system

Abstract

Provide a system. [Solution] A means for collecting content posted by users on a social networking service; A method that uses generative artificial intelligence to analyze the collected posts and identify the intention behind the posts, A means using generative artificial intelligence to generate an appropriate reply sentence based on the identified intention; means for automatically transmitting the generated reply text to a social networking service; A system including:

Description

[Technical Field]

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

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

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] When a post on a social networking service (SNS) goes viral, it can spread unintentionally and result in a large number of malicious replies, making it difficult for users to accurately convey the intention of the post. This can lead to misunderstandings and disputes, ultimately forcing users to close their accounts. To solve this problem, a method is needed that can accurately analyze the intention of a user's post and generate and send appropriate replies. [Means for solving the problem]

[0005] To solve the above problems, the present invention provides the following means: A system including: means for collecting content posted by users on a social networking service; means for analyzing the collected posted content and identifying the intention of the post using generative artificial intelligence; means for generating appropriate reply messages based on the identified intention using generative artificial intelligence; and means for automatically transmitting the generated reply messages to the social networking service. The system also includes means for generating responses to malicious replies and negative comments based on the results of identifying the intention of the collected posted content, and filtering means for checking the quality of the generated reply messages and correcting them as necessary, thereby enabling the user's intention to be accurately conveyed.

[0006] "User" means an individual or organization that uses a social networking service and posts or replies.

[0007] A "social networking service" is a service that allows users to share information and interact with each other over the Internet.

[0008] "Posted content" refers to information such as text, images, and videos that users make public on social networking services.

[0009] "Means of collection" refers to the function that allows the system to acquire the content posted by users and store it for analysis.

[0010] "Generative AI" is AI that uses natural language processing and machine learning to analyze text and generate sentences.

[0011] "Analysis" is the process of breaking down and analyzing the text and metadata of a post to determine its intent and meaning.

[0012] "Identifying the intent" means understanding the background and purpose of the post and, based on that information, clarifying the main point the user wanted to convey.

[0013] A "reply" is a generated response message in response to another user's post or comment.

[0014] "Checking the quality" is the process of verifying that the generated reply is appropriate and not misleading.

[0015] The "filtering means" is a function that identifies inappropriate parts of the generated reply text and makes corrections as necessary. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0017] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0018] First, the terms used in the following description will be explained.

[0019] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0020] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0021] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0022] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0023] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0024] [First embodiment]

[0025] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0026] 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.

[0027] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0028] 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.

[0029] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0030] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0031] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0032] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0033] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0034] 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.

[0035] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0036] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0037] The present invention relates to a system for automatically generating and sending replies that accurately convey the intention of a user in response to a post made by a user on a social networking service (SNS). Specific embodiments of the system are described below.

[0038] 1. Collecting User Submissions

[0039] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0040] The device collects the posted content and sends it to the server. The collected data includes the text of the post, the time of posting, and the poster's account information.

[0041] 2. Post Intention Analysis

[0042] The server analyzes the received posts, using a natural language processing (NLP) engine to perform basic processing such as word segmentation and part-of-speech analysis.

[0043] The analyzed data is passed to a generative AI, which understands the context and intent of the post. For example, for a post that says "We've released a new product," the intent is extracted as "This is an introduction to a new product."

[0044] 3. Reply Generation

[0045] The server generates a reply on social media based on the intent analysis results. For example, if another user comments, "Why is it so expensive?", the comment is passed to the generative AI.

[0046] Generative AI generates an appropriate reply, such as, "We use high-quality materials, which is why we set our prices."

[0047] The generated replies are filtered on the server side for quality checks, where misleading phrases and inappropriate language are corrected.

[0048] 4. Automatic sending

[0049] The server automatically sends the filtered reply to the original commenter using an API or interface provided by the social media platform.

[0050] The device logs into the SNS, checks the reply text, and posts it. This process ensures that a reply that accurately conveys the user's intention is posted quickly.

[0051] Example: Actual flow

[0052] For example, a user might post, "I've published a new blog post!", and then another user might comment, "This blog post is completely different from the last one."

[0053] The device collects these posts and comments and sends them to a server. The server uses generative artificial intelligence to analyze the intent of the post and understands that it is an introduction to a new blog post. At the same time, it generates and automatically sends a reply to the comment, such as, "This blog post is written from a new perspective and offers different content from the previous post."

[0054] This system reduces the burden on users of manually replying to each individual user, and also allows the intention of the post to be accurately conveyed, effectively preventing malicious replies and unintended spreading.

[0055] The processing flow will be explained below.

[0056] Step 1:

[0057] A user posts on a social networking service (SNS). This post includes content such as text, images, and videos. For example, a user posts content such as "We have released a new product."

[0058] Step 2:

[0059] The device collects this posting content and sends it to the server, including metadata such as the posting text, posting time, and user account information.

[0060] Step 3:

[0061] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine to segment the post and analyze parts of speech. The analyzed data is then passed to generative artificial intelligence (AI).

[0062] Step 4:

[0063] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it identifies the intent as "This is an introduction to the new product."

[0064] Step 5:

[0065] The server starts the reply generation process based on the analysis results. Replies and comments from other users are also collected and passed to the generative AI.

[0066] Step 6:

[0067] Generative AI generates replies. For example, if another user comments, "Why is it so expensive?", the system generates a reply such as, "We use high-quality materials, which is why we set the price that way."

[0068] Step 7:

[0069] The server checks the quality of the generated replies, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering.

[0070] Step 8:

[0071] The server automatically sends the filtered reply to the social networking site via API or an interface provided by the social networking site platform.

[0072] Step 9:

[0073] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[0074] For example, if a user posts "I've published a new blog post!", other users may comment, "This blog post is completely different from the last one." The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and generate an appropriate reply. The reply is then filtered and finally automatically sent to the social networking site.

[0075] Example 1

[0076] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0077] In conventional social networking services, manually generating and posting appropriate replies to user posts is a time-consuming and labor-intensive task. Furthermore, there is a risk that inappropriate replies or malicious comments may be posted as is, significantly impairing the user experience and the quality of communication. Therefore, there is a need for a system that can properly analyze the intent of posts and generate and send replies quickly and accurately.

[0078] 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.

[0079] In this invention, the server includes means for collecting content posted by users on a social networking service, means for transmitting the collected posted content from a terminal to the server, means for analyzing the collected posted content using natural language processing and identifying the intention of the post, means for using generative artificial intelligence to generate appropriate reply messages based on the intention including comments from other users, means for filtering and correcting the generated reply messages, and means for automatically transmitting the filtered reply messages to the social networking service. This makes it possible to automatically generate and transmit prompt and appropriate replies to user posts, improving the user experience and effectively suppressing inappropriate replies and malicious comments.

[0080] "User" refers to an individual or organization that posts using a social networking service.

[0081] "Device" means a device capable of collecting and transmitting User Posts to a server, including computers, smartphones, tablets, etc.

[0082] "Server" refers to a computer system that has the function of analyzing collected posted data and generating and sending replies.

[0083] "Social Networking Service" means an online platform that enables users to share information and communicate with other users. Examples include Twitter and Facebook.

[0084] "Natural language processing" refers to the technology that enables computers to understand and analyze human language and extract meaning, including word segmentation, part-of-speech analysis, and context understanding.

[0085] "Generative AI" refers to AI technology that can generate sentences and intentions based on input data. For example, models such as GPT-4 (registered trademark) fall into this category.

[0086] A "reply message" refers to a message that is sent in response to a post made by a user.

[0087] "Filtering" refers to the process of checking generated replies and correcting them if they contain inappropriate or misleading content.

[0088] "API" stands for Application Programming Interface and refers to a protocol for communication between different software applications. In this case, this includes APIs provided by social media platforms.

[0089] A "prompt sentence" refers to input text used to give instructions to generative artificial intelligence.

[0090] The present invention relates to a system for automatically generating and sending replies that accurately convey the intention of a user in response to a post made by a user on a social networking service (SNS). Specific embodiments of the system are described below.

[0091] User submission collection

[0092] A user posts on a social networking site. For example, they may post something like, "We've released a new product." The device collects this content and sends it to the server. The collected data includes the text of the post, the time of posting, and the poster's account information. The device can be a computer, smartphone, tablet, or other device.

[0093] Post Intention Analysis

[0094] The server receives the posted data from the device and analyzes it using a natural language processing (NLP) engine. NLP tools such as Stanford NLP and spaCy are used here. The analysis involves tokenizing the posted text and analyzing parts of speech. This analyzed data is passed to a generative artificial intelligence model. A high-performance language model such as GPT-4 is used as the generative artificial intelligence model. This model understands the intention of the post based on the prompt text. An example of a prompt text is, "A user posted, 'We have released a new product.' Please analyze the intention of this post."

[0095] Reply Generation

[0096] The server generates a reply based on the analyzed intent. For example, if another user comments, "Why is it so expensive?", that comment is input into the generative AI model. The generative AI model generates an appropriate reply using a prompt such as, "Another user commented, 'Why is it so expensive?' Please generate a reply to this comment." A specific example of a prompt that is generated is, "We set the price that way because we use high-quality materials."

[0097] The generated replies are filtered on the server side using regular expressions and custom rule-based filtering techniques to remove misleading or inappropriate phrases.

[0098] Automatic transmission

[0099] The server automatically sends the filtered reply to the original commenter. This is done using an API or an interface provided by the SNS platform (e.g., Twitter API, Facebook Graph API). The device logs into the SNS, checks the reply, and posts it. This process ensures that a reply that accurately conveys the user's intention is posted quickly.

[0100] Specific examples

[0101] For example, suppose a user posts, "I've published a new blog post!". Then, another user comments, "This blog post is completely different from the last one." The device collects this post and comment and sends them to the server. The server uses generative artificial intelligence to analyze the intent of the post and understands that it is an introduction to a new blog post. At the same time, it generates a reply to the comment, such as, "This blog post is written from a new perspective and offers different content from the last one," and automatically sends it.

[0102] This system reduces the burden on users of manually replying to each individual user, and also allows the intention of the post to be accurately conveyed, effectively preventing malicious replies and unintended spreading.

[0103] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0104] Step 1: Collect user submissions

[0105] A user logs into a social networking service (SNS) and posts, "We've released a new product."

[0106] The device uses the SNS API to collect the content of the post, including the text of the post, the time of posting, and the poster's account information.

[0107] The terminal sends the collected posting data to the server using HTTPS.

[0108] Input: User's SNS post text, posting time, poster's account information

[0109] Output: Post data sent to the server

[0110] Specific behavior:

[0111] A user posts using a social networking application.

[0112] The device obtains the posting data using an SNS API (e.g., Twitter API) and sends it to the server.

[0113] Step 2: Analyze the intent of the post

[0114] The server passes the posted data received from the device to a natural language processing (NLP) engine (e.g., spaCy).

[0115] The NLP engine tokenizes the posted text and performs part-of-speech analysis.

[0116] The analyzed data is input into a generative artificial intelligence model (e.g., GPT-4).

[0117] The generative artificial intelligence model receives the prompt, "A user posted, 'We have released a new product.' Please analyze the intent of this post." It analyzes the intent and extracts the intent, "This is an introduction to a new product."

[0118] Input: Post data (text, time, account information)

[0119] Output: Parsed intent data

[0120] Specific behavior:

[0121] The server passes the posted data to the NLP engine, which performs tokenization and part-of-speech analysis.

[0122] The server passes the prompt sentence to a generative artificial intelligence model, which analyzes the intent.

[0123] Step 3: Reply Generation

[0124] The server inputs comments from other users into the generative AI model. For example, if a comment is added, "Why is it so expensive?"

[0125] The server passes the prompt text "Another user commented, 'Why is it so expensive?' Please generate a reply to this comment" to the generative artificial intelligence model.

[0126] The generative artificial intelligence model generates the appropriate reply: "We use high-quality materials, that's why we price our products that way."

[0127] Input: Comments from other users, intent data

[0128] Output: The generated reply

[0129] Specific behavior:

[0130] The server collects comments from other users and passes them to the generative artificial intelligence model.

[0131] The server uses the prompt to cause the generative artificial intelligence model to generate a reply sentence.

[0132] Step 4: Filtering replies

[0133] The server filters the generated reply.

[0134] Filtering uses regular expressions and custom rule-based filtering techniques to detect and correct inappropriate or misleading content.

[0135] Input: Generated reply text

[0136] Output: filtered reply

[0137] Specific behavior:

[0138] The server checks the generated reply and makes corrections if necessary.

[0139] Step 5: Automated reply

[0140] The server uses the API of the SNS (for example, Twitter API) to send the filtered reply text to the SNS.

[0141] The device logs into the SNS, checks the reply text, and posts it.

[0142] Input: filtered reply text

[0143] Output: Reply text posted on SNS

[0144] Specific behavior:

[0145] The server sends a reply using the SNS API.

[0146] The device logs into the social networking site and automatically posts a reply.

[0147] The above steps realize a system that automatically generates and sends quick and appropriate replies to user posts.

[0148] (Application example 1)

[0149] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0150] Modern companies need to respond quickly and appropriately to user comments when advertising or running campaigns on social media. However, manually creating consistent and accurate replies is extremely time-consuming, especially when a large number of comments are received. Effective management is also difficult, as companies are required to respond to malicious and negative comments. This reduces the efficiency of their marketing activities and puts their brand image at risk.

[0151] 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.

[0152] In this invention, the server

[0153] A means for collecting content posted by users on a social networking service;

[0154] A method that uses generative artificial intelligence to analyze the collected posts and identify the intention behind the posts,

[0155] A means for analyzing collected advertisement and campaign posts and user comments thereon and generating appropriate reply sentences using generative artificial intelligence;

[0156] means for automatically transmitting the generated reply text to a social networking service;

[0157] A way to generate responses to malicious replies and negative comments;

[0158] means for providing automated filtering and correction functions for the purpose of improving the quality of replies related to advertisements and campaigns;

[0159] Includes:

[0160] This will enable companies to respond quickly and appropriately to user comments on advertisements and campaigns on social media, which is expected to improve the efficiency of marketing activities and enhance brand image.

[0161] "Means for collecting content posted by users on social networking services" refers to a function that automatically acquires posted data on SNS and sends it to a server for subsequent processing.

[0162] "Means using generative artificial intelligence to analyze collected post content and identify the intention of the post" refers to a system that uses natural language processing technology to analyze post content and extract the intention and purpose contained within it.

[0163] "Means for analyzing collected advertisement and campaign posts and user comments on them, and generating appropriate reply messages using generative artificial intelligence" refers to a system that analyzes advertisement and campaign-related posts and user comments, and automatically generates appropriate reply messages based on their content.

[0164] "Means for automatically sending the generated reply text on a social networking service" refers to a function that automatically posts the generated reply text in the comment section or reply section of the SNS.

[0165] The "means for generating responses to malicious replies and negative comments" is a system that automatically generates appropriate responses and replies to comments on social media that show hostility or negative reactions.

[0166] "Means for automatic filtering and correction aimed at improving the quality of reply messages related to advertisements and campaigns" refers to a system that has the function of checking the quality of generated reply messages and automatically correcting inappropriate or misleading content.

[0167] A specific system configuration and an operation method thereof will be described below as an embodiment of the present invention.

[0168] System Program

[0169] This system has the ability to collect content posted by users on social networking services (SNS), analyze the collected content, generate appropriate replies using generative artificial intelligence, and automatically send the generated replies to the SNS. It also has a means to analyze the collected advertisement and campaign posts and user comments on them, and generate appropriate replies using generative artificial intelligence. It also automatically generates responses to malicious replies and negative comments, and has automatic filtering and correction functions aimed at improving the quality of replies related to advertisements and campaigns.

[0170] Processing Details

[0171] The server first collects users' social media posts. The collected data includes text, posting time, account information, and more. This data is sent to the server and analyzed by a natural language processing (NLP) engine. The analyzed data is then passed to a generative artificial intelligence (AI) to identify the intention of the post. For example, for a post that says "We've released a new product," the intent is extracted as "This is an introduction to a new product."

[0172] Generative AI generates appropriate replies based on the identified intent. For example, based on user comments on an advertisement or campaign, the generated replies are automatically posted to the social media comment section. At this stage, they are filtered and edited for quality control. At this stage, misleading or inappropriate content is filtered out and edited as necessary.

[0173] Hardware and Software

[0174] This system is implemented using the following hardware and software:

[0175] SNS API: An interface for collecting posts on social media.

[0176] NLP engine: Performs natural language processing and analyzes post content.

[0177] Generative AI: Used to identify intent and generate appropriate replies. A specific example is the OpenAI (registered trademark) API.

[0178] Automatic filtering and correction: Quality checks generated replies and corrects inappropriate language.

[0179] Specific examples

[0180] For example, if a user posts "I've published a new blog post!" and another user subsequently comments "This post is completely different from the last one," the system would act as follows:

[0181] 1. Collecting posts and comments: The device collects posts and comments and sends them to the server.

[0182] 2. Intent analysis and reply generation: The server uses an NLP engine to analyze the intent of the post and determine that it is an introduction to a new blog post. At the same time, it generates a reply to the comment, saying that this blog post is written from a new perspective and offers different content from the previous post.

[0183] 3. Automatic sending and filtering: The generated replies are automatically posted to social media, with quality checks and necessary corrections being carried out during this process.

[0184] Prompt Sentence Examples

[0185] text

[0186] Please respond to the comment below based on the intent of the post.

[0187] Post Intent: Promoting a new product

[0188] Comments: Where can I buy this product?

[0189] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0190] Step 1:

[0191] A user posts to a social networking site. This post includes text, the time of posting, and account information. The device collects the post content and sends it to the server. Specifically, it uses the social networking site's API to obtain the post data and sends the post content to the server in JSON format or similar. The input is the SNS post data, and the output is the collected data sent to the server.

[0192] Step 2:

[0193] The server analyzes the collected post text using a natural language processing (NLP) engine. Here, basic processing such as word segmentation and part-of-speech analysis is performed to identify the intention of the post. Specifically, the post text is passed to the NLP engine, which analyzes the meaning and context of words. The input is the text data of the SNS post, and the output is the analysis results in which the intention has been identified.

[0194] Step 3:

[0195] The server uses a generative artificial intelligence (generative AI model) to generate an appropriate reply based on the identified intent. For example, a reply to a user's comment is generated based on the intent of "promoting a new product." Specifically, the intent analysis result is input into the generative AI model as a prompt, and a reply is generated. The input is the intent analysis result and the prompt, and the output is the generated reply.

[0196] Step 4:

[0197] The server performs filtering to check the quality of the generated reply and correct it if necessary. Specifically, it re-evaluates the content of the reply and corrects inappropriate or potentially misleading phrases. The input is the generated reply, and the output is the quality-checked and corrected reply.

[0198] Step 5:

[0199] The server automatically sends the reply text that has passed the quality check to the SNS. The API interface of the SNS platform is used for sending. Specifically, the reply text is posted as a comment using the API. The input is the reply text that has passed the quality check, and the output is the reply posted on the SNS.

[0200] Step 6:

[0201] The server analyzes user comments and generates replies to advertisement and campaign posts. The server collects advertisement-related post data and user comments and performs similar processing. As a specific example, in response to a post saying "We've released a new product" and a comment saying "Where can I buy this product?", the server generates a reply saying "This product can be purchased on our official website and at affiliated retailers." The input is advertisement-related post data and user comments, and the output is the generated reply text.

[0202] Step 7:

[0203] Generates responses to malicious replies and negative comments. The server automatically generates appropriate countermeasures and replies based on the identified negative intent. Specifically, a prompt sentence is input into the generative AI model to generate an appropriate response to malicious comments. The input is a negative comment and a prompt sentence, and the output is a corresponding reply sentence.

[0204] Step 8:

[0205] It applies automatic filtering and correction functions to improve the quality of replies related to ads and campaigns. The server re-evaluates the generated replies and corrects them as necessary to improve quality. Specifically, it re-analyzes them using a generative AI model or NLP engine to ensure quality. The input is the generated reply, and the output is the improved reply.

[0206] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0207] The present invention relates to a system that accurately conveys the intention of a user in response to a post made on a social networking service (SNS), recognizes the user's emotions, generates a reply, and automatically sends the reply. Specific embodiments of the system are described below.

[0208] 1. Collecting User Submissions

[0209] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0210] The device collects the content of these posts and sends it to the server. The collected data includes metadata such as the text of the post, the time of posting, and the user's account information.

[0211] 2. Post Intention Analysis

[0212] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine to segment the post and analyze parts of speech. The analyzed data is then passed to a generative artificial intelligence (AI).

[0213] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[0214] 3. Emotion recognition

[0215] The server applies an emotion engine to the content of the post to recognize the user's emotions. For example, in a post saying "We've released a new product," the emotion engine detects positive emotions such as "excitement" and "expectation."

[0216] The sentiment engine is also applied to comments to recognize whether the comment contains negative sentiment. For example, if a comment says "This product is too expensive," the comment is detected as having a "negative" sentiment.

[0217] 4. Reply Generation

[0218] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI is used to generate the reply.

[0219] The content of the reply text is adjusted based on the emotion recognized by the emotion engine. For example, in response to a negative comment such as "This product is too expensive," a more gentle reply such as "We set the price at that level because we use high-quality materials" is generated.

[0220] 5. Check the quality of replies

[0221] The server checks the quality of the generated replies, using rule-based filtering to correct any misleading or inappropriate phrases.

[0222] 6. Automatic Transmission

[0223] The server automatically sends the filtered reply to the social networking site using an API or an interface provided by the social networking site platform.

[0224] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[0225] Example: Actual flow

[0226] For example, a user might post, "I've published a new blog post!", and then another user might comment, "This blog post is completely different from the last one."

[0227] The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and recognizes the emotion of the comment using an emotion engine. For example, if a comment is recognized as containing "confusion" or "negative" emotions, a gentle reply such as "This is written from a new perspective and offers content that differs from the previous article" is generated.

[0228] These replies are filtered and then automatically sent to the social networking site, allowing for replies that accurately convey the user's intentions and are sensitive to their feelings to be posted quickly.

[0229] The processing flow will be explained below.

[0230] Step 1:

[0231] A user posts on a social networking service (SNS). This post includes content such as text, images, and videos. For example, a user posts content such as "We have released a new product."

[0232] Step 2:

[0233] The device collects this posting content and sends it to the server, including metadata such as the posting text, posting time, and user account information.

[0234] Step 3:

[0235] The server analyzes the received post. First, it uses a natural language processing (NLP) engine to segment the post content and perform basic processing such as part-of-speech analysis.

[0236] Step 4:

[0237] The server passes the analyzed data to a generative artificial intelligence (AI). The generative AI understands the content of the post and identifies its intent. For example, for a post saying "We've released a new product," it identifies the intent as "This is an introduction to the new product."

[0238] Step 5:

[0239] The server applies an emotion engine to the content of the post to recognize the user's emotions. For example, if a post says "We've released a new product," the emotion engine will detect positive emotions such as "excitement" and "expectation."

[0240] Step 6:

[0241] The server collects replies and comments from other users, such as "Why is it so expensive?"

[0242] Step 7:

[0243] The server applies an emotion engine to the collected comments to recognize the emotions contained in the comments. For example, it detects a "negative" emotion in the comment "Why is it so expensive?"

[0244] Step 8:

[0245] Generative AI generates appropriate replies based on the results of intent analysis and emotion recognition. For example, in response to a negative comment like "Why is it so expensive?", it generates a reply such as "We set the price at that because we use high-quality materials."

[0246] Step 9:

[0247] The server checks the quality of the generated replies, using rule-based filtering to check for misleading phrases or inappropriate language, and corrects them if necessary.

[0248] Step 10:

[0249] The server automatically sends the reply text that has passed the quality check to the SNS, using the API or interface provided by the SNS platform.

[0250] Step 11:

[0251] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[0252] As a concrete example, suppose a user posts, "I've published a new blog post!". Later, another user comments, "This blog post is completely different from the last one." The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and recognizes the emotion of the comment using an emotion engine. For example, if the comment is recognized as containing emotions such as "confusion" or "negative," a gentle reply such as "This is written from a new perspective and offers different content from the last post" is generated. This reply is then filtered and automatically sent to the social networking site. This allows replies to be posted quickly that accurately convey the user's intention and take their emotions into consideration.

[0253] Example 2

[0254] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0255] In conventional social networking services, it is difficult to reply appropriately to posts made by users, and in particular, it is difficult to generate replies that accurately reflect the user's intentions and emotions. Furthermore, when the process of checking the quality of generated replies is done manually, it takes time and effort, and there is also the problem that it is not possible to respond quickly. A system that solves these problems is needed.

[0256] 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.

[0257] In this invention, the server includes means for collecting content posted by users on a social networking service, means for analyzing the collected posted content and identifying the intention of the post using a natural language processing engine and generative artificial intelligence, means for using an emotion engine to recognize the emotion of the posted content and comments, means for using generative artificial intelligence to generate appropriate replies based on the identified intention and the recognized emotion, filtering means for checking the quality of the generated replies and correcting them as necessary, and means for automatically sending the filtered replies to the social networking service. This makes it possible to accurately understand the intention and emotion of the user and automatically generate and send prompt and appropriate replies.

[0258] "User" refers to an individual or entity that posts content or messages on a social networking service.

[0259] "Social networking service" refers to an online platform that allows users to interact with each other and share information over the Internet.

[0260] "Means of collection" refers to hardware or software functionality for obtaining content posted by users on social networking services and their metadata.

[0261] A "natural language processing engine" refers to software or algorithms that analyze collected posts and understand the structure and meaning of language.

[0262] "Generative AI" refers to AI technologies and systems that have the ability to generate text based on input information.

[0263] "Emotion Engine" refers to software or algorithms that recognize and classify user emotions from text data in posts and comments.

[0264] "Filtering Measures" means hardware or software features that check the quality of generated replies and modify or remove misleading or inappropriate phrases.

[0265] "Automatic sending means" refers to hardware or software functionality for posting the generated reply on a social networking service.

[0266] "Intention" refers to the purpose or meaning of the information or message a user wants to convey through their post.

[0267] "Emotion" refers to the psychological state expressed by a user when posting or commenting.

[0268] A "reply" refers to a reply message created in response to a user's post.

[0269] The present invention relates to a system that accurately conveys the intention of a user in response to a post made on a social networking service (SNS), recognizes the user's emotions, generates a reply, and automatically sends the reply. Specific embodiments of the present invention are described in detail below.

[0270] User submission collection

[0271] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0272] The device collects the content of these posts and sends it to the server via the social media platform's API. The collected data includes metadata such as the post text, posting time, and user account information.

[0273] Specific working example:

[0274] A user posts a new product release on social media

[0275] The device calls the SNS API and retrieves the post data.

[0276] Send the acquired data to the server

[0277] Post Intention Analysis

[0278] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine (e.g., Google® Cloud Natural Language API) to segment the post and analyze parts of speech. The analyzed data is then passed to a generative artificial intelligence (AI) (e.g., OpenAI GPT-3®).

[0279] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[0280] Specific working example:

[0281] The server calls the Google Cloud Natural Language API to parse the post text.

[0282] The analysis results are passed to OpenAI GPT-3 to understand the intent.

[0283] The generative AI returns the intent, "This is an introduction to a new product."

[0284] emotion recognition

[0285] The server applies an emotion engine (e.g., IBM Watson (registered trademark) Tone Analyzer) to the posted content to recognize the user's emotions. For example, the emotion engine detects positive emotions such as "excitement" or "expectation" in a post saying "We've released a new product." The emotion engine is also applied to comments to recognize whether the comment contains negative emotions. For example, if there is a comment saying "This product is too expensive," the server detects a "negative" emotion in the comment.

[0286] Specific working example:

[0287] The server calls IBM Watson Tone Analyzer to analyze the sentiment of posts and comments.

[0288] Detecting "excitement" from the post "We've released a new product"

[0289] Detecting "negativity" from the comment "This product is too expensive"

[0290] Reply Generation

[0291] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI (e.g., OpenAI GPT-3) is used to generate replies. The content of the reply is adjusted based on the emotion recognized by the emotion engine. For example, in response to a negative comment such as "This product is too expensive," a more gentle reply such as "We set the price at that level because we use high-quality materials" is generated.

[0292] Specific working example:

[0293] The server sends a prompt to OpenAI GPT-3 and generates an appropriate reply.

[0294] The generated reply is "We use high-quality materials, that's why we price it that way."

[0295] Check the quality of replies

[0296] The server checks the quality of the generated reply text, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering (e.g., the Profanity Filter API).

[0297] Specific working example:

[0298] The server calls the Profanity Filter API and checks the reply

[0299] If profanity is detected, it will be automatically corrected or an administrator will be alerted

[0300] Automatic transmission

[0301] The server automatically sends the filtered reply to the SNS using an API or an interface provided by the SNS platform. The device then posts the reply to the SNS, allowing the reply to be displayed on the SNS promptly and accurately conveying the user's intention.

[0302] Specific working example:

[0303] The server sends a reply using the API of the social networking platform (e.g., Twitter API).

[0304] The device receives the message and posts a reply on social media.

[0305] The reply "We use high-quality materials, that's why we price it that way" appears on social media.

[0306] Example prompts for generative AI models

[0307] For example, when a user posts "I published a new blog post!", here's an example prompt:

[0308] A user posts that they've published a new blog post. Another user comments, "This blog post is completely different from the last one." Recognize the sentiment of this comment and generate an appropriate reply.

[0309] This allows the generative AI model to understand the user's intentions and emotions and generate appropriate replies.

[0310] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0311] Step 1:

[0312] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0313] Input: User-submitted social media content (text, images, metadata, etc.)

[0314] Output: Post data obtained from the SNS API

[0315] The device collects the posted content and sends it to the server via the SNS platform's API. As a specific example of operation, the device calls the SNS API, obtains the posted data (post text, posting time, user account information, etc.), and sends it to the server.

[0316] Step 2:

[0317] The server analyzes the received post. First, it uses a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to segment the post and analyze parts of speech.

[0318] Input: Post data received from the SNS API

[0319] Output: Parsed structured data of post content

[0320] Specifically, the server calls the Google Cloud Natural Language API, analyzes the posted text, and performs word segmentation and part-of-speech analysis. The analysis results are then passed to a generative artificial intelligence (AI) (e.g., OpenAI GPT-3).

[0321] Step 3:

[0322] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[0323] Input: Structured data parsed by an NLP engine

[0324] Output: The intent of the post (e.g., "New product introduction")

[0325] As a specific example of how it works, the analysis results are passed to OpenAI GPT-3 to perform intent understanding. The generative AI identifies the intent of the post and returns the intent to the server.

[0326] Step 4:

[0327] The server applies an emotion engine (e.g., IBM Watson Tone Analyzer) to the posted content to recognize the user's emotions.

[0328] Input: Post data received from SNS API and intent identification results by generative AI

[0329] Output: Sentiment contained in the post and comments (e.g., "excited," "negative," etc.)

[0330] As a specific example of how it works, the server calls IBM Watson Tone Analyzer to analyze the sentiment of posts and comments. For example, it detects "excitement" from "We've released a new product" and "negativity" from "This product is too expensive."

[0331] Step 5:

[0332] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI (e.g., OpenAI GPT-3) is used to generate the reply.

[0333] Input: Intention identification results and emotion recognition results

[0334] Output: Generated reply (e.g., "We use high-quality materials, which is why we set our prices this way.")

[0335] As a specific example of how it works, the server sends a prompt to OpenAI GPT-3, which generates an appropriate reply based on the intent and emotion. The generated reply is then returned to the server.

[0336] Step 6:

[0337] The server checks the quality of the generated reply text, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering (e.g., the Profanity Filter API).

[0338] Input: Generated reply text

[0339] Output: Filtered reply (or revised reply if needed)

[0340] For example, the server calls the Profanity Filter API to check the reply text, and if inappropriate language is detected, it automatically corrects it or sends an alert to the administrator.

[0341] Step 7:

[0342] The server automatically sends the filtered reply to the social networking site using an API or an interface provided by the social networking site platform.

[0343] Input: filtered reply text

[0344] Output: Reply text posted on SNS

[0345] As a specific example of how this works, the server sends a reply using the API of the SNS platform (e.g., Twitter API). The device receives the reply and posts it on the SNS. The reply, "We use high-quality materials, which is why we set our prices this way," is then displayed on the SNS.

[0346] (Application example 2)

[0347] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0348] Modern information distribution platforms often face the problem of not providing replies to user posts promptly and appropriately. It is also difficult to respond appropriately to negative comments or malicious replies, which can detract from the user experience. Furthermore, if quality control of generated replies is insufficient, they may contain misleading phrases or inappropriate expressions, which can reduce user trust. A system that solves these problems is needed.

[0349] 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.

[0350] In this invention, the server includes means for collecting content posted by users to an information distribution platform, means for analyzing the collected posted content and identifying the intention of the post using generative artificial intelligence, means for generating appropriate reply messages based on the identified intention using generative artificial intelligence, means for automatically transmitting the generated reply messages to the information distribution platform, filtering means for checking the quality of the generated reply messages and correcting them as necessary, means for analyzing the collected comments and using an emotion recognition device to recognize the emotions contained in the comments, and means for adjusting the content of the reply messages based on the identified intention and the recognized emotion. This makes it possible to provide quick and appropriate replies, appropriately respond to negative comments and malicious replies, and improve the quality of the generated reply messages.

[0351] "User" means an individual or corporation that posts or comments using the information distribution platform.

[0352] An "information distribution platform" is a service for disseminating and sharing information online, such as blogs and video platforms.

[0353] "Means of collection" refers to software or hardware functions that automatically obtain user posts and comments.

[0354] "Generative artificial intelligence" refers to machine learning models or algorithms that can generate sentences and analyze intent in a human-like manner.

[0355] "Identifying intent" means analyzing and understanding the purpose and intent of posts and comments from their content.

[0356] A "reply" is a document that is automatically generated as a reply to a user's post or comment.

[0357] "Automatic sending means" means software or hardware functionality for automatically posting the generated reply text on a designated platform.

[0358] "Filtering means" is a function that checks the content of generated replies and corrects inappropriate expressions or misleading phrases.

[0359] An "emotion recognizer" is a device or software that analyzes emotions in text and identifies whether the emotions are positive, negative, or neutral.

[0360] "Adjustment" refers to the ability to change or fine-tune the content and tone of the generated reply based on the identified intent and recognized emotion.

[0361] The present invention relates to a system that collects content posted by users on an information distribution platform, analyzes the intention and sentiment of the user in response to the comment, generates appropriate replies, and automatically sends them. Specific embodiments of the system are described below.

[0362] System Configuration

[0363] The system consists of the following main components:

[0364] 1. User terminal: A device (e.g., smartphone, PC) used to access the information distribution platform and make posts and comments.

[0365] 2. Cloud server: A central processing unit that analyzes posts and comments and generates reply messages.

[0366] NLP engine: Uses Google Cloud Natural Language API and IBM Watson Natural Language Understanding.

[0367] Generative AI model: Uses OpenAI's GPT-3 and GPT-4.

[0368] Emotion recognition engine: Uses Microsoft(registered trademark) Azure(registered trademark) Text Analytics API.

[0369] 3. Filtering module: A system for checking the quality of generated replies and correcting inappropriate expressions.

[0370] Processing flow

[0371] 1. Collecting submissions:

[0372] A user posts to an information distribution platform.

[0373] The server automatically collects user posts and comments and stores them in a database.

[0374] 2. Intent Analysis:

[0375] The server analyzes the collected posts using an NLP engine to identify the intention of the post.

[0376] Example: For a post that says "We've released a new product," identify the intent as "This is an introduction to a new product."

[0377] 3. Emotion recognition:

[0378] The server uses an emotion recognition engine to analyze the emotions contained in the collected comments.

[0379] For example, recognize the "negative sentiment" in response to the comment "This product is too expensive."

[0380] 4. Reply Generation:

[0381] The server generates a reply sentence using a generative AI model based on the identified intention and emotion recognition results.

[0382] Example: In response to a comment like "This product is too expensive," generate a gentle reply such as "We charge that price because we use high-quality materials."

[0383] 5. Quality check:

[0384] A filtering module checks the quality of the generated replies and corrects any misleading or inappropriate phrases.

[0385] 6. Automated Submission:

[0386] The server automatically sends the filtered reply text to the information distribution platform.

[0387] The specific hardware and software used

[0388] Smartphone: A device that allows users to post and comment.

[0389] Cloud server: Uses Google Cloud or AWS (registered trademark).

[0390] NLP engines: Google Cloud Natural Language API, IBM Watson Natural Language Understanding.

[0391] Generative AI model: OpenAI GPT-3 / GPT-4.

[0392] Sentiment recognition engine: Microsoft Azure Text Analytics API.

[0393] Filtering module: Rule-based system (e.g. Python script).

[0394] Specific examples of operations

[0395] For example, suppose a user posts "I've published a new blog post!" on an information distribution platform. Later, another user comments, "This blog post is completely different from the last one." The server collects these posts and comments, analyzes the intention of the post using a generative AI model, and recognizes the emotion of the comment using an emotion recognition engine. If the comment is recognized as containing "confusion" or "negative" emotions, it generates a gentle reply such as, "This is written from a new perspective and offers content that differs from the previous post." This reply is filtered and eventually automatically sent to the information distribution platform.

[0396] Prompt Sentence Examples

[0397] "Generate a reply to the following comment based on the comment intent 'announce a new blog post' and sentiment 'negative': 'This blog post is completely different from the last one.'"

[0398] In this way, the present invention realizes a system that provides quick and appropriate replies to user posts and also responds appropriately to negative comments.

[0399] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0400] Step 1:

[0401] The server collects content posted by users to the information distribution platform. It uses an API to collect data, including metadata such as the post text, posting time, and user account information. The input is the user's post, and the output is the collected posting data.

[0402] Step 2:

[0403] The server analyzes the collected posts using a natural language processing (NLP) engine. Specifically, it performs word segmentation and part-of-speech analysis. The input is the post data collected in step 1, and the output is the analyzed text data.

[0404] Step 3:

[0405] The server passes the analyzed post content to a generative artificial intelligence (AI) to identify the intent. For example, from the sentence "We have released a new product," it extracts the intent "Introducing a new product." The input is the analyzed text data, and the output is the extracted intent.

[0406] Step 4:

[0407] The server analyzes the collected comments using an emotion recognition engine. It classifies the emotions contained in the comments as positive, negative, or neutral. The input is the comment data, and the output is the recognized emotion information.

[0408] Step 5:

[0409] The server uses generative AI to generate a reply based on the identified intent and the analyzed emotional information. For example, in response to a negative comment such as "This product is too expensive," it generates a reply such as "We set the price at that level because we use high-quality materials." The input is intent information and emotional information, and the output is the generated reply.

[0410] Step 6:

[0411] The server passes the generated reply to a filtering module for quality checks. If it contains misleading phrases or inappropriate expressions, it corrects them. The input is the generated reply, and the output is the reply that passes the quality check.

[0412] Step 7:

[0413] The server automatically sends the filtered reply text to the information distribution platform using an API or an interface provided by the information distribution platform. The input is the reply text that has passed the quality check, and the output is the reply posted on the information distribution platform.

[0414] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0415] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0416] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0417] [Second embodiment]

[0418] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0419] 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.

[0420] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0421] 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.

[0422] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0423] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0424] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0425] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0426] 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.

[0427] 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.

[0428] 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.

[0429] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0430] The present invention relates to a system for automatically generating and sending replies that accurately convey the intention of a user in response to a post made by a user on a social networking service (SNS). Specific embodiments of the system are described below.

[0431] 1. Collecting User Submissions

[0432] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0433] The device collects the posted content and sends it to the server. The collected data includes the text of the post, the time of posting, and the poster's account information.

[0434] 2. Post Intention Analysis

[0435] The server analyzes the received posts, using a natural language processing (NLP) engine to perform basic processing such as word segmentation and part-of-speech analysis.

[0436] The analyzed data is passed to a generative AI, which understands the context and intent of the post. For example, for a post that says "We've released a new product," the intent is extracted as "This is an introduction to a new product."

[0437] 3. Reply Generation

[0438] The server generates a reply on social media based on the intent analysis results. For example, if another user comments, "Why is it so expensive?", the comment is passed to the generative AI.

[0439] Generative AI generates an appropriate reply, such as, "We use high-quality materials, which is why we set our prices."

[0440] The generated replies are filtered on the server side for quality checks, where misleading phrases and inappropriate language are corrected.

[0441] 4. Automatic sending

[0442] The server automatically sends the filtered reply to the original commenter using an API or interface provided by the social media platform.

[0443] The device logs into the SNS, checks the reply text, and posts it. This process ensures that a reply that accurately conveys the user's intention is posted quickly.

[0444] Example: Actual flow

[0445] For example, a user might post, "I've published a new blog post!", and then another user might comment, "This blog post is completely different from the last one."

[0446] The device collects these posts and comments and sends them to a server. The server uses generative artificial intelligence to analyze the intent of the post and understands that it is an introduction to a new blog post. At the same time, it generates and automatically sends a reply to the comment, such as, "This blog post is written from a new perspective and offers different content from the previous post."

[0447] This system reduces the burden on users of manually replying to each individual user, and also allows the intention of the post to be accurately conveyed, effectively preventing malicious replies and unintended spreading.

[0448] The processing flow will be explained below.

[0449] Step 1:

[0450] A user posts on a social networking service (SNS). This post includes content such as text, images, and videos. For example, a user posts content such as "We have released a new product."

[0451] Step 2:

[0452] The device collects this posting content and sends it to the server, including metadata such as the posting text, posting time, and user account information.

[0453] Step 3:

[0454] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine to segment the post and analyze parts of speech. The analyzed data is then passed to generative artificial intelligence (AI).

[0455] Step 4:

[0456] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it identifies the intent as "This is an introduction to the new product."

[0457] Step 5:

[0458] The server starts the reply generation process based on the analysis results. Replies and comments from other users are also collected and passed to the generative AI.

[0459] Step 6:

[0460] Generative AI generates replies. For example, if another user comments, "Why is it so expensive?", the system generates a reply such as, "We use high-quality materials, which is why we set the price that way."

[0461] Step 7:

[0462] The server checks the quality of the generated replies, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering.

[0463] Step 8:

[0464] The server automatically sends the filtered reply to the social networking site via API or an interface provided by the social networking site platform.

[0465] Step 9:

[0466] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[0467] For example, if a user posts "I've published a new blog post!", other users may comment, "This blog post is completely different from the last one." The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and generate an appropriate reply. The reply is then filtered and finally automatically sent to the social networking site.

[0468] Example 1

[0469] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0470] In conventional social networking services, manually generating and posting appropriate replies to user posts is a time-consuming and labor-intensive task. Furthermore, there is a risk that inappropriate replies or malicious comments may be posted as is, significantly impairing the user experience and the quality of communication. Therefore, there is a need for a system that can properly analyze the intent of posts and generate and send replies quickly and accurately.

[0471] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0472] In this invention, the server includes means for collecting content posted by users on a social networking service, means for transmitting the collected posted content from a terminal to the server, means for analyzing the collected posted content using natural language processing and identifying the intention of the post, means for using generative artificial intelligence to generate appropriate reply messages based on the intention including comments from other users, means for filtering and correcting the generated reply messages, and means for automatically transmitting the filtered reply messages to the social networking service. This makes it possible to automatically generate and transmit prompt and appropriate replies to user posts, improving the user experience and effectively suppressing inappropriate replies and malicious comments.

[0473] "User" refers to an individual or organization that posts using a social networking service.

[0474] "Device" means a device capable of collecting and transmitting User Posts to a server, including computers, smartphones, tablets, etc.

[0475] "Server" refers to a computer system that has the function of analyzing collected posted data and generating and sending replies.

[0476] "Social Networking Service" means an online platform that enables users to share information and communicate with other users. Examples include Twitter and Facebook.

[0477] "Natural language processing" refers to the technology that enables computers to understand and analyze human language and extract meaning, including word segmentation, part-of-speech analysis, and context understanding.

[0478] "Generative AI" refers to AI technology that can generate sentences and intentions based on input data, such as models like GPT-4.

[0479] A "reply message" refers to a message that is sent in response to a post made by a user.

[0480] "Filtering" refers to the process of checking generated replies and correcting them if they contain inappropriate or misleading content.

[0481] "API" stands for Application Programming Interface and refers to a protocol for communication between different software applications. In this case, this includes APIs provided by social media platforms.

[0482] A "prompt sentence" refers to input text used to give instructions to generative artificial intelligence.

[0483] The present invention relates to a system for automatically generating and sending replies that accurately convey the intention of a user in response to a post made by a user on a social networking service (SNS). Specific embodiments of the system are described below.

[0484] User submission collection

[0485] A user posts on a social networking site. For example, they may post something like, "We've released a new product." The device collects this content and sends it to the server. The collected data includes the text of the post, the time of posting, and the poster's account information. The device can be a computer, smartphone, tablet, or other device.

[0486] Post Intention Analysis

[0487] The server receives the posted data from the device and analyzes it using a natural language processing (NLP) engine. NLP tools such as Stanford NLP and spaCy are used here. The analysis involves tokenizing the posted text and analyzing parts of speech. This analyzed data is passed to a generative artificial intelligence model. A high-performance language model such as GPT-4 is used as the generative artificial intelligence model. This model understands the intention of the post based on the prompt text. An example of a prompt text is, "A user posted, 'We have released a new product.' Please analyze the intention of this post."

[0488] Reply Generation

[0489] The server generates a reply based on the analyzed intent. For example, if another user comments, "Why is it so expensive?", that comment is input into the generative AI model. The generative AI model generates an appropriate reply using a prompt such as, "Another user commented, 'Why is it so expensive?' Please generate a reply to this comment." A specific example of a prompt that is generated is, "We set the price that way because we use high-quality materials."

[0490] The generated replies are filtered on the server side using regular expressions and custom rule-based filtering techniques to remove misleading or inappropriate phrases.

[0491] Automatic transmission

[0492] The server automatically sends the filtered reply to the original commenter. This is done using an API or an interface provided by the SNS platform (e.g., Twitter API, Facebook Graph API). The device logs into the SNS, checks the reply, and posts it. This process ensures that a reply that accurately conveys the user's intention is posted quickly.

[0493] Specific examples

[0494] For example, suppose a user posts, "I've published a new blog post!". Then, another user comments, "This blog post is completely different from the last one." The device collects this post and comment and sends them to the server. The server uses generative artificial intelligence to analyze the intent of the post and understands that it is an introduction to a new blog post. At the same time, it generates a reply to the comment, such as, "This blog post is written from a new perspective and offers different content from the last one," and automatically sends it.

[0495] This system reduces the burden on users of manually replying to each individual user, and also allows the intention of the post to be accurately conveyed, effectively preventing malicious replies and unintended spreading.

[0496] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0497] Step 1: Collect user submissions

[0498] A user logs into a social networking service (SNS) and posts, "We've released a new product."

[0499] The device uses the SNS API to collect the content of the post, including the text of the post, the time of posting, and the poster's account information.

[0500] The terminal sends the collected posting data to the server using HTTPS.

[0501] Input: User's SNS post text, posting time, poster's account information

[0502] Output: Post data sent to the server

[0503] Specific behavior:

[0504] A user posts using a social networking application.

[0505] The device obtains the posting data using an SNS API (e.g., Twitter API) and sends it to the server.

[0506] Step 2: Analyze the intent of the post

[0507] The server passes the posted data received from the device to a natural language processing (NLP) engine (e.g., spaCy).

[0508] The NLP engine tokenizes the posted text and performs part-of-speech analysis.

[0509] The analyzed data is input into a generative artificial intelligence model (e.g., GPT-4).

[0510] The generative artificial intelligence model receives the prompt, "A user posted, 'We have released a new product.' Please analyze the intent of this post." It analyzes the intent and extracts the intent, "This is an introduction to a new product."

[0511] Input: Post data (text, time, account information)

[0512] Output: Parsed intent data

[0513] Specific behavior:

[0514] The server passes the posted data to the NLP engine, which performs tokenization and part-of-speech analysis.

[0515] The server passes the prompt sentence to a generative artificial intelligence model, which analyzes the intent.

[0516] Step 3: Reply Generation

[0517] The server inputs comments from other users into the generative AI model. For example, if a comment is added, "Why is it so expensive?"

[0518] The server passes the prompt text "Another user commented, 'Why is it so expensive?' Please generate a reply to this comment" to the generative artificial intelligence model.

[0519] The generative artificial intelligence model generates the appropriate reply: "We use high-quality materials, that's why we price our products that way."

[0520] Input: Comments from other users, intent data

[0521] Output: The generated reply

[0522] Specific behavior:

[0523] The server collects comments from other users and passes them to the generative artificial intelligence model.

[0524] The server uses the prompt to cause the generative artificial intelligence model to generate a reply sentence.

[0525] Step 4: Filtering replies

[0526] The server filters the generated reply.

[0527] Filtering uses regular expressions and custom rule-based filtering techniques to detect and correct inappropriate or misleading content.

[0528] Input: Generated reply text

[0529] Output: filtered reply

[0530] Specific behavior:

[0531] The server checks the generated reply and makes corrections if necessary.

[0532] Step 5: Automated reply

[0533] The server uses the API of the SNS (for example, Twitter API) to send the filtered reply text to the SNS.

[0534] The device logs into the SNS, checks the reply text, and posts it.

[0535] Input: filtered reply text

[0536] Output: Reply text posted on SNS

[0537] Specific behavior:

[0538] The server sends a reply using the SNS API.

[0539] The device logs into the social networking site and automatically posts a reply.

[0540] The above steps realize a system that automatically generates and sends quick and appropriate replies to user posts.

[0541] (Application example 1)

[0542] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0543] Modern companies need to respond quickly and appropriately to user comments when advertising or running campaigns on social media. However, manually creating consistent and accurate replies is extremely time-consuming, especially when a large number of comments are received. Effective management is also difficult, as companies are required to respond to malicious and negative comments. This reduces the efficiency of their marketing activities and puts their brand image at risk.

[0544] 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.

[0545] In this invention, the server

[0546] A means for collecting content posted by users on a social networking service;

[0547] A method that uses generative artificial intelligence to analyze the collected posts and identify the intention behind the posts,

[0548] A means for analyzing collected advertisement and campaign posts and user comments thereon and generating appropriate reply sentences using generative artificial intelligence;

[0549] means for automatically transmitting the generated reply text to a social networking service;

[0550] A way to generate responses to malicious replies and negative comments;

[0551] means for providing automated filtering and correction functions for the purpose of improving the quality of replies related to advertisements and campaigns;

[0552] Includes:

[0553] This will enable companies to respond quickly and appropriately to user comments on advertisements and campaigns on social media, which is expected to improve the efficiency of marketing activities and enhance brand image.

[0554] "Means for collecting content posted by users on social networking services" refers to a function that automatically acquires posted data on SNS and sends it to a server for subsequent processing.

[0555] "Means using generative artificial intelligence to analyze collected post content and identify the intention of the post" refers to a system that uses natural language processing technology to analyze post content and extract the intention and purpose contained within it.

[0556] "Means for analyzing collected advertisement and campaign posts and user comments on them, and generating appropriate reply messages using generative artificial intelligence" refers to a system that analyzes advertisement and campaign-related posts and user comments, and automatically generates appropriate reply messages based on their content.

[0557] "Means for automatically sending the generated reply text on a social networking service" refers to a function that automatically posts the generated reply text in the comment section or reply section of the SNS.

[0558] The "means for generating responses to malicious replies and negative comments" is a system that automatically generates appropriate responses and replies to comments on social media that show hostility or negative reactions.

[0559] "Means for automatic filtering and correction aimed at improving the quality of reply messages related to advertisements and campaigns" refers to a system that has the function of checking the quality of generated reply messages and automatically correcting inappropriate or misleading content.

[0560] A specific system configuration and an operation method thereof will be described below as an embodiment of the present invention.

[0561] System Program

[0562] This system has the ability to collect content posted by users on social networking services (SNS), analyze the collected content, generate appropriate replies using generative artificial intelligence, and automatically send the generated replies to the SNS. It also has a means to analyze the collected advertisement and campaign posts and user comments on them, and generate appropriate replies using generative artificial intelligence. It also automatically generates responses to malicious replies and negative comments, and has automatic filtering and correction functions aimed at improving the quality of replies related to advertisements and campaigns.

[0563] Processing Details

[0564] The server first collects users' social media posts. The collected data includes text, posting time, account information, and more. This data is sent to the server and analyzed by a natural language processing (NLP) engine. The analyzed data is then passed to a generative artificial intelligence (AI) to identify the intention of the post. For example, for a post that says "We've released a new product," the intent is extracted as "This is an introduction to a new product."

[0565] Generative AI generates appropriate replies based on the identified intent. For example, based on user comments on an advertisement or campaign, the generated replies are automatically posted to the social media comment section. At this stage, they are filtered and edited for quality control. At this stage, misleading or inappropriate content is filtered out and edited as necessary.

[0566] Hardware and Software

[0567] This system is implemented using the following hardware and software:

[0568] SNS API: An interface for collecting posts on social media.

[0569] NLP engine: Performs natural language processing and analyzes post content.

[0570] Generative AI: Used to identify intent and generate appropriate replies. An example is the OpenAI API.

[0571] Automatic filtering and correction: Quality checks generated replies and corrects inappropriate language.

[0572] Specific examples

[0573] For example, if a user posts "I've published a new blog post!" and another user subsequently comments "This post is completely different from the last one," the system would act as follows:

[0574] 1. Collecting posts and comments: The device collects posts and comments and sends them to the server.

[0575] 2. Intent analysis and reply generation: The server uses an NLP engine to analyze the intent of the post and determine that it is an introduction to a new blog post. At the same time, it generates a reply to the comment, saying that this blog post is written from a new perspective and offers different content from the previous post.

[0576] 3. Automatic sending and filtering: The generated replies are automatically posted to social media, with quality checks and necessary corrections being carried out during this process.

[0577] Prompt Sentence Examples

[0578] text

[0579] Please respond to the comment below based on the intent of the post.

[0580] Post Intent: Promoting a new product

[0581] Comments: Where can I buy this product?

[0582] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0583] Step 1:

[0584] A user posts to a social networking site. This post includes text, the time of posting, and account information. The device collects the post content and sends it to the server. Specifically, it uses the social networking site's API to obtain the post data and sends the post content to the server in JSON format or similar. The input is the SNS post data, and the output is the collected data sent to the server.

[0585] Step 2:

[0586] The server analyzes the collected post text using a natural language processing (NLP) engine. Here, basic processing such as word segmentation and part-of-speech analysis is performed to identify the intention of the post. Specifically, the post text is passed to the NLP engine, which analyzes the meaning and context of words. The input is the text data of the SNS post, and the output is the analysis results in which the intention has been identified.

[0587] Step 3:

[0588] The server uses a generative artificial intelligence (generative AI model) to generate an appropriate reply based on the identified intent. For example, a reply to a user's comment is generated based on the intent of "promoting a new product." Specifically, the intent analysis result is input into the generative AI model as a prompt, and a reply is generated. The input is the intent analysis result and the prompt, and the output is the generated reply.

[0589] Step 4:

[0590] The server performs filtering to check the quality of the generated reply and correct it if necessary. Specifically, it re-evaluates the content of the reply and corrects inappropriate or potentially misleading phrases. The input is the generated reply, and the output is the quality-checked and corrected reply.

[0591] Step 5:

[0592] The server automatically sends the reply text that has passed the quality check to the SNS. The API interface of the SNS platform is used for sending. Specifically, the reply text is posted as a comment using the API. The input is the reply text that has passed the quality check, and the output is the reply posted on the SNS.

[0593] Step 6:

[0594] The server analyzes user comments and generates replies to advertisement and campaign posts. The server collects advertisement-related post data and user comments and performs similar processing. As a specific example, in response to a post saying "We've released a new product" and a comment saying "Where can I buy this product?", the server generates a reply saying "This product can be purchased on our official website and at affiliated retailers." The input is advertisement-related post data and user comments, and the output is the generated reply text.

[0595] Step 7:

[0596] Generates responses to malicious replies and negative comments. The server automatically generates appropriate countermeasures and replies based on the identified negative intent. Specifically, a prompt sentence is input into the generative AI model to generate an appropriate response to malicious comments. The input is a negative comment and a prompt sentence, and the output is a corresponding reply sentence.

[0597] Step 8:

[0598] It applies automatic filtering and correction functions to improve the quality of replies related to ads and campaigns. The server re-evaluates the generated replies and corrects them as necessary to improve quality. Specifically, it re-analyzes them using a generative AI model or NLP engine to ensure quality. The input is the generated reply, and the output is the improved reply.

[0599] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0600] The present invention relates to a system that accurately conveys the intention of a user in response to a post made on a social networking service (SNS), recognizes the user's emotions, generates a reply, and automatically sends the reply. Specific embodiments of the system are described below.

[0601] 1. Collecting User Submissions

[0602] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0603] The device collects the content of these posts and sends it to the server. The collected data includes metadata such as the text of the post, the time of posting, and the user's account information.

[0604] 2. Post Intention Analysis

[0605] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine to segment the post and analyze parts of speech. The analyzed data is then passed to a generative artificial intelligence (AI).

[0606] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[0607] 3. Emotion recognition

[0608] The server applies an emotion engine to the content of the post to recognize the user's emotions. For example, in a post saying "We've released a new product," the emotion engine detects positive emotions such as "excitement" and "expectation."

[0609] The sentiment engine is also applied to comments to recognize whether the comment contains negative sentiment. For example, if a comment says "This product is too expensive," the comment is detected as having a "negative" sentiment.

[0610] 4. Reply Generation

[0611] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI is used to generate the reply.

[0612] The content of the reply text is adjusted based on the emotion recognized by the emotion engine. For example, in response to a negative comment such as "This product is too expensive," a more gentle reply such as "We set the price at that level because we use high-quality materials" is generated.

[0613] 5. Check the quality of replies

[0614] The server checks the quality of the generated replies, using rule-based filtering to correct any misleading or inappropriate phrases.

[0615] 6. Automatic Transmission

[0616] The server automatically sends the filtered reply to the social networking site using an API or an interface provided by the social networking site platform.

[0617] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[0618] Example: Actual flow

[0619] For example, a user might post, "I've published a new blog post!", and then another user might comment, "This blog post is completely different from the last one."

[0620] The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and recognizes the emotion of the comment using an emotion engine. For example, if a comment is recognized as containing "confusion" or "negative" emotions, a gentle reply such as "This is written from a new perspective and offers content that differs from the previous article" is generated.

[0621] These replies are filtered and then automatically sent to the social networking site, allowing for replies that accurately convey the user's intentions and are sensitive to their feelings to be posted quickly.

[0622] The processing flow will be explained below.

[0623] Step 1:

[0624] A user posts on a social networking service (SNS). This post includes content such as text, images, and videos. For example, a user posts content such as "We have released a new product."

[0625] Step 2:

[0626] The device collects this posting content and sends it to the server, including metadata such as the posting text, posting time, and user account information.

[0627] Step 3:

[0628] The server analyzes the received post. First, it uses a natural language processing (NLP) engine to segment the post content and perform basic processing such as part-of-speech analysis.

[0629] Step 4:

[0630] The server passes the analyzed data to a generative artificial intelligence (AI). The generative AI understands the content of the post and identifies its intent. For example, for a post saying "We've released a new product," it identifies the intent as "This is an introduction to the new product."

[0631] Step 5:

[0632] The server applies an emotion engine to the content of the post to recognize the user's emotions. For example, if a post says "We've released a new product," the emotion engine will detect positive emotions such as "excitement" and "expectation."

[0633] Step 6:

[0634] The server collects replies and comments from other users, such as "Why is it so expensive?"

[0635] Step 7:

[0636] The server applies an emotion engine to the collected comments to recognize the emotions contained in the comments. For example, it detects a "negative" emotion in the comment "Why is it so expensive?"

[0637] Step 8:

[0638] Generative AI generates appropriate replies based on the results of intent analysis and emotion recognition. For example, in response to a negative comment like "Why is it so expensive?", it generates a reply such as "We set the price at that because we use high-quality materials."

[0639] Step 9:

[0640] The server checks the quality of the generated replies, using rule-based filtering to check for misleading phrases or inappropriate language, and corrects them if necessary.

[0641] Step 10:

[0642] The server automatically sends the reply text that has passed the quality check to the SNS, using the API or interface provided by the SNS platform.

[0643] Step 11:

[0644] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[0645] As a concrete example, suppose a user posts, "I've published a new blog post!". Later, another user comments, "This blog post is completely different from the last one." The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and recognizes the emotion of the comment using an emotion engine. For example, if the comment is recognized as containing emotions such as "confusion" or "negative," a gentle reply such as "This is written from a new perspective and offers different content from the last post" is generated. This reply is then filtered and automatically sent to the social networking site. This allows replies to be posted quickly that accurately convey the user's intention and take their emotions into consideration.

[0646] Example 2

[0647] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0648] In conventional social networking services, it is difficult to reply appropriately to posts made by users, and in particular, it is difficult to generate replies that accurately reflect the user's intentions and emotions. Furthermore, when the process of checking the quality of generated replies is done manually, it takes time and effort, and there is also the problem that it is not possible to respond quickly. A system that solves these problems is needed.

[0649] 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.

[0650] In this invention, the server includes means for collecting content posted by users on a social networking service, means for analyzing the collected posted content and identifying the intention of the post using a natural language processing engine and generative artificial intelligence, means for using an emotion engine to recognize the emotion of the posted content and comments, means for using generative artificial intelligence to generate appropriate replies based on the identified intention and the recognized emotion, filtering means for checking the quality of the generated replies and correcting them as necessary, and means for automatically sending the filtered replies to the social networking service. This makes it possible to accurately understand the intention and emotion of the user and automatically generate and send prompt and appropriate replies.

[0651] "User" refers to an individual or entity that posts content or messages on a social networking service.

[0652] "Social networking service" refers to an online platform that allows users to interact with each other and share information over the Internet.

[0653] "Means of collection" refers to hardware or software functionality for obtaining content posted by users on social networking services and their metadata.

[0654] A "natural language processing engine" refers to software or algorithms that analyze collected posts and understand the structure and meaning of language.

[0655] "Generative AI" refers to AI technologies and systems that have the ability to generate text based on input information.

[0656] "Emotion Engine" refers to software or algorithms that recognize and classify user emotions from text data in posts and comments.

[0657] "Filtering Measures" means hardware or software features that check the quality of generated replies and modify or remove misleading or inappropriate phrases.

[0658] "Automatic sending means" refers to hardware or software functionality for posting the generated reply on a social networking service.

[0659] "Intention" refers to the purpose or meaning of the information or message a user wants to convey through their post.

[0660] "Emotion" refers to the psychological state expressed by a user when posting or commenting.

[0661] A "reply" refers to a reply message created in response to a user's post.

[0662] The present invention relates to a system that accurately conveys the intention of a user in response to a post made on a social networking service (SNS), recognizes the user's emotions, generates a reply, and automatically sends the reply. Specific embodiments of the present invention are described in detail below.

[0663] User submission collection

[0664] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0665] The device collects the content of these posts and sends it to the server via the social media platform's API. The collected data includes metadata such as the post text, posting time, and user account information.

[0666] Specific working example:

[0667] A user posts a new product release on social media

[0668] The device calls the SNS API and retrieves the post data.

[0669] Send the acquired data to the server

[0670] Post Intention Analysis

[0671] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to segment the post and analyze parts of speech. The analyzed data is then passed to a generative artificial intelligence (AI) (e.g., OpenAI GPT-3).

[0672] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[0673] Specific working example:

[0674] The server calls the Google Cloud Natural Language API to parse the post text.

[0675] The analysis results are passed to OpenAI GPT-3 to understand the intent.

[0676] The generative AI returns the intent, "This is an introduction to a new product."

[0677] emotion recognition

[0678] The server applies an emotion engine (e.g., IBM Watson Tone Analyzer) to the content of posts to recognize the user's emotions. For example, for a post saying "We've released a new product," the emotion engine will detect positive emotions such as "excitement" and "expectation." The emotion engine is also applied to comments to recognize whether the comment contains negative emotions. For example, if there is a comment saying "This product is too expensive," the comment will be detected as having a "negative" emotion.

[0679] Specific working example:

[0680] The server calls IBM Watson Tone Analyzer to analyze the sentiment of posts and comments.

[0681] Detecting "excitement" from the post "We've released a new product"

[0682] Detecting "negativity" from the comment "This product is too expensive"

[0683] Reply Generation

[0684] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI (e.g., OpenAI GPT-3) is used to generate replies. The content of the reply is adjusted based on the emotion recognized by the emotion engine. For example, in response to a negative comment such as "This product is too expensive," a more gentle reply such as "We set the price at that level because we use high-quality materials" is generated.

[0685] Specific working example:

[0686] The server sends a prompt to OpenAI GPT-3 and generates an appropriate reply.

[0687] The generated reply is "We use high-quality materials, that's why we price it that way."

[0688] Check the quality of replies

[0689] The server checks the quality of the generated reply text, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering (e.g., the Profanity Filter API).

[0690] Specific working example:

[0691] The server calls the Profanity Filter API and checks the reply

[0692] If profanity is detected, it will be automatically corrected or an administrator will be alerted

[0693] Automatic transmission

[0694] The server automatically sends the filtered reply to the SNS using an API or an interface provided by the SNS platform. The device then posts the reply to the SNS, allowing the reply to be displayed on the SNS promptly and accurately conveying the user's intention.

[0695] Specific working example:

[0696] The server sends a reply using the API of the social networking platform (e.g., Twitter API).

[0697] The device receives the message and posts a reply on social media.

[0698] The reply "We use high-quality materials, that's why we price it that way" appears on social media.

[0699] Example prompts for generative AI models

[0700] For example, when a user posts "I published a new blog post!", here's an example prompt:

[0701] A user posts that they've published a new blog post. Another user comments, "This blog post is completely different from the last one." Recognize the sentiment of this comment and generate an appropriate reply.

[0702] This allows the generative AI model to understand the user's intentions and emotions and generate appropriate replies.

[0703] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0704] Step 1:

[0705] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0706] Input: User-submitted social media content (text, images, metadata, etc.)

[0707] Output: Post data obtained from the SNS API

[0708] The device collects the posted content and sends it to the server via the SNS platform's API. As a specific example of operation, the device calls the SNS API, obtains the posted data (post text, posting time, user account information, etc.), and sends it to the server.

[0709] Step 2:

[0710] The server analyzes the received post. First, it uses a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to segment the post and analyze parts of speech.

[0711] Input: Post data received from the SNS API

[0712] Output: Parsed structured data of post content

[0713] Specifically, the server calls the Google Cloud Natural Language API, analyzes the posted text, and performs word segmentation and part-of-speech analysis. The analysis results are then passed to a generative artificial intelligence (AI) (e.g., OpenAI GPT-3).

[0714] Step 3:

[0715] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[0716] Input: Structured data parsed by an NLP engine

[0717] Output: The intent of the post (e.g., "New product introduction")

[0718] As a specific example of how it works, the analysis results are passed to OpenAI GPT-3 to perform intent understanding. The generative AI identifies the intent of the post and returns the intent to the server.

[0719] Step 4:

[0720] The server applies an emotion engine (e.g., IBM Watson Tone Analyzer) to the posted content to recognize the user's emotions.

[0721] Input: Post data received from SNS API and intent identification results by generative AI

[0722] Output: Sentiment contained in the post and comments (e.g., "excited," "negative," etc.)

[0723] As a specific example of how it works, the server calls IBM Watson Tone Analyzer to analyze the sentiment of posts and comments. For example, it detects "excitement" from "We've released a new product" and "negativity" from "This product is too expensive."

[0724] Step 5:

[0725] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI (e.g., OpenAI GPT-3) is used to generate the reply.

[0726] Input: Intention identification results and emotion recognition results

[0727] Output: Generated reply (e.g., "We use high-quality materials, which is why we set our prices this way.")

[0728] As a specific example of how it works, the server sends a prompt to OpenAI GPT-3, which generates an appropriate reply based on the intent and emotion. The generated reply is then returned to the server.

[0729] Step 6:

[0730] The server checks the quality of the generated reply text, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering (e.g., the Profanity Filter API).

[0731] Input: Generated reply text

[0732] Output: Filtered reply (or revised reply if needed)

[0733] For example, the server calls the Profanity Filter API to check the reply text, and if inappropriate language is detected, it automatically corrects it or sends an alert to the administrator.

[0734] Step 7:

[0735] The server automatically sends the filtered reply to the social networking site using an API or an interface provided by the social networking site platform.

[0736] Input: filtered reply text

[0737] Output: Reply text posted on SNS

[0738] As a specific example of how this works, the server sends a reply using the API of the SNS platform (e.g., Twitter API). The device receives the reply and posts it on the SNS. The reply, "We use high-quality materials, which is why we set our prices this way," is then displayed on the SNS.

[0739] (Application example 2)

[0740] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0741] Modern information distribution platforms often face the problem of not providing replies to user posts promptly and appropriately. It is also difficult to respond appropriately to negative comments or malicious replies, which can detract from the user experience. Furthermore, if quality control of generated replies is insufficient, they may contain misleading phrases or inappropriate expressions, which can reduce user trust. A system that solves these problems is needed.

[0742] 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.

[0743] In this invention, the server includes means for collecting content posted by users to an information distribution platform, means for analyzing the collected posted content and identifying the intention of the post using generative artificial intelligence, means for generating appropriate reply messages based on the identified intention using generative artificial intelligence, means for automatically transmitting the generated reply messages to the information distribution platform, filtering means for checking the quality of the generated reply messages and correcting them as necessary, means for analyzing the collected comments and using an emotion recognition device to recognize the emotions contained in the comments, and means for adjusting the content of the reply messages based on the identified intention and the recognized emotion. This makes it possible to provide quick and appropriate replies, appropriately respond to negative comments and malicious replies, and improve the quality of the generated reply messages.

[0744] "User" means an individual or corporation that posts or comments using the information distribution platform.

[0745] An "information distribution platform" is a service for disseminating and sharing information online, such as blogs and video platforms.

[0746] "Means of collection" refers to software or hardware functions that automatically obtain user posts and comments.

[0747] "Generative artificial intelligence" refers to machine learning models or algorithms that can generate sentences and analyze intent in a human-like manner.

[0748] "Identifying intent" means analyzing and understanding the purpose and intent of posts and comments from their content.

[0749] A "reply" is a document that is automatically generated as a reply to a user's post or comment.

[0750] "Automatic sending means" means software or hardware functionality for automatically posting the generated reply text on a designated platform.

[0751] "Filtering means" is a function that checks the content of generated replies and corrects inappropriate expressions or misleading phrases.

[0752] An "emotion recognizer" is a device or software that analyzes emotions in text and identifies whether the emotions are positive, negative, or neutral.

[0753] "Adjustment" refers to the ability to change or fine-tune the content and tone of the generated reply based on the identified intent and recognized emotion.

[0754] The present invention relates to a system that collects content posted by users on an information distribution platform, analyzes the intention and sentiment of the user in response to the comment, generates appropriate replies, and automatically sends them. Specific embodiments of the system are described below.

[0755] System Configuration

[0756] The system consists of the following main components:

[0757] 1. User terminal: A device (e.g., smartphone, PC) used to access the information distribution platform and make posts and comments.

[0758] 2. Cloud server: A central processing unit that analyzes posts and comments and generates reply messages.

[0759] NLP engine: Uses Google Cloud Natural Language API and IBM Watson Natural Language Understanding.

[0760] Generative AI model: Uses OpenAI's GPT-3 and GPT-4.

[0761] Emotion recognition engine: Uses Microsoft Azure's Text Analytics API.

[0762] 3. Filtering module: A system for checking the quality of generated replies and correcting inappropriate expressions.

[0763] Processing flow

[0764] 1. Collecting submissions:

[0765] A user posts to an information distribution platform.

[0766] The server automatically collects user posts and comments and stores them in a database.

[0767] 2. Intent Analysis:

[0768] The server analyzes the collected posts using an NLP engine to identify the intention of the post.

[0769] Example: For a post that says "We've released a new product," identify the intent as "This is an introduction to a new product."

[0770] 3. Emotion recognition:

[0771] The server uses an emotion recognition engine to analyze the emotions contained in the collected comments.

[0772] For example, recognize the "negative sentiment" in response to the comment "This product is too expensive."

[0773] 4. Reply Generation:

[0774] The server generates a reply sentence using a generative AI model based on the identified intention and emotion recognition results.

[0775] Example: In response to a comment like "This product is too expensive," generate a gentle reply such as "We charge that price because we use high-quality materials."

[0776] 5. Quality check:

[0777] A filtering module checks the quality of the generated replies and corrects any misleading or inappropriate phrases.

[0778] 6. Automated Submission:

[0779] The server automatically sends the filtered reply text to the information distribution platform.

[0780] The specific hardware and software used

[0781] Smartphone: A device that allows users to post and comment.

[0782] Cloud server: Uses Google Cloud or AWS.

[0783] NLP engines: Google Cloud Natural Language API, IBM Watson Natural Language Understanding.

[0784] Generative AI model: OpenAI GPT-3 / GPT-4.

[0785] Sentiment recognition engine: Microsoft Azure Text Analytics API.

[0786] Filtering module: Rule-based system (e.g. Python script).

[0787] Specific examples of operations

[0788] For example, suppose a user posts "I've published a new blog post!" on an information distribution platform. Later, another user comments, "This blog post is completely different from the last one." The server collects these posts and comments, analyzes the intention of the post using a generative AI model, and recognizes the emotion of the comment using an emotion recognition engine. If the comment is recognized as containing "confusion" or "negative" emotions, it generates a gentle reply such as, "This is written from a new perspective and offers content that differs from the previous post." This reply is filtered and eventually automatically sent to the information distribution platform.

[0789] Prompt Sentence Examples

[0790] "Generate a reply to the following comment based on the comment intent 'announce a new blog post' and sentiment 'negative': 'This blog post is completely different from the last one.'"

[0791] In this way, the present invention realizes a system that provides quick and appropriate replies to user posts and also responds appropriately to negative comments.

[0792] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0793] Step 1:

[0794] The server collects content posted by users to the information distribution platform. It uses an API to collect data, including metadata such as the post text, posting time, and user account information. The input is the user's post, and the output is the collected posting data.

[0795] Step 2:

[0796] The server analyzes the collected posts using a natural language processing (NLP) engine. Specifically, it performs word segmentation and part-of-speech analysis. The input is the post data collected in step 1, and the output is the analyzed text data.

[0797] Step 3:

[0798] The server passes the analyzed post content to a generative artificial intelligence (AI) to identify the intent. For example, from the sentence "We have released a new product," it extracts the intent "Introducing a new product." The input is the analyzed text data, and the output is the extracted intent.

[0799] Step 4:

[0800] The server analyzes the collected comments using an emotion recognition engine. It classifies the emotions contained in the comments as positive, negative, or neutral. The input is the comment data, and the output is the recognized emotion information.

[0801] Step 5:

[0802] The server uses generative AI to generate a reply based on the identified intent and the analyzed emotional information. For example, in response to a negative comment such as "This product is too expensive," it generates a reply such as "We set the price at that level because we use high-quality materials." The input is intent information and emotional information, and the output is the generated reply.

[0803] Step 6:

[0804] The server passes the generated reply to a filtering module for quality checks. If it contains misleading phrases or inappropriate expressions, it corrects them. The input is the generated reply, and the output is the reply that passes the quality check.

[0805] Step 7:

[0806] The server automatically sends the filtered reply text to the information distribution platform using an API or an interface provided by the information distribution platform. The input is the reply text that has passed the quality check, and the output is the reply posted on the information distribution platform.

[0807] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0808] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0809] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0810] [Third embodiment]

[0811] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0812] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0813] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0814] 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.

[0815] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0816] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0817] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0818] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0819] 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.

[0820] 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.

[0821] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0822] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0823] The present invention relates to a system for automatically generating and sending replies that accurately convey the intention of a user in response to a post made by a user on a social networking service (SNS). Specific embodiments of the system are described below.

[0824] 1. Collecting User Submissions

[0825] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0826] The device collects the posted content and sends it to the server. The collected data includes the text of the post, the time of posting, and the poster's account information.

[0827] 2. Post Intention Analysis

[0828] The server analyzes the received posts, using a natural language processing (NLP) engine to perform basic processing such as word segmentation and part-of-speech analysis.

[0829] The analyzed data is passed to a generative AI, which understands the context and intent of the post. For example, for a post that says "We've released a new product," the intent is extracted as "This is an introduction to a new product."

[0830] 3. Reply Generation

[0831] The server generates a reply on social media based on the intent analysis results. For example, if another user comments, "Why is it so expensive?", the comment is passed to the generative AI.

[0832] Generative AI generates an appropriate reply, such as, "We use high-quality materials, which is why we set our prices."

[0833] The generated replies are filtered on the server side for quality checks, where misleading phrases and inappropriate language are corrected.

[0834] 4. Automatic sending

[0835] The server automatically sends the filtered reply to the original commenter using an API or interface provided by the social media platform.

[0836] The device logs into the SNS, checks the reply text, and posts it. This process ensures that a reply that accurately conveys the user's intention is posted quickly.

[0837] Example: Actual flow

[0838] For example, a user might post, "I've published a new blog post!", and then another user might comment, "This blog post is completely different from the last one."

[0839] The device collects these posts and comments and sends them to a server. The server uses generative artificial intelligence to analyze the intent of the post and understands that it is an introduction to a new blog post. At the same time, it generates and automatically sends a reply to the comment, such as, "This blog post is written from a new perspective and offers different content from the previous post."

[0840] This system reduces the burden on users of manually replying to each individual user, and also allows the intention of the post to be accurately conveyed, effectively preventing malicious replies and unintended spreading.

[0841] The processing flow will be explained below.

[0842] Step 1:

[0843] A user posts on a social networking service (SNS). This post includes content such as text, images, and videos. For example, a user posts content such as "We have released a new product."

[0844] Step 2:

[0845] The device collects this posting content and sends it to the server, including metadata such as the posting text, posting time, and user account information.

[0846] Step 3:

[0847] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine to segment the post and analyze parts of speech. The analyzed data is then passed to generative artificial intelligence (AI).

[0848] Step 4:

[0849] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it identifies the intent as "This is an introduction to the new product."

[0850] Step 5:

[0851] The server starts the reply generation process based on the analysis results. Replies and comments from other users are also collected and passed to the generative AI.

[0852] Step 6:

[0853] Generative AI generates replies. For example, if another user comments, "Why is it so expensive?", the system generates a reply such as, "We use high-quality materials, which is why we set the price that way."

[0854] Step 7:

[0855] The server checks the quality of the generated replies, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering.

[0856] Step 8:

[0857] The server automatically sends the filtered reply to the social networking site via API or an interface provided by the social networking site platform.

[0858] Step 9:

[0859] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[0860] For example, if a user posts "I've published a new blog post!", other users may comment, "This blog post is completely different from the last one." The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and generate an appropriate reply. The reply is then filtered and finally automatically sent to the social networking site.

[0861] Example 1

[0862] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0863] In conventional social networking services, manually generating and posting appropriate replies to user posts is a time-consuming and labor-intensive task. Furthermore, there is a risk that inappropriate replies or malicious comments may be posted as is, significantly impairing the user experience and the quality of communication. Therefore, there is a need for a system that can properly analyze the intent of posts and generate and send replies quickly and accurately.

[0864] 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.

[0865] In this invention, the server includes means for collecting content posted by users on a social networking service, means for transmitting the collected posted content from a terminal to the server, means for analyzing the collected posted content using natural language processing and identifying the intention of the post, means for using generative artificial intelligence to generate appropriate reply messages based on the intention including comments from other users, means for filtering and correcting the generated reply messages, and means for automatically transmitting the filtered reply messages to the social networking service. This makes it possible to automatically generate and transmit prompt and appropriate replies to user posts, improving the user experience and effectively suppressing inappropriate replies and malicious comments.

[0866] "User" refers to an individual or organization that posts using a social networking service.

[0867] "Device" means a device capable of collecting and transmitting User Posts to a server, including computers, smartphones, tablets, etc.

[0868] "Server" refers to a computer system that has the function of analyzing collected posted data and generating and sending replies.

[0869] "Social Networking Service" means an online platform that enables users to share information and communicate with other users. Examples include Twitter and Facebook.

[0870] "Natural language processing" refers to the technology that enables computers to understand and analyze human language and extract meaning, including word segmentation, part-of-speech analysis, and context understanding.

[0871] "Generative AI" refers to AI technology that can generate sentences and intentions based on input data, such as models like GPT-4.

[0872] A "reply message" refers to a message that is sent in response to a post made by a user.

[0873] "Filtering" refers to the process of checking generated replies and correcting them if they contain inappropriate or misleading content.

[0874] "API" stands for Application Programming Interface and refers to a protocol for communication between different software applications. In this case, this includes APIs provided by social media platforms.

[0875] A "prompt sentence" refers to input text used to give instructions to generative artificial intelligence.

[0876] The present invention relates to a system for automatically generating and sending replies that accurately convey the intention of a user in response to a post made by a user on a social networking service (SNS). Specific embodiments of the system are described below.

[0877] User submission collection

[0878] A user posts on a social networking site. For example, they may post something like, "We've released a new product." The device collects this content and sends it to the server. The collected data includes the text of the post, the time of posting, and the poster's account information. The device can be a computer, smartphone, tablet, or other device.

[0879] Post Intention Analysis

[0880] The server receives the posted data from the device and analyzes it using a natural language processing (NLP) engine. NLP tools such as Stanford NLP and spaCy are used here. The analysis involves tokenizing the posted text and analyzing parts of speech. This analyzed data is passed to a generative artificial intelligence model. A high-performance language model such as GPT-4 is used as the generative artificial intelligence model. This model understands the intention of the post based on the prompt text. An example of a prompt text is, "A user posted, 'We have released a new product.' Please analyze the intention of this post."

[0881] Reply Generation

[0882] The server generates a reply based on the analyzed intent. For example, if another user comments, "Why is it so expensive?", that comment is input into the generative AI model. The generative AI model generates an appropriate reply using a prompt such as, "Another user commented, 'Why is it so expensive?' Please generate a reply to this comment." A specific example of a prompt that is generated is, "We set the price that way because we use high-quality materials."

[0883] The generated replies are filtered on the server side using regular expressions and custom rule-based filtering techniques to remove misleading or inappropriate phrases.

[0884] Automatic transmission

[0885] The server automatically sends the filtered reply to the original commenter. This is done using an API or an interface provided by the SNS platform (e.g., Twitter API, Facebook Graph API). The device logs into the SNS, checks the reply, and posts it. This process ensures that a reply that accurately conveys the user's intention is posted quickly.

[0886] Specific examples

[0887] For example, suppose a user posts, "I've published a new blog post!". Then, another user comments, "This blog post is completely different from the last one." The device collects this post and comment and sends them to the server. The server uses generative artificial intelligence to analyze the intent of the post and understands that it is an introduction to a new blog post. At the same time, it generates a reply to the comment, such as, "This blog post is written from a new perspective and offers different content from the last one," and automatically sends it.

[0888] This system reduces the burden on users of manually replying to each individual user, and also allows the intention of the post to be accurately conveyed, effectively preventing malicious replies and unintended spreading.

[0889] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0890] Step 1: Collect user submissions

[0891] A user logs into a social networking service (SNS) and posts, "We've released a new product."

[0892] The device uses the SNS API to collect the content of the post, including the text of the post, the time of posting, and the poster's account information.

[0893] The terminal sends the collected posting data to the server using HTTPS.

[0894] Input: User's SNS post text, posting time, poster's account information

[0895] Output: Post data sent to the server

[0896] Specific behavior:

[0897] A user posts using a social networking application.

[0898] The device obtains the posting data using an SNS API (e.g., Twitter API) and sends it to the server.

[0899] Step 2: Analyze the intent of the post

[0900] The server passes the posted data received from the device to a natural language processing (NLP) engine (e.g., spaCy).

[0901] The NLP engine tokenizes the posted text and performs part-of-speech analysis.

[0902] The analyzed data is input into a generative artificial intelligence model (e.g., GPT-4).

[0903] The generative artificial intelligence model receives the prompt, "A user posted, 'We have released a new product.' Please analyze the intent of this post." It analyzes the intent and extracts the intent, "This is an introduction to a new product."

[0904] Input: Post data (text, time, account information)

[0905] Output: Parsed intent data

[0906] Specific behavior:

[0907] The server passes the posted data to the NLP engine, which performs tokenization and part-of-speech analysis.

[0908] The server passes the prompt sentence to a generative artificial intelligence model, which analyzes the intent.

[0909] Step 3: Reply Generation

[0910] The server inputs comments from other users into the generative AI model. For example, if a comment is added, "Why is it so expensive?"

[0911] The server passes the prompt text "Another user commented, 'Why is it so expensive?' Please generate a reply to this comment" to the generative artificial intelligence model.

[0912] The generative artificial intelligence model generates the appropriate reply: "We use high-quality materials, that's why we price our products that way."

[0913] Input: Comments from other users, intent data

[0914] Output: The generated reply

[0915] Specific behavior:

[0916] The server collects comments from other users and passes them to the generative artificial intelligence model.

[0917] The server uses the prompt to cause the generative artificial intelligence model to generate a reply sentence.

[0918] Step 4: Filtering replies

[0919] The server filters the generated reply.

[0920] Filtering uses regular expressions and custom rule-based filtering techniques to detect and correct inappropriate or misleading content.

[0921] Input: Generated reply text

[0922] Output: filtered reply

[0923] Specific behavior:

[0924] The server checks the generated reply and makes corrections if necessary.

[0925] Step 5: Automated reply

[0926] The server uses the API of the SNS (for example, Twitter API) to send the filtered reply text to the SNS.

[0927] The device logs into the SNS, checks the reply text, and posts it.

[0928] Input: filtered reply text

[0929] Output: Reply text posted on SNS

[0930] Specific behavior:

[0931] The server sends a reply using the SNS API.

[0932] The device logs into the social networking site and automatically posts a reply.

[0933] The above steps realize a system that automatically generates and sends quick and appropriate replies to user posts.

[0934] (Application example 1)

[0935] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0936] Modern companies need to respond quickly and appropriately to user comments when advertising or running campaigns on social media. However, manually creating consistent and accurate replies is extremely time-consuming, especially when a large number of comments are received. Effective management is also difficult, as companies are required to respond to malicious and negative comments. This reduces the efficiency of their marketing activities and puts their brand image at risk.

[0937] 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.

[0938] In this invention, the server

[0939] A means for collecting content posted by users on a social networking service;

[0940] A method that uses generative artificial intelligence to analyze the collected posts and identify the intention behind the posts,

[0941] A means for analyzing collected advertisement and campaign posts and user comments thereon and generating appropriate reply sentences using generative artificial intelligence;

[0942] means for automatically transmitting the generated reply text to a social networking service;

[0943] A way to generate responses to malicious replies and negative comments;

[0944] means for providing automated filtering and correction functions for the purpose of improving the quality of replies related to advertisements and campaigns;

[0945] Includes:

[0946] This will enable companies to respond quickly and appropriately to user comments on advertisements and campaigns on social media, which is expected to improve the efficiency of marketing activities and enhance brand image.

[0947] "Means for collecting content posted by users on social networking services" refers to a function that automatically acquires posted data on SNS and sends it to a server for subsequent processing.

[0948] "Means using generative artificial intelligence to analyze collected post content and identify the intention of the post" refers to a system that uses natural language processing technology to analyze post content and extract the intention and purpose contained within it.

[0949] "Means for analyzing collected advertisement and campaign posts and user comments on them, and generating appropriate reply messages using generative artificial intelligence" refers to a system that analyzes advertisement and campaign-related posts and user comments, and automatically generates appropriate reply messages based on their content.

[0950] "Means for automatically sending the generated reply text on a social networking service" refers to a function that automatically posts the generated reply text in the comment section or reply section of the SNS.

[0951] The "means for generating responses to malicious replies and negative comments" is a system that automatically generates appropriate responses and replies to comments on social media that show hostility or negative reactions.

[0952] "Means for automatic filtering and correction aimed at improving the quality of reply messages related to advertisements and campaigns" refers to a system that has the function of checking the quality of generated reply messages and automatically correcting inappropriate or misleading content.

[0953] A specific system configuration and an operation method thereof will be described below as an embodiment of the present invention.

[0954] System Program

[0955] This system has the ability to collect content posted by users on social networking services (SNS), analyze the collected content, generate appropriate replies using generative artificial intelligence, and automatically send the generated replies to the SNS. It also has a means to analyze the collected advertisement and campaign posts and user comments on them, and generate appropriate replies using generative artificial intelligence. It also automatically generates responses to malicious replies and negative comments, and has automatic filtering and correction functions aimed at improving the quality of replies related to advertisements and campaigns.

[0956] Processing Details

[0957] The server first collects users' social media posts. The collected data includes text, posting time, account information, and more. This data is sent to the server and analyzed by a natural language processing (NLP) engine. The analyzed data is then passed to a generative artificial intelligence (AI) to identify the intention of the post. For example, for a post that says "We've released a new product," the intent is extracted as "This is an introduction to a new product."

[0958] Generative AI generates appropriate replies based on the identified intent. For example, based on user comments on an advertisement or campaign, the generated replies are automatically posted to the social media comment section. At this stage, they are filtered and edited for quality control. At this stage, misleading or inappropriate content is filtered out and edited as necessary.

[0959] Hardware and Software

[0960] This system is implemented using the following hardware and software:

[0961] SNS API: An interface for collecting posts on social media.

[0962] NLP engine: Performs natural language processing and analyzes post content.

[0963] Generative AI: Used to identify intent and generate appropriate replies. An example is the OpenAI API.

[0964] Automatic filtering and correction: Quality checks generated replies and corrects inappropriate language.

[0965] Specific examples

[0966] For example, if a user posts "I've published a new blog post!" and another user subsequently comments "This post is completely different from the last one," the system would act as follows:

[0967] 1. Collecting posts and comments: The device collects posts and comments and sends them to the server.

[0968] 2. Intent analysis and reply generation: The server uses an NLP engine to analyze the intent of the post and determine that it is an introduction to a new blog post. At the same time, it generates a reply to the comment, saying that this blog post is written from a new perspective and offers different content from the previous post.

[0969] 3. Automatic sending and filtering: The generated replies are automatically posted to social media, with quality checks and necessary corrections being carried out during this process.

[0970] Prompt Sentence Examples

[0971] text

[0972] Please respond to the comment below based on the intent of the post.

[0973] Post Intent: Promoting a new product

[0974] Comments: Where can I buy this product?

[0975] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0976] Step 1:

[0977] A user posts to a social networking site. This post includes text, the time of posting, and account information. The device collects the post content and sends it to the server. Specifically, it uses the social networking site's API to obtain the post data and sends the post content to the server in JSON format or similar. The input is the SNS post data, and the output is the collected data sent to the server.

[0978] Step 2:

[0979] The server analyzes the collected post text using a natural language processing (NLP) engine. Here, basic processing such as word segmentation and part-of-speech analysis is performed to identify the intention of the post. Specifically, the post text is passed to the NLP engine, which analyzes the meaning and context of words. The input is the text data of the SNS post, and the output is the analysis results in which the intention has been identified.

[0980] Step 3:

[0981] The server uses a generative artificial intelligence (generative AI model) to generate an appropriate reply based on the identified intent. For example, a reply to a user's comment is generated based on the intent of "promoting a new product." Specifically, the intent analysis result is input into the generative AI model as a prompt, and a reply is generated. The input is the intent analysis result and the prompt, and the output is the generated reply.

[0982] Step 4:

[0983] The server performs filtering to check the quality of the generated reply and correct it if necessary. Specifically, it re-evaluates the content of the reply and corrects inappropriate or potentially misleading phrases. The input is the generated reply, and the output is the quality-checked and corrected reply.

[0984] Step 5:

[0985] The server automatically sends the reply text that has passed the quality check to the SNS. The API interface of the SNS platform is used for sending. Specifically, the reply text is posted as a comment using the API. The input is the reply text that has passed the quality check, and the output is the reply posted on the SNS.

[0986] Step 6:

[0987] The server analyzes user comments and generates replies to advertisement and campaign posts. The server collects advertisement-related post data and user comments and performs similar processing. As a specific example, in response to a post saying "We've released a new product" and a comment saying "Where can I buy this product?", the server generates a reply saying "This product can be purchased on our official website and at affiliated retailers." The input is advertisement-related post data and user comments, and the output is the generated reply text.

[0988] Step 7:

[0989] Generates responses to malicious replies and negative comments. The server automatically generates appropriate countermeasures and replies based on the identified negative intent. Specifically, a prompt sentence is input into the generative AI model to generate an appropriate response to malicious comments. The input is a negative comment and a prompt sentence, and the output is a corresponding reply sentence.

[0990] Step 8:

[0991] It applies automatic filtering and correction functions to improve the quality of replies related to ads and campaigns. The server re-evaluates the generated replies and corrects them as necessary to improve quality. Specifically, it re-analyzes them using a generative AI model or NLP engine to ensure quality. The input is the generated reply, and the output is the improved reply.

[0992] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0993] The present invention relates to a system that accurately conveys the intention of a user in response to a post made on a social networking service (SNS), recognizes the user's emotions, generates a reply, and automatically sends the reply. Specific embodiments of the system are described below.

[0994] 1. Collecting User Submissions

[0995] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[0996] The device collects the content of these posts and sends it to the server. The collected data includes metadata such as the text of the post, the time of posting, and the user's account information.

[0997] 2. Post Intention Analysis

[0998] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine to segment the post and analyze parts of speech. The analyzed data is then passed to a generative artificial intelligence (AI).

[0999] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[1000] 3. Emotion recognition

[1001] The server applies an emotion engine to the content of the post to recognize the user's emotions. For example, in a post saying "We've released a new product," the emotion engine detects positive emotions such as "excitement" and "expectation."

[1002] The sentiment engine is also applied to comments to recognize whether the comment contains negative sentiment. For example, if a comment says "This product is too expensive," the comment is detected as having a "negative" sentiment.

[1003] 4. Reply Generation

[1004] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI is used to generate the reply.

[1005] The content of the reply text is adjusted based on the emotion recognized by the emotion engine. For example, in response to a negative comment such as "This product is too expensive," a more gentle reply such as "We set the price at that level because we use high-quality materials" is generated.

[1006] 5. Check the quality of replies

[1007] The server checks the quality of the generated replies, using rule-based filtering to correct any misleading or inappropriate phrases.

[1008] 6. Automatic Transmission

[1009] The server automatically sends the filtered reply to the social networking site using an API or an interface provided by the social networking site platform.

[1010] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[1011] Example: Actual flow

[1012] For example, a user might post, "I've published a new blog post!", and then another user might comment, "This blog post is completely different from the last one."

[1013] The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and recognizes the emotion of the comment using an emotion engine. For example, if a comment is recognized as containing "confusion" or "negative" emotions, a gentle reply such as "This is written from a new perspective and offers content that differs from the previous article" is generated.

[1014] These replies are filtered and then automatically sent to the social networking site, allowing for replies that accurately convey the user's intentions and are sensitive to their feelings to be posted quickly.

[1015] The processing flow will be explained below.

[1016] Step 1:

[1017] A user posts on a social networking service (SNS). This post includes content such as text, images, and videos. For example, a user posts content such as "We have released a new product."

[1018] Step 2:

[1019] The device collects this posting content and sends it to the server, including metadata such as the posting text, posting time, and user account information.

[1020] Step 3:

[1021] The server analyzes the received post. First, it uses a natural language processing (NLP) engine to segment the post content and perform basic processing such as part-of-speech analysis.

[1022] Step 4:

[1023] The server passes the analyzed data to a generative artificial intelligence (AI). The generative AI understands the content of the post and identifies its intent. For example, for a post saying "We've released a new product," it identifies the intent as "This is an introduction to the new product."

[1024] Step 5:

[1025] The server applies an emotion engine to the content of the post to recognize the user's emotions. For example, if a post says "We've released a new product," the emotion engine will detect positive emotions such as "excitement" and "expectation."

[1026] Step 6:

[1027] The server collects replies and comments from other users, such as "Why is it so expensive?"

[1028] Step 7:

[1029] The server applies an emotion engine to the collected comments to recognize the emotions contained in the comments. For example, it detects a "negative" emotion in the comment "Why is it so expensive?"

[1030] Step 8:

[1031] Generative AI generates appropriate replies based on the results of intent analysis and emotion recognition. For example, in response to a negative comment like "Why is it so expensive?", it generates a reply such as "We set the price at that because we use high-quality materials."

[1032] Step 9:

[1033] The server checks the quality of the generated replies, using rule-based filtering to check for misleading phrases or inappropriate language, and corrects them if necessary.

[1034] Step 10:

[1035] The server automatically sends the reply text that has passed the quality check to the SNS, using the API or interface provided by the SNS platform.

[1036] Step 11:

[1037] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[1038] As a concrete example, suppose a user posts, "I've published a new blog post!". Later, another user comments, "This blog post is completely different from the last one." The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and recognizes the emotion of the comment using an emotion engine. For example, if the comment is recognized as containing emotions such as "confusion" or "negative," a gentle reply such as "This is written from a new perspective and offers different content from the last post" is generated. This reply is then filtered and automatically sent to the social networking site. This allows replies to be posted quickly that accurately convey the user's intention and take their emotions into consideration.

[1039] Example 2

[1040] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1041] In conventional social networking services, it is difficult to reply appropriately to posts made by users, and in particular, it is difficult to generate replies that accurately reflect the user's intentions and emotions. Furthermore, when the process of checking the quality of generated replies is done manually, it takes time and effort, and there is also the problem that it is not possible to respond quickly. A system that solves these problems is needed.

[1042] 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.

[1043] In this invention, the server includes means for collecting content posted by users on a social networking service, means for analyzing the collected posted content and identifying the intention of the post using a natural language processing engine and generative artificial intelligence, means for using an emotion engine to recognize the emotion of the posted content and comments, means for using generative artificial intelligence to generate appropriate replies based on the identified intention and the recognized emotion, filtering means for checking the quality of the generated replies and correcting them as necessary, and means for automatically sending the filtered replies to the social networking service. This makes it possible to accurately understand the intention and emotion of the user and automatically generate and send prompt and appropriate replies.

[1044] "User" refers to an individual or entity that posts content or messages on a social networking service.

[1045] "Social networking service" refers to an online platform that allows users to interact with each other and share information over the Internet.

[1046] "Means of collection" refers to hardware or software functionality for obtaining content posted by users on social networking services and their metadata.

[1047] A "natural language processing engine" refers to software or algorithms that analyze collected posts and understand the structure and meaning of language.

[1048] "Generative AI" refers to AI technologies and systems that have the ability to generate text based on input information.

[1049] "Emotion Engine" refers to software or algorithms that recognize and classify user emotions from text data in posts and comments.

[1050] "Filtering Measures" means hardware or software features that check the quality of generated replies and modify or remove misleading or inappropriate phrases.

[1051] "Automatic sending means" refers to hardware or software functionality for posting the generated reply on a social networking service.

[1052] "Intention" refers to the purpose or meaning of the information or message a user wants to convey through their post.

[1053] "Emotion" refers to the psychological state expressed by a user when posting or commenting.

[1054] A "reply" refers to a reply message created in response to a user's post.

[1055] The present invention relates to a system that accurately conveys the intention of a user in response to a post made on a social networking service (SNS), recognizes the user's emotions, generates a reply, and automatically sends the reply. Specific embodiments of the present invention are described in detail below.

[1056] User submission collection

[1057] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[1058] The device collects the content of these posts and sends it to the server via the social media platform's API. The collected data includes metadata such as the post text, posting time, and user account information.

[1059] Specific working example:

[1060] A user posts a new product release on social media

[1061] The device calls the SNS API and retrieves the post data.

[1062] Send the acquired data to the server

[1063] Post Intention Analysis

[1064] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to segment the post and analyze parts of speech. The analyzed data is then passed to a generative artificial intelligence (AI) (e.g., OpenAI GPT-3).

[1065] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[1066] Specific working example:

[1067] The server calls the Google Cloud Natural Language API to parse the post text.

[1068] The analysis results are passed to OpenAI GPT-3 to understand the intent.

[1069] The generative AI returns the intent, "This is an introduction to a new product."

[1070] emotion recognition

[1071] The server applies an emotion engine (e.g., IBM Watson Tone Analyzer) to the content of posts to recognize the user's emotions. For example, for a post saying "We've released a new product," the emotion engine will detect positive emotions such as "excitement" and "expectation." The emotion engine is also applied to comments to recognize whether the comment contains negative emotions. For example, if there is a comment saying "This product is too expensive," the comment will be detected as having a "negative" emotion.

[1072] Specific working example:

[1073] The server calls IBM Watson Tone Analyzer to analyze the sentiment of posts and comments.

[1074] Detecting "excitement" from the post "We've released a new product"

[1075] Detecting "negativity" from the comment "This product is too expensive"

[1076] Reply Generation

[1077] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI (e.g., OpenAI GPT-3) is used to generate replies. The content of the reply is adjusted based on the emotion recognized by the emotion engine. For example, in response to a negative comment such as "This product is too expensive," a more gentle reply such as "We set the price at that level because we use high-quality materials" is generated.

[1078] Specific working example:

[1079] The server sends a prompt to OpenAI GPT-3 and generates an appropriate reply.

[1080] The generated reply is "We use high-quality materials, that's why we price it that way."

[1081] Check the quality of replies

[1082] The server checks the quality of the generated reply text, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering (e.g., the Profanity Filter API).

[1083] Specific working example:

[1084] The server calls the Profanity Filter API and checks the reply

[1085] If profanity is detected, it will be automatically corrected or an administrator will be alerted

[1086] Automatic transmission

[1087] The server automatically sends the filtered reply to the SNS using an API or an interface provided by the SNS platform. The device then posts the reply to the SNS, allowing the reply to be displayed on the SNS promptly and accurately conveying the user's intention.

[1088] Specific working example:

[1089] The server sends a reply using the API of the social networking platform (e.g., Twitter API).

[1090] The device receives the message and posts a reply on social media.

[1091] The reply "We use high-quality materials, that's why we price it that way" appears on social media.

[1092] Example prompts for generative AI models

[1093] For example, when a user posts "I published a new blog post!", here's an example prompt:

[1094] A user posts that they've published a new blog post. Another user comments, "This blog post is completely different from the last one." Recognize the sentiment of this comment and generate an appropriate reply.

[1095] This allows the generative AI model to understand the user's intentions and emotions and generate appropriate replies.

[1096] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1097] Step 1:

[1098] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[1099] Input: User-submitted social media content (text, images, metadata, etc.)

[1100] Output: Post data obtained from the SNS API

[1101] The device collects the posted content and sends it to the server via the SNS platform's API. As a specific example of operation, the device calls the SNS API, obtains the posted data (post text, posting time, user account information, etc.), and sends it to the server.

[1102] Step 2:

[1103] The server analyzes the received post. First, it uses a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to segment the post and analyze parts of speech.

[1104] Input: Post data received from the SNS API

[1105] Output: Parsed structured data of post content

[1106] Specifically, the server calls the Google Cloud Natural Language API, analyzes the posted text, and performs word segmentation and part-of-speech analysis. The analysis results are then passed to a generative artificial intelligence (AI) (e.g., OpenAI GPT-3).

[1107] Step 3:

[1108] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[1109] Input: Structured data parsed by an NLP engine

[1110] Output: The intent of the post (e.g., "New product introduction")

[1111] As a specific example of how it works, the analysis results are passed to OpenAI GPT-3 to perform intent understanding. The generative AI identifies the intent of the post and returns the intent to the server.

[1112] Step 4:

[1113] The server applies an emotion engine (e.g., IBM Watson Tone Analyzer) to the posted content to recognize the user's emotions.

[1114] Input: Post data received from SNS API and intent identification results by generative AI

[1115] Output: Sentiment contained in the post and comments (e.g., "excited," "negative," etc.)

[1116] As a specific example of how it works, the server calls IBM Watson Tone Analyzer to analyze the sentiment of posts and comments. For example, it detects "excitement" from "We've released a new product" and "negativity" from "This product is too expensive."

[1117] Step 5:

[1118] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI (e.g., OpenAI GPT-3) is used to generate the reply.

[1119] Input: Intention identification results and emotion recognition results

[1120] Output: Generated reply (e.g., "We use high-quality materials, which is why we set our prices this way.")

[1121] As a specific example of how it works, the server sends a prompt to OpenAI GPT-3, which generates an appropriate reply based on the intent and emotion. The generated reply is then returned to the server.

[1122] Step 6:

[1123] The server checks the quality of the generated reply text, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering (e.g., the Profanity Filter API).

[1124] Input: Generated reply text

[1125] Output: Filtered reply (or revised reply if needed)

[1126] For example, the server calls the Profanity Filter API to check the reply text, and if inappropriate language is detected, it automatically corrects it or sends an alert to the administrator.

[1127] Step 7:

[1128] The server automatically sends the filtered reply to the social networking site using an API or an interface provided by the social networking site platform.

[1129] Input: filtered reply text

[1130] Output: Reply text posted on SNS

[1131] As a specific example of how this works, the server sends a reply using the API of the SNS platform (e.g., Twitter API). The device receives the reply and posts it on the SNS. The reply, "We use high-quality materials, which is why we set our prices this way," is then displayed on the SNS.

[1132] (Application example 2)

[1133] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1134] Modern information distribution platforms often face the problem of not providing replies to user posts promptly and appropriately. It is also difficult to respond appropriately to negative comments or malicious replies, which can detract from the user experience. Furthermore, if quality control of generated replies is insufficient, they may contain misleading phrases or inappropriate expressions, which can reduce user trust. A system that solves these problems is needed.

[1135] 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.

[1136] In this invention, the server includes means for collecting content posted by users to an information distribution platform, means for analyzing the collected posted content and identifying the intention of the post using generative artificial intelligence, means for generating appropriate reply messages based on the identified intention using generative artificial intelligence, means for automatically transmitting the generated reply messages to the information distribution platform, filtering means for checking the quality of the generated reply messages and correcting them as necessary, means for analyzing the collected comments and using an emotion recognition device to recognize the emotions contained in the comments, and means for adjusting the content of the reply messages based on the identified intention and the recognized emotion. This makes it possible to provide quick and appropriate replies, appropriately respond to negative comments and malicious replies, and improve the quality of the generated reply messages.

[1137] "User" means an individual or corporation that posts or comments using the information distribution platform.

[1138] An "information distribution platform" is a service for disseminating and sharing information online, such as blogs and video platforms.

[1139] "Means of collection" refers to software or hardware functions that automatically obtain user posts and comments.

[1140] "Generative artificial intelligence" refers to machine learning models or algorithms that can generate sentences and analyze intent in a human-like manner.

[1141] "Identifying intent" means analyzing and understanding the purpose and intent of posts and comments from their content.

[1142] A "reply" is a document that is automatically generated as a reply to a user's post or comment.

[1143] "Automatic sending means" means software or hardware functionality for automatically posting the generated reply text on a designated platform.

[1144] "Filtering means" is a function that checks the content of generated replies and corrects inappropriate expressions or misleading phrases.

[1145] An "emotion recognizer" is a device or software that analyzes emotions in text and identifies whether the emotions are positive, negative, or neutral.

[1146] "Adjustment" refers to the ability to change or fine-tune the content and tone of the generated reply based on the identified intent and recognized emotion.

[1147] The present invention relates to a system that collects content posted by users on an information distribution platform, analyzes the intention and sentiment of the user in response to the comment, generates appropriate replies, and automatically sends them. Specific embodiments of the system are described below.

[1148] System Configuration

[1149] The system consists of the following main components:

[1150] 1. User terminal: A device (e.g., smartphone, PC) used to access the information distribution platform and make posts and comments.

[1151] 2. Cloud server: A central processing unit that analyzes posts and comments and generates reply messages.

[1152] NLP engine: Uses Google Cloud Natural Language API and IBM Watson Natural Language Understanding.

[1153] Generative AI model: Uses OpenAI's GPT-3 and GPT-4.

[1154] Emotion recognition engine: Uses Microsoft Azure's Text Analytics API.

[1155] 3. Filtering module: A system for checking the quality of generated replies and correcting inappropriate expressions.

[1156] Processing flow

[1157] 1. Collecting submissions:

[1158] A user posts to an information distribution platform.

[1159] The server automatically collects user posts and comments and stores them in a database.

[1160] 2. Intent Analysis:

[1161] The server analyzes the collected posts using an NLP engine to identify the intention of the post.

[1162] Example: For a post that says "We've released a new product," identify the intent as "This is an introduction to a new product."

[1163] 3. Emotion recognition:

[1164] The server uses an emotion recognition engine to analyze the emotions contained in the collected comments.

[1165] For example, recognize the "negative sentiment" in response to the comment "This product is too expensive."

[1166] 4. Reply Generation:

[1167] The server generates a reply sentence using a generative AI model based on the identified intention and emotion recognition results.

[1168] Example: In response to a comment like "This product is too expensive," generate a gentle reply such as "We charge that price because we use high-quality materials."

[1169] 5. Quality check:

[1170] A filtering module checks the quality of the generated replies and corrects any misleading or inappropriate phrases.

[1171] 6. Automated Submission:

[1172] The server automatically sends the filtered reply text to the information distribution platform.

[1173] The specific hardware and software used

[1174] Smartphone: A device that allows users to post and comment.

[1175] Cloud server: Uses Google Cloud or AWS.

[1176] NLP engines: Google Cloud Natural Language API, IBM Watson Natural Language Understanding.

[1177] Generative AI model: OpenAI GPT-3 / GPT-4.

[1178] Sentiment recognition engine: Microsoft Azure Text Analytics API.

[1179] Filtering module: Rule-based system (e.g. Python script).

[1180] Specific examples of operations

[1181] For example, suppose a user posts "I've published a new blog post!" on an information distribution platform. Later, another user comments, "This blog post is completely different from the last one." The server collects these posts and comments, analyzes the intention of the post using a generative AI model, and recognizes the emotion of the comment using an emotion recognition engine. If the comment is recognized as containing "confusion" or "negative" emotions, it generates a gentle reply such as, "This is written from a new perspective and offers content that differs from the previous post." This reply is filtered and eventually automatically sent to the information distribution platform.

[1182] Prompt Sentence Examples

[1183] "Generate a reply to the following comment based on the comment intent 'announce a new blog post' and sentiment 'negative': 'This blog post is completely different from the last one.'"

[1184] In this way, the present invention realizes a system that provides quick and appropriate replies to user posts and also responds appropriately to negative comments.

[1185] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1186] Step 1:

[1187] The server collects content posted by users to the information distribution platform. It uses an API to collect data, including metadata such as the post text, posting time, and user account information. The input is the user's post, and the output is the collected posting data.

[1188] Step 2:

[1189] The server analyzes the collected posts using a natural language processing (NLP) engine. Specifically, it performs word segmentation and part-of-speech analysis. The input is the post data collected in step 1, and the output is the analyzed text data.

[1190] Step 3:

[1191] The server passes the analyzed post content to a generative artificial intelligence (AI) to identify the intent. For example, from the sentence "We have released a new product," it extracts the intent "Introducing a new product." The input is the analyzed text data, and the output is the extracted intent.

[1192] Step 4:

[1193] The server analyzes the collected comments using an emotion recognition engine. It classifies the emotions contained in the comments as positive, negative, or neutral. The input is the comment data, and the output is the recognized emotion information.

[1194] Step 5:

[1195] The server uses generative AI to generate a reply based on the identified intent and the analyzed emotional information. For example, in response to a negative comment such as "This product is too expensive," it generates a reply such as "We set the price at that level because we use high-quality materials." The input is intent information and emotional information, and the output is the generated reply.

[1196] Step 6:

[1197] The server passes the generated reply to a filtering module for quality checks. If it contains misleading phrases or inappropriate expressions, it corrects them. The input is the generated reply, and the output is the reply that passes the quality check.

[1198] Step 7:

[1199] The server automatically sends the filtered reply text to the information distribution platform using an API or an interface provided by the information distribution platform. The input is the reply text that has passed the quality check, and the output is the reply posted on the information distribution platform.

[1200] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1201] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1202] 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.

[1203] [Fourth embodiment]

[1204] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1205] 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.

[1206] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1207] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1208] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1209] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1210] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1211] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1212] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1213] 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.

[1214] 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.

[1215] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1216] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1217] The present invention relates to a system for automatically generating and sending replies that accurately convey the intention of a user in response to a post made by a user on a social networking service (SNS). Specific embodiments of the system are described below.

[1218] 1. Collecting User Submissions

[1219] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[1220] The device collects the posted content and sends it to the server. The collected data includes the text of the post, the time of posting, and the poster's account information.

[1221] 2. Post Intention Analysis

[1222] The server analyzes the received posts, using a natural language processing (NLP) engine to perform basic processing such as word segmentation and part-of-speech analysis.

[1223] The analyzed data is passed to a generative AI, which understands the context and intent of the post. For example, for a post that says "We've released a new product," the intent is extracted as "This is an introduction to a new product."

[1224] 3. Reply Generation

[1225] The server generates a reply on social media based on the intent analysis results. For example, if another user comments, "Why is it so expensive?", the comment is passed to the generative AI.

[1226] Generative AI generates an appropriate reply, such as, "We use high-quality materials, which is why we set our prices."

[1227] The generated replies are filtered on the server side for quality checks, where misleading phrases and inappropriate language are corrected.

[1228] 4. Automatic sending

[1229] The server automatically sends the filtered reply to the original commenter using an API or interface provided by the social media platform.

[1230] The device logs into the SNS, checks the reply text, and posts it. This process ensures that a reply that accurately conveys the user's intention is posted quickly.

[1231] Example: Actual flow

[1232] For example, a user might post, "I've published a new blog post!", and then another user might comment, "This blog post is completely different from the last one."

[1233] The device collects these posts and comments and sends them to a server. The server uses generative artificial intelligence to analyze the intent of the post and understands that it is an introduction to a new blog post. At the same time, it generates and automatically sends a reply to the comment, such as, "This blog post is written from a new perspective and offers different content from the previous post."

[1234] This system reduces the burden on users of manually replying to each individual user, and also allows the intention of the post to be accurately conveyed, effectively preventing malicious replies and unintended spreading.

[1235] The processing flow will be explained below.

[1236] Step 1:

[1237] A user posts on a social networking service (SNS). This post includes content such as text, images, and videos. For example, a user posts content such as "We have released a new product."

[1238] Step 2:

[1239] The device collects this posting content and sends it to the server, including metadata such as the posting text, posting time, and user account information.

[1240] Step 3:

[1241] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine to segment the post and analyze parts of speech. The analyzed data is then passed to generative artificial intelligence (AI).

[1242] Step 4:

[1243] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it identifies the intent as "This is an introduction to the new product."

[1244] Step 5:

[1245] The server starts the reply generation process based on the analysis results. Replies and comments from other users are also collected and passed to the generative AI.

[1246] Step 6:

[1247] Generative AI generates replies. For example, if another user comments, "Why is it so expensive?", the system generates a reply such as, "We use high-quality materials, which is why we set the price that way."

[1248] Step 7:

[1249] The server checks the quality of the generated replies, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering.

[1250] Step 8:

[1251] The server automatically sends the filtered reply to the social networking site via API or an interface provided by the social networking site platform.

[1252] Step 9:

[1253] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[1254] For example, if a user posts "I've published a new blog post!", other users may comment, "This blog post is completely different from the last one." The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and generate an appropriate reply. The reply is then filtered and finally automatically sent to the social networking site.

[1255] Example 1

[1256] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1257] In conventional social networking services, manually generating and posting appropriate replies to user posts is a time-consuming and labor-intensive task. Furthermore, there is a risk that inappropriate replies or malicious comments may be posted as is, significantly impairing the user experience and the quality of communication. Therefore, there is a need for a system that can properly analyze the intent of posts and generate and send replies quickly and accurately.

[1258] 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.

[1259] In this invention, the server includes means for collecting content posted by users on a social networking service, means for transmitting the collected posted content from a terminal to the server, means for analyzing the collected posted content using natural language processing and identifying the intention of the post, means for using generative artificial intelligence to generate appropriate reply messages based on the intention including comments from other users, means for filtering and correcting the generated reply messages, and means for automatically transmitting the filtered reply messages to the social networking service. This makes it possible to automatically generate and transmit prompt and appropriate replies to user posts, improving the user experience and effectively suppressing inappropriate replies and malicious comments.

[1260] "User" refers to an individual or organization that posts using a social networking service.

[1261] "Device" means a device capable of collecting and transmitting User Posts to a server, including computers, smartphones, tablets, etc.

[1262] "Server" refers to a computer system that has the function of analyzing collected posted data and generating and sending replies.

[1263] "Social Networking Service" means an online platform that enables users to share information and communicate with other users. Examples include Twitter and Facebook.

[1264] "Natural language processing" refers to the technology that enables computers to understand and analyze human language and extract meaning, including word segmentation, part-of-speech analysis, and context understanding.

[1265] "Generative AI" refers to AI technology that can generate sentences and intentions based on input data, such as models like GPT-4.

[1266] A "reply message" refers to a message that is sent in response to a post made by a user.

[1267] "Filtering" refers to the process of checking generated replies and correcting them if they contain inappropriate or misleading content.

[1268] "API" stands for Application Programming Interface and refers to a protocol for communication between different software applications. In this case, this includes APIs provided by social media platforms.

[1269] A "prompt sentence" refers to input text used to give instructions to generative artificial intelligence.

[1270] The present invention relates to a system for automatically generating and sending replies that accurately convey the intention of a user in response to a post made by a user on a social networking service (SNS). Specific embodiments of the system are described below.

[1271] User submission collection

[1272] A user posts on a social networking site. For example, they may post something like, "We've released a new product." The device collects this content and sends it to the server. The collected data includes the text of the post, the time of posting, and the poster's account information. The device can be a computer, smartphone, tablet, or other device.

[1273] Post Intention Analysis

[1274] The server receives the posted data from the device and analyzes it using a natural language processing (NLP) engine. NLP tools such as Stanford NLP and spaCy are used here. The analysis involves tokenizing the posted text and analyzing parts of speech. This analyzed data is passed to a generative artificial intelligence model. A high-performance language model such as GPT-4 is used as the generative artificial intelligence model. This model understands the intention of the post based on the prompt text. An example of a prompt text is, "A user posted, 'We have released a new product.' Please analyze the intention of this post."

[1275] Reply Generation

[1276] The server generates a reply based on the analyzed intent. For example, if another user comments, "Why is it so expensive?", that comment is input into the generative AI model. The generative AI model generates an appropriate reply using a prompt such as, "Another user commented, 'Why is it so expensive?' Please generate a reply to this comment." A specific example of a prompt that is generated is, "We set the price that way because we use high-quality materials."

[1277] The generated replies are filtered on the server side using regular expressions and custom rule-based filtering techniques to remove misleading or inappropriate phrases.

[1278] Automatic transmission

[1279] The server automatically sends the filtered reply to the original commenter. This is done using an API or an interface provided by the SNS platform (e.g., Twitter API, Facebook Graph API). The device logs into the SNS, checks the reply, and posts it. This process ensures that a reply that accurately conveys the user's intention is posted quickly.

[1280] Specific examples

[1281] For example, suppose a user posts, "I've published a new blog post!". Then, another user comments, "This blog post is completely different from the last one." The device collects this post and comment and sends them to the server. The server uses generative artificial intelligence to analyze the intent of the post and understands that it is an introduction to a new blog post. At the same time, it generates a reply to the comment, such as, "This blog post is written from a new perspective and offers different content from the last one," and automatically sends it.

[1282] This system reduces the burden on users of manually replying to each individual user, and also allows the intention of the post to be accurately conveyed, effectively preventing malicious replies and unintended spreading.

[1283] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1284] Step 1: Collect user submissions

[1285] A user logs into a social networking service (SNS) and posts, "We've released a new product."

[1286] The device uses the SNS API to collect the content of the post, including the text of the post, the time of posting, and the poster's account information.

[1287] The terminal sends the collected posting data to the server using HTTPS.

[1288] Input: User's SNS post text, posting time, poster's account information

[1289] Output: Post data sent to the server

[1290] Specific behavior:

[1291] A user posts using a social networking application.

[1292] The device obtains the posting data using an SNS API (e.g., Twitter API) and sends it to the server.

[1293] Step 2: Analyze the intent of the post

[1294] The server passes the posted data received from the device to a natural language processing (NLP) engine (e.g., spaCy).

[1295] The NLP engine tokenizes the posted text and performs part-of-speech analysis.

[1296] The analyzed data is input into a generative artificial intelligence model (e.g., GPT-4).

[1297] The generative artificial intelligence model receives the prompt, "A user posted, 'We have released a new product.' Please analyze the intent of this post." It analyzes the intent and extracts the intent, "This is an introduction to a new product."

[1298] Input: Post data (text, time, account information)

[1299] Output: Parsed intent data

[1300] Specific behavior:

[1301] The server passes the posted data to the NLP engine, which performs tokenization and part-of-speech analysis.

[1302] The server passes the prompt sentence to a generative artificial intelligence model, which analyzes the intent.

[1303] Step 3: Reply Generation

[1304] The server inputs comments from other users into the generative AI model. For example, if a comment is added, "Why is it so expensive?"

[1305] The server passes the prompt text "Another user commented, 'Why is it so expensive?' Please generate a reply to this comment" to the generative artificial intelligence model.

[1306] The generative artificial intelligence model generates the appropriate reply: "We use high-quality materials, that's why we price our products that way."

[1307] Input: Comments from other users, intent data

[1308] Output: The generated reply

[1309] Specific behavior:

[1310] The server collects comments from other users and passes them to the generative artificial intelligence model.

[1311] The server uses the prompt to cause the generative artificial intelligence model to generate a reply sentence.

[1312] Step 4: Filtering replies

[1313] The server filters the generated reply.

[1314] Filtering uses regular expressions and custom rule-based filtering techniques to detect and correct inappropriate or misleading content.

[1315] Input: Generated reply text

[1316] Output: filtered reply

[1317] Specific behavior:

[1318] The server checks the generated reply and makes corrections if necessary.

[1319] Step 5: Automated reply

[1320] The server uses the API of the SNS (for example, Twitter API) to send the filtered reply text to the SNS.

[1321] The device logs into the SNS, checks the reply text, and posts it.

[1322] Input: filtered reply text

[1323] Output: Reply text posted on SNS

[1324] Specific behavior:

[1325] The server sends a reply using the SNS API.

[1326] The device logs into the social networking site and automatically posts a reply.

[1327] The above steps realize a system that automatically generates and sends quick and appropriate replies to user posts.

[1328] (Application example 1)

[1329] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1330] Modern companies need to respond quickly and appropriately to user comments when advertising or running campaigns on social media. However, manually creating consistent and accurate replies is extremely time-consuming, especially when a large number of comments are received. Effective management is also difficult, as companies are required to respond to malicious and negative comments. This reduces the efficiency of their marketing activities and puts their brand image at risk.

[1331] 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.

[1332] In this invention, the server

[1333] A means for collecting content posted by users on a social networking service;

[1334] A method that uses generative artificial intelligence to analyze the collected posts and identify the intention behind the posts,

[1335] A means for analyzing collected advertisement and campaign posts and user comments thereon and generating appropriate reply sentences using generative artificial intelligence;

[1336] means for automatically transmitting the generated reply text to a social networking service;

[1337] A way to generate responses to malicious replies and negative comments;

[1338] means for providing automated filtering and correction functions for the purpose of improving the quality of replies related to advertisements and campaigns;

[1339] Includes:

[1340] This will enable companies to respond quickly and appropriately to user comments on advertisements and campaigns on social media, which is expected to improve the efficiency of marketing activities and enhance brand image.

[1341] "Means for collecting content posted by users on social networking services" refers to a function that automatically acquires posted data on SNS and sends it to a server for subsequent processing.

[1342] "Means using generative artificial intelligence to analyze collected post content and identify the intention of the post" refers to a system that uses natural language processing technology to analyze post content and extract the intention and purpose contained within it.

[1343] "Means for analyzing collected advertisement and campaign posts and user comments on them, and generating appropriate reply messages using generative artificial intelligence" refers to a system that analyzes advertisement and campaign-related posts and user comments, and automatically generates appropriate reply messages based on their content.

[1344] "Means for automatically sending the generated reply text on a social networking service" refers to a function that automatically posts the generated reply text in the comment section or reply section of the SNS.

[1345] The "means for generating responses to malicious replies and negative comments" is a system that automatically generates appropriate responses and replies to comments on social media that show hostility or negative reactions.

[1346] "Means for automatic filtering and correction aimed at improving the quality of reply messages related to advertisements and campaigns" refers to a system that has the function of checking the quality of generated reply messages and automatically correcting inappropriate or misleading content.

[1347] A specific system configuration and an operation method thereof will be described below as an embodiment of the present invention.

[1348] System Program

[1349] This system has the ability to collect content posted by users on social networking services (SNS), analyze the collected content, generate appropriate replies using generative artificial intelligence, and automatically send the generated replies to the SNS. It also has a means to analyze the collected advertisement and campaign posts and user comments on them, and generate appropriate replies using generative artificial intelligence. It also automatically generates responses to malicious replies and negative comments, and has automatic filtering and correction functions aimed at improving the quality of replies related to advertisements and campaigns.

[1350] Processing Details

[1351] The server first collects users' social media posts. The collected data includes text, posting time, account information, and more. This data is sent to the server and analyzed by a natural language processing (NLP) engine. The analyzed data is then passed to a generative artificial intelligence (AI) to identify the intention of the post. For example, for a post that says "We've released a new product," the intent is extracted as "This is an introduction to a new product."

[1352] Generative AI generates appropriate replies based on the identified intent. For example, based on user comments on an advertisement or campaign, the generated replies are automatically posted to the social media comment section. At this stage, they are filtered and edited for quality control. At this stage, misleading or inappropriate content is filtered out and edited as necessary.

[1353] Hardware and Software

[1354] This system is implemented using the following hardware and software:

[1355] SNS API: An interface for collecting posts on social media.

[1356] NLP engine: Performs natural language processing and analyzes post content.

[1357] Generative AI: Used to identify intent and generate appropriate replies. An example is the OpenAI API.

[1358] Automatic filtering and correction: Quality checks generated replies and corrects inappropriate language.

[1359] Specific examples

[1360] For example, if a user posts "I've published a new blog post!" and another user subsequently comments "This post is completely different from the last one," the system would act as follows:

[1361] 1. Collecting posts and comments: The device collects posts and comments and sends them to the server.

[1362] 2. Intent analysis and reply generation: The server uses an NLP engine to analyze the intent of the post and determine that it is an introduction to a new blog post. At the same time, it generates a reply to the comment, saying that this blog post is written from a new perspective and offers different content from the previous post.

[1363] 3. Automatic sending and filtering: The generated replies are automatically posted to social media, with quality checks and necessary corrections being carried out during this process.

[1364] Prompt Sentence Examples

[1365] text

[1366] Please respond to the comment below based on the intent of the post.

[1367] Post Intent: Promoting a new product

[1368] Comments: Where can I buy this product?

[1369] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1370] Step 1:

[1371] A user posts to a social networking site. This post includes text, the time of posting, and account information. The device collects the post content and sends it to the server. Specifically, it uses the social networking site's API to obtain the post data and sends the post content to the server in JSON format or similar. The input is the SNS post data, and the output is the collected data sent to the server.

[1372] Step 2:

[1373] The server analyzes the collected post text using a natural language processing (NLP) engine. Here, basic processing such as word segmentation and part-of-speech analysis is performed to identify the intention of the post. Specifically, the post text is passed to the NLP engine, which analyzes the meaning and context of words. The input is the text data of the SNS post, and the output is the analysis results in which the intention has been identified.

[1374] Step 3:

[1375] The server uses a generative artificial intelligence (generative AI model) to generate an appropriate reply based on the identified intent. For example, a reply to a user's comment is generated based on the intent of "promoting a new product." Specifically, the intent analysis result is input into the generative AI model as a prompt, and a reply is generated. The input is the intent analysis result and the prompt, and the output is the generated reply.

[1376] Step 4:

[1377] The server performs filtering to check the quality of the generated reply and correct it if necessary. Specifically, it re-evaluates the content of the reply and corrects inappropriate or potentially misleading phrases. The input is the generated reply, and the output is the quality-checked and corrected reply.

[1378] Step 5:

[1379] The server automatically sends the reply text that has passed the quality check to the SNS. The API interface of the SNS platform is used for sending. Specifically, the reply text is posted as a comment using the API. The input is the reply text that has passed the quality check, and the output is the reply posted on the SNS.

[1380] Step 6:

[1381] The server analyzes user comments and generates replies to advertisement and campaign posts. The server collects advertisement-related post data and user comments and performs similar processing. As a specific example, in response to a post saying "We've released a new product" and a comment saying "Where can I buy this product?", the server generates a reply saying "This product can be purchased on our official website and at affiliated retailers." The input is advertisement-related post data and user comments, and the output is the generated reply text.

[1382] Step 7:

[1383] Generates responses to malicious replies and negative comments. The server automatically generates appropriate countermeasures and replies based on the identified negative intent. Specifically, a prompt sentence is input into the generative AI model to generate an appropriate response to malicious comments. The input is a negative comment and a prompt sentence, and the output is a corresponding reply sentence.

[1384] Step 8:

[1385] It applies automatic filtering and correction functions to improve the quality of replies related to ads and campaigns. The server re-evaluates the generated replies and corrects them as necessary to improve quality. Specifically, it re-analyzes them using a generative AI model or NLP engine to ensure quality. The input is the generated reply, and the output is the improved reply.

[1386] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1387] The present invention relates to a system that accurately conveys the intention of a user in response to a post made on a social networking service (SNS), recognizes the user's emotions, generates a reply, and automatically sends the reply. Specific embodiments of the system are described below.

[1388] 1. Collecting User Submissions

[1389] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[1390] The device collects the content of these posts and sends it to the server. The collected data includes metadata such as the text of the post, the time of posting, and the user's account information.

[1391] 2. Post Intention Analysis

[1392] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine to segment the post and analyze parts of speech. The analyzed data is then passed to a generative artificial intelligence (AI).

[1393] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[1394] 3. Emotion recognition

[1395] The server applies an emotion engine to the content of the post to recognize the user's emotions. For example, in a post saying "We've released a new product," the emotion engine detects positive emotions such as "excitement" and "expectation."

[1396] The sentiment engine is also applied to comments to recognize whether the comment contains negative sentiment. For example, if a comment says "This product is too expensive," the comment is detected as having a "negative" sentiment.

[1397] 4. Reply Generation

[1398] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI is used to generate the reply.

[1399] The content of the reply text is adjusted based on the emotion recognized by the emotion engine. For example, in response to a negative comment such as "This product is too expensive," a more gentle reply such as "We set the price at that level because we use high-quality materials" is generated.

[1400] 5. Check the quality of replies

[1401] The server checks the quality of the generated replies, using rule-based filtering to correct any misleading or inappropriate phrases.

[1402] 6. Automatic Transmission

[1403] The server automatically sends the filtered reply to the social networking site using an API or an interface provided by the social networking site platform.

[1404] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[1405] Example: Actual flow

[1406] For example, a user might post, "I've published a new blog post!", and then another user might comment, "This blog post is completely different from the last one."

[1407] The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and recognizes the emotion of the comment using an emotion engine. For example, if a comment is recognized as containing "confusion" or "negative" emotions, a gentle reply such as "This is written from a new perspective and offers content that differs from the previous article" is generated.

[1408] These replies are filtered and then automatically sent to the social networking site, allowing for replies that accurately convey the user's intentions and are sensitive to their feelings to be posted quickly.

[1409] The processing flow will be explained below.

[1410] Step 1:

[1411] A user posts on a social networking service (SNS). This post includes content such as text, images, and videos. For example, a user posts content such as "We have released a new product."

[1412] Step 2:

[1413] The device collects this posting content and sends it to the server, including metadata such as the posting text, posting time, and user account information.

[1414] Step 3:

[1415] The server analyzes the received post. First, it uses a natural language processing (NLP) engine to segment the post content and perform basic processing such as part-of-speech analysis.

[1416] Step 4:

[1417] The server passes the analyzed data to a generative artificial intelligence (AI). The generative AI understands the content of the post and identifies its intent. For example, for a post saying "We've released a new product," it identifies the intent as "This is an introduction to the new product."

[1418] Step 5:

[1419] The server applies an emotion engine to the content of the post to recognize the user's emotions. For example, if a post says "We've released a new product," the emotion engine will detect positive emotions such as "excitement" and "expectation."

[1420] Step 6:

[1421] The server collects replies and comments from other users, such as "Why is it so expensive?"

[1422] Step 7:

[1423] The server applies an emotion engine to the collected comments to recognize the emotions contained in the comments. For example, it detects a "negative" emotion in the comment "Why is it so expensive?"

[1424] Step 8:

[1425] Generative AI generates appropriate replies based on the results of intent analysis and emotion recognition. For example, in response to a negative comment like "Why is it so expensive?", it generates a reply such as "We set the price at that because we use high-quality materials."

[1426] Step 9:

[1427] The server checks the quality of the generated replies, using rule-based filtering to check for misleading phrases or inappropriate language, and corrects them if necessary.

[1428] Step 10:

[1429] The server automatically sends the reply text that has passed the quality check to the SNS, using the API or interface provided by the SNS platform.

[1430] Step 11:

[1431] The device posts the sent reply to the SNS, allowing a reply that accurately conveys the user's intention to be displayed promptly on the SNS.

[1432] As a concrete example, suppose a user posts, "I've published a new blog post!". Later, another user comments, "This blog post is completely different from the last one." The device collects these posts and comments, and the server uses generative AI to analyze the intention of the post and recognizes the emotion of the comment using an emotion engine. For example, if the comment is recognized as containing emotions such as "confusion" or "negative," a gentle reply such as "This is written from a new perspective and offers different content from the last post" is generated. This reply is then filtered and automatically sent to the social networking site. This allows replies to be posted quickly that accurately convey the user's intention and take their emotions into consideration.

[1433] Example 2

[1434] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1435] In conventional social networking services, it is difficult to reply appropriately to posts made by users, and in particular, it is difficult to generate replies that accurately reflect the user's intentions and emotions. Furthermore, when the process of checking the quality of generated replies is done manually, it takes time and effort, and there is also the problem that it is not possible to respond quickly. A system that solves these problems is needed.

[1436] 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.

[1437] In this invention, the server includes means for collecting content posted by users on a social networking service, means for analyzing the collected posted content and identifying the intention of the post using a natural language processing engine and generative artificial intelligence, means for using an emotion engine to recognize the emotion of the posted content and comments, means for using generative artificial intelligence to generate appropriate replies based on the identified intention and the recognized emotion, filtering means for checking the quality of the generated replies and correcting them as necessary, and means for automatically sending the filtered replies to the social networking service. This makes it possible to accurately understand the intention and emotion of the user and automatically generate and send prompt and appropriate replies.

[1438] "User" refers to an individual or entity that posts content or messages on a social networking service.

[1439] "Social networking service" refers to an online platform that allows users to interact with each other and share information over the Internet.

[1440] "Means of collection" refers to hardware or software functionality for obtaining content posted by users on social networking services and their metadata.

[1441] A "natural language processing engine" refers to software or algorithms that analyze collected posts and understand the structure and meaning of language.

[1442] "Generative AI" refers to AI technologies and systems that have the ability to generate text based on input information.

[1443] "Emotion Engine" refers to software or algorithms that recognize and classify user emotions from text data in posts and comments.

[1444] "Filtering Measures" means hardware or software features that check the quality of generated replies and modify or remove misleading or inappropriate phrases.

[1445] "Automatic sending means" refers to hardware or software functionality for posting the generated reply on a social networking service.

[1446] "Intention" refers to the purpose or meaning of the information or message a user wants to convey through their post.

[1447] "Emotion" refers to the psychological state expressed by a user when posting or commenting.

[1448] A "reply" refers to a reply message created in response to a user's post.

[1449] The present invention relates to a system that accurately conveys the intention of a user in response to a post made on a social networking service (SNS), recognizes the user's emotions, generates a reply, and automatically sends the reply. Specific embodiments of the present invention are described in detail below.

[1450] User submission collection

[1451] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[1452] The device collects the content of these posts and sends it to the server via the social media platform's API. The collected data includes metadata such as the post text, posting time, and user account information.

[1453] Specific working example:

[1454] A user posts a new product release on social media

[1455] The device calls the SNS API and retrieves the post data.

[1456] Send the acquired data to the server

[1457] Post Intention Analysis

[1458] The server analyzes the received posts. First, it uses a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to segment the post and analyze parts of speech. The analyzed data is then passed to a generative artificial intelligence (AI) (e.g., OpenAI GPT-3).

[1459] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[1460] Specific working example:

[1461] The server calls the Google Cloud Natural Language API to parse the post text.

[1462] The analysis results are passed to OpenAI GPT-3 to understand the intent.

[1463] The generative AI returns the intent, "This is an introduction to a new product."

[1464] emotion recognition

[1465] The server applies an emotion engine (e.g., IBM Watson Tone Analyzer) to the content of posts to recognize the user's emotions. For example, for a post saying "We've released a new product," the emotion engine will detect positive emotions such as "excitement" and "expectation." The emotion engine is also applied to comments to recognize whether the comment contains negative emotions. For example, if there is a comment saying "This product is too expensive," the comment will be detected as having a "negative" emotion.

[1466] Specific working example:

[1467] The server calls IBM Watson Tone Analyzer to analyze the sentiment of posts and comments.

[1468] Detecting "excitement" from the post "We've released a new product"

[1469] Detecting "negativity" from the comment "This product is too expensive"

[1470] Reply Generation

[1471] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI (e.g., OpenAI GPT-3) is used to generate replies. The content of the reply is adjusted based on the emotion recognized by the emotion engine. For example, in response to a negative comment such as "This product is too expensive," a more gentle reply such as "We set the price at that level because we use high-quality materials" is generated.

[1472] Specific working example:

[1473] The server sends a prompt to OpenAI GPT-3 and generates an appropriate reply.

[1474] The generated reply is "We use high-quality materials, that's why we price it that way."

[1475] Check the quality of replies

[1476] The server checks the quality of the generated reply text, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering (e.g., the Profanity Filter API).

[1477] Specific working example:

[1478] The server calls the Profanity Filter API and checks the reply

[1479] If profanity is detected, it will be automatically corrected or an administrator will be alerted

[1480] Automatic transmission

[1481] The server automatically sends the filtered reply to the SNS using an API or an interface provided by the SNS platform. The device then posts the reply to the SNS, allowing the reply to be displayed on the SNS promptly and accurately conveying the user's intention.

[1482] Specific working example:

[1483] The server sends a reply using the API of the social networking platform (e.g., Twitter API).

[1484] The device receives the message and posts a reply on social media.

[1485] The reply "We use high-quality materials, that's why we price it that way" appears on social media.

[1486] Example prompts for generative AI models

[1487] For example, when a user posts "I published a new blog post!", here's an example prompt:

[1488] A user posts that they've published a new blog post. Another user comments, "This blog post is completely different from the last one." Recognize the sentiment of this comment and generate an appropriate reply.

[1489] This allows the generative AI model to understand the user's intentions and emotions and generate appropriate replies.

[1490] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1491] Step 1:

[1492] A user posts on a social networking site. For example, they may post something like "We've released a new product."

[1493] Input: User-submitted social media content (text, images, metadata, etc.)

[1494] Output: Post data obtained from the SNS API

[1495] The device collects the posted content and sends it to the server via the SNS platform's API. As a specific example of operation, the device calls the SNS API, obtains the posted data (post text, posting time, user account information, etc.), and sends it to the server.

[1496] Step 2:

[1497] The server analyzes the received post. First, it uses a natural language processing (NLP) engine (e.g., Google Cloud Natural Language API) to segment the post and analyze parts of speech.

[1498] Input: Post data received from the SNS API

[1499] Output: Parsed structured data of post content

[1500] Specifically, the server calls the Google Cloud Natural Language API, analyzes the posted text, and performs word segmentation and part-of-speech analysis. The analysis results are then passed to a generative artificial intelligence (AI) (e.g., OpenAI GPT-3).

[1501] Step 3:

[1502] Generative AI analyzes the content of posts and understands their intent. For example, if a post says "We've released a new product," it can identify the intent as "This is an introduction to the new product."

[1503] Input: Structured data parsed by an NLP engine

[1504] Output: The intent of the post (e.g., "New product introduction")

[1505] As a specific example of how it works, the analysis results are passed to OpenAI GPT-3 to perform intent understanding. The generative AI identifies the intent of the post and returns the intent to the server.

[1506] Step 4:

[1507] The server applies an emotion engine (e.g., IBM Watson Tone Analyzer) to the posted content to recognize the user's emotions.

[1508] Input: Post data received from SNS API and intent identification results by generative AI

[1509] Output: Sentiment contained in the post and comments (e.g., "excited," "negative," etc.)

[1510] As a specific example of how it works, the server calls IBM Watson Tone Analyzer to analyze the sentiment of posts and comments. For example, it detects "excitement" from "We've released a new product" and "negativity" from "This product is too expensive."

[1511] Step 5:

[1512] The server starts the reply generation process based on the intent analysis and emotion recognition results. A generative AI (e.g., OpenAI GPT-3) is used to generate the reply.

[1513] Input: Intention identification results and emotion recognition results

[1514] Output: Generated reply (e.g., "We use high-quality materials, which is why we set our prices this way.")

[1515] As a specific example of how it works, the server sends a prompt to OpenAI GPT-3, which generates an appropriate reply based on the intent and emotion. The generated reply is then returned to the server.

[1516] Step 6:

[1517] The server checks the quality of the generated reply text, and at this stage, any misleading or inappropriate phrases are corrected using rule-based filtering (e.g., the Profanity Filter API).

[1518] Input: Generated reply text

[1519] Output: Filtered reply (or revised reply if needed)

[1520] For example, the server calls the Profanity Filter API to check the reply text, and if inappropriate language is detected, it automatically corrects it or sends an alert to the administrator.

[1521] Step 7:

[1522] The server automatically sends the filtered reply to the social networking site using an API or an interface provided by the social networking site platform.

[1523] Input: filtered reply text

[1524] Output: Reply text posted on SNS

[1525] As a specific example of how this works, the server sends a reply using the API of the SNS platform (e.g., Twitter API). The device receives the reply and posts it on the SNS. The reply, "We use high-quality materials, which is why we set our prices this way," is then displayed on the SNS.

[1526] (Application example 2)

[1527] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1528] Modern information distribution platforms often face the problem of not providing replies to user posts promptly and appropriately. It is also difficult to respond appropriately to negative comments or malicious replies, which can detract from the user experience. Furthermore, if quality control of generated replies is insufficient, they may contain misleading phrases or inappropriate expressions, which can reduce user trust. A system that solves these problems is needed.

[1529] 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.

[1530] In this invention, the server includes means for collecting content posted by users to an information distribution platform, means for analyzing the collected posted content and identifying the intention of the post using generative artificial intelligence, means for generating appropriate reply messages based on the identified intention using generative artificial intelligence, means for automatically transmitting the generated reply messages to the information distribution platform, filtering means for checking the quality of the generated reply messages and correcting them as necessary, means for analyzing the collected comments and using an emotion recognition device to recognize the emotions contained in the comments, and means for adjusting the content of the reply messages based on the identified intention and the recognized emotion. This makes it possible to provide quick and appropriate replies, appropriately respond to negative comments and malicious replies, and improve the quality of the generated reply messages.

[1531] "User" means an individual or corporation that posts or comments using the information distribution platform.

[1532] An "information distribution platform" is a service for disseminating and sharing information online, such as blogs and video platforms.

[1533] "Means of collection" refers to software or hardware functions that automatically obtain user posts and comments.

[1534] "Generative artificial intelligence" refers to machine learning models or algorithms that can generate sentences and analyze intent in a human-like manner.

[1535] "Identifying intent" means analyzing and understanding the purpose and intent of posts and comments from their content.

[1536] A "reply" is a document that is automatically generated as a reply to a user's post or comment.

[1537] "Automatic sending means" means software or hardware functionality for automatically posting the generated reply text on a designated platform.

[1538] "Filtering means" is a function that checks the content of generated replies and corrects inappropriate expressions or misleading phrases.

[1539] An "emotion recognizer" is a device or software that analyzes emotions in text and identifies whether the emotions are positive, negative, or neutral.

[1540] "Adjustment" refers to the ability to change or fine-tune the content and tone of the generated reply based on the identified intent and recognized emotion.

[1541] The present invention relates to a system that collects content posted by users on an information distribution platform, analyzes the intention and sentiment of the user in response to the comment, generates appropriate replies, and automatically sends them. Specific embodiments of the system are described below.

[1542] System Configuration

[1543] The system consists of the following main components:

[1544] 1. User terminal: A device (e.g., smartphone, PC) used to access the information distribution platform and make posts and comments.

[1545] 2. Cloud server: A central processing unit that analyzes posts and comments and generates reply messages.

[1546] NLP engine: Uses Google Cloud Natural Language API and IBM Watson Natural Language Understanding.

[1547] Generative AI model: Uses OpenAI's GPT-3 and GPT-4.

[1548] Emotion recognition engine: Uses Microsoft Azure's Text Analytics API.

[1549] 3. Filtering module: A system for checking the quality of generated replies and correcting inappropriate expressions.

[1550] Processing flow

[1551] 1. Collecting submissions:

[1552] A user posts to an information distribution platform.

[1553] The server automatically collects user posts and comments and stores them in a database.

[1554] 2. Intent Analysis:

[1555] The server analyzes the collected posts using an NLP engine to identify the intention of the post.

[1556] Example: For a post that says "We've released a new product," identify the intent as "This is an introduction to a new product."

[1557] 3. Emotion recognition:

[1558] The server uses an emotion recognition engine to analyze the emotions contained in the collected comments.

[1559] For example, recognize the "negative sentiment" in response to the comment "This product is too expensive."

[1560] 4. Reply Generation:

[1561] The server generates a reply sentence using a generative AI model based on the identified intention and emotion recognition results.

[1562] Example: In response to a comment like "This product is too expensive," generate a gentle reply such as "We charge that price because we use high-quality materials."

[1563] 5. Quality check:

[1564] A filtering module checks the quality of the generated replies and corrects any misleading or inappropriate phrases.

[1565] 6. Automated Submission:

[1566] The server automatically sends the filtered reply text to the information distribution platform.

[1567] The specific hardware and software used

[1568] Smartphone: A device that allows users to post and comment.

[1569] Cloud server: Uses Google Cloud or AWS.

[1570] NLP engines: Google Cloud Natural Language API, IBM Watson Natural Language Understanding.

[1571] Generative AI model: OpenAI GPT-3 / GPT-4.

[1572] Sentiment recognition engine: Microsoft Azure Text Analytics API.

[1573] Filtering module: Rule-based system (e.g. Python script).

[1574] Specific examples of operations

[1575] For example, suppose a user posts "I've published a new blog post!" on an information distribution platform. Later, another user comments, "This blog post is completely different from the last one." The server collects these posts and comments, analyzes the intention of the post using a generative AI model, and recognizes the emotion of the comment using an emotion recognition engine. If the comment is recognized as containing "confusion" or "negative" emotions, it generates a gentle reply such as, "This is written from a new perspective and offers content that differs from the previous post." This reply is filtered and eventually automatically sent to the information distribution platform.

[1576] Prompt Sentence Examples

[1577] "Generate a reply to the following comment based on the comment intent 'announce a new blog post' and sentiment 'negative': 'This blog post is completely different from the last one.'"

[1578] In this way, the present invention realizes a system that provides quick and appropriate replies to user posts and also responds appropriately to negative comments.

[1579] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1580] Step 1:

[1581] The server collects content posted by users to the information distribution platform. It uses an API to collect data, including metadata such as the post text, posting time, and user account information. The input is the user's post, and the output is the collected posting data.

[1582] Step 2:

[1583] The server analyzes the collected posts using a natural language processing (NLP) engine. Specifically, it performs word segmentation and part-of-speech analysis. The input is the post data collected in step 1, and the output is the analyzed text data.

[1584] Step 3:

[1585] The server passes the analyzed post content to a generative artificial intelligence (AI) to identify the intent. For example, from the sentence "We have released a new product," it extracts the intent "Introducing a new product." The input is the analyzed text data, and the output is the extracted intent.

[1586] Step 4:

[1587] The server analyzes the collected comments using an emotion recognition engine. It classifies the emotions contained in the comments as positive, negative, or neutral. The input is the comment data, and the output is the recognized emotion information.

[1588] Step 5:

[1589] The server uses generative AI to generate a reply based on the identified intent and the analyzed emotional information. For example, in response to a negative comment such as "This product is too expensive," it generates a reply such as "We set the price at that level because we use high-quality materials." The input is intent information and emotional information, and the output is the generated reply.

[1590] Step 6:

[1591] The server passes the generated reply to a filtering module for quality checks. If it contains misleading phrases or inappropriate expressions, it corrects them. The input is the generated reply, and the output is the reply that passes the quality check.

[1592] Step 7:

[1593] The server automatically sends the filtered reply text to the information distribution platform using an API or an interface provided by the information distribution platform. The input is the reply text that has passed the quality check, and the output is the reply posted on the information distribution platform.

[1594] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1595] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1596] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1597] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1598] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1599] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1600] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1601] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1602] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1603] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1604] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1605] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1606] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1607] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1608] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1609] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1610] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1611] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1612] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1613] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1614] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1615] The following is further disclosed regarding the above embodiment.

[1616] (Claim 1)

[1617] A means for collecting content posted by users on a social networking service;

[1618] A method that uses generative artificial intelligence to analyze the collected posts and identify the intention behind the posts,

[1619] A means using generative artificial intelligence to generate an appropriate reply sentence based on the identified intention;

[1620] means for automatically transmitting the generated reply text to a social networking service;

[1621] A system including:

[1622] (Claim 2)

[1623] A means for generating responses to malicious replies and negative comments based on the results of identifying the intentions of the collected post content;

[1624] The system of claim 1 further comprising:

[1625] (Claim 3)

[1626] A filtering method for checking the quality of the generated reply text and correcting it if necessary;

[1627] The system of claim 1 further comprising:

[1628] "Example 1"

[1629] (Claim 1)

[1630] A means for collecting content posted by users on a social networking service;

[1631] A means for transmitting the posted content collected from the terminal to a server;

[1632] A method that uses generative artificial intelligence to analyze the collected posts using natural language processing and identify the intention of the posts;

[1633] A method using generative artificial intelligence to generate appropriate reply sentences based on the intentions of users, including comments from other users;

[1634] a means for filtering and modifying the generated reply text;

[1635] means for automatically sending the filtered reply text on a social networking service;

[1636] A system including:

[1637] (Claim 2)

[1638] A means for generating responses to malicious replies and negative comments based on the results of identifying the intentions of the collected post content;

[1639] The system of claim 1 further comprising:

[1640] (Claim 3)

[1641] A filtering method for checking the quality of the generated reply text and correcting it if necessary;

[1642] The system of claim 1 further comprising:

[1643] "Application Example 1"

[1644] (Claim 1)

[1645] A means for collecting content posted by users on a social networking service;

[1646] A method that uses generative artificial intelligence to analyze the collected posts and identify the intention behind the posts,

[1647] A means using generative artificial intelligence to generate an appropriate reply sentence based on the identified intention;

[1648] means for automatically transmitting the generated reply text to a social networking service;

[1649] A means for analyzing collected advertisement and campaign posts and user comments thereon and generating appropriate reply sentences using generative artificial intelligence;

[1650] A system including:

[1651] (Claim 2)

[1652] A means for generating responses to malicious replies and negative comments based on the results of identifying the intentions of the collected post content;

[1653] A means for analyzing the intent of advertisements and campaigns, generating high-quality replies to user comments based on the analysis, and performing filtering and quality checks;

[1654] 10. The system of claim 1, further comprising:

[1655] (Claim 3)

[1656] A filtering method for checking the quality of the generated reply text and correcting it if necessary;

[1657] means for providing automated filtering and correction functions for the purpose of improving the quality of replies related to advertisements and campaigns;

[1658] The system of claim 1 further comprising:

[1659] "Example 2: Combining Emotion Engines"

[1660] (Claim 1)

[1661] A means for collecting content posted by users on a social networking service;

[1662] A means using a natural language processing engine and generative artificial intelligence to analyze the collected post content and identify the intention of the post;

[1663] a means using an emotion engine to recognize the sentiment of posts and comments;

[1664] A means using generative artificial intelligence to generate an appropriate reply sentence based on the identified intention and the recognized emotion;

[1665] A filtering method for checking the quality of the generated reply text and correcting it if necessary;

[1666] means for automatically sending the filtered reply text on a social networking service;

[1667] A system including:

[1668] (Claim 2)

[1669] A means for generating responses to malicious replies and negative comments based on the collected results of identifying the intention of the post content and the recognized emotions;

[1670] The system of claim 1 further comprising:

[1671] (Claim 3)

[1672] a means for checking the quality of the generated reply using a filtering means and correcting it if necessary before sending the reply;

[1673] The system of claim 1 further comprising:

[1674] "Application example 2 when combining emotion engines"

[1675] (Claim 1)

[1676] A means for collecting content posted by users on the information distribution platform;

[1677] A method that uses generative artificial intelligence to analyze the collected posts and identify the intention behind the posts,

[1678] A means using generative artificial intelligence to generate an appropriate reply sentence based on the identified intention;

[1679] A means for automatically transmitting the generated reply text to an information distribution platform;

[1680] A filtering method for checking the quality of the generated reply text and correcting it if necessary;

[1681] A means for analyzing the collected comments and using an emotion recognition device to recognize emotions contained in the comments;

[1682] means for adjusting the content of a reply based on the identified intent and the recognized emotion;

[1683] A system including:

[1684] (Claim 2)

[1685] A means for generating responses to malicious replies and negative comments based on the collected comment intention identification results and emotion recognition results;

[1686] The system of claim 1 further comprising:

[1687] (Claim 3)

[1688] A filtering method to check the quality of generated replies before they are sent and to correct misleading or inappropriate phrases;

[1689] The system of claim 1 further comprising: [Explanation of symbols]

[1690] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A means for collecting content posted by users on a social networking service; A method that uses generative artificial intelligence to analyze the collected posts and identify the intention behind the posts, A means using generative artificial intelligence to generate an appropriate reply sentence based on the identified intention; means for automatically transmitting the generated reply text to a social networking service; A system including:

2. A means for generating responses to malicious replies and negative comments based on the results of identifying the intentions of the collected post content; The system of claim 1 further comprising:

3. A filtering method for checking the quality of the generated reply text and correcting it if necessary; The system of claim 1 further comprising:

Images