this comment text.

[1524] Step 2:

[1525] The server analyzes the received comment text and determines whether it is offensive. Specifically, it uses a natural language processing library (e.g., NLTK, spaCy) to analyze the tone and sentiment of the comment. The input data is the received comment text, and the output data is the judgment result of whether the comment is offensive or not.

[1526] Step 3:

[1527] The server converts comments that are judged as offensive into positive and kind words. Specifically, it uses OpenAI's generation API to convert aggressive comments into positive expressions. The input data is the comment text that is judged as offensive, and the output data is the comment text that has been converted into positive expressions.

[1528] Step 4:

[1529] The server transmits the converted comment text to the user terminal and displays it. The input data is the comment text converted into a positive comment, and the output data is the converted comment text displayed on the user terminal. The user terminal displays the received converted comment in the review section.

[1530] Through this series of processes, aggressive comments are transformed into positive expressions, which can then be displayed in the review section of the online shopping site.

[1531] In addition, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1532] "Example 1"

[1533] In the second embodiment of the present invention, in addition to the means for receiving comments from users, the means for determining whether the received comments are aggressive, and the means for converting aggressive comments into positive and kind words, an emotion engine for recognizing the emotions of users is built in. This emotion engine uses natural language processing technology and machine learning to read emotions from user comments. For example, emotions such as "anger" and "dissatisfaction" are recognized from a comment such as "your service is the worst." Then, based on this emotion information, it is determined with higher accuracy whether the comment is aggressive. Furthermore, the emotion engine is also utilized when converting aggressive comments. Specifically, a comment including the emotion of "anger" such as "your service is the worst" is converted into an expression that softens the emotional element, such as "your service has room for improvement." This makes it possible to properly understand the emotions of the user and take appropriate measures based on them.

[1534] "Example 2"

[1535] In the second embodiment of the present invention, in addition to the means for receiving comments from users, the means for determining whether the received comments are aggressive, and the means for converting aggressive comments into positive and kind words, an emotion engine for recognizing the emotions of users is built in. This emotion engine uses natural language processing technology and machine learning to read emotions from user comments. For example, emotions such as "anger" and "dissatisfaction" are recognized from a comment such as "your service is the worst." Then, based on this emotion information, it is determined with higher accuracy whether the comment is aggressive. Furthermore, the emotion engine is also utilized when converting aggressive comments. Specifically, a comment including the emotion of "anger" such as "your service is the worst" is converted into an expression that softens the emotional element, such as "your service has room for improvement." This makes it possible to properly understand the emotions of the user and take appropriate measures based on them.

[1536] The process flow of each embodiment will be described below.

[1537] "Example 1"

[1538] Step 1: Receive user comments. These comments can come in a variety of forms, such as a feedback form on your website or a comment section on social media.

[1539] Step 2: The received comments are input into the emotion engine, which uses natural language processing technology and machine learning to read the user's emotions from the comments.

[1540] Step 3: Based on the emotion information obtained from the emotion engine, determine whether the comment is offensive. This determination is made based on the type and intensity of the emotion recognized by the emotion engine.

[1541] Step 4: If a comment is deemed offensive, we transform it into positive, kinder language, taking into account the emotions recognized by the sentiment engine, and converting it into a more toned-down version of the comment.

[1542] Step 5: The converted comment is output. This comment retains the content of the original comment but removes any offensive elements. This allows the service provider to reduce the mental burden when receiving feedback from users.

[1543] Example 1

[1544] Next, a description will be given of Example 1 of the embodiment 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal".

[1545] In communication on the Internet, comments from users often contain offensive content, which can cause discomfort and stress to other users and service providers. It is also difficult to properly process aggressive comments and convert them into positive, kind words. Another challenge is to accurately recognize the sentiment of the comments and respond appropriately based on that sentiment.

[1546] The identification process by the identification processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means. In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are aggressive, a means for converting aggressive comments into positive and kind words, a means for analyzing the tone and emotions of the comments using natural language processing technology, and a means for converting the expression while maintaining the content of the comments using a generation API. This makes it possible to appropriately analyze the user's comments and convert aggressive content into positive expressions. In addition, by using an emotion engine, it is possible to accurately recognize the emotions of the comments and to make more accurate judgments and conversions based on the emotions.

[1547] "User" means an individual or entity that submits comments using the System.

[1548] "Comment" means a text message that a User sends via a feedback form on the Website or Application or through a comment section on a social networking site.

[1549] A "server" is a computer system for receiving, parsing, and optionally transforming comments from users.

[1550] "Natural language processing technology" is a technology that enables computers to understand, analyze, and generate human language.

[1551] A "machine learning model" is an algorithm that learns from data and makes predictions and classifications.

[1552] A "generation API" is an application programming interface for generating natural language text for particular input.

[1553] An "emotion engine" is a system that recognizes emotions from user comments and analyzes and converts the comments based on that information.

[1554] "Tone" refers to the overall mood or feeling of a comment.

[1555] An "offensive comment" is a comment that includes content that causes discomfort or stress to other users or service providers.

[1556] "Positive and kind words" are those that soften aggressive comments and turn them into constructive, friendly language.

[1557] The present invention relates to a system for receiving comments from users and converting aggressive comments into positive and kind words. A specific embodiment of this system will be described below.

[1558] Hardware and Software Configuration

[1559] The server is a computer system for receiving, analyzing, and optionally converting comments from users. The server uses the following software and libraries:

[1560] Natural Language Processing (NLP): Analyze the tone and sentiment of comments using Python libraries NLTK and spaCy.

[1561] Machine learning model: As an emotion engine, we use the Python libraries TextBlob and VADER to recognize emotions from user comments.

[1562] Generative API: Uses generative AI models such as OpenAI’s GPT-3 to transform comments while preserving their content.

[1563] System Operation

[1564] 1. Receiving comments

[1565] Users can submit comments via feedback forms on websites and applications, or through comment sections on social media.

[1566] The server receives comments from users through these means. Specifically, the server receives an HTTP request and extracts the comment data.

[1567] 2. Comment Analysis

[1568] The server analyzes the received comments using natural language processing (NLP) and machine learning models. Specifically, the server tokenizes the comment text and inputs it into a sentiment analysis model.

[1569] An emotion engine is incorporated to recognize emotions such as "anger" or "dissatisfaction" from user comments.

[1570] 3. Judging offensive comments

[1571] The server uses the emotion engine to more accurately determine whether a comment is offensive. Specifically, if the emotion engine detects emotions such as "anger" or "dissatisfaction," the server determines the comment is offensive.

[1572] 4. Converting offensive comments

[1573] When the server detects an offensive comment, it uses the generation API to convert the comment's expression while preserving its content. Specifically, the server sends the offensive comment to the generation API as a prompt text and receives the converted comment.

[1574] Examples

[1575] Example of an embodiment

[1576] User comment: "Your service is the worst."

[1577] Server action:

[1578] Receive comments.

[1579] Comments are analyzed using NLTK and spaCy and deemed offensive.

[1580] Send the prompt text "Please convert the following comment into positive and kind words. Comment: 'Your service is the worst'" to the generation API.

[1581] Receive the converted comment "Your service has room for improvement" from the generation API.

[1582] Example of the second form

[1583] User comment: "Your service is the worst."

[1584] Server action:

[1585] Receive comments.

[1586] Comments are analyzed using TextBlob and VADER to recognize emotions such as "anger" and "dissatisfaction."

[1587] The sentiment engine determines that the comment is offensive.

[1588] Send the prompt text "Please convert the following comment into positive and kind words. Comment: 'Your service is the worst'" to the generation API.

[1589] Receive the converted comment "Your service has room for improvement" from the generation API.

[1590] In this way, we have created a system that properly analyzes user comments and converts offensive content into positive expressions. By using an emotion engine, we can accurately recognize the emotions in comments and make more accurate judgments and conversions based on those emotions.

[1591] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1592] Step 1: Receiving comments

[1593] Users can submit comments via feedback forms on websites and applications, or through comment sections on social media.

[1594] The server receives the HTTP request and extracts the comment data.

[1595] Input: User submitted comment (e.g. "Your service is terrible")

[1596] Output: Extracted comment data

[1597] Step 2: Analyzing comments

[1598] The server analyzes the received comments using natural language processing technology (NLP) and machine learning models.

[1599] How it works: The server uses Python libraries NLTK and spaCy to tokenize the comment text and input it into a sentiment analysis model.

[1600] Input: Extracted comment data

[1601] Output: An analysis of the comments' tone and sentiment.

[1602] Step 3: Recognize emotions

[1603] The server uses an emotion engine to recognize emotions such as "anger" or "dissatisfaction" from user comments.

[1604] Specific operation: The server analyzes sentiment using Python libraries TextBlob and VADER.

[1605] Input: Analysis of the comment's tone and sentiment

[1606] Output: Recognized emotion information (e.g. anger, frustration)

[1607] Step 4: Identifying offensive comments

[1608] The server uses a sentiment engine to determine whether a comment is offensive.

[1609] What it does: If the emotion engine detects emotions like "anger" or "frustration," it marks the comment as offensive.

[1610] Input: Recognized emotion information

[1611] Output: Judgment result of whether the comment is offensive (e.g., offensive)

[1612] Step 5: Convert offensive comments

[1613] When the server detects an offensive comment, it uses a generation API to convert the expression while preserving the content of the comment.

[1614] Specific behavior: The server sends the offensive comment as a prompt to the generation API and receives the converted comment.

[1615] Input: Offensive comment (e.g. "Your service is terrible")

[1616] Output: The converted comment (e.g. "Your service could be improved")

[1617] In this way, we have created a system that properly analyzes user comments and converts offensive content into positive expressions. By using an emotion engine, we can accurately recognize the emotions in comments and make more accurate judgments and conversions based on those emotions.

[1618] (Application example 1)

[1619] Next, a description will be given of Application Example 1 of Form Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal".

[1620] Conventional customer support systems had problems with poor user experience due to difficulties in responding appropriately to aggressive comments from users. In addition, there was a lack of technology to accurately recognize users' emotions and convert comments based on them, so users' dissatisfaction could not be effectively alleviated.

[1621] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1622] In this invention, the server includes a means for receiving comments from a user, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, a means for recognizing the user's emotions, a means for converting the comments based on the recognized emotions, and a means for being installed in a smartphone application. This makes it possible to convert the offensive comments from the user into positive expressions and to appropriately understand and respond to the user's emotions.

[1623] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[1624] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[1625] "Methods for transforming aggressive comments into positive and kind words" are techniques for transforming aggressive comments into more positive and kind expressions while preserving their content.

[1626] "Means for recognizing user emotions" refers to technology for reading emotions from user comments and identifying those emotions.

[1627] The "means for converting a comment based on a recognized emotion" is a technique for converting a comment into an appropriate expression based on the recognized emotion of the user.

[1628] "Means for being installed in a smartphone application" refers to technology for being incorporated into an application running on a smartphone.

[1629] As a form for implementing this invention, a system is constructed for a customer support chatbot on an online shopping site that receives comments from users and converts aggressive comments into positive expressions.

[1630] System Programming

[1631] The system consists of the following main components:

[1632] 1. Comment Receiving Module: An interface for receiving comments from users. This is implemented as a chatbot function in a smartphone application.

[1633] 2. Sentiment Analysis Module: Natural language processing techniques to determine whether the comments received are offensive or not. Specifically, it uses natural language processing libraries such as spaCy and NLTK.

[1634] 3. Comment Transformation Module: A generative AI model for transforming offensive comments into positive ones. Specifically, we use generative APIs such as OpenAI's GPT-3.

[1635] 4. Emotion Recognition Module: Technology to read and identify emotions from user comments. This also uses natural language processing libraries.

[1636] 5. Smartphone application: A platform to integrate the above modules and make them accessible to users.

[1637] Processing Description

[1638] When the server receives a comment from a user, it first analyzes the content of the comment using the sentiment analysis module. If the comment is determined to be offensive based on the analysis results, the comment conversion module converts the comment into a positive expression using a generative AI model. At this time, the emotion recognition module identifies the user's emotion and performs the conversion based on that emotion. Finally, the converted comment is returned to the user through the smartphone application.

[1639] Hardware and Software Used

[1640] Hardware: Servers, smartphones

[1641] Software: spaCy, NLTK, OpenAI GPT-3, smartphone applications

[1642] Examples

[1643] For example, if a user submits a comment saying, "Your service is terrible," the sentiment analysis module determines the comment as "aggressive." The emotion recognition module then identifies that the comment contains the emotion of "anger." The comment transformation module uses a generative AI model to transform the comment into a positive expression, such as, "Your service could use some improvement."

[1644] Examples of prompt statements

[1645] Please convert the following offensive comments into positive ones: Your service is awful.

[1646] In this way, it is possible to convert aggressive comments from users into positive expressions and appropriately understand and respond to the user's feelings.

[1647] The flow of the specific process in the application example 1 will be described with reference to FIG.

[1648] Step 1:

[1649] A user inputs and submits a comment through a smartphone application. The input comment is sent to the server. The input data is the user's comment text.

[1650] Step 2:

[1651] The server passes the received comments to a sentiment analysis module, which uses a natural language processing library (e.g., spaCy or NLTK) to analyze the content of the comment and determine whether it is offensive. The input data is the user's comment text, and the output data is the judgment result of whether the comment is offensive or not.

[1652] Step 3:

[1653] The server receives the output of the sentiment analysis module, and if the comment is determined to be offensive, passes the comment to the emotion recognition module, which uses natural language processing techniques to identify the user's emotion from the comment. The input data is the user's comment text, and the output data is the identified emotion information.

[1654] Step 4:

[1655] The server receives the output of the emotion recognition module and passes the comment and emotion information to the comment conversion module, which uses a generative AI model (e.g., OpenAI's GPT-3) to convert aggressive comments into positive expressions. The input data is the user's comment text and emotion information, and the output data is the converted positive comment text.

[1656] Step 5:

[1657] The server returns the converted comment to the smartphone application, which displays the converted comment to the user. The input data is the converted positive comment text, and the output data is the comment displayed on the user's screen.

[1658] In this way, it is possible to convert aggressive comments from users into positive expressions and appropriately understand and respond to the user's feelings.

[1659] Example 2

[1660] Next, a second embodiment of the second embodiment will be described. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal".

[1661] When a user's comment on the Internet contains offensive content, it can cause discomfort and stress to other users and service providers. There is also the issue that it is difficult to properly process offensive comments and convert them into positive expressions. Furthermore, there is a need to accurately recognize the sentiment of the comments and respond appropriately based on that.

[1662] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1663] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are offensive, a means for converting the offensive comments into positive and kind words, a means for recognizing emotions from the user's comments, a means for determining whether the comments are offensive based on the recognized emotional information, and a means for converting the offensive comments based on the recognized emotional information. This makes it possible to appropriately understand the emotions of the user and convert the offensive comments into positive expressions.

[1664] "User" means an individual or entity that submits comments using the System.

[1665] A "comment" is text data such as an opinion, impression, or feedback that a user sends through the system.

[1666] A "server" is a computer system that receives comments from users, analyzes them, and performs the necessary processing.

[1667] The "means for receiving" is a function that enables the server to receive comments sent by users.

[1668] The "means of judgment" is a function for analyzing and determining whether a received comment is offensive or not.

[1669] "Means of transformation" is a function for transforming aggressive comments into positive and kind words.

[1670] "Means for recognizing emotions" is a function for analyzing and recognizing emotions from user comments.

[1671] “Generation API” means an application programming interface for utilizing external generative AI models to generate or transform text.

[1672] A "database" is a data management system for storing data such as comments and analysis results.

[1673] This invention is a system that receives comments from users, judges whether the comments are offensive, and converts the offensive comments into positive and kind words. It also includes a function to recognize emotions from the user's comments, judge whether the comments are offensive based on the emotional information, and convert them appropriately.

[1674] Hardware and Software Used

[1675] The system uses the following hardware and software:

[1676] Server: A computer system used to receive, parse, and transform comments, including a web server (e.g., Apache, Nginx) to process HTTP requests, and a database (e.g., MySQL, PostgreSQL).

[1677] Natural language processing technology: Libraries and APIs for analyzing the content of comments and determining their tone and sentiment. Specifically, we use the Python NLTK library and the Google Cloud Natural Language API.

[1678] Generative API: A generative AI model for converting offensive comments into positive and kind words. Specifically, it uses OpenAI's GPT-3 and GPT-4.

[1679] Explanation of program processing

[1680] 1. Receiving comments

[1681] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site.

[1682] The server receives the HTTP request, extracts the comment data, and stores it in a database.

[1683] 2. Judging offensive comments

[1684] The server retrieves the comments from the database and analyzes the content of the comments using natural language processing techniques.

[1685] The server uses Python's NLTK library and the Google Cloud Natural Language API to determine the tone and sentiment of the comments.

[1686] The server determines whether a comment is offensive and stores the result as a flag in a database.

[1687] 3. Emotion Recognition

[1688] The server retrieves comments from the database and analyzes their sentiment using natural language processing technology and machine learning.

[1689] The server uses Python's NLTK library and the Google Cloud Natural Language API to recognize emotions such as "anger" or "dissatisfaction" from the comments.

[1690] The server stores the recognized emotion information in a database.

[1691] 4. Converting offensive comments

[1692] The server sends comments determined to be offensive to the generation API.

[1693] The server uses OpenAI's GPT-3 and GPT-4 to convert offensive comments into positive, kind words.

[1694] The server stores the converted comment in a database, replacing the original comment.

[1695] Examples of concrete examples and prompts

[1696] Examples

[1697] A user enters the comment "Your service is terrible" into a feedback form on a website and clicks the submit button.

[1698] The server receives the HTTP POST request, extracts the comment data, and stores it in a database.

[1699] The server uses Python's NLTK library to analyze the tone and sentiment of comments and determine which ones are offensive.

[1700] The server uses OpenAI's GPT-3 to translate the comment into "Your service could use some improvement."

[1701] The server stores the converted comment in a database, replacing the original comment.

[1702] Examples of prompt statements

[1703] "Please translate the following comment into something positive and kind: 'Your service is awful.'"

[1704] "Recognize the emotion in the following comment and tone down the offensive language: 'Your service is terrible.'"

[1705] In this way, it is possible to properly understand the user's emotions and convert aggressive comments into positive expressions.

[1706] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1707] Step 1:

[1708] Receiving comments

[1709] A user submits a comment through a feedback form on a website or application, or through a comment section on a social networking site.

[1710] Input: Comment entered by the user (e.g. "Your service is terrible").

[1711] The server receives the HTTP POST request and extracts the comment data.

[1712] Output: Extracted comment data (e.g. "Your service is the worst").

[1713] Specific operation: The server stores the received comments in a database.

[1714] Step 2:

[1715] Judging offensive comments

[1716] The server retrieves the comments from the database.

[1717] Input: Comment data stored in database (e.g. "Your service is terrible").

[1718] The server analyzes the content of the comments using natural language processing technology.

[1719] How it works: The server uses Python's NLTK library and the Google Cloud Natural Language API to determine the tone and sentiment of the comments.

[1720] Output: A judgement as to whether the comment is offensive (e.g. "Offensive").

[1721] The server stores the result of the determination in a database as a flag.

[1722] Step 3:

[1723] Emotion recognition

[1724] The server retrieves the comments from the database.

[1725] Input: Comment data stored in database (e.g. "Your service is terrible").

[1726] The server uses natural language processing technology and machine learning to analyze the sentiment of the comments.

[1727] How it works: The server uses Python's NLTK library and the Google Cloud Natural Language API to recognize emotions such as "anger" or "dissatisfaction" from comments.

[1728] Output: Recognized emotion information (e.g., "anger").

[1729] The server stores the recognized emotion information in a database.

[1730] Step 4:

[1731] Converting offensive comments

[1732] The server sends comments determined to be offensive to the generation API.

[1733] Input: aggressive comments (e.g., "your service is the worst") and perceived emotion information (e.g., "anger").

[1734] The server uses generative AI models (e.g., OpenAI's GPT-3 and GPT-4) to transform offensive comments into positive, kind words.

[1735] Specific operation: The server sends a prompt text to the generation API and receives the converted comment.

[1736] Output: The transformed comment (e.g. "Your service could be improved").

[1737] The server stores the converted comment in its database, replacing the original comment.

[1738] (Application example 2)

[1739] Next, a description will be given of Application Example 2 of Form Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal".

[1740] Conventional systems have difficulty in responding appropriately to aggressive comments from users, and responses to comments that contain emotional elements are often insufficient. In addition, when converting aggressive comments into positive expressions, it is necessary to convert them into appropriate expressions while retaining the content of the comment, but this is difficult to achieve. Furthermore, there is a demand to implement these functions in smartphone applications.

[1741] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1742] In this invention, the server includes a means for receiving comments from users, a means for determining whether the received comments are aggressive, a means for converting aggressive comments into positive and kind words, a means for recognizing the user's emotions, a means for determining the aggressiveness of the comments based on the emotion information, a means for converting aggressive comments based on the emotion information, and a means for being installed in a smartphone application. This makes it possible to appropriately respond to aggressive comments from users, accurately determine comments that include emotional elements, and convert them into positive expressions.

[1743] The "means for receiving comments from a user" refers to an interface that enables the system to receive comments entered by a user.

[1744] "Means for determining whether a received comment is offensive" refers to an algorithm or technology for analyzing the content of a received comment and determining whether the comment is offensive.

[1745] "Means of converting aggressive comments into positive and kind words" refers to a technique for converting aggressive comments into positive and kind expressions while retaining their content.

[1746] "Means for recognizing user emotions" refers to technology for reading emotions from users' comments and analyzing those emotions.

[1747] The "means for determining the aggressiveness of a comment based on emotional information" is a technique for determining with higher accuracy whether a comment is aggressive or not based on the emotional information of a user.

[1748] The "means for converting aggressive comments based on emotional information" is a technique for converting aggressive comments into positive expressions by taking into account the emotional information of the user.

[1749] "Means to be installed in a smartphone application" refers to technology for incorporating this system into an application running on a smartphone.

[1750] As an embodiment of the present invention, a method for installing a system in a smartphone application that receives comments from users, determines whether the comments are offensive, and converts the offensive comments into positive, kind words will be described.

[1751] First, a user inputs a comment through a smartphone application. The comment is then sent to the server through an interface within the application. The server then analyzes the received comment using natural language processing technology (e.g., spaCy) to understand the content of the comment.

[1752] The server then uses a machine learning model (e.g., the sentim...

Claims

1. A means for receiving comments from users; an emotion engine that recognizes the emotion of the user who made the comment; a determination means for calculating a score indicating whether the user has a positive or negative emotion based on the emotion, and determining whether the comment is offensive based on the score; A conversion means for converting the offensive comment into a specific comment by inputting a specific prompt sentence into a generative AI model if the comment is offensive; A means for displaying the converted comment on a terminal of the user; A system including:

2. The system according to claim 1 , wherein the conversion means converts the expression of the comment from the user while preserving the content of the comment.

3. The system according to claim 2 , wherein the conversion means converts offensive comments into positive, kindly expressed comments.

Images