System

Abstract

To provide a system capable of quickly and properly replying to word-of-mouth posted on a network.SOLUTION: A system comprising: user input means; transmission means for transmitting word-of-mouth data input by the user input means to a server; analysis means for analyzing the word-of-mouth data and extracting positive information and negative information; generation means for generating a reply sentence based on the extracted information; and reply transmission means for transmitting the generated reply sentence to a user.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] In recent years, the number of reviews posted online has been increasing, reaching hundreds of millions per year. It is difficult to individually respond to such a large number of reviews, and standard responses make it difficult to create a positive impression. Furthermore, the content of reviews varies widely, and flexible and appropriate responses are required to address this diversity. However, doing this manually is not practical in terms of effort and time. Therefore, a system that can quickly and appropriately respond to reviews is needed. [Means for solving the problem]

[0005] The present invention is a system including a user input means, a transmission means for transmitting review data input by the user input means to a server, an analysis means for analyzing the review data and extracting positive and negative information, a generation means for generating a reply message based on the extracted information, and a reply sending means for sending the generated reply message to the user. This makes it possible to quickly analyze reviews input by users and generate natural and individualized replies using generative AI, thereby automatically providing replies that are appropriate and leave a positive impression on reviews.

[0006] "User input means" refers to an interface or device that allows a user to input a review.

[0007] The "transmission means" refers to a module or process for transmitting word-of-mouth data input by the user input means to the server.

[0008] "Server" refers to a computer system that receives and analyzes review data, and generates and sends replies.

[0009] "Review data" refers to the text information of reviews entered by users.

[0010] "Analysis means" refers to software or algorithms used to analyze word-of-mouth data and extract positive and negative information.

[0011] "Positive information" refers to the positive content contained in the word-of-mouth data.

[0012] "Negative information" refers to negative content contained in word-of-mouth data.

[0013] "Generation means" refers to the algorithm or AI model used to generate a reply message based on the extracted information.

[0014] "Reply sending means" refers to a module or process for sending the generated reply message to the user.

[0015] "Natural language processing technology" refers to technology that enables computers to understand and analyze human language.

[0016] A "machine learning model" refers to an AI algorithm that can perform a specific task by learning from large amounts of data. [Brief explanation of the drawings]

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

[0018] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

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

[0020] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0021] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0022] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0023] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0024] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0025] [First embodiment]

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

[0027] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0028] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0029] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0030] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0031] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0032] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

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

[0034] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0035] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0036] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0037] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0038] The system of the present invention includes a user input means, a transmission means, an analysis means, a generation means, and a reply transmission means. The program of this system is used and implemented as follows.

[0039] The implementation of this system begins when a user inputs a review using a terminal. For example, the user might input, "The food at this restaurant was delicious, but the service was a little poor."

[0040] The input word-of-mouth data is transmitted from the terminal to the server. The transmitting means plays this role and transmits the word-of-mouth data input at the terminal to the server.

[0041] The server analyzes the received review data. The analysis method used here is natural language processing (NLP). Specifically, the review text is tokenized (divided into words and sentences) and then grammatically analyzed to extract positive and negative information. For example, parts such as "The food was delicious" are extracted as positive, and parts such as "The service was a little poor" are extracted as negative.

[0042] Once the analysis is complete, the server uses a generative method to generate a reply. This can be done using machine learning models or generative AI. The generative AI creates a natural and appropriate reply based on the analysis results. For example, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we apologize that our service did not meet your expectations."

[0043] The generated reply message is sent to the user's terminal by the reply sending means. When the reply message is sent from the server to the terminal, the terminal displays it to the user. The user can check the reply message displayed on their own terminal and feel that they have received a natural and polite response.

[0044] In this way, the system can provide quick and natural replies to reviews entered by users. In addition, by using AI, it is possible to respond to individual reviews and automatically provide high-quality replies to many reviews.

[0045] The processing flow will be explained below.

[0046] Step 1:

[0047] A user uses a device to enter a review. For example, the user might enter, "The food at this restaurant was good, but the service was a bit poor."

[0048] Step 2:

[0049] The terminal receives the user's input and transmits the review data to the server. The transmission means plays this role and checks the amount of input and the connection status to confirm the success of the transmission.

[0050] Step 3:

[0051] The server receives the review data sent from the device, temporarily stores the received data, and checks whether the data format is correct.

[0052] Step 4:

[0053] The server uses a natural language processing (NLP) engine to analyze the review text, specifically tokenizing it (dividing it into words and sentences), performing grammatical analysis, and extracting positive and negative information.

[0054] Step 5:

[0055] The server then requests the AI ​​to generate a reply based on the analysis results. For example, it sends data in the form of "Positive: The food was delicious" or "Negative: The service was a bit poor."

[0056] Step 6:

[0057] Based on the information it receives, generative AI generates natural and appropriate responses, such as, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0058] Step 7:

[0059] The server receives the reply text generated by the generative AI and saves it along with the review data.

[0060] Step 8:

[0061] The server sends the saved reply to the device, and then takes steps to verify that the data was sent and received correctly.

[0062] Step 9:

[0063] The reply received by the terminal from the server is displayed to the user in an appropriate format for easy viewing.

[0064] Step 10:

[0065] The user checks the reply displayed on their device and evaluates whether the generated reply feels natural.

[0066] Example 1

[0067] 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."

[0068] Conventional review reply systems have had difficulty in providing fast, natural replies to review content. In particular, responding to many reviews individually requires high manpower and time costs, and the quality of the replies cannot be maintained. Furthermore, technology for appropriately analyzing the sentiment of review content and generating replies based on the results has not been sufficiently developed. To solve these problems, the present invention aims to provide a system that automatically analyzes review content and provides fast, natural replies.

[0069] 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.

[0070] In this invention, the server includes a user input means for a user to input reviews using a terminal, a transmission means for transmitting the review data input by the user input means to the server via a network, an analysis means for receiving the review data and extracting positive and negative information using natural language processing technology, a generation means for generating a reply message using a generative AI model based on the extracted information, a reply transmission means for transmitting the generated reply message to the user's terminal via the network, and a means for displaying the reply message on the user's terminal. This enables automatic, quick, and natural replies to reviews. By analyzing positive and negative information, it is possible to generate appropriate replies that match the user's emotions, thereby providing high-quality service.

[0071] "User input means" refers to the function of a device or software that allows a user to input a review.

[0072] The "transmission means" is a function for transmitting the word-of-mouth data entered by the user to the server via the network.

[0073] The "analysis means" is a technology for processing the received word-of-mouth data and extracting analysis results such as positive and negative information.

[0074] The "generation means" is a function for generating an appropriate reply message based on the analysis results, and mainly uses a generative AI model.

[0075] The "reply sending means" is a function for sending the generated reply message to the user's terminal.

[0076] "Natural language processing technology" is a technology that enables computers to understand and appropriately process human language.

[0077] A "generative AI model" is an algorithm or software that uses machine learning techniques to generate appropriate replies from data.

[0078] A "prompt sentence" is text data that is input into a generative AI model and serves to guide the generation of a reply sentence.

[0079] The system of the present invention allows users to input reviews using a terminal, and automatically generates replies to those reviews, providing a quick and natural response. To implement this system, the following elements are required:

[0080] First, a user inputs a review using a device (such as a smartphone or PC). The user input means plays the role of receiving this review data. Specifically, a review entered by a user such as "The food at this restaurant was delicious, but the service was a bit poor" is included.

[0081] The device then sends the review data to the server via a transmission mechanism that sends the data to the server over the network. The data is sent in JSON format using an HTTP POST request.

[0082] When the server receives the review data, it uses analytical tools to analyze the content of the review. The natural language processing (NLP) technology used here performs tokenization, grammar analysis, and sentiment analysis. For example, it extracts parts such as "The food was delicious" as positive information and parts such as "The service was a bit poor" as negative information.

[0083] Once the analysis is complete, the server generates a reply using a generation means. This generation means uses a generative AI model (e.g., GPT-3). Based on the analysis results, a prompt is constructed and input into the generative AI model to generate an appropriate reply. Examples of prompts are as follows:

[0084] User review: "This cafe had great coffee, but limited seating."

[0085] Positive: "The coffee was great."

[0086] Negative: "There weren't many seats"

[0087] Generate a reply.

[0088] The generated reply is then sent from the server to the user's device. The reply is sent in JSON format as an HTTP response via the reply sending means. The user's device analyzes the received reply and displays it on the screen. This display means allows the user to see the reply quickly and naturally.

[0089] For example, if a user's review said, "This cafe had great coffee, but limited seating," the generated response might look like this:

[0090] "Thank you for visiting us. We're very happy that you enjoyed our coffee. However, we apologize for any inconvenience caused by the limited seating."

[0091] This system automatically provides high-quality replies to reviews, enabling users to respond quickly and appropriately.

[0092] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0093] Step 1:

[0094] A user inputs a review using a terminal. For example, the user inputs, "The food at this restaurant was delicious, but the service was a little poor." This review is processed as text data.

[0095] input:

[0096] User review text

[0097] output:

[0098] Review text data

[0099] Step 2:

[0100] The device sends the review data to the server. Specifically, the review text data is sent to the server using an HTTP POST request. At this time, the data is sent in JSON format.

[0101] input:

[0102] Review text data

[0103] output:

[0104] Review data sent to the server (JSON format)

[0105] Step 3:

[0106] The server receives the review data and analyzes it using an analysis tool. First, the review text is tokenized, and then grammatical and sentiment analysis is performed. For example, "The food was delicious" is extracted as a positive, and "The service was a bit poor" is extracted as a negative.

[0107] input:

[0108] Review data sent to the server (JSON format)

[0109] output:

[0110] Positive and negative information

[0111] Step 4:

[0112] The server constructs a prompt sentence based on the extracted information. Specifically, it generates a prompt sentence that combines the review text, positive information, and negative information.

[0113] input:

[0114] Positive and negative information

[0115] output:

[0116] Prompt statement

[0117] Step 5:

[0118] The server generates a reply using a generation method. The prompt is input to a generative AI model (e.g., GPT-3) to generate a reply. For example, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0119] input:

[0120] Prompt statement

[0121] output:

[0122] Generated reply

[0123] Step 6:

[0124] The server sends the generated reply to the user's device. The reply is sent in JSON format as an HTTP response.

[0125] input:

[0126] Generated reply

[0127] output:

[0128] Reply text sent to the user's device (JSON format)

[0129] Step 7:

[0130] The device analyzes the received reply and displays it to the user. Specifically, the reply is displayed to the user through a mobile application or a web browser. The user can check the displayed reply and feel satisfied.

[0131] input:

[0132] Reply text sent to the user's device (JSON format)

[0133] output:

[0134] Reply text displayed on the device

[0135] (Application example 1)

[0136] 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."

[0137] In traditional customer service at brick-and-mortar stores, it has been difficult to collect and analyze customer reviews and feedback in real time and respond appropriately on the spot. This often delays the improvement of customer satisfaction and service. Furthermore, manually analyzing and responding to large amounts of feedback is inefficient and results in inconsistent quality. To solve these issues, a system is needed that can collect customer feedback in real time and respond quickly and appropriately.

[0138] 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.

[0139] In this invention, the server includes a user input means, a transmission means for converting voice data input by the user input means into text data and transmitting the text data to the server, an analysis means for analyzing the text data to extract positive and negative information, a generation means for generating a reply message based on the extracted information, and a display means for transmitting and displaying the generated reply message to the user. This makes it possible to collect and analyze feedback from customers in real time and respond quickly and appropriately.

[0140] "User input means" refers to a device or interface for a user to input voice data or text data.

[0141] The "transmission means" is a device or interface for converting input voice data into text data and transmitting it to the server.

[0142] The "analysis means" is a device or software that analyzes the transmitted text data using natural language processing technology and extracts positive and negative information.

[0143] "Generation means" refers to a device or software that uses a generative AI model to generate reply messages or prompt messages based on the information extracted by the analysis means.

[0144] The "display means" is a device or interface for displaying the generated reply message to the user.

[0145] "Natural language processing technology" is a technology for understanding and analyzing human language, and is used when analyzing text data.

[0146] A "generative AI model" refers to an artificial intelligence model that uses machine learning and deep learning technologies to generate natural language based on input data.

[0147] A "prompt sentence" is an input sentence that gives instructions to a generative AI model and specifies what kind of output the model should generate.

[0148] This invention relates to a system including a user input means, a transmission means, an analysis means, a generation means, and a display means. The invention particularly relates to a system in which staff in brick-and-mortar stores wear smart glasses and analyze and display voice feedback from customers in real time, thereby enabling prompt and appropriate customer service.

[0149] The user input means is a device or interface for the user to input voice data or text data, specifically, a microphone in the smart glasses is used to input voice feedback.

[0150] The transmitting means is a device or interface for converting input voice data into text data and transmitting it to the server. Voice recognition software is used to convert the voice into text and the text is transmitted to the server via the network.

[0151] The analysis means is a device or software that analyzes the transmitted text data using natural language processing (NLP) technology to extract positive and negative information. This analysis uses natural language processing (NLP) technology to tokenize sentences and perform grammatical analysis.

[0152] The generation means is a device or software that uses a generative AI model to generate replies and prompts based on the information extracted by the analysis means. The generative AI model uses machine learning and deep learning techniques to generate appropriate replies based on the input data.

[0153] The display means is a device or interface for displaying the generated reply to the user. The smart glasses display is used to display the reply to the staff in real time.

[0154] For example, if a customer inputs voice feedback such as "The food was delicious, but the wait time was too long," the microphone in the smart glasses recognizes this voice, and the transmission means converts the voice into text and sends it to the server. On the server, the analysis means analyzes the feedback and extracts positive and negative aspects. The generation means then uses the generative AI model to create an appropriate reply, such as "Thank you for your patronage. We are very pleased that you enjoyed your meal. We apologize for the long wait, but we will strive to improve in the future." Finally, this reply is displayed on the smart glasses' display, allowing the staff to immediately respond appropriately to the customer.

[0155] An example of a prompt is as follows:

[0156] "Please enter your feedback by voice."

[0157] "Parse the feedback text."

[0158] "Performs sentiment analysis and generates replies."

[0159] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0160] Step 1:

[0161] The user provides verbal feedback.

[0162] Input: User's spoken feedback

[0163] How it works: The microphone in the smart glasses picks up sound and records it as audio data.

[0164] Output: Audio data

[0165] Step 2:

[0166] Convert the audio data into text data.

[0167] Input: Audio data

[0168] How it works: The smart glasses' transmitter uses speech recognition software (e.g., the SpeechRecognition library) to convert the voice data into text data.

[0169] Output: Text data

[0170] Step 3:

[0171] Sends text data to the server.

[0172] Input: Text data

[0173] Operation: The transmission means of the smart glasses transmits text data to the server via the network.

[0174] Output: Text data sent to the server

[0175] Step 4:

[0176] The server parses the text data.

[0177] Input: Text data sent to the server

[0178] How it works: The server's analysis means uses natural language processing techniques (e.g., the spaCy library) to analyze the text data and extract positive and negative information.

[0179] Output: Positive and negative information

[0180] Step 5:

[0181] A reply message is generated based on the extracted information.

[0182] Input: Positive and negative information

[0183] How it works: The server's generator uses a generative AI model (e.g., the transformers library) to generate an appropriate reply based on the prompt.

[0184] Output: The generated reply

[0185] Step 6:

[0186] The generated reply is sent to the smart glasses and displayed.

[0187] Input: Generated reply

[0188] Operation: The display means of the server transmits the generated reply message to the smart glasses via the network, and displays the reply message on the display of the smart glasses.

[0189] Output: Reply displayed on staff member's smart glasses

[0190] This allows users (staff) to instantly check the appropriate response to customer feedback and respond quickly.

[0191] 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.

[0192] The system of the present invention includes a user input means, a transmission means, an analysis means, a generation means, a reply transmission means, and an emotion engine. The program of this system is used and implemented as follows.

[0193] First, a user uses a terminal to input a review. For example, the user may input, "The food at this restaurant was delicious, but the service was a little poor."

[0194] The input word-of-mouth data is transmitted from the terminal to the server. The transmitting means plays this role and transmits the word-of-mouth data input at the terminal to the server.

[0195] The server analyzes the received review data. The analysis method used here uses natural language processing (NLP). Specifically, the text is tokenized (divided into words and sentences) and grammatical analysis is performed to extract positive and negative information. For example, parts such as "The food was delicious" are extracted as positive, and parts such as "The service was a little poor" are extracted as negative.

[0196] Furthermore, an emotion engine is used to recognize user emotions contained in the review data. The emotion engine analyzes the text of the review and identifies positive and negative emotions. For example, positive emotions are identified from the "delicious" part and negative emotions are identified from the "bad" part.

[0197] The server uses a generation means to generate a reply based on the information obtained by the analysis means and emotion engine. The generation means uses machine learning models and generative AI to create a reply taking into account the emotion score. For example, a reply such as "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we apologize that our service did not meet your expectations" may be generated.

[0198] The generated reply message is sent to the user's terminal by the reply sending means. When the reply message is sent from the server to the terminal, the terminal displays it to the user. The user can check the reply message displayed on their own terminal and feel that they have received a natural and polite response.

[0199] In this way, the system can provide quick and natural replies to user-submitted reviews. In particular, the use of an emotion engine allows for more personalized responses that reflect the user's emotions, enabling high-quality replies to be automatically generated for many reviews. Specific embodiments of the system can enhance a positive user experience.

[0200] The processing flow will be explained below.

[0201] Step 1:

[0202] A user uses a device to enter a review. For example, the user might enter, "The food at this restaurant was good, but the service was a bit poor."

[0203] Step 2:

[0204] The terminal receives the user's input and transmits the review data to the server. The transmission means plays this role and checks the amount of input and the connection status to confirm the success of the transmission.

[0205] Step 3:

[0206] The server receives the review data sent from the device, temporarily stores the received data, and checks whether the data format is correct.

[0207] Step 4:

[0208] The server uses a natural language processing (NLP) engine to analyze the review text. Specifically, it tokenizes the text (divides it into words and sentences), performs grammatical analysis, and extracts positive and negative information. For example, it extracts "The food was delicious" as a positive and "The service was a bit poor" as a negative.

[0209] Step 5:

[0210] The server uses an analysis means and an emotion engine to recognize emotions contained in the review data. The emotion engine uses a language model to generate an emotion score from the review text. For example, it identifies positive emotions from the "delicious" part and negative emotions from the "bad" part.

[0211] Step 6:

[0212] The server requests the generative AI to generate a reply based on the analysis results and emotion score. For example, it sends data in the form of "Positive: The food was delicious" and "Negative: The service was a little poor", with "Positive emotion score: 80" and "Negative emotion score: 20".

[0213] Step 7:

[0214] Based on the information it receives, generative AI generates natural and appropriate responses, such as, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0215] Step 8:

[0216] The server receives the reply text generated by the generative AI and saves it along with the review data.

[0217] Step 9:

[0218] The server sends the saved reply to the device, and then takes steps to verify that the data was sent and received correctly.

[0219] Step 10:

[0220] The reply received by the terminal from the server is displayed to the user in an appropriate format for easy viewing.

[0221] Step 11:

[0222] The user checks the reply displayed on their device and evaluates whether the generated reply feels natural.

[0223] Example 2

[0224] 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."

[0225] Conventional review response systems have had difficulty automatically generating quick and natural replies to reviews entered by users. They also struggled to generate replies that accurately reflected the user's feelings, resulting in a problem of only being able to provide uniform replies. This resulted in a poor user experience and made it difficult to achieve satisfaction.

[0226] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: means for a user to input a review using a terminal; means for transmitting review data from the terminal to the server; means for the server to receive the review data and perform data preprocessing; means for analyzing the review data and extracting positive information and negative information; means for identifying emotions using an emotion engine based on the extracted information; means for generating a reply message using a generative AI model taking into account the identified emotions; means for transmitting the generated reply message to the terminal; and means for the terminal to display the reply message to the user. This makes it possible to automatically generate a high-quality reply message that is quick, natural, and reflects emotions in response to a review input by a user, thereby improving user satisfaction.

[0227] "User" means an individual or organization that uses the system to enter reviews.

[0228] A "terminal" is a device used by a user, such as a smartphone, tablet, or computer.

[0229] "Word-of-mouth data" refers to text data of reviews and comments that users input and submit via their terminals.

[0230] The "server" is a back-end system that receives and analyzes word-of-mouth data and generates replies.

[0231] "Preprocessing" refers to processes such as normalization and cleaning that are performed before analyzing data.

[0232] The "analysis means" is a means for analyzing word-of-mouth data and extracting positive and negative information.

[0233] "Positive information" is information that indicates positive content or evaluations within the word-of-mouth data.

[0234] "Negative information" is information that indicates negative content or evaluations within the word-of-mouth data.

[0235] An "emotion engine" is software for identifying and classifying user emotions within review data.

[0236] A "generative AI model" is an artificial intelligence model that generates natural-looking sentences based on input data and conditions.

[0237] A "reply" is a reply text that the server generates based on word-of-mouth data and sends to the user.

[0238] The system of the present invention includes a user input means, a sending means, an analyzing means, a generating means, a reply sending means, and an emotion engine. This system operates in the following procedure.

[0239] First, a user uses a device to enter a review. For example, a user might write, "The curry at this restaurant was excellent, but the waiter's service was disappointing." The user enters this information through a review app using a device such as a smartphone or computer.

[0240] Next, the entered review data is sent from the device to the server. The device sends the review data entered by the user to the server as an HTTP request. At this time, the data is sent in JSON format, and HTTPS is used as the communication protocol. For example, the sent data has the following format:

[0241] POST / api / reviews

[0242] {

[0243] "review": "The curry at this restaurant was great, but the waiter's service was disappointing."

[0244] }

[0245] The server analyzes the received review data. This analysis uses natural language processing (NLP) technology. Specifically, for example, Google's BERT model is used to tokenize the sentences and analyze the meaning of each token. Next, positive and negative information is extracted. For example, "The curry was excellent" is extracted as positive information, and "The waiter's service was disappointing" is extracted as negative information.

[0246] The server then uses an emotion engine to identify user emotions contained in the review data. The emotion engine uses models such as the Hugging Face emotion analysis model. This allows positive and negative emotions to be identified in the review data. Specifically, "It was great" is classified as a positive emotion, and "It was disappointing" is classified as a negative emotion.

[0247] Next, the server generates a reply using a generative AI model, such as OpenAI's GPT-3, based on the analysis results and sentiment score. To take into account the content and sentiment of the review during this generation process, the prompt text is specified as follows:

[0248] Prompt: "A user wrote a review saying, 'The curry at this restaurant was great, but the waiter's service was disappointing.' Generate a response that reflects that sentiment."

[0249] The generated reply has specific content such as, "Thank you for visiting us. We are glad that you enjoyed our curry. We apologize that our waiter's service did not meet your expectations."

[0250] Finally, the server sends the generated reply to the terminal. The generated reply is sent to the terminal again in JSON format, and the terminal displays it to the user. The user can check the reply and feel that the response was natural and polite.

[0251] In this way, the system can provide quick and natural replies to reviews entered by users. In particular, by using an emotion engine and generative AI model, it is possible to provide personalized responses that reflect the user's emotions, and to automatically generate high-quality replies to a large number of reviews. Specific implementations of the system can provide a positive user experience and improve customer satisfaction.

[0252] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0253] Step 1:

[0254] The user uses the device to input a review. For example, the user might input, "The curry at this restaurant was excellent, but the waiter's service was disappointing." The review entered by the user is acquired as input data. The user then presses the send button, and the review data is sent.

[0255] Step 2:

[0256] The terminal sends the entered review data to the server. The terminal sends the review data entered by the user to the server as an HTTP request. Specifically, the review data is converted into JSON format and sent to the server via the HTTPS protocol. The input data is passed from the client to the server.

[0257] Step 3:

[0258] The server receives the review data and performs the necessary preprocessing. The server deserializes the received JSON format review data and extracts the text data. Next, preprocessing such as normalization and removal of unnecessary whitespace and special characters is performed. The input data is unprocessed text data, and the output data is preprocessed text data.

[0259] Step 4:

[0260] The server analyzes the review data and extracts positive and negative information. The server uses a natural language processing (NLP) model to tokenize and analyze grammar. It then analyzes using the BERT model to extract positive information (e.g., "The curry was excellent") and negative information (e.g., "The waiter's service was disappointing"). Through this process, the input data is preprocessed text data, and the output data is the extracted information.

[0261] Step 5:

[0262] The server uses an emotion engine to identify emotions. The emotion engine analyzes the text data and calculates an emotion score. Using Hugging Face's emotion analysis model, it classifies the emotion of each part. For example, the part "It was great" is identified as positive, and the part "It was disappointing" is identified as negative. The input data is the analyzed text data, and the output data is the emotion score and emotion classification results.

[0263] Step 6:

[0264] The server generates a reply using a generative AI model. Based on the analyzed emotion data and prompt, a reply is generated using generative AI (for example, GPT-3). An example of a prompt is input to the model in the form: "A user posted a review saying, 'The curry at this restaurant was excellent, but the waiter's service was disappointing.' Please generate a reply that reflects the emotion in response to this review." The output is a natural reply that reflects the emotion.

[0265] Step 7:

[0266] The server sends the generated reply to the terminal. The server converts the generated reply back to JSON format and sends it to the terminal as an HTTP response. The output data is the generated reply, which is passed to the client.

[0267] Step 8:

[0268] The terminal displays the reply to the user. The terminal analyzes the reply received from the server and displays it on the screen. The user can check the reply on their own terminal and feel that they have received a natural and polite response. The input data is the reply from the server, and the output data is the display to the user.

[0269] (Application example 2)

[0270] 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."

[0271] For modern online shopping sites, responding appropriately and promptly to user reviews is an important factor in improving customer satisfaction. However, manually responding to many reviews takes time and effort, making it difficult to manage efficiently. Furthermore, conventional automated response systems have the problem of not being able to capture user sentiment effectively, resulting in boilerplate replies that give users an impersonal impression.

[0272] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a user input means, a sending means, an analysis means, an emotion engine, a generation means, and a reply sending means. This makes it possible to appropriately analyze the emotions in the user's review and generate a reply message that can be individually tailored.

[0273] The "user input means" refers to a device and method for a user to input word-of-mouth data.

[0274] The "transmission means" refers to a device and method for transmitting word-of-mouth data entered by a user to a server.

[0275] The "analysis means" refers to a device and method that analyzes the word-of-mouth data sent to the server and extracts positive and negative information.

[0276] A "sentiment engine" is an apparatus and method for identifying sentiment based on analyzed word-of-mouth data.

[0277] The "generator" is a device and method for generating a reply based on the analyzed information and sentiment score.

[0278] The "reply sending means" refers to a device and method for sending the generated reply to the user.

[0279] "Natural language processing technology" is a technology that understands the structure and meaning of language used to analyze word-of-mouth data.

[0280] A "generative AI model" is a model used to generate replies using machine learning and artificial intelligence techniques.

[0281] The "emotion score" is a numerical representation of the positive and negative emotions contained in the review data.

[0282] A "prompt sentence" is an input sentence used to derive a generated reply sentence.

[0283] The present invention provides a system for automatically and appropriately and quickly responding to user reviews on an online shopping site. A specific embodiment of this system will be described below.

[0284] First, the user inputs review data using a smartphone application. This review data is then sent to the server via the user input means. The sending means is realized using a smartphone and AWS API Gateway.

[0285] Once the review data reaches the server, it is processed by an analytical tool that uses natural language processing technology (using TensorFlow and spaCy) to tokenize the review data and extract positive and negative information.

[0286] Next, the sentiment engine identifies the sentiment of the parsed review data. The sentiment engine uses the BERT model to identify positive and negative sentiment from review text.

[0287] Based on the analysis results and the emotion score, a generator generates a reply message. The generator uses GPT-3 (OpenAI's generative AI model) and creates a reply message using a prompt that takes the emotion score into account.

[0288] The generated reply is sent to the user's smartphone by a reply sending means, which uses AWS SNS (Simple Notification Service).

[0289] The operating procedure and data flow of this system will be explained using a concrete example. Suppose a user enters a review in the app saying, "This product was very good, but delivery was too slow." The entered review is sent to the server, and the analysis means extracts "very good" as positive information and "delivery was too slow" as negative information. The emotion engine then calculates an emotion score, and the generation means generates a reply message based on the following prompt message.

[0290] plaintext

[0291] User review: "This product was very good, but delivery was too slow."

[0292] Positive: "This product was very good"

[0293] Negative: "Delivery was too slow"

[0294] Sentiment score: Positive=0.7, Negative=0.3

[0295] Using the information above, generate a natural response like this:

[0296] "Thank you for your purchase. We are very happy that you were satisfied with the product. We apologize for the delay in delivery. We will strive to improve in the future."

[0297] In this way, users can check natural and polite replies on their smartphones. This system will improve the efficiency of responding to user reviews and increase customer satisfaction.

[0298] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0299] Step 1:

[0300] A user inputs a review using a smartphone application. After the user inputs the review, the device receives the review data as input data. This input data includes review information in text format.

[0301] Step 2:

[0302] The terminal sends the entered review data to the server. This process uses AWS API Gateway, which sends the entered text data in a format that transfers it to the API. The input is the text of the user review, and the output is the text data transferred to the server.

[0303] Step 3:

[0304] The server analyzes the received review data. It uses an NLP library (e.g., TensorFlow or spaCy) to tokenize the review data and perform grammatical analysis. The input is the text data received by the server, and the output is the tokenized data from which positive and negative information has been extracted.

[0305] Step 4:

[0306] The server passes the parsed data to a sentiment engine, which uses the BERT model to calculate sentiment scores from the parsed data. The input is the tokenized and parsed data, and the output is positive and negative sentiment scores.

[0307] Step 5:

[0308] The server generates a reply using a generation method based on the emotion score obtained by the emotion engine. The generation method uses a generative AI model such as GPT-3. The emotion score and extracted information are input into the generative AI model as a prompt to generate a natural reply. The input is the prompt and emotion score, and the output is the reply.

[0309] Step 6:

[0310] The server passes the generated reply to the reply sending means and sends it to the user's smartphone. AWS SNS (Simple Notification Service) is used as the reply sending means to send a notification to the user's device. The input is the generated reply, and the output is the reply message displayed on the user's smartphone.

[0311] In this way, a series of processes from when the user inputs a review to when the generated reply message is displayed is realized.

[0312] 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.

[0313] 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.

[0314] 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.

[0315] [Second embodiment]

[0316] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0317] 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.

[0318] 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).

[0319] 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.

[0320] 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.

[0321] 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).

[0322] 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.

[0323] 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.

[0324] 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.

[0325] 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.

[0326] 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.

[0327] 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."

[0328] The system of the present invention includes a user input means, a transmission means, an analysis means, a generation means, and a reply transmission means. The program of this system is used and implemented as follows.

[0329] The implementation of this system begins when a user inputs a review using a terminal. For example, the user might input, "The food at this restaurant was delicious, but the service was a little poor."

[0330] The input word-of-mouth data is transmitted from the terminal to the server. The transmitting means plays this role and transmits the word-of-mouth data input at the terminal to the server.

[0331] The server analyzes the received review data. The analysis method used here is natural language processing (NLP). Specifically, the review text is tokenized (divided into words and sentences) and then grammatically analyzed to extract positive and negative information. For example, parts such as "The food was delicious" are extracted as positive, and parts such as "The service was a little poor" are extracted as negative.

[0332] Once the analysis is complete, the server uses a generative method to generate a reply. This can be done using machine learning models or generative AI. The generative AI creates a natural and appropriate reply based on the analysis results. For example, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we apologize that our service did not meet your expectations."

[0333] The generated reply message is sent to the user's terminal by the reply sending means. When the reply message is sent from the server to the terminal, the terminal displays it to the user. The user can check the reply message displayed on their own terminal and feel that they have received a natural and polite response.

[0334] In this way, the system can provide quick and natural replies to reviews entered by users. In addition, by using AI, it is possible to respond to individual reviews and automatically provide high-quality replies to many reviews.

[0335] The processing flow will be explained below.

[0336] Step 1:

[0337] A user uses a device to enter a review. For example, the user might enter, "The food at this restaurant was good, but the service was a bit poor."

[0338] Step 2:

[0339] The terminal receives the user's input and transmits the review data to the server. The transmission means plays this role and checks the amount of input and the connection status to confirm the success of the transmission.

[0340] Step 3:

[0341] The server receives the review data sent from the device, temporarily stores the received data, and checks whether the data format is correct.

[0342] Step 4:

[0343] The server uses a natural language processing (NLP) engine to analyze the review text, specifically tokenizing it (dividing it into words and sentences), performing grammatical analysis, and extracting positive and negative information.

[0344] Step 5:

[0345] The server then requests the AI ​​to generate a reply based on the analysis results. For example, it sends data in the form of "Positive: The food was delicious" or "Negative: The service was a bit poor."

[0346] Step 6:

[0347] Based on the information it receives, generative AI generates natural and appropriate responses, such as, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0348] Step 7:

[0349] The server receives the reply text generated by the generative AI and saves it along with the review data.

[0350] Step 8:

[0351] The server sends the saved reply to the device, and then takes steps to verify that the data was sent and received correctly.

[0352] Step 9:

[0353] The reply received by the terminal from the server is displayed to the user in an appropriate format for easy viewing.

[0354] Step 10:

[0355] The user checks the reply displayed on their device and evaluates whether the generated reply feels natural.

[0356] Example 1

[0357] 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."

[0358] Conventional review reply systems have had difficulty in providing fast, natural replies to review content. In particular, responding to many reviews individually requires high manpower and time costs, and the quality of the replies cannot be maintained. Furthermore, technology for appropriately analyzing the sentiment of review content and generating replies based on the results has not been sufficiently developed. To solve these problems, the present invention aims to provide a system that automatically analyzes review content and provides fast, natural replies.

[0359] 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.

[0360] In this invention, the server includes a user input means for a user to input reviews using a terminal, a transmission means for transmitting the review data input by the user input means to the server via a network, an analysis means for receiving the review data and extracting positive and negative information using natural language processing technology, a generation means for generating a reply message using a generative AI model based on the extracted information, a reply transmission means for transmitting the generated reply message to the user's terminal via the network, and a means for displaying the reply message on the user's terminal. This enables automatic, quick, and natural replies to reviews. By analyzing positive and negative information, it is possible to generate appropriate replies that match the user's emotions, thereby providing high-quality service.

[0361] "User input means" refers to the function of a device or software that allows a user to input a review.

[0362] The "transmission means" is a function for transmitting the word-of-mouth data entered by the user to the server via the network.

[0363] The "analysis means" is a technology for processing the received word-of-mouth data and extracting analysis results such as positive and negative information.

[0364] The "generation means" is a function for generating an appropriate reply message based on the analysis results, and mainly uses a generative AI model.

[0365] The "reply sending means" is a function for sending the generated reply message to the user's terminal.

[0366] "Natural language processing technology" is a technology that enables computers to understand and appropriately process human language.

[0367] A "generative AI model" is an algorithm or software that uses machine learning techniques to generate appropriate replies from data.

[0368] A "prompt sentence" is text data that is input into a generative AI model and serves to guide the generation of a reply sentence.

[0369] The system of the present invention allows users to input reviews using a terminal, and automatically generates replies to those reviews, providing a quick and natural response. To implement this system, the following elements are required:

[0370] First, a user inputs a review using a device (such as a smartphone or PC). The user input means plays the role of receiving this review data. Specifically, a review entered by a user such as "The food at this restaurant was delicious, but the service was a bit poor" is included.

[0371] The device then sends the review data to the server via a transmission mechanism that sends the data to the server over the network. The data is sent in JSON format using an HTTP POST request.

[0372] When the server receives the review data, it uses analytical tools to analyze the content of the review. The natural language processing (NLP) technology used here performs tokenization, grammar analysis, and sentiment analysis. For example, it extracts parts such as "The food was delicious" as positive information and parts such as "The service was a bit poor" as negative information.

[0373] Once the analysis is complete, the server generates a reply using a generation means. This generation means uses a generative AI model (e.g., GPT-3). Based on the analysis results, a prompt is constructed and input into the generative AI model to generate an appropriate reply. Examples of prompts are as follows:

[0374] User review: "This cafe had great coffee, but limited seating."

[0375] Positive: "The coffee was great."

[0376] Negative: "There weren't many seats"

[0377] Generate a reply.

[0378] The generated reply is then sent from the server to the user's device. The reply is sent in JSON format as an HTTP response via the reply sending means. The user's device analyzes the received reply and displays it on the screen. This display means allows the user to see the reply quickly and naturally.

[0379] For example, if a user's review said, "This cafe had great coffee, but limited seating," the generated response might look like this:

[0380] "Thank you for visiting us. We're very happy that you enjoyed our coffee. However, we apologize for any inconvenience caused by the limited seating."

[0381] This system automatically provides high-quality replies to reviews, enabling users to respond quickly and appropriately.

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

[0383] Step 1:

[0384] A user inputs a review using a terminal. For example, the user inputs, "The food at this restaurant was delicious, but the service was a little poor." This review is processed as text data.

[0385] input:

[0386] User review text

[0387] output:

[0388] Review text data

[0389] Step 2:

[0390] The device sends the review data to the server. Specifically, the review text data is sent to the server using an HTTP POST request. At this time, the data is sent in JSON format.

[0391] input:

[0392] Review text data

[0393] output:

[0394] Review data sent to the server (JSON format)

[0395] Step 3:

[0396] The server receives the review data and analyzes it using an analysis tool. First, the review text is tokenized, and then grammatical and sentiment analysis is performed. For example, "The food was delicious" is extracted as a positive, and "The service was a bit poor" is extracted as a negative.

[0397] input:

[0398] Review data sent to the server (JSON format)

[0399] output:

[0400] Positive and negative information

[0401] Step 4:

[0402] The server constructs a prompt sentence based on the extracted information. Specifically, it generates a prompt sentence that combines the review text, positive information, and negative information.

[0403] input:

[0404] Positive and negative information

[0405] output:

[0406] Prompt statement

[0407] Step 5:

[0408] The server generates a reply using a generation method. The prompt is input to a generative AI model (e.g., GPT-3) to generate a reply. For example, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0409] input:

[0410] Prompt statement

[0411] output:

[0412] Generated reply

[0413] Step 6:

[0414] The server sends the generated reply to the user's device. The reply is sent in JSON format as an HTTP response.

[0415] input:

[0416] Generated reply

[0417] output:

[0418] Reply text sent to the user's device (JSON format)

[0419] Step 7:

[0420] The device analyzes the received reply and displays it to the user. Specifically, the reply is displayed to the user through a mobile application or a web browser. The user can check the displayed reply and feel satisfied.

[0421] input:

[0422] Reply text sent to the user's device (JSON format)

[0423] output:

[0424] Reply text displayed on the device

[0425] (Application example 1)

[0426] 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."

[0427] In traditional customer service at brick-and-mortar stores, it has been difficult to collect and analyze customer reviews and feedback in real time and respond appropriately on the spot. This often delays the improvement of customer satisfaction and service. Furthermore, manually analyzing and responding to large amounts of feedback is inefficient and results in inconsistent quality. To solve these issues, a system is needed that can collect customer feedback in real time and respond quickly and appropriately.

[0428] 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.

[0429] In this invention, the server includes a user input means, a transmission means for converting voice data input by the user input means into text data and transmitting the text data to the server, an analysis means for analyzing the text data to extract positive and negative information, a generation means for generating a reply message based on the extracted information, and a display means for transmitting and displaying the generated reply message to the user. This makes it possible to collect and analyze feedback from customers in real time and respond quickly and appropriately.

[0430] "User input means" refers to a device or interface for a user to input voice data or text data.

[0431] The "transmission means" is a device or interface for converting input voice data into text data and transmitting it to the server.

[0432] The "analysis means" is a device or software that analyzes the transmitted text data using natural language processing technology and extracts positive and negative information.

[0433] "Generation means" refers to a device or software that uses a generative AI model to generate reply messages or prompt messages based on the information extracted by the analysis means.

[0434] The "display means" is a device or interface for displaying the generated reply message to the user.

[0435] "Natural language processing technology" is a technology for understanding and analyzing human language, and is used when analyzing text data.

[0436] A "generative AI model" refers to an artificial intelligence model that uses machine learning and deep learning technologies to generate natural language based on input data.

[0437] A "prompt sentence" is an input sentence that gives instructions to a generative AI model and specifies what kind of output the model should generate.

[0438] This invention relates to a system including a user input means, a transmission means, an analysis means, a generation means, and a display means. The invention particularly relates to a system in which staff in brick-and-mortar stores wear smart glasses and analyze and display voice feedback from customers in real time, thereby enabling prompt and appropriate customer service.

[0439] The user input means is a device or interface for the user to input voice data or text data, specifically, a microphone in the smart glasses is used to input voice feedback.

[0440] The transmitting means is a device or interface for converting input voice data into text data and transmitting it to the server. Voice recognition software is used to convert the voice into text and the text is transmitted to the server via the network.

[0441] The analysis means is a device or software that analyzes the transmitted text data using natural language processing (NLP) technology to extract positive and negative information. This analysis uses natural language processing (NLP) technology to tokenize sentences and perform grammatical analysis.

[0442] The generation means is a device or software that uses a generative AI model to generate replies and prompts based on the information extracted by the analysis means. The generative AI model uses machine learning and deep learning techniques to generate appropriate replies based on the input data.

[0443] The display means is a device or interface for displaying the generated reply to the user. The smart glasses display is used to display the reply to the staff in real time.

[0444] For example, if a customer inputs voice feedback such as "The food was delicious, but the wait time was too long," the microphone in the smart glasses recognizes this voice, and the transmission means converts the voice into text and sends it to the server. On the server, the analysis means analyzes this feedback and extracts positive and negative aspects. The generation means then uses the generative AI model to create an appropriate reply, such as "Thank you for your patronage. We are very pleased that you enjoyed your meal. We apologize for the long wait, but we will strive to improve in the future." Finally, this reply is displayed on the smart glasses' display, allowing the staff to immediately respond appropriately to the customer.

[0445] An example of a prompt is as follows:

[0446] "Please enter your feedback by voice."

[0447] "Parse the feedback text."

[0448] "Performs sentiment analysis and generates replies."

[0449] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0450] Step 1:

[0451] The user provides verbal feedback.

[0452] Input: User's spoken feedback

[0453] How it works: The microphone in the smart glasses picks up sound and records it as audio data.

[0454] Output: Audio data

[0455] Step 2:

[0456] Convert the audio data into text data.

[0457] Input: Audio data

[0458] How it works: The smart glasses' transmitter uses speech recognition software (e.g., the SpeechRecognition library) to convert the voice data into text data.

[0459] Output: Text data

[0460] Step 3:

[0461] Sends text data to the server.

[0462] Input: Text data

[0463] Operation: The transmission means of the smart glasses transmits text data to the server via the network.

[0464] Output: Text data sent to the server

[0465] Step 4:

[0466] The server parses the text data.

[0467] Input: Text data sent to the server

[0468] How it works: The server's analysis means uses natural language processing techniques (e.g., the spaCy library) to analyze the text data and extract positive and negative information.

[0469] Output: Positive and negative information

[0470] Step 5:

[0471] A reply message is generated based on the extracted information.

[0472] Input: Positive and negative information

[0473] How it works: The server's generator uses a generative AI model (e.g., the transformers library) to generate an appropriate reply based on the prompt.

[0474] Output: The generated reply

[0475] Step 6:

[0476] The generated reply is sent to the smart glasses and displayed.

[0477] Input: Generated reply

[0478] Operation: The display means of the server transmits the generated reply message to the smart glasses via the network, and displays the reply message on the display of the smart glasses.

[0479] Output: Reply displayed on staff member's smart glasses

[0480] This allows users (staff) to instantly check the appropriate response to customer feedback and respond quickly.

[0481] 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.

[0482] The system of the present invention includes a user input means, a transmission means, an analysis means, a generation means, a reply transmission means, and an emotion engine. The program of this system is used and implemented as follows.

[0483] First, a user uses a terminal to input a review. For example, the user may input, "The food at this restaurant was delicious, but the service was a little poor."

[0484] The input word-of-mouth data is transmitted from the terminal to the server. The transmitting means plays this role and transmits the word-of-mouth data input at the terminal to the server.

[0485] The server analyzes the received review data. The analysis method used here uses natural language processing (NLP). Specifically, the text is tokenized (divided into words and sentences) and grammatical analysis is performed to extract positive and negative information. For example, parts such as "The food was delicious" are extracted as positive, and parts such as "The service was a little poor" are extracted as negative.

[0486] Furthermore, an emotion engine is used to recognize user emotions contained in the review data. The emotion engine analyzes the text of the review and identifies positive and negative emotions. For example, positive emotions are identified from the "delicious" part and negative emotions are identified from the "bad" part.

[0487] The server uses a generation means to generate a reply based on the information obtained by the analysis means and emotion engine. The generation means uses machine learning models and generative AI to create a reply taking into account the emotion score. For example, a reply such as "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we apologize that our service did not meet your expectations" may be generated.

[0488] The generated reply message is sent to the user's terminal by the reply sending means. When the reply message is sent from the server to the terminal, the terminal displays it to the user. The user can check the reply message displayed on their own terminal and feel that they have received a natural and polite response.

[0489] In this way, the system can provide quick and natural replies to user-submitted reviews. In particular, the use of an emotion engine allows for more personalized responses that reflect the user's emotions, enabling high-quality replies to be automatically generated for many reviews. Specific embodiments of the system can enhance a positive user experience.

[0490] The processing flow will be explained below.

[0491] Step 1:

[0492] A user uses a device to enter a review. For example, the user might enter, "The food at this restaurant was good, but the service was a bit poor."

[0493] Step 2:

[0494] The terminal receives the user's input and transmits the review data to the server. The transmission means plays this role and checks the amount of input and the connection status to confirm the success of the transmission.

[0495] Step 3:

[0496] The server receives the review data sent from the device, temporarily stores the received data, and checks whether the data format is correct.

[0497] Step 4:

[0498] The server uses a natural language processing (NLP) engine to analyze the review text. Specifically, it tokenizes the text (divides it into words and sentences), performs grammatical analysis, and extracts positive and negative information. For example, it extracts "The food was delicious" as a positive and "The service was a bit poor" as a negative.

[0499] Step 5:

[0500] The server uses an analysis means and an emotion engine to recognize emotions contained in the review data. The emotion engine uses a language model to generate an emotion score from the review text. For example, it identifies positive emotions from the "delicious" part and negative emotions from the "bad" part.

[0501] Step 6:

[0502] The server requests the generative AI to generate a reply based on the analysis results and emotion score. For example, it sends data in the form of "Positive: The food was delicious" and "Negative: The service was a little poor", with "Positive emotion score: 80" and "Negative emotion score: 20".

[0503] Step 7:

[0504] Based on the information it receives, generative AI generates natural and appropriate responses, such as, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0505] Step 8:

[0506] The server receives the reply text generated by the generative AI and saves it along with the review data.

[0507] Step 9:

[0508] The server sends the saved reply to the device, and then takes steps to verify that the data was sent and received correctly.

[0509] Step 10:

[0510] The reply received by the terminal from the server is displayed to the user in an appropriate format for easy viewing.

[0511] Step 11:

[0512] The user checks the reply displayed on their device and evaluates whether the generated reply feels natural.

[0513] Example 2

[0514] 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."

[0515] Conventional review response systems have had difficulty automatically generating quick and natural replies to reviews entered by users. They also struggled to generate replies that accurately reflected the user's feelings, resulting in a problem of only being able to provide uniform replies. This resulted in a poor user experience and made it difficult to achieve satisfaction.

[0516] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: means for a user to input a review using a terminal; means for transmitting review data from the terminal to the server; means for the server to receive the review data and perform data preprocessing; means for analyzing the review data and extracting positive information and negative information; means for identifying emotions using an emotion engine based on the extracted information; means for generating a reply message using a generative AI model taking into account the identified emotions; means for transmitting the generated reply message to the terminal; and means for the terminal to display the reply message to the user. This makes it possible to automatically generate a high-quality reply message that is quick, natural, and reflects emotions in response to a review input by a user, thereby improving user satisfaction.

[0517] "User" means an individual or organization that uses the system to enter reviews.

[0518] A "terminal" is a device used by a user, such as a smartphone, tablet, or computer.

[0519] "Word-of-mouth data" refers to text data of reviews and comments that users input and submit via their terminals.

[0520] The "server" is a back-end system that receives and analyzes word-of-mouth data and generates replies.

[0521] "Preprocessing" refers to processes such as normalization and cleaning that are performed before analyzing data.

[0522] The "analysis means" is a means for analyzing word-of-mouth data and extracting positive and negative information.

[0523] "Positive information" is information that indicates positive content or evaluations within the word-of-mouth data.

[0524] "Negative information" is information that indicates negative content or evaluations within the word-of-mouth data.

[0525] An "emotion engine" is software for identifying and classifying user emotions within review data.

[0526] A "generative AI model" is an artificial intelligence model that generates natural-looking sentences based on input data and conditions.

[0527] A "reply" is a reply text that the server generates based on word-of-mouth data and sends to the user.

[0528] The system of the present invention includes a user input means, a sending means, an analyzing means, a generating means, a reply sending means, and an emotion engine. This system operates in the following procedure.

[0529] First, a user uses a device to enter a review. For example, a user might write, "The curry at this restaurant was excellent, but the waiter's service was disappointing." The user enters this information through a review app using a device such as a smartphone or computer.

[0530] Next, the entered review data is sent from the device to the server. The device sends the review data entered by the user to the server as an HTTP request. At this time, the data is sent in JSON format, and HTTPS is used as the communication protocol. For example, the sent data has the following format:

[0531] POST / api / reviews

[0532] {

[0533] "review": "The curry at this restaurant was great, but the waiter's service was disappointing."

[0534] }

[0535] The server analyzes the received review data. This analysis uses natural language processing (NLP) technology. Specifically, for example, Google's BERT model is used to tokenize the sentences and analyze the meaning of each token. Next, positive and negative information is extracted. For example, "The curry was excellent" is extracted as positive information, and "The waiter's service was disappointing" is extracted as negative information.

[0536] The server then uses an emotion engine to identify user emotions contained in the review data. The emotion engine uses models such as the Hugging Face emotion analysis model. This allows positive and negative emotions to be identified in the review data. Specifically, "It was great" is classified as a positive emotion, and "It was disappointing" is classified as a negative emotion.

[0537] Next, the server generates a reply using a generative AI model, such as OpenAI's GPT-3, based on the analysis results and sentiment score. To take into account the content and sentiment of the review during this generation process, the prompt text is specified as follows:

[0538] Prompt: "A user wrote a review saying, 'The curry at this restaurant was great, but the waiter's service was disappointing.' Generate a response that reflects that sentiment."

[0539] The generated reply has specific content such as, "Thank you for visiting us. We are glad that you enjoyed our curry. We apologize that our waiter's service did not meet your expectations."

[0540] Finally, the server sends the generated reply to the terminal. The generated reply is sent to the terminal again in JSON format, and the terminal displays it to the user. The user can check the reply and feel that the response was natural and polite.

[0541] In this way, the system can provide quick and natural replies to reviews entered by users. In particular, by using an emotion engine and generative AI model, it is possible to provide personalized responses that reflect the user's emotions, and to automatically generate high-quality replies to a large number of reviews. Specific implementations of the system can provide a positive user experience and improve customer satisfaction.

[0542] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0543] Step 1:

[0544] The user uses the device to input a review. For example, the user might input, "The curry at this restaurant was excellent, but the waiter's service was disappointing." The review entered by the user is acquired as input data. The user then presses the send button, and the review data is sent.

[0545] Step 2:

[0546] The terminal sends the entered review data to the server. The terminal sends the review data entered by the user to the server as an HTTP request. Specifically, the review data is converted into JSON format and sent to the server via the HTTPS protocol. The input data is passed from the client to the server.

[0547] Step 3:

[0548] The server receives the review data and performs the necessary preprocessing. The server deserializes the received JSON format review data and extracts the text data. Next, preprocessing such as normalization and removal of unnecessary whitespace and special characters is performed. The input data is unprocessed text data, and the output data is preprocessed text data.

[0549] Step 4:

[0550] The server analyzes the review data and extracts positive and negative information. The server uses a natural language processing (NLP) model to tokenize and analyze grammar. It then analyzes using the BERT model to extract positive information (e.g., "The curry was excellent") and negative information (e.g., "The waiter's service was disappointing"). Through this process, the input data is preprocessed text data, and the output data is the extracted information.

[0551] Step 5:

[0552] The server uses an emotion engine to identify emotions. The emotion engine analyzes the text data and calculates an emotion score. Using Hugging Face's emotion analysis model, it classifies the emotion of each part. For example, the part "It was great" is identified as positive, and the part "It was disappointing" is identified as negative. The input data is the analyzed text data, and the output data is the emotion score and emotion classification results.

[0553] Step 6:

[0554] The server generates a reply using a generative AI model. Based on the analyzed emotion data and prompt, a reply is generated using generative AI (for example, GPT-3). An example of a prompt is input to the model in the form: "A user posted a review saying, 'The curry at this restaurant was excellent, but the waiter's service was disappointing.' Please generate a reply that reflects the emotion in response to this review." The output is a natural reply that reflects the emotion.

[0555] Step 7:

[0556] The server sends the generated reply to the terminal. The server converts the generated reply back to JSON format and sends it to the terminal as an HTTP response. The output data is the generated reply, which is passed to the client.

[0557] Step 8:

[0558] The terminal displays the reply to the user. The terminal analyzes the reply received from the server and displays it on the screen. The user can check the reply on their own terminal and feel that they have received a natural and polite response. The input data is the reply from the server, and the output data is the display to the user.

[0559] (Application example 2)

[0560] 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."

[0561] For modern online shopping sites, responding appropriately and promptly to user reviews is an important factor in improving customer satisfaction. However, manually responding to many reviews takes time and effort, making it difficult to manage efficiently. Furthermore, conventional automated response systems have the problem of not being able to capture user sentiment effectively, resulting in boilerplate replies that give users an impersonal impression.

[0562] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a user input means, a sending means, an analysis means, an emotion engine, a generation means, and a reply sending means. This makes it possible to appropriately analyze the emotions in the user's review and generate a reply message that can be individually tailored.

[0563] The "user input means" refers to a device and method for a user to input word-of-mouth data.

[0564] The "transmission means" refers to a device and method for transmitting word-of-mouth data entered by a user to a server.

[0565] The "analysis means" refers to a device and method that analyzes the word-of-mouth data sent to the server and extracts positive and negative information.

[0566] A "sentiment engine" is an apparatus and method for identifying sentiment based on analyzed word-of-mouth data.

[0567] The "generator" is a device and method for generating a reply based on the analyzed information and sentiment score.

[0568] The "reply sending means" refers to a device and method for sending the generated reply to the user.

[0569] "Natural language processing technology" is a technology that understands the structure and meaning of language used to analyze word-of-mouth data.

[0570] A "generative AI model" is a model used to generate replies using machine learning and artificial intelligence techniques.

[0571] The "emotion score" is a numerical representation of the positive and negative emotions contained in the review data.

[0572] A "prompt sentence" is an input sentence used to derive a generated reply sentence.

[0573] The present invention provides a system for automatically and appropriately and quickly responding to user reviews on an online shopping site. A specific embodiment of this system will be described below.

[0574] First, the user inputs review data using a smartphone application. This review data is then sent to the server via the user input means. The sending means is realized using a smartphone and AWS API Gateway.

[0575] Once the review data reaches the server, it is processed by an analytical tool that uses natural language processing technology (using TensorFlow and spaCy) to tokenize the review data and extract positive and negative information.

[0576] Next, the sentiment engine identifies the sentiment of the parsed review data. The sentiment engine uses the BERT model to identify positive and negative sentiment from review text.

[0577] Based on the analysis results and the emotion score, a generator generates a reply message. The generator uses GPT-3 (OpenAI's generative AI model) and creates a reply message using a prompt that takes the emotion score into account.

[0578] The generated reply is sent to the user's smartphone by a reply sending means, which uses AWS SNS (Simple Notification Service).

[0579] The operating procedure and data flow of this system will be explained using a concrete example. Suppose a user enters a review in the app saying, "This product was very good, but delivery was too slow." The entered review is sent to the server, and the analysis means extracts "very good" as positive information and "delivery was too slow" as negative information. The emotion engine then calculates an emotion score, and the generation means generates a reply message based on the following prompt message.

[0580] plaintext

[0581] User review: "This product was very good, but delivery was too slow."

[0582] Positive: "This product was very good"

[0583] Negative: "Delivery was too slow"

[0584] Sentiment score: Positive=0.7, Negative=0.3

[0585] Using the information above, generate a natural response like this:

[0586] "Thank you for your purchase. We are very happy that you were satisfied with the product. We apologize for the delay in delivery. We will strive to improve in the future."

[0587] In this way, users can check natural and polite replies on their smartphones. This system will improve the efficiency of responding to user reviews and increase customer satisfaction.

[0588] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0589] Step 1:

[0590] A user inputs a review using a smartphone application. After the user inputs the review, the device receives the review data as input data. This input data includes review information in text format.

[0591] Step 2:

[0592] The terminal sends the entered review data to the server. This process uses AWS API Gateway, which sends the entered text data in a format that transfers it to the API. The input is the text of the user review, and the output is the text data transferred to the server.

[0593] Step 3:

[0594] The server analyzes the received review data. It uses an NLP library (e.g., TensorFlow or spaCy) to tokenize the review data and perform grammatical analysis. The input is the text data received by the server, and the output is the tokenized data from which positive and negative information has been extracted.

[0595] Step 4:

[0596] The server passes the parsed data to a sentiment engine, which uses the BERT model to calculate sentiment scores from the parsed data. The input is the tokenized and parsed data, and the output is positive and negative sentiment scores.

[0597] Step 5:

[0598] The server generates a reply using a generation method based on the emotion score obtained by the emotion engine. The generation method uses a generative AI model such as GPT-3. The emotion score and extracted information are input into the generative AI model as a prompt to generate a natural reply. The input is the prompt and emotion score, and the output is the reply.

[0599] Step 6:

[0600] The server passes the generated reply to the reply sending means and sends it to the user's smartphone. AWS SNS (Simple Notification Service) is used as the reply sending means to send a notification to the user's device. The input is the generated reply, and the output is the reply message displayed on the user's smartphone.

[0601] In this way, a series of processes from when the user inputs a review to when the generated reply message is displayed is realized.

[0602] 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.

[0603] 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.

[0604] 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.

[0605] [Third embodiment]

[0606] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0607] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0608] 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).

[0609] 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.

[0610] 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.

[0611] 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).

[0612] 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.

[0613] 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.

[0614] 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.

[0615] 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.

[0616] 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.

[0617] 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."

[0618] The system of the present invention includes a user input means, a transmission means, an analysis means, a generation means, and a reply transmission means. The program of this system is used and implemented as follows.

[0619] The implementation of this system begins when a user inputs a review using a terminal. For example, the user might input, "The food at this restaurant was delicious, but the service was a little poor."

[0620] The input word-of-mouth data is transmitted from the terminal to the server. The transmitting means plays this role and transmits the word-of-mouth data input at the terminal to the server.

[0621] The server analyzes the received review data. The analysis method used here is natural language processing (NLP). Specifically, the review text is tokenized (divided into words and sentences) and then grammatically analyzed to extract positive and negative information. For example, parts such as "The food was delicious" are extracted as positive, and parts such as "The service was a little poor" are extracted as negative.

[0622] Once the analysis is complete, the server uses a generative method to generate a reply. This can be done using machine learning models or generative AI. The generative AI creates a natural and appropriate reply based on the analysis results. For example, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we apologize that our service did not meet your expectations."

[0623] The generated reply message is sent to the user's terminal by the reply sending means. When the reply message is sent from the server to the terminal, the terminal displays it to the user. The user can check the reply message displayed on their own terminal and feel that they have received a natural and polite response.

[0624] In this way, the system can provide quick and natural replies to reviews entered by users. In addition, by using AI, it is possible to respond to individual reviews and automatically provide high-quality replies to many reviews.

[0625] The processing flow will be explained below.

[0626] Step 1:

[0627] A user uses a device to enter a review. For example, the user might enter, "The food at this restaurant was good, but the service was a bit poor."

[0628] Step 2:

[0629] The terminal receives the user's input and transmits the review data to the server. The transmission means plays this role and checks the amount of input and the connection status to confirm the success of the transmission.

[0630] Step 3:

[0631] The server receives the review data sent from the device, temporarily stores the received data, and checks whether the data format is correct.

[0632] Step 4:

[0633] The server uses a natural language processing (NLP) engine to analyze the review text, specifically tokenizing it (dividing it into words and sentences), performing grammatical analysis, and extracting positive and negative information.

[0634] Step 5:

[0635] The server then requests the AI ​​to generate a reply based on the analysis results. For example, it sends data in the form of "Positive: The food was delicious" or "Negative: The service was a bit poor."

[0636] Step 6:

[0637] Based on the information it receives, generative AI generates natural and appropriate responses, such as, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0638] Step 7:

[0639] The server receives the reply text generated by the generative AI and saves it along with the review data.

[0640] Step 8:

[0641] The server sends the saved reply to the device, and then takes steps to verify that the data was sent and received correctly.

[0642] Step 9:

[0643] The reply received by the terminal from the server is displayed to the user in an appropriate format for easy viewing.

[0644] Step 10:

[0645] The user checks the reply displayed on their device and evaluates whether the generated reply feels natural.

[0646] Example 1

[0647] 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."

[0648] Conventional review reply systems have had difficulty in providing fast, natural replies to review content. In particular, responding to many reviews individually requires high manpower and time costs, and the quality of the replies cannot be maintained. Furthermore, technology for appropriately analyzing the sentiment of review content and generating replies based on the results has not been sufficiently developed. To solve these problems, the present invention aims to provide a system that automatically analyzes review content and provides fast, natural replies.

[0649] 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.

[0650] In this invention, the server includes a user input means for a user to input reviews using a terminal, a transmission means for transmitting the review data input by the user input means to the server via a network, an analysis means for receiving the review data and extracting positive and negative information using natural language processing technology, a generation means for generating a reply message using a generative AI model based on the extracted information, a reply transmission means for transmitting the generated reply message to the user's terminal via the network, and a means for displaying the reply message on the user's terminal. This enables automatic, quick, and natural replies to reviews. By analyzing positive and negative information, it is possible to generate appropriate replies that match the user's emotions, thereby providing high-quality service.

[0651] "User input means" refers to the function of a device or software that allows a user to input a review.

[0652] The "transmission means" is a function for transmitting the word-of-mouth data entered by the user to the server via the network.

[0653] The "analysis means" is a technology for processing the received word-of-mouth data and extracting analysis results such as positive and negative information.

[0654] The "generation means" is a function for generating an appropriate reply message based on the analysis results, and mainly uses a generative AI model.

[0655] The "reply sending means" is a function for sending the generated reply message to the user's terminal.

[0656] "Natural language processing technology" is a technology that enables computers to understand and appropriately process human language.

[0657] A "generative AI model" is an algorithm or software that uses machine learning techniques to generate appropriate replies from data.

[0658] A "prompt sentence" is text data that is input into a generative AI model and serves to guide the generation of a reply sentence.

[0659] The system of the present invention allows users to input reviews using a terminal, and automatically generates replies to those reviews, providing a quick and natural response. To implement this system, the following elements are required:

[0660] First, a user inputs a review using a device (such as a smartphone or PC). The user input means plays the role of receiving this review data. Specifically, a review entered by a user such as "The food at this restaurant was delicious, but the service was a bit poor" is included.

[0661] The device then sends the review data to the server via a transmission mechanism that sends the data to the server over the network. The data is sent in JSON format using an HTTP POST request.

[0662] When the server receives the review data, it uses analytical tools to analyze the content of the review. The natural language processing (NLP) technology used here performs tokenization, grammar analysis, and sentiment analysis. For example, it extracts parts such as "The food was delicious" as positive information and parts such as "The service was a bit poor" as negative information.

[0663] Once the analysis is complete, the server generates a reply using a generation means. This generation means uses a generative AI model (e.g., GPT-3). Based on the analysis results, a prompt is constructed and input into the generative AI model to generate an appropriate reply. Examples of prompts are as follows:

[0664] User review: "This cafe had great coffee, but limited seating."

[0665] Positive: "The coffee was great."

[0666] Negative: "There weren't many seats"

[0667] Generate a reply.

[0668] The generated reply is then sent from the server to the user's device. The reply is sent in JSON format as an HTTP response via the reply sending means. The user's device analyzes the received reply and displays it on the screen. This display means allows the user to see the reply quickly and naturally.

[0669] For example, if a user's review said, "This cafe had great coffee, but limited seating," the generated response might look like this:

[0670] "Thank you for visiting us. We're very happy that you enjoyed our coffee. However, we apologize for any inconvenience caused by the limited seating."

[0671] This system automatically provides high-quality replies to reviews, enabling users to respond quickly and appropriately.

[0672] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0673] Step 1:

[0674] A user inputs a review using a terminal. For example, the user inputs, "The food at this restaurant was delicious, but the service was a little poor." This review is processed as text data.

[0675] input:

[0676] User review text

[0677] output:

[0678] Review text data

[0679] Step 2:

[0680] The device sends the review data to the server. Specifically, the review text data is sent to the server using an HTTP POST request. At this time, the data is sent in JSON format.

[0681] input:

[0682] Review text data

[0683] output:

[0684] Review data sent to the server (JSON format)

[0685] Step 3:

[0686] The server receives the review data and analyzes it using an analysis tool. First, the review text is tokenized, and then grammatical and sentiment analysis is performed. For example, "The food was delicious" is extracted as a positive, and "The service was a bit poor" is extracted as a negative.

[0687] input:

[0688] Review data sent to the server (JSON format)

[0689] output:

[0690] Positive and negative information

[0691] Step 4:

[0692] The server constructs a prompt sentence based on the extracted information. Specifically, it generates a prompt sentence that combines the review text, positive information, and negative information.

[0693] input:

[0694] Positive and negative information

[0695] output:

[0696] Prompt statement

[0697] Step 5:

[0698] The server generates a reply using a generation method. The prompt is input to a generative AI model (e.g., GPT-3) to generate a reply. For example, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0699] input:

[0700] Prompt statement

[0701] output:

[0702] Generated reply

[0703] Step 6:

[0704] The server sends the generated reply to the user's device. The reply is sent in JSON format as an HTTP response.

[0705] input:

[0706] Generated reply

[0707] output:

[0708] Reply text sent to the user's device (JSON format)

[0709] Step 7:

[0710] The device analyzes the received reply and displays it to the user. Specifically, the reply is displayed to the user through a mobile application or a web browser. The user can check the displayed reply and feel satisfied.

[0711] input:

[0712] Reply text sent to the user's device (JSON format)

[0713] output:

[0714] Reply text displayed on the device

[0715] (Application example 1)

[0716] 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."

[0717] In traditional customer service at brick-and-mortar stores, it has been difficult to collect and analyze customer reviews and feedback in real time and respond appropriately on the spot. This often delays the improvement of customer satisfaction and service. Furthermore, manually analyzing and responding to large amounts of feedback is inefficient and results in inconsistent quality. To solve these issues, a system is needed that can collect customer feedback in real time and respond quickly and appropriately.

[0718] 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.

[0719] In this invention, the server includes a user input means, a transmission means for converting voice data input by the user input means into text data and transmitting the text data to the server, an analysis means for analyzing the text data to extract positive and negative information, a generation means for generating a reply message based on the extracted information, and a display means for transmitting and displaying the generated reply message to the user. This makes it possible to collect and analyze feedback from customers in real time and respond quickly and appropriately.

[0720] "User input means" refers to a device or interface for a user to input voice data or text data.

[0721] The "transmission means" is a device or interface for converting input voice data into text data and transmitting it to the server.

[0722] The "analysis means" is a device or software that analyzes the transmitted text data using natural language processing technology and extracts positive and negative information.

[0723] "Generation means" refers to a device or software that uses a generative AI model to generate reply messages or prompt messages based on the information extracted by the analysis means.

[0724] The "display means" is a device or interface for displaying the generated reply message to the user.

[0725] "Natural language processing technology" is a technology for understanding and analyzing human language, and is used when analyzing text data.

[0726] A "generative AI model" refers to an artificial intelligence model that uses machine learning and deep learning technologies to generate natural language based on input data.

[0727] A "prompt sentence" is an input sentence that gives instructions to a generative AI model and specifies what kind of output the model should generate.

[0728] This invention relates to a system including a user input means, a transmission means, an analysis means, a generation means, and a display means. The invention particularly relates to a system in which staff in brick-and-mortar stores wear smart glasses and analyze and display voice feedback from customers in real time, thereby enabling prompt and appropriate customer service.

[0729] The user input means is a device or interface for the user to input voice data or text data, specifically, a microphone in the smart glasses is used to input voice feedback.

[0730] The transmitting means is a device or interface for converting input voice data into text data and transmitting it to the server. Voice recognition software is used to convert the voice into text and the text is transmitted to the server via the network.

[0731] The analysis means is a device or software that analyzes the transmitted text data using natural language processing (NLP) technology to extract positive and negative information. This analysis uses natural language processing (NLP) technology to tokenize sentences and perform grammatical analysis.

[0732] The generation means is a device or software that uses a generative AI model to generate replies and prompts based on the information extracted by the analysis means. The generative AI model uses machine learning and deep learning techniques to generate appropriate replies based on the input data.

[0733] The display means is a device or interface for displaying the generated reply to the user. The smart glasses display is used to display the reply to the staff in real time.

[0734] For example, if a customer inputs voice feedback such as "The food was delicious, but the wait time was too long," the microphone in the smart glasses recognizes this voice, and the transmission means converts the voice into text and sends it to the server. On the server, the analysis means analyzes this feedback and extracts positive and negative aspects. The generation means then uses the generative AI model to create an appropriate reply, such as "Thank you for your patronage. We are very pleased that you enjoyed your meal. We apologize for the long wait, but we will strive to improve in the future." Finally, this reply is displayed on the smart glasses' display, allowing the staff to immediately respond appropriately to the customer.

[0735] An example of a prompt is as follows:

[0736] "Please enter your feedback by voice."

[0737] "Parse the feedback text."

[0738] "Performs sentiment analysis and generates replies."

[0739] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0740] Step 1:

[0741] The user provides verbal feedback.

[0742] Input: User's spoken feedback

[0743] How it works: The microphone in the smart glasses picks up sound and records it as audio data.

[0744] Output: Audio data

[0745] Step 2:

[0746] Convert the audio data into text data.

[0747] Input: Audio data

[0748] How it works: The smart glasses' transmitter uses speech recognition software (e.g., the SpeechRecognition library) to convert the voice data into text data.

[0749] Output: Text data

[0750] Step 3:

[0751] Sends text data to the server.

[0752] Input: Text data

[0753] Operation: The transmission means of the smart glasses transmits text data to the server via the network.

[0754] Output: Text data sent to the server

[0755] Step 4:

[0756] The server parses the text data.

[0757] Input: Text data sent to the server

[0758] How it works: The server's analysis means uses natural language processing techniques (e.g., the spaCy library) to analyze the text data and extract positive and negative information.

[0759] Output: Positive and negative information

[0760] Step 5:

[0761] A reply message is generated based on the extracted information.

[0762] Input: Positive and negative information

[0763] How it works: The server's generator uses a generative AI model (e.g., the transformers library) to generate an appropriate reply based on the prompt.

[0764] Output: The generated reply

[0765] Step 6:

[0766] The generated reply is sent to the smart glasses and displayed.

[0767] Input: Generated reply

[0768] Operation: The display means of the server transmits the generated reply message to the smart glasses via the network, and displays the reply message on the display of the smart glasses.

[0769] Output: Reply displayed on staff member's smart glasses

[0770] This allows users (staff) to instantly check the appropriate response to customer feedback and respond quickly.

[0771] 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.

[0772] The system of the present invention includes a user input means, a transmission means, an analysis means, a generation means, a reply transmission means, and an emotion engine. The program of this system is used and implemented as follows.

[0773] First, a user uses a terminal to input a review. For example, the user may input, "The food at this restaurant was delicious, but the service was a little poor."

[0774] The input word-of-mouth data is transmitted from the terminal to the server. The transmitting means plays this role and transmits the word-of-mouth data input at the terminal to the server.

[0775] The server analyzes the received review data. The analysis method used here uses natural language processing (NLP). Specifically, the text is tokenized (divided into words and sentences) and grammatical analysis is performed to extract positive and negative information. For example, parts such as "The food was delicious" are extracted as positive, and parts such as "The service was a little poor" are extracted as negative.

[0776] Furthermore, an emotion engine is used to recognize user emotions contained in the review data. The emotion engine analyzes the text of the review and identifies positive and negative emotions. For example, positive emotions are identified from the "delicious" part and negative emotions are identified from the "bad" part.

[0777] The server uses a generation means to generate a reply based on the information obtained by the analysis means and emotion engine. The generation means uses machine learning models and generative AI to create a reply taking into account the emotion score. For example, a reply such as "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we apologize that our service did not meet your expectations" may be generated.

[0778] The generated reply message is sent to the user's terminal by the reply sending means. When the reply message is sent from the server to the terminal, the terminal displays it to the user. The user can check the reply message displayed on their own terminal and feel that they have received a natural and polite response.

[0779] In this way, the system can provide quick and natural replies to user-submitted reviews. In particular, the use of an emotion engine allows for more personalized responses that reflect the user's emotions, enabling high-quality replies to be automatically generated for many reviews. Specific embodiments of the system can enhance a positive user experience.

[0780] The processing flow will be explained below.

[0781] Step 1:

[0782] A user uses a device to enter a review. For example, the user might enter, "The food at this restaurant was good, but the service was a bit poor."

[0783] Step 2:

[0784] The terminal receives the user's input and transmits the review data to the server. The transmission means plays this role and checks the amount of input and the connection status to confirm the success of the transmission.

[0785] Step 3:

[0786] The server receives the review data sent from the device, temporarily stores the received data, and checks whether the data format is correct.

[0787] Step 4:

[0788] The server uses a natural language processing (NLP) engine to analyze the review text. Specifically, it tokenizes the text (divides it into words and sentences), performs grammatical analysis, and extracts positive and negative information. For example, it extracts "The food was delicious" as a positive and "The service was a bit poor" as a negative.

[0789] Step 5:

[0790] The server uses an analysis means and an emotion engine to recognize emotions contained in the review data. The emotion engine uses a language model to generate an emotion score from the review text. For example, it identifies positive emotions from the "delicious" part and negative emotions from the "bad" part.

[0791] Step 6:

[0792] The server requests the generative AI to generate a reply based on the analysis results and emotion score. For example, it sends data in the form of "Positive: The food was delicious" and "Negative: The service was a little poor", with "Positive emotion score: 80" and "Negative emotion score: 20".

[0793] Step 7:

[0794] Based on the information it receives, generative AI generates natural and appropriate responses, such as, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0795] Step 8:

[0796] The server receives the reply text generated by the generative AI and saves it along with the review data.

[0797] Step 9:

[0798] The server sends the saved reply to the device, and then takes steps to verify that the data was sent and received correctly.

[0799] Step 10:

[0800] The reply received by the terminal from the server is displayed to the user in an appropriate format for easy viewing.

[0801] Step 11:

[0802] The user checks the reply displayed on their device and evaluates whether the generated reply feels natural.

[0803] Example 2

[0804] 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."

[0805] Conventional review response systems have had difficulty automatically generating quick and natural replies to reviews entered by users. They also struggled to generate replies that accurately reflected the user's feelings, resulting in a problem of only being able to provide uniform replies. This resulted in a poor user experience and made it difficult to achieve satisfaction.

[0806] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: means for a user to input a review using a terminal; means for transmitting review data from the terminal to the server; means for the server to receive the review data and perform data preprocessing; means for analyzing the review data and extracting positive information and negative information; means for identifying emotions using an emotion engine based on the extracted information; means for generating a reply message using a generative AI model taking into account the identified emotions; means for transmitting the generated reply message to the terminal; and means for the terminal to display the reply message to the user. This makes it possible to automatically generate a high-quality reply message that is quick, natural, and reflects emotions in response to a review input by a user, thereby improving user satisfaction.

[0807] "User" means an individual or organization that uses the system to enter reviews.

[0808] A "terminal" is a device used by a user, such as a smartphone, tablet, or computer.

[0809] "Word-of-mouth data" refers to text data of reviews and comments that users input and submit via their terminals.

[0810] The "server" is a back-end system that receives and analyzes word-of-mouth data and generates replies.

[0811] "Preprocessing" refers to processes such as normalization and cleaning that are performed before analyzing data.

[0812] The "analysis means" is a means for analyzing word-of-mouth data and extracting positive and negative information.

[0813] "Positive information" is information that indicates positive content or evaluations within the word-of-mouth data.

[0814] "Negative information" is information that indicates negative content or evaluations within the word-of-mouth data.

[0815] An "emotion engine" is software for identifying and classifying user emotions within review data.

[0816] A "generative AI model" is an artificial intelligence model that generates natural-looking sentences based on input data and conditions.

[0817] A "reply" is a reply text that the server generates based on word-of-mouth data and sends to the user.

[0818] The system of the present invention includes a user input means, a sending means, an analyzing means, a generating means, a reply sending means, and an emotion engine. This system operates in the following procedure.

[0819] First, a user uses a device to enter a review. For example, a user might write, "The curry at this restaurant was excellent, but the waiter's service was disappointing." The user enters this information through a review app using a device such as a smartphone or computer.

[0820] Next, the entered review data is sent from the device to the server. The device sends the review data entered by the user to the server as an HTTP request. At this time, the data is sent in JSON format, and HTTPS is used as the communication protocol. For example, the sent data has the following format:

[0821] POST / api / reviews

[0822] {

[0823] "review": "The curry at this restaurant was great, but the waiter's service was disappointing."

[0824] }

[0825] The server analyzes the received review data. This analysis uses natural language processing (NLP) technology. Specifically, for example, Google's BERT model is used to tokenize the sentences and analyze the meaning of each token. Next, positive and negative information is extracted. For example, "The curry was excellent" is extracted as positive information, and "The waiter's service was disappointing" is extracted as negative information.

[0826] The server then uses an emotion engine to identify user emotions contained in the review data. The emotion engine uses models such as the Hugging Face emotion analysis model. This allows positive and negative emotions to be identified in the review data. Specifically, "It was great" is classified as a positive emotion, and "It was disappointing" is classified as a negative emotion.

[0827] Next, the server generates a reply using a generative AI model, such as OpenAI's GPT-3, based on the analysis results and sentiment score. To take into account the content and sentiment of the review during this generation process, the prompt text is specified as follows:

[0828] Prompt: "A user wrote a review saying, 'The curry at this restaurant was great, but the waiter's service was disappointing.' Generate a response that reflects that sentiment."

[0829] The generated reply has specific content such as, "Thank you for visiting us. We are glad that you enjoyed our curry. We apologize that our waiter's service did not meet your expectations."

[0830] Finally, the server sends the generated reply to the terminal. The generated reply is sent to the terminal again in JSON format, and the terminal displays it to the user. The user can check the reply and feel that the response was natural and polite.

[0831] In this way, the system can provide quick and natural replies to reviews entered by users. In particular, by using an emotion engine and generative AI model, it is possible to provide personalized responses that reflect the user's emotions, and to automatically generate high-quality replies to a large number of reviews. Specific implementations of the system can provide a positive user experience and improve customer satisfaction.

[0832] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0833] Step 1:

[0834] The user uses the device to input a review. For example, the user might input, "The curry at this restaurant was excellent, but the waiter's service was disappointing." The review entered by the user is acquired as input data. The user then presses the send button, and the review data is sent.

[0835] Step 2:

[0836] The terminal sends the entered review data to the server. The terminal sends the review data entered by the user to the server as an HTTP request. Specifically, the review data is converted into JSON format and sent to the server via the HTTPS protocol. The input data is passed from the client to the server.

[0837] Step 3:

[0838] The server receives the review data and performs the necessary preprocessing. The server deserializes the received JSON format review data and extracts the text data. Next, preprocessing such as normalization and removal of unnecessary whitespace and special characters is performed. The input data is unprocessed text data, and the output data is preprocessed text data.

[0839] Step 4:

[0840] The server analyzes the review data and extracts positive and negative information. The server uses a natural language processing (NLP) model to tokenize and analyze grammar. It then analyzes using the BERT model to extract positive information (e.g., "The curry was excellent") and negative information (e.g., "The waiter's service was disappointing"). Through this process, the input data is preprocessed text data, and the output data is the extracted information.

[0841] Step 5:

[0842] The server uses an emotion engine to identify emotions. The emotion engine analyzes the text data and calculates an emotion score. Using Hugging Face's emotion analysis model, it classifies the emotion of each part. For example, the part "It was great" is identified as positive, and the part "It was disappointing" is identified as negative. The input data is the analyzed text data, and the output data is the emotion score and emotion classification results.

[0843] Step 6:

[0844] The server generates a reply using a generative AI model. Based on the analyzed emotion data and prompt, a reply is generated using generative AI (for example, GPT-3). An example of a prompt is input to the model in the form: "A user posted a review saying, 'The curry at this restaurant was excellent, but the waiter's service was disappointing.' Please generate a reply that reflects the emotion in response to this review." The output is a natural reply that reflects the emotion.

[0845] Step 7:

[0846] The server sends the generated reply to the terminal. The server converts the generated reply back to JSON format and sends it to the terminal as an HTTP response. The output data is the generated reply, which is passed to the client.

[0847] Step 8:

[0848] The terminal displays the reply to the user. The terminal analyzes the reply received from the server and displays it on the screen. The user can check the reply on their own terminal and feel that they have received a natural and polite response. The input data is the reply from the server, and the output data is the display to the user.

[0849] (Application example 2)

[0850] 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."

[0851] For modern online shopping sites, responding appropriately and promptly to user reviews is an important factor in improving customer satisfaction. However, manually responding to many reviews takes time and effort, making it difficult to manage efficiently. Furthermore, conventional automated response systems have the problem of not being able to capture user sentiment effectively, resulting in boilerplate replies that give users an impersonal impression.

[0852] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a user input means, a sending means, an analysis means, an emotion engine, a generation means, and a reply sending means. This makes it possible to appropriately analyze the emotions in the user's review and generate a reply message that can be individually tailored.

[0853] The "user input means" refers to a device and method for a user to input word-of-mouth data.

[0854] The "transmission means" refers to a device and method for transmitting word-of-mouth data entered by a user to a server.

[0855] The "analysis means" refers to a device and method that analyzes the word-of-mouth data sent to the server and extracts positive and negative information.

[0856] A "sentiment engine" is an apparatus and method for identifying sentiment based on analyzed word-of-mouth data.

[0857] The "generator" is a device and method for generating a reply based on the analyzed information and sentiment score.

[0858] The "reply sending means" refers to a device and method for sending the generated reply to the user.

[0859] "Natural language processing technology" is a technology that understands the structure and meaning of language used to analyze word-of-mouth data.

[0860] A "generative AI model" is a model used to generate replies using machine learning and artificial intelligence techniques.

[0861] The "emotion score" is a numerical representation of the positive and negative emotions contained in the review data.

[0862] A "prompt sentence" is an input sentence used to derive a generated reply sentence.

[0863] The present invention provides a system for automatically and appropriately and quickly responding to user reviews on an online shopping site. A specific embodiment of this system will be described below.

[0864] First, the user inputs review data using a smartphone application. This review data is then sent to the server via the user input means. The sending means is realized using a smartphone and AWS API Gateway.

[0865] Once the review data reaches the server, it is processed by an analytical tool that uses natural language processing technology (using TensorFlow and spaCy) to tokenize the review data and extract positive and negative information.

[0866] Next, the sentiment engine identifies the sentiment of the parsed review data. The sentiment engine uses the BERT model to identify positive and negative sentiment from review text.

[0867] Based on the analysis results and the emotion score, a generator generates a reply message. The generator uses GPT-3 (OpenAI's generative AI model) and creates a reply message using a prompt that takes the emotion score into account.

[0868] The generated reply is sent to the user's smartphone by a reply sending means, which uses AWS SNS (Simple Notification Service).

[0869] The operating procedure and data flow of this system will be explained using a concrete example. Suppose a user enters a review in the app saying, "This product was very good, but delivery was too slow." The entered review is sent to the server, and the analysis means extracts "very good" as positive information and "delivery was too slow" as negative information. The emotion engine then calculates an emotion score, and the generation means generates a reply message based on the following prompt message.

[0870] plaintext

[0871] User review: "This product was very good, but delivery was too slow."

[0872] Positive: "This product was very good"

[0873] Negative: "Delivery was too slow"

[0874] Sentiment score: Positive=0.7, Negative=0.3

[0875] Using the information above, generate a natural response like this:

[0876] "Thank you for your purchase. We are very happy that you were satisfied with the product. We apologize for the delay in delivery. We will strive to improve in the future."

[0877] In this way, users can check natural and polite replies on their smartphones. This system will improve the efficiency of responding to user reviews and increase customer satisfaction.

[0878] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0879] Step 1:

[0880] A user inputs a review using a smartphone application. After the user inputs the review, the device receives the review data as input data. This input data includes review information in text format.

[0881] Step 2:

[0882] The terminal sends the entered review data to the server. This process uses AWS API Gateway, which sends the entered text data in a format that transfers it to the API. The input is the text of the user review, and the output is the text data transferred to the server.

[0883] Step 3:

[0884] The server analyzes the received review data. It uses an NLP library (e.g., TensorFlow or spaCy) to tokenize the review data and perform grammatical analysis. The input is the text data received by the server, and the output is the tokenized data from which positive and negative information has been extracted.

[0885] Step 4:

[0886] The server passes the parsed data to a sentiment engine, which uses the BERT model to calculate sentiment scores from the parsed data. The input is the tokenized and parsed data, and the output is positive and negative sentiment scores.

[0887] Step 5:

[0888] The server generates a reply using a generation method based on the emotion score obtained by the emotion engine. The generation method uses a generative AI model such as GPT-3. The emotion score and extracted information are input into the generative AI model as a prompt to generate a natural reply. The input is the prompt and emotion score, and the output is the reply.

[0889] Step 6:

[0890] The server passes the generated reply to the reply sending means and sends it to the user's smartphone. AWS SNS (Simple Notification Service) is used as the reply sending means to send a notification to the user's device. The input is the generated reply, and the output is the reply message displayed on the user's smartphone.

[0891] In this way, a series of processes from when the user inputs a review to when the generated reply message is displayed is realized.

[0892] 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.

[0893] 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.

[0894] 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.

[0895] [Fourth embodiment]

[0896] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[0897] 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.

[0898] 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).

[0899] 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.

[0900] 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.

[0901] 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).

[0902] 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.

[0903] 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.

[0904] 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.

[0905] 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.

[0906] 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.

[0907] 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.

[0908] 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."

[0909] The system of the present invention includes a user input means, a transmission means, an analysis means, a generation means, and a reply transmission means. The program of this system is used and implemented as follows.

[0910] The implementation of this system begins when a user inputs a review using a terminal. For example, the user might input, "The food at this restaurant was delicious, but the service was a little poor."

[0911] The input word-of-mouth data is transmitted from the terminal to the server. The transmitting means plays this role and transmits the word-of-mouth data input at the terminal to the server.

[0912] The server analyzes the received review data. The analysis method used here is natural language processing (NLP). Specifically, the review text is tokenized (divided into words and sentences) and then grammatically analyzed to extract positive and negative information. For example, parts such as "The food was delicious" are extracted as positive, and parts such as "The service was a little poor" are extracted as negative.

[0913] Once the analysis is complete, the server uses a generative method to generate a reply. This can be done using machine learning models or generative AI. The generative AI creates a natural and appropriate reply based on the analysis results. For example, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we apologize that our service did not meet your expectations."

[0914] The generated reply message is sent to the user's terminal by the reply sending means. When the reply message is sent from the server to the terminal, the terminal displays it to the user. The user can check the reply message displayed on their own terminal and feel that they have received a natural and polite response.

[0915] In this way, the system can provide quick and natural replies to reviews entered by users. In addition, by using AI, it is possible to respond to individual reviews and automatically provide high-quality replies to many reviews.

[0916] The processing flow will be explained below.

[0917] Step 1:

[0918] A user uses a device to enter a review. For example, the user might enter, "The food at this restaurant was good, but the service was a bit poor."

[0919] Step 2:

[0920] The terminal receives the user's input and transmits the review data to the server. The transmission means plays this role and checks the amount of input and the connection status to confirm the success of the transmission.

[0921] Step 3:

[0922] The server receives the review data sent from the device, temporarily stores the received data, and checks whether the data format is correct.

[0923] Step 4:

[0924] The server uses a natural language processing (NLP) engine to analyze the review text, specifically tokenizing it (dividing it into words and sentences), performing grammatical analysis, and extracting positive and negative information.

[0925] Step 5:

[0926] The server then requests the AI ​​to generate a reply based on the analysis results. For example, it sends data in the form of "Positive: The food was delicious" or "Negative: The service was a bit poor."

[0927] Step 6:

[0928] Based on the information it receives, generative AI generates natural and appropriate responses, such as, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0929] Step 7:

[0930] The server receives the reply text generated by the generative AI and saves it along with the review data.

[0931] Step 8:

[0932] The server sends the saved reply to the device, and then takes steps to verify that the data was sent and received correctly.

[0933] Step 9:

[0934] The reply received by the terminal from the server is displayed to the user in an appropriate format for easy viewing.

[0935] Step 10:

[0936] The user checks the reply displayed on their device and evaluates whether the generated reply feels natural.

[0937] Example 1

[0938] 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."

[0939] Conventional review reply systems have had difficulty in providing fast, natural replies to review content. In particular, responding to many reviews individually requires high manpower and time costs, and the quality of the replies cannot be maintained. Furthermore, technology for appropriately analyzing the sentiment of review content and generating replies based on the results has not been sufficiently developed. To solve these problems, the present invention aims to provide a system that automatically analyzes review content and provides fast, natural replies.

[0940] 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.

[0941] In this invention, the server includes a user input means for a user to input reviews using a terminal, a transmission means for transmitting the review data input by the user input means to the server via a network, an analysis means for receiving the review data and extracting positive and negative information using natural language processing technology, a generation means for generating a reply message using a generative AI model based on the extracted information, a reply transmission means for transmitting the generated reply message to the user's terminal via the network, and a means for displaying the reply message on the user's terminal. This enables automatic, quick, and natural replies to reviews. By analyzing positive and negative information, it is possible to generate appropriate replies that match the user's emotions, thereby providing high-quality service.

[0942] "User input means" refers to the function of a device or software that allows a user to input a review.

[0943] The "transmission means" is a function for transmitting the word-of-mouth data entered by the user to the server via the network.

[0944] The "analysis means" is a technology for processing the received word-of-mouth data and extracting analysis results such as positive and negative information.

[0945] The "generation means" is a function for generating an appropriate reply message based on the analysis results, and mainly uses a generative AI model.

[0946] The "reply sending means" is a function for sending the generated reply message to the user's terminal.

[0947] "Natural language processing technology" is a technology that enables computers to understand and appropriately process human language.

[0948] A "generative AI model" is an algorithm or software that uses machine learning techniques to generate appropriate replies from data.

[0949] A "prompt sentence" is text data that is input into a generative AI model and serves to guide the generation of a reply sentence.

[0950] The system of the present invention allows users to input reviews using a terminal, and automatically generates replies to those reviews, providing a quick and natural response. To implement this system, the following elements are required:

[0951] First, a user inputs a review using a device (such as a smartphone or PC). The user input means plays the role of receiving this review data. Specifically, a review entered by a user such as "The food at this restaurant was delicious, but the service was a bit poor" is included.

[0952] The device then sends the review data to the server via a transmission mechanism that sends the data to the server over the network. The data is sent in JSON format using an HTTP POST request.

[0953] When the server receives the review data, it uses analytical tools to analyze the content of the review. The natural language processing (NLP) technology used here performs tokenization, grammar analysis, and sentiment analysis. For example, it extracts parts such as "The food was delicious" as positive information and parts such as "The service was a bit poor" as negative information.

[0954] Once the analysis is complete, the server generates a reply using a generation means. This generation means uses a generative AI model (e.g., GPT-3). Based on the analysis results, a prompt is constructed and input into the generative AI model to generate an appropriate reply. Examples of prompts are as follows:

[0955] User review: "This cafe had great coffee, but limited seating."

[0956] Positive: "The coffee was great."

[0957] Negative: "There weren't many seats"

[0958] Generate a reply.

[0959] The generated reply is then sent from the server to the user's device. The reply is sent in JSON format as an HTTP response via the reply sending means. The user's device analyzes the received reply and displays it on the screen. This display means allows the user to see the reply quickly and naturally.

[0960] For example, if a user's review said, "This cafe had great coffee, but limited seating," the generated response might look like this:

[0961] "Thank you for visiting us. We're very happy that you enjoyed our coffee. However, we apologize for any inconvenience caused by the limited seating."

[0962] This system automatically provides high-quality replies to reviews, enabling users to respond quickly and appropriately.

[0963] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0964] Step 1:

[0965] A user inputs a review using a terminal. For example, the user inputs, "The food at this restaurant was delicious, but the service was a little poor." This review is processed as text data.

[0966] input:

[0967] User review text

[0968] output:

[0969] Review text data

[0970] Step 2:

[0971] The device sends the review data to the server. Specifically, the review text data is sent to the server using an HTTP POST request. At this time, the data is sent in JSON format.

[0972] input:

[0973] Review text data

[0974] output:

[0975] Review data sent to the server (JSON format)

[0976] Step 3:

[0977] The server receives the review data and analyzes it using an analysis tool. First, the review text is tokenized, and then grammatical and sentiment analysis is performed. For example, "The food was delicious" is extracted as a positive, and "The service was a bit poor" is extracted as a negative.

[0978] input:

[0979] Review data sent to the server (JSON format)

[0980] output:

[0981] Positive and negative information

[0982] Step 4:

[0983] The server constructs a prompt sentence based on the extracted information. Specifically, it generates a prompt sentence that combines the review text, positive information, and negative information.

[0984] input:

[0985] Positive and negative information

[0986] output:

[0987] Prompt statement

[0988] Step 5:

[0989] The server generates a reply using a generation method. The prompt is input to a generative AI model (e.g., GPT-3) to generate a reply. For example, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[0990] input:

[0991] Prompt statement

[0992] output:

[0993] Generated reply

[0994] Step 6:

[0995] The server sends the generated reply to the user's device. The reply is sent in JSON format as an HTTP response.

[0996] input:

[0997] Generated reply

[0998] output:

[0999] Reply text sent to the user's device (JSON format)

[1000] Step 7:

[1001] The device analyzes the received reply and displays it to the user. Specifically, the reply is displayed to the user through a mobile application or a web browser. The user can check the displayed reply and feel satisfied.

[1002] input:

[1003] Reply text sent to the user's device (JSON format)

[1004] output:

[1005] Reply text displayed on the device

[1006] (Application example 1)

[1007] 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."

[1008] In traditional customer service at brick-and-mortar stores, it has been difficult to collect and analyze customer reviews and feedback in real time and respond appropriately on the spot. This often delays the improvement of customer satisfaction and service. Furthermore, manually analyzing and responding to large amounts of feedback is inefficient and results in inconsistent quality. To solve these issues, a system is needed that can collect customer feedback in real time and respond quickly and appropriately.

[1009] 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.

[1010] In this invention, the server includes a user input means, a transmission means for converting voice data input by the user input means into text data and transmitting the text data to the server, an analysis means for analyzing the text data to extract positive and negative information, a generation means for generating a reply message based on the extracted information, and a display means for transmitting and displaying the generated reply message to the user. This makes it possible to collect and analyze feedback from customers in real time and respond quickly and appropriately.

[1011] "User input means" refers to a device or interface for a user to input voice data or text data.

[1012] The "transmission means" is a device or interface for converting input voice data into text data and transmitting it to the server.

[1013] The "analysis means" is a device or software that analyzes the transmitted text data using natural language processing technology and extracts positive and negative information.

[1014] "Generation means" refers to a device or software that uses a generative AI model to generate reply messages or prompt messages based on the information extracted by the analysis means.

[1015] The "display means" is a device or interface for displaying the generated reply message to the user.

[1016] "Natural language processing technology" is a technology for understanding and analyzing human language, and is used when analyzing text data.

[1017] A "generative AI model" refers to an artificial intelligence model that uses machine learning and deep learning technologies to generate natural language based on input data.

[1018] A "prompt sentence" is an input sentence that gives instructions to a generative AI model and specifies what kind of output the model should generate.

[1019] This invention relates to a system including a user input means, a transmission means, an analysis means, a generation means, and a display means. The invention particularly relates to a system in which staff in brick-and-mortar stores wear smart glasses and analyze and display voice feedback from customers in real time, thereby enabling prompt and appropriate customer service.

[1020] The user input means is a device or interface for the user to input voice data or text data, specifically, a microphone in the smart glasses is used to input voice feedback.

[1021] The transmitting means is a device or interface for converting input voice data into text data and transmitting it to the server. Voice recognition software is used to convert the voice into text and the text is transmitted to the server via the network.

[1022] The analysis means is a device or software that analyzes the transmitted text data using natural language processing (NLP) technology to extract positive and negative information. This analysis uses natural language processing (NLP) technology to tokenize sentences and perform grammatical analysis.

[1023] The generation means is a device or software that uses a generative AI model to generate replies and prompts based on the information extracted by the analysis means. The generative AI model uses machine learning and deep learning techniques to generate appropriate replies based on the input data.

[1024] The display means is a device or interface for displaying the generated reply to the user. The smart glasses display is used to display the reply to the staff in real time.

[1025] For example, if a customer inputs voice feedback such as "The food was delicious, but the wait time was too long," the microphone in the smart glasses recognizes this voice, and the transmission means converts the voice into text and sends it to the server. On the server, the analysis means analyzes this feedback and extracts positive and negative aspects. The generation means then uses the generative AI model to create an appropriate reply, such as "Thank you for your patronage. We are very pleased that you enjoyed your meal. We apologize for the long wait, but we will strive to improve in the future." Finally, this reply is displayed on the smart glasses' display, allowing the staff to immediately respond appropriately to the customer.

[1026] An example of a prompt is as follows:

[1027] "Please enter your feedback by voice."

[1028] "Parse the feedback text."

[1029] "Performs sentiment analysis and generates replies."

[1030] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1031] Step 1:

[1032] The user provides verbal feedback.

[1033] Input: User's spoken feedback

[1034] How it works: The microphone in the smart glasses picks up sound and records it as audio data.

[1035] Output: Audio data

[1036] Step 2:

[1037] Convert the audio data into text data.

[1038] Input: Audio data

[1039] How it works: The smart glasses' transmitter uses speech recognition software (e.g., the SpeechRecognition library) to convert the voice data into text data.

[1040] Output: Text data

[1041] Step 3:

[1042] Sends text data to the server.

[1043] Input: Text data

[1044] Operation: The transmission means of the smart glasses transmits text data to the server via the network.

[1045] Output: Text data sent to the server

[1046] Step 4:

[1047] The server parses the text data.

[1048] Input: Text data sent to the server

[1049] How it works: The server's analysis means uses natural language processing techniques (e.g., the spaCy library) to analyze the text data and extract positive and negative information.

[1050] Output: Positive and negative information

[1051] Step 5:

[1052] A reply message is generated based on the extracted information.

[1053] Input: Positive and negative information

[1054] How it works: The server's generator uses a generative AI model (e.g., the transformers library) to generate an appropriate reply based on the prompt.

[1055] Output: The generated reply

[1056] Step 6:

[1057] The generated reply is sent to the smart glasses and displayed.

[1058] Input: Generated reply

[1059] Operation: The display means of the server transmits the generated reply message to the smart glasses via the network, and displays the reply message on the display of the smart glasses.

[1060] Output: Reply displayed on staff member's smart glasses

[1061] This allows users (staff) to instantly check the appropriate response to customer feedback and respond quickly.

[1062] 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.

[1063] The system of the present invention includes a user input means, a transmission means, an analysis means, a generation means, a reply transmission means, and an emotion engine. The program of this system is used and implemented as follows.

[1064] First, a user uses a terminal to input a review. For example, the user may input, "The food at this restaurant was delicious, but the service was a little poor."

[1065] The input word-of-mouth data is transmitted from the terminal to the server. The transmitting means plays this role and transmits the word-of-mouth data input at the terminal to the server.

[1066] The server analyzes the received review data. The analysis method used here uses natural language processing (NLP). Specifically, the text is tokenized (divided into words and sentences) and grammatical analysis is performed to extract positive and negative information. For example, parts such as "The food was delicious" are extracted as positive, and parts such as "The service was a little poor" are extracted as negative.

[1067] Furthermore, an emotion engine is used to recognize user emotions contained in the review data. The emotion engine analyzes the text of the review and identifies positive and negative emotions. For example, positive emotions are identified from the "delicious" part and negative emotions are identified from the "bad" part.

[1068] The server uses a generation means to generate a reply based on the information obtained by the analysis means and emotion engine. The generation means uses machine learning models and generative AI to create a reply taking into account the emotion score. For example, a reply such as "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we apologize that our service did not meet your expectations" may be generated.

[1069] The generated reply message is sent to the user's terminal by the reply sending means. When the reply message is sent from the server to the terminal, the terminal displays it to the user. The user can check the reply message displayed on their own terminal and feel that they have received a natural and polite response.

[1070] In this way, the system can provide quick and natural replies to user-submitted reviews. In particular, the use of an emotion engine allows for more personalized responses that reflect the user's emotions, enabling high-quality replies to be automatically generated for many reviews. Specific embodiments of the system can enhance a positive user experience.

[1071] The processing flow will be explained below.

[1072] Step 1:

[1073] A user uses a device to enter a review. For example, the user might enter, "The food at this restaurant was good, but the service was a bit poor."

[1074] Step 2:

[1075] The terminal receives the user's input and transmits the review data to the server. The transmission means plays this role and checks the amount of input and the connection status to confirm the success of the transmission.

[1076] Step 3:

[1077] The server receives the review data sent from the device, temporarily stores the received data, and checks whether the data format is correct.

[1078] Step 4:

[1079] The server uses a natural language processing (NLP) engine to analyze the review text. Specifically, it tokenizes the text (divides it into words and sentences), performs grammatical analysis, and extracts positive and negative information. For example, it extracts "The food was delicious" as a positive and "The service was a bit poor" as a negative.

[1080] Step 5:

[1081] The server uses an analysis means and an emotion engine to recognize emotions contained in the review data. The emotion engine uses a language model to generate an emotion score from the review text. For example, it identifies positive emotions from the "delicious" part and negative emotions from the "bad" part.

[1082] Step 6:

[1083] The server requests the generative AI to generate a reply based on the analysis results and emotion score. For example, it sends data in the form of "Positive: The food was delicious" and "Negative: The service was a little poor", with "Positive emotion score: 80" and "Negative emotion score: 20".

[1084] Step 7:

[1085] Based on the information it receives, generative AI generates natural and appropriate responses, such as, "Thank you for visiting us. We are very pleased that you enjoyed our food. However, we are sorry that our service did not meet your expectations."

[1086] Step 8:

[1087] The server receives the reply text generated by the generative AI and saves it along with the review data.

[1088] Step 9:

[1089] The server sends the saved reply to the device, and then takes steps to verify that the data was sent and received correctly.

[1090] Step 10:

[1091] The reply received by the terminal from the server is displayed to the user in an appropriate format for easy viewing.

[1092] Step 11:

[1093] The user checks the reply displayed on their device and evaluates whether the generated reply feels natural.

[1094] Example 2

[1095] 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."

[1096] Conventional review response systems have had difficulty automatically generating quick and natural replies to reviews entered by users. They also struggled to generate replies that accurately reflected the user's feelings, resulting in a problem of only being able to provide uniform replies. This resulted in a poor user experience and made it difficult to achieve satisfaction.

[1097] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes: means for a user to input a review using a terminal; means for transmitting review data from the terminal to the server; means for the server to receive the review data and perform data preprocessing; means for analyzing the review data and extracting positive information and negative information; means for identifying emotions using an emotion engine based on the extracted information; means for generating a reply message using a generative AI model taking into account the identified emotions; means for transmitting the generated reply message to the terminal; and means for the terminal to display the reply message to the user. This makes it possible to automatically generate a high-quality reply message that is quick, natural, and reflects emotions in response to a review input by a user, thereby improving user satisfaction.

[1098] "User" means an individual or organization that uses the system to enter reviews.

[1099] A "terminal" is a device used by a user, such as a smartphone, tablet, or computer.

[1100] "Word-of-mouth data" refers to text data of reviews and comments that users input and submit via their terminals.

[1101] The "server" is a back-end system that receives and analyzes word-of-mouth data and generates replies.

[1102] "Preprocessing" refers to processes such as normalization and cleaning that are performed before analyzing data.

[1103] The "analysis means" is a means for analyzing word-of-mouth data and extracting positive and negative information.

[1104] "Positive information" is information that indicates positive content or evaluations within the word-of-mouth data.

[1105] "Negative information" is information that indicates negative content or evaluations within the word-of-mouth data.

[1106] An "emotion engine" is software for identifying and classifying user emotions within review data.

[1107] A "generative AI model" is an artificial intelligence model that generates natural-looking sentences based on input data and conditions.

[1108] A "reply" is a reply text that the server generates based on word-of-mouth data and sends to the user.

[1109] The system of the present invention includes a user input means, a sending means, an analyzing means, a generating means, a reply sending means, and an emotion engine. This system operates in the following procedure.

[1110] First, a user uses a device to enter a review. For example, a user might write, "The curry at this restaurant was excellent, but the waiter's service was disappointing." The user enters this information through a review app using a device such as a smartphone or computer.

[1111] Next, the entered review data is sent from the device to the server. The device sends the review data entered by the user to the server as an HTTP request. At this time, the data is sent in JSON format, and HTTPS is used as the communication protocol. For example, the sent data has the following format:

[1112] POST / api / reviews

[1113] {

[1114] "review": "The curry at this restaurant was great, but the waiter's service was disappointing."

[1115] }

[1116] The server analyzes the received review data. This analysis uses natural language processing (NLP) technology. Specifically, for example, Google's BERT model is used to tokenize the sentences and analyze the meaning of each token. Next, positive and negative information is extracted. For example, "The curry was excellent" is extracted as positive information, and "The waiter's service was disappointing" is extracted as negative information.

[1117] The server then uses an emotion engine to identify user emotions contained in the review data. The emotion engine uses models such as the Hugging Face emotion analysis model. This allows positive and negative emotions to be identified in the review data. Specifically, "It was great" is classified as a positive emotion, and "It was disappointing" is classified as a negative emotion.

[1118] Next, the server generates a reply using a generative AI model, such as OpenAI's GPT-3, based on the analysis results and sentiment score. To take into account the content and sentiment of the review during this generation process, the prompt text is specified as follows:

[1119] Prompt: "A user wrote a review saying, 'The curry at this restaurant was great, but the waiter's service was disappointing.' Generate a response that reflects that sentiment."

[1120] The generated reply has specific content such as, "Thank you for visiting us. We are glad that you enjoyed our curry. We apologize that our waiter's service did not meet your expectations."

[1121] Finally, the server sends the generated reply to the terminal. The generated reply is sent to the terminal again in JSON format, and the terminal displays it to the user. The user can check the reply and feel that the response was natural and polite.

[1122] In this way, the system can provide quick and natural replies to reviews entered by users. In particular, by using an emotion engine and generative AI model, it is possible to provide personalized responses that reflect the user's emotions, and to automatically generate high-quality replies to a large number of reviews. Specific implementations of the system can provide a positive user experience and improve customer satisfaction.

[1123] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1124] Step 1:

[1125] The user uses the device to input a review. For example, the user might input, "The curry at this restaurant was excellent, but the waiter's service was disappointing." The review entered by the user is acquired as input data. The user then presses the send button, and the review data is sent.

[1126] Step 2:

[1127] The terminal sends the entered review data to the server. The terminal sends the review data entered by the user to the server as an HTTP request. Specifically, the review data is converted into JSON format and sent to the server via the HTTPS protocol. The input data is passed from the client to the server.

[1128] Step 3:

[1129] The server receives the review data and performs the necessary preprocessing. The server deserializes the received JSON format review data and extracts the text data. Next, preprocessing such as normalization and removal of unnecessary whitespace and special characters is performed. The input data is unprocessed text data, and the output data is preprocessed text data.

[1130] Step 4:

[1131] The server analyzes the review data and extracts positive and negative information. The server uses a natural language processing (NLP) model to tokenize and analyze grammar. It then analyzes using the BERT model to extract positive information (e.g., "The curry was excellent") and negative information (e.g., "The waiter's service was disappointing"). Through this process, the input data is preprocessed text data, and the output data is the extracted information.

[1132] Step 5:

[1133] The server uses an emotion engine to identify emotions. The emotion engine analyzes the text data and calculates an emotion score. Using Hugging Face's emotion analysis model, it classifies the emotion of each part. For example, the part "It was great" is identified as positive, and the part "It was disappointing" is identified as negative. The input data is the analyzed text data, and the output data is the emotion score and emotion classification results.

[1134] Step 6:

[1135] The server generates a reply using a generative AI model. Based on the analyzed emotion data and prompt, a reply is generated using generative AI (for example, GPT-3). An example of a prompt is input to the model in the form: "A user posted a review saying, 'The curry at this restaurant was excellent, but the waiter's service was disappointing.' Please generate a reply that reflects the emotion in response to this review." The output is a natural reply that reflects the emotion.

[1136] Step 7:

[1137] The server sends the generated reply to the terminal. The server converts the generated reply back to JSON format and sends it to the terminal as an HTTP response. The output data is the generated reply, which is passed to the client.

[1138] Step 8:

[1139] The terminal displays the reply to the user. The terminal analyzes the reply received from the server and displays it on the screen. The user can check the reply on their own terminal and feel that they have received a natural and polite response. The input data is the reply from the server, and the output data is the display to the user.

[1140] (Application example 2)

[1141] 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."

[1142] For modern online shopping sites, responding appropriately and promptly to user reviews is an important factor in improving customer satisfaction. However, manually responding to many reviews takes time and effort, making it difficult to manage efficiently. Furthermore, conventional automated response systems have the problem of not being able to capture user sentiment effectively, resulting in boilerplate replies that give users an impersonal impression.

[1143] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a user input means, a sending means, an analysis means, an emotion engine, a generation means, and a reply sending means. This makes it possible to appropriately analyze the emotions in the user's review and generate a reply message that can be individually tailored.

[1144] The "user input means" refers to a device and method for a user to input word-of-mouth data.

[1145] The "transmission means" refers to a device and method for transmitting word-of-mouth data entered by a user to a server.

[1146] The "analysis means" refers to a device and method that analyzes the word-of-mouth data sent to the server and extracts positive and negative information.

[1147] A "sentiment engine" is an apparatus and method for identifying sentiment based on analyzed word-of-mouth data.

[1148] The "generator" is a device and method for generating a reply based on the analyzed information and sentiment score.

[1149] The "reply sending means" refers to a device and method for sending the generated reply to the user.

[1150] "Natural language processing technology" is a technology that understands the structure and meaning of language used to analyze word-of-mouth data.

[1151] A "generative AI model" is a model used to generate replies using machine learning and artificial intelligence techniques.

[1152] The "emotion score" is a numerical representation of the positive and negative emotions contained in the review data.

[1153] A "prompt sentence" is an input sentence used to derive a generated reply sentence.

[1154] The present invention provides a system for automatically and appropriately and quickly responding to user reviews on an online shopping site. A specific embodiment of this system will be described below.

[1155] First, the user inputs review data using a smartphone application. This review data is then sent to the server via the user input means. The sending means is realized using a smartphone and AWS API Gateway.

[1156] Once the review data reaches the server, it is processed by an analytical tool that uses natural language processing technology (using TensorFlow and spaCy) to tokenize the review data and extract positive and negative information.

[1157] Next, the sentiment engine identifies the sentiment of the parsed review data. The sentiment engine uses the BERT model to identify positive and negative sentiment from review text.

[1158] Based on the analysis results and the emotion score, a generator generates a reply message. The generator uses GPT-3 (OpenAI's generative AI model) and creates a reply message using a prompt that takes the emotion score into account.

[1159] The generated reply is sent to the user's smartphone by a reply sending means, which uses AWS SNS (Simple Notification Service).

[1160] The operating procedure and data flow of this system will be explained using a concrete example. Suppose a user enters a review in the app saying, "This product was very good, but delivery was too slow." The entered review is sent to the server, and the analysis means extracts "very good" as positive information and "delivery was too slow" as negative information. The emotion engine then calculates an emotion score, and the generation means generates a reply message based on the following prompt message.

[1161] plaintext

[1162] User review: "This product was very good, but delivery was too slow."

[1163] Positive: "This product was very good"

[1164] Negative: "Delivery was too slow"

[1165] Sentiment score: Positive=0.7, Negative=0.3

[1166] Using the information above, generate a natural response like this:

[1167] "Thank you for your purchase. We are very happy that you were satisfied with the product. We apologize for the delay in delivery. We will strive to improve in the future."

[1168] In this way, users can check natural and polite replies on their smartphones. This system will improve the efficiency of responding to user reviews and increase customer satisfaction.

[1169] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1170] Step 1:

[1171] A user inputs a review using a smartphone application. After the user inputs the review, the device receives the review data as input data. This input data includes review information in text format.

[1172] Step 2:

[1173] The terminal sends the entered review data to the server. This process uses AWS API Gateway, which sends the entered text data in a format that transfers it to the API. The input is the text of the user review, and the output is the text data transferred to the server.

[1174] Step 3:

[1175] The server analyzes the received review data. It uses an NLP library (e.g., TensorFlow or spaCy) to tokenize the review data and perform grammatical analysis. The input is the text data received by the server, and the output is the tokenized data from which positive and negative information has been extracted.

[1176] Step 4:

[1177] The server passes the parsed data to a sentiment engine, which uses the BERT model to calculate sentiment scores from the parsed data. The input is the tokenized and parsed data, and the output is positive and negative sentiment scores.

[1178] Step 5:

[1179] The server generates a reply using a generation method based on the emotion score obtained by the emotion engine. The generation method uses a generative AI model such as GPT-3. The emotion score and extracted information are input into the generative AI model as a prompt to generate a natural reply. The input is the prompt and emotion score, and the output is the reply.

[1180] Step 6:

[1181] The server passes the generated reply to the reply sending means and sends it to the user's smartphone. AWS SNS (Simple Notification Service) is used as the reply sending means to send a notification to the user's device. The input is the generated reply, and the output is the reply message displayed on the user's smartphone.

[1182] In this way, a series of processes from when the user inputs a review to when the generated reply message is displayed is realized.

[1183] 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.

[1184] 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.

[1185] 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.

[1186] 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.

[1187] FIG. 9 illustrates 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 behaviors 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.

[1188] 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.

[1189] 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).

[1190] 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.

[1191] 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."

[1192] 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.

[1193] 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).

[1194] 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.

[1195] 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.

[1196] 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.

[1197] 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.

[1198] 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.

[1199] 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.

[1200] 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.

[1201] 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.

[1202] 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.

[1203] 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.

[1204] The following is further disclosed regarding the above embodiment.

[1205] (Claim 1)

[1206] a user input means;

[1207] a transmission means for transmitting the word-of-mouth data input by the user input means to a server;

[1208] an analysis means for analyzing the word-of-mouth data to extract positive information and negative information;

[1209] a generating means for generating a reply message based on the extracted information;

[1210] a reply sending means for sending the generated reply message to a user;

[1211] A system including:

[1212] (Claim 2)

[1213] 2. The system according to claim 1, wherein the analyzing means analyzes the word-of-mouth data by using a natural language processing technique.

[1214] (Claim 3)

[1215] The system according to claim 1, wherein the generating means generates a reply message using a machine learning model.

[1216] "Example 1"

[1217] (Claim 1)

[1218] a user input means for allowing a user to input a word-of-mouth comment using a terminal;

[1219] a transmission means for transmitting the word-of-mouth data input by the user input means to a server via a network;

[1220] an analysis means for receiving the word-of-mouth data and extracting positive and negative information using natural language processing technology;

[1221] A generation means for generating a reply message using a generation AI model based on the extracted information;

[1222] a reply sending means for sending the generated reply message to a user terminal via a network;

[1223] means for displaying the reply message on a user's terminal;

[1224] A system including:

[1225] (Claim 2)

[1226] 2. The system according to claim 1, wherein the analyzing means extracts positive and negative information by tokenizing and grammatically analyzing the review text.

[1227] (Claim 3)

[1228] 2. The system according to claim 1, wherein the generating means generates a prompt sentence and uses the prompt sentence to generate a reply sentence using a generative AI model.

[1229] "Application Example 1"

[1230] (Claim 1)

[1231] a user input means;

[1232] a transmitting means for converting the voice data input by the user input means into text data and transmitting the text data to a server;

[1233] an analysis means for analyzing the text data to extract positive information and negative information;

[1234] a generating means for generating a reply message based on the extracted information;

[1235] The system further includes a display means for transmitting and displaying the generated reply message to a user.

[1236] (Claim 2)

[1237] 2. The system according to claim 1, wherein the analyzing means analyzes the text data and performs sentiment analysis by using natural language processing technology.

[1238] (Claim 3)

[1239] The system of claim 1, wherein the generating means generates a reply sentence based on a prompt sentence using a generative AI model.

[1240] "Example 2: Combining Emotion Engines"

[1241] (Claim 1)

[1242] A means for a user to input a review using a terminal;

[1243] means for transmitting word-of-mouth data from the terminal to a server;

[1244] a means for the server to receive word-of-mouth data and pre-process the data;

[1245] means for analyzing the word-of-mouth data and extracting positive and negative information;

[1246] means for identifying emotions based on the extracted information using an emotion engine;

[1247] A means for generating a reply using a generative AI model taking into account the identified emotion;

[1248] means for transmitting the generated reply message to a terminal;

[1249] means for displaying a reply message to a user on the terminal;

[1250] A system including:

[1251] (Claim 2)

[1252] 2. The system according to claim 1, wherein the analyzing means analyzes the word-of-mouth data by using a natural language processing technique.

[1253] (Claim 3)

[1254] The system according to claim 1, wherein the generating means generates the reply message using machine learning technology.

[1255] "Application example 2 when combining emotion engines"

[1256] Rewritten claims

[1257] (Claim 1)

[1258] a user input means;

[1259] a transmission means for transmitting the word-of-mouth data input by the user input means to a server;

[1260] an analysis means for analyzing the word-of-mouth data to extract positive information and negative information;

[1261] an emotion engine that identifies emotions based on the analyzed data;

[1262] a generation means for generating a reply message based on the extracted information and emotion score;

[1263] a reply sending means for sending the generated reply message to a user;

[1264] A system including:

[1265] (Claim 2)

[1266] the analysis means analyzes the word-of-mouth data by using natural language processing technology,

[1267] The system according to claim 1, wherein the generating means generates a reply message using a generative AI model.

[1268] (Claim 3)

[1269] 2. The system of claim 1, wherein the generating means generates a reply sentence using a prompt sentence based on the emotion score. [Explanation of symbols]

[1270] 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 user input means; a transmission means for transmitting the word-of-mouth data input by the user input means to a server; an analysis means for analyzing the word-of-mouth data to extract positive information and negative information; a generating means for generating a reply message based on the extracted information; a reply sending means for sending the generated reply message to a user; A system including:

2. 2. The system according to claim 1, wherein the analyzing means analyzes the word-of-mouth data by using a natural language processing technique.

3. The system according to claim 1 , wherein the generating means generates a reply sentence using a machine learning model.

Images