system

Abstract

Traditional methods often result in long response times to customer inquiries, leading to decreased customer satisfaction. Furthermore, it reduces the productivity of sales representatives and places a heavy load on product servers and systems. This system addresses these issues by providing a rapid and accurate customer response solution. [Solution] The system receives input from the user, tokenizes the user's input, passes the tokenized data to a generative AI model to generate a response, converts the generated response into JSON format, and sends the converted response back to the user.

Description

Technical Field

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

Background Art

[0002] Patent Document 1 discloses a persona chatbot control method performed by at least one processor, including steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a 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

Summary of the Invention

Problems to be Solved by the Invention

[0004] In response to inquiries from customers, conventional methods often take a long time to answer, resulting in a problem of reduced customer satisfaction. In addition, the productivity of sales staff also decreases, and there is also a problem of burden on the product server and system. There is a need for a method to quickly and accurately respond to customers to address these problems.

Means for Solving the Problems

[0005] The present invention solves the above problems by providing a system that receives input from a user, tokenizes that input, passes the tokenized data to a generative AI model to generate a response, converts that response to JSON format, and sends it back to the user. Furthermore, this system uses a large-scale language model as the generative AI model and includes means for displaying user input in a chat window, thereby achieving rapid and accurate customer service.

[0006] A "user" is an individual or group that uses the system to make an inquiry.

[0007] "Input" refers to the text data of questions and requests that users provide to the system.

[0008] "Tokenization" is the process of dividing input text data and breaking it down into individual words and phrases.

[0009] A "generative AI model" is an artificial intelligence algorithm that generates a response based on given input data, and large-scale language models fall into this category.

[0010] "Response" refers to the text data output generated by a generative AI model.

[0011] "JSON format" is an abbreviation for JavaScript (registered trademark) Object Notation, and is a format for structuring, storing, and transferring data.

[0012] A "chat window" is an interface that allows users and the system to communicate in real time using text. [Brief explanation of the drawing]

[0013] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2]It is a conceptual diagram showing an example of the main functions of a data processing device and a smart device according to the first embodiment. [Figure 3] It is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] It is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] It is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] It is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] It is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] It is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. <000007​​​​​​​​​​​​​​​​​​​​​​​​​​​​First, the terms used in the following description will be explained.

[0016] In the following embodiments, the labeled processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), and the like.

[0017] In the following embodiments, the labeled RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0018] In the following embodiments, the labeled storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0019] In the following embodiments, the labeled communication I / F (Interface) is an interface including a communication processor and 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), or Bluetooth (registered trademark), etc.

[0020] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0021] [First Embodiment]

[0022] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0023] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0024] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0025] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0026] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0027] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0028] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0029] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0030] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0031] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0032] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0033] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0034] As an embodiment of the present invention, the process of constructing a chat system using a generative AI model will be described. This system significantly improves the efficiency of customer service by providing quick and accurate responses to questions entered by users.

[0035] System Configuration

[0036] This system consists of the following main components:

[0037] 1. User interface for receiving inquiries

[0038] 2. Terminals that receive and process data from the user interface.

[0039] 3. Servers that perform data processing and response generation.

[0040] 4. Communication means for sending the generated response back to the user.

[0041] Specific steps

[0042] 1. User input:

[0043] A user enters a question in the chat window such as, "How does KOALA work?"

[0044] 2. Sending a request from the terminal to the server:

[0045] The terminal converts the user's input into JSON format and sends a POST request to the server. This request contains the user's question data.

[0046] 3. Server-side processing:

[0047] The server receives the request and tokenizes the user's input. The tokenized data is passed to a generative AI model (e.g., a large-scale language model) to generate an appropriate response.

[0048] 4. Response generation and transformation:

[0049] Convert the generated response text to JSON format. The converted data will have the following structure: "response": "KOALA processes the data in real time using advanced analytical algorithms."

[0050] 5. Sending a response from the server to the terminal:

[0051] The server sends this JSON data back to the device.

[0052] 6. Displaying responses to the user:

[0053] The terminal displays the response data received from the server in the chat window. The user can review this and continue asking further questions.

[0054] Specific example

[0055] If a user enters the question "What are the main functions of KOALA?", it will behave as follows:

[0056] 1. The user enters the question into the device.

[0057] 2. The terminal sends the question data to the server.

[0058] 3. The server processes the data and generates a response using a generative AI model.

[0059] 4. The server converts the generated response into JSON format and sends it to the terminal.

[0060] 5. The device displays the response in the chat window.

[0061] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[0062] The implementation of this system will dramatically improve the response speed to user inquiries, leading to increased productivity for sales representatives and reduced workload on the product. Furthermore, faster response times will improve customer satisfaction.

[0063] The following describes the processing flow.

[0064] Step 1:

[0065] The user types a question into the chat window. For example, they might type a specific question like, "How does KOALA work?"

[0066] Step 2:

[0067] The terminal receives user input and converts this input data into JSON format. The converted data has the following JSON structure:

[0068] json

[0069] {

[0070] "user_input": "How does KOALA work?"

[0071] }

[0072] Step 3:

[0073] The device sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[0074] Step 4:

[0075] The server processes the POST request received from the terminal. At this time, the / chat endpoint of the Flask application receives the request.

[0076] Step 5:

[0077] The server extracts user input (user_input) from the received request data. Specifically, it extracts the data as follows:

[0078] python

[0079] data = request.json

[0080] user_input = data['user_input']

[0081] Step 6:

[0082] The server tokenizes the extracted user input. For this purpose, it uses a tokenizer from a pre-loaded generative AI model.

[0083] python

[0084] inputs = tokenizer.encode(user_input, return_tensors='pt')

[0085] Step 7:

[0086] The server passes the tokenized data to a generative AI model (large-scale language model) to generate response text.

[0087] python

[0088] outputs = model.generate(inputs, max_length=200, num_return_sequences=1)

[0089] response = tokenizer.decode(outputs[0], skip_special_tokens=True)

[0090] Step 8:

[0091] The server converts the generated response into JSON format. The converted JSON data has the following structure:

[0092] json

[0093] {

[0094] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[0095] }

[0096] Step 9:

[0097] The server sends the generated response data back to the terminal. This response data is sent again in JSON format.

[0098] Step 10:

[0099] The terminal analyzes the response data received from the server and displays it in the chat window. The user reviews this response and, if they have further questions, enter them again in the chat window.

[0100] Through these steps, the system provides fast and accurate responses to user inquiries.

[0101] (Example 1)

[0102] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0103] In today's digital transformation, there is a demand for systems that can respond quickly and accurately to user inquiries. Conventional systems often process inquiries manually, which is time-consuming and labor-intensive. Furthermore, the quality of responses can vary depending on the person handling them, leading to decreased customer satisfaction. Therefore, this invention aims to solve these problems by providing a system that uses a generative AI model to deliver consistent, high-quality responses quickly.

[0104] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0105] In this invention, the server includes means for tokenizing user input and creating a prompt, means for passing the tokenized data to a generation AI model to generate a response, and means for converting the generated response into JSON format. This enables a rapid and accurate response to user inquiries.

[0106] A "user" refers to a person who uses the system to make an inquiry.

[0107] "Means for receiving input" refers to a device or software that has the function of receiving inquiries from users.

[0108] "JSON format" is an abbreviation for JavaScript Object Notation, and refers to a format that structures data using key-value pairs and represents it in an easy-to-read format.

[0109] "Tokenization" refers to the process of dividing input text into smaller units such as words and phrases.

[0110] A "prompt" refers to the initial conditions or question-style sentences that a generative AI model uses to receive input.

[0111] A "generative AI model" refers to a model that uses machine learning techniques to generate responses based on user input.

[0112] A "large-scale language model" refers to a model that is trained using a large amount of text data to understand and generate complex language patterns.

[0113] "Means for generating responses" refers to devices or software that create appropriate responses to user inquiries based on tokenized data.

[0114] "Means for sending back a response" refers to a device or software that has the function of sending back the response generated by the server to the user.

[0115] A "chat window" refers to an interface for users to input and display text-based inquiries and responses.

[0116] As an embodiment of the present invention, a method for constructing a chat system using a generative AI model will be described. This system significantly improves the efficiency of customer service by providing quick and accurate responses to questions entered by users.

[0117] System Configuration

[0118] This system consists of the following main components:

[0119] 1. User Interface: Receives inquiries from users.

[0120] 2. Terminal: Receives data from the user interface and sends it to the server.

[0121] 3. Server: Performs data processing and response generation.

[0122] 4. Communication method: The generated response is sent back to the user.

[0123] Hardware and software to be used

[0124] Hardware: Computers and smartphones used by users, and server machines on which servers are installed.

[0125] Software: Web pages and applications with user interfaces, software for sending requests used by devices, generative AI models running on servers (e.g., GPT-3®), and Python libraries used for data processing (e.g., NLTK, spaCy, etc.).

[0126] System Operation Description

[0127] The system operates as follows:

[0128] 1. User input: The user enters their question into the chat window. In this example, they type, "How does KOALA work?"

[0129] 2. Terminal generation and transmission of requests: The terminal converts the user's input into JSON format and sends it to the server as a POST request.

[0130] 3. Receiving and processing requests by the server: The server receives the request and tokenizes the user's input. This process uses natural language processing libraries such as NLTK (Natural Language Toolkit).

[0131] 4. Response generation by generative AI model: The server inputs tokenized data into the generative AI model to generate an appropriate response. In this example, the response "KOALA processes data in real time using advanced analytical algorithms" is obtained.

[0132] 5. Server-side response conversion and transmission: The generated response is converted to JSON format and sent back to the terminal.

[0133] 6. Displaying responses via the terminal: The terminal displays the response received from the server in the chat window. This allows the user to see the generated response.

[0134] Specific example

[0135] For example, if a user enters the question "What are the main functions of KOALA?", the system will behave as follows:

[0136] 1. User input: The user types "What are the main functions of KOALA?" into the chat window.

[0137] 2. Terminal generation and sending of requests: The terminal converts the question into JSON format and sends it to the server.

[0138] 3. Server receives and processes the request: The server receives the query and performs tokenization.

[0139] 4. Response generation using a generative AI model: The server uses a generative AI model to generate a response that says, "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[0140] 5. Server conversion and transmission of response: The server converts the generated response into JSON format and sends it to the terminal.

[0141] 6. Display of responses by the device: The device displays the response in the chat window.

[0142] Example of a prompt

[0143] "What are the main functions of KOALA?"

[0144] This system enables quick and accurate responses to user inquiries, leading to increased productivity for sales representatives and improved customer satisfaction.

[0145] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0146] Step 1:

[0147] User input

[0148] The user enters a question into the chat window. In this example, they type, "How does KOALA work?". The user directly types the text on the keyboard, and it appears in the chat window. This input is then processed.

[0149] Specific operation: The user accesses a chat window in a web browser or application and types their inquiry using the keyboard. The user's input is displayed in the chat window in real time.

[0150] Step 2:

[0151] Request generation by the terminal

[0152] The terminal converts the user's input into JSON format and sends a POST request to the server. The input is the user's question text, "How does KOALA work?". Converting this to JSON format results in the following:

[0153] json

[0154] {

[0155] "Question": "How does KOALA work?"

[0156] }

[0157] Specific operation: Use JavaScript's XMLHttpRequest and fetch APIs to convert user input data into JSON format and send a POST request to the server.

[0158] Step 3:

[0159] Server receives request

[0160] The server receives a request in JSON format sent from the terminal. The input is data in JSON format.

[0161] json

[0162] {

[0163] "Question": "How does KOALA work?"

[0164] }

[0165] The server receives this data for analysis.

[0166] Specific operation: The server uses a web framework such as Flask or Django in Python to receive POST requests from the terminal and parse their contents.

[0167] Step 4:

[0168] Execute tokenization process

[0169] The server tokenizes the user's question received. The input is the user's question text. Tokenization divides this into units such as words and phrases. Natural language processing libraries such as NLTK and spaCy are used for this process. For example, it might be divided into "KOALA", "is", "how", "works", "of", "is it?".

[0170] Specific operation: The server uses libraries such as NLTK and spaCy to tokenize the user's input text.

[0171] Step 5:

[0172] Response generation using generative AI models

[0173] The server inputs tokenized data into a generative AI model, which then generates an appropriate response. The input is tokenized data. Based on this, the generative AI model (e.g., GPT-3) generates a response such as, "KOALA uses advanced analytical algorithms to process the data in real time."

[0174] Specific operation: The server uses the OpenAI® API and other tools to send prompts to the generated AI model, causing it to generate responses.

[0175] Step 6:

[0176] Response transformation

[0177] The server converts the generated response into JSON format. The input is the generated response text. When converted to JSON format, it looks like this:

[0178] json

[0179] {

[0180] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[0181] }

[0182] Specific operation: Use Python's standard library or other JSON libraries to convert the generated response into JSON format.

[0183] Step 7:

[0184] Sending a response

[0185] The server returns the converted JSON data to the terminal. The input is a response data in JSON format. This is then sent to the terminal.

[0186] Specific operation: The server uses features from Python's Flask or Django to return JSON data as an HTTP response.

[0187] Step 8:

[0188] Display of response

[0189] The terminal displays the response data received from the server in the chat window. The input is response data in JSON format.

[0190] json

[0191] {

[0192] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[0193] }

[0194] The device analyzes this information and displays it in the chat window in a format that is easy for the user to understand.

[0195] Specific operation: Using JavaScript, the response text is extracted from the received JSON data and inserted into an HTML element for display.

[0196] (Application Example 1)

[0197] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0198] In food delivery services, responding quickly and accurately to customer inquiries directly leads to improved customer satisfaction and operational efficiency. However, traditional systems often suffer from delayed responses or inappropriate answers depending on the nature of the inquiry. Furthermore, providing real-time information on delivery progress and menu item details is difficult. This results in increased customer dissatisfaction and a decline in service quality.

[0199] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0200] In this invention, the server includes means for receiving input from a user, means for tokenizing the user's input, means for passing the tokenized data to a generative AI model to generate a response, means for converting the generated response into JSON format, means for returning the response to the user, means for generating a response specific to food delivery services, and means for providing real-time information on the delivery progress and menu item details of food deliveries. This makes it possible to respond quickly and accurately to user inquiries and to provide real-time information on delivery progress and menu details.

[0201] "Means of receiving user input" refers to providing an interface for users of a food delivery service to enter inquiries or questions using a smartphone or other device.

[0202] "Methods for tokenizing user input" refer to processes that analyze the text entered by the user and divide it into words and phrases.

[0203] "Means of passing tokenized data to a generative AI model to generate a response" refers to the process by which a generative AI model generates an appropriate response based on tokenized input.

[0204] "Means for converting the generated response to JSON format" refers to the process of converting the generated text response into JSON, a common data exchange format.

[0205] "Means of sending the response back to the user" refers to sending the converted response data to the user's device so that the user can confirm the response.

[0206] "Means for generating responses specifically for food delivery services" refers to processing that provides professional and appropriate responses to questions and inquiries related to food delivery.

[0207] "A means of providing real-time responses regarding the progress of food delivery and details of menu items" refers to a system that immediately provides users with real-time information regarding the current progress of delivery and specific menu details.

[0208] Modes for carrying out the invention

[0209] Program processing

[0210] The system that embodies this application is a chat support system specifically designed for food delivery services, providing quick and accurate responses to user inquiries through a smartphone application.

[0211] Hardware and software to be used

[0212] Hardware: Cloud servers or on-premises servers

[0213] Software: Flask (Web framework), OpenAI API

[0214] Data processing and data calculation

[0215] 1. Receiving user input:

[0216] The server receives questions entered by users through a smartphone application. Users initiate an inquiry by typing their questions into the application's chat window.

[0217] 2. Tokenization of user input:

[0218] The server tokenizes the user's input. Tokenization is the process of analyzing the user's text and dividing it into words and phrases.

[0219] 3. Pass the tokenized data to a generative AI model to generate a response:

[0220] The tokenized data is input into a generative AI model (e.g., GPT-3) to generate an appropriate response. The generative AI model generates the most appropriate text response to the user's question.

[0221] 4. Converting the generated response to JSON format:

[0222] The generated response is converted to JSON format. JSON is a common format for data exchange and allows for proper formatting of the response.

[0223] 5. Sending the response back to the user:

[0224] The server sends the generated JSON response back to the user's smartphone application. The user can then view the response in the application's chat window.

[0225] 6. Generating responses specifically for food delivery services:

[0226] The generative AI model has the capability to generate responses specifically tailored to food delivery services. It quickly provides specific information to users in response to their questions, such as delivery progress and details of menu items.

[0227] 7. Real-time responses regarding delivery progress and menu item details:

[0228] The server provides real-time information on the progress of food deliveries and details of menu items. This allows users to quickly obtain information about the current delivery status and menu items.

[0229] Adding specific examples

[0230] As a concrete example, let's explain the behavior when a user enters the question, "How long will it take for my ordered pizza to arrive?"

[0231] 1. The user enters their question into the chat window of the smartphone application.

[0232] 2. The server receives this question and tokenizes the input text.

[0233] 3. Pass the tokenized data to the generative AI model (GPT-3) and use the following prompts to generate a response:

[0234] As a support bot for a food delivery service, answer user inquiries.

[0235] User: How long will it take for my pizza to arrive?

[0236] Bot:

[0237] 4. The generative AI model generates the response, "Pizza delivery usually takes 30-40 minutes, but the current delivery time can be specifically tracked based on GPS."

[0238] 5. Convert this response to JSON format and send it back to the user's smartphone application.

[0239] 6. Users can check the response on the application and obtain information about the delivery status.

[0240] In this application, users can receive quick and accurate answers to their inquiries, and real-time information on delivery progress and details of menu items can also be provided.

[0241] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0242] Program processing steps

[0243] Step 1:

[0244] The user enters their question into the chat window of the smartphone application. When the user types a message such as "How long will it take for my ordered pizza to arrive?" and presses the send button, the input data is sent to the server. At this stage, the input is the user's question text.

[0245] Step 2:

[0246] The server receives input text from the user and tokenizes it. Specifically, it divides the input text into words and phrases, converting the data into a format that is easy to analyze. In this example, the question text received as input, "How long will it take for my ordered pizza to arrive?", is output as tokenized data.

[0247] Step 3:

[0248] The server passes the tokenized data to a generative AI model, which then generates a response. The generative AI model (e.g., GPT-3) generates an appropriate response according to the format of the prompt. A concrete example of a prompt is as follows:

[0249] As a support bot for a food delivery service, answer user inquiries.

[0250] User: How long will it take for my pizza to arrive?

[0251] Bot:

[0252] Given tokenized data as input, a generative AI model generates a response such as, "Pizza delivery usually takes 30-40 minutes, but the current delivery time can be specifically tracked based on GPS." This is the output data.

[0253] Step 4:

[0254] The server converts the generated response into JSON format. This conversion process transforms the data into the standard JSON format. The generated response text is converted into JSON data like the following:

[0255] json

[0256] {

[0257] "Response": "Pizza delivery usually takes 30-40 minutes, but current delivery times can be tracked specifically based on GPS."

[0258] }

[0259] It receives the generated response text as input and outputs data in JSON format.

[0260] Step 5:

[0261] The server sends this JSON data back to the user's device. The device (smartphone application) receives this data and processes it to display it in the chat window. The output is receiving JSON data as input and displaying the response in the user's chat window.

[0262] Step 6:

[0263] The user checks the response displayed in the chat window of the smartphone application. This response allows the user to receive real-time information about the delivery progress and details of the food delivery menu. Specifically, the user reads and understands the displayed response. The input in this step is the response text displayed in the application, and the output is the user's understanding and satisfaction.

[0264] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0265] This invention relates to a chat system that utilizes a generative AI model combined with an emotion engine that recognizes user emotions. This system significantly improves the efficiency and satisfaction of customer service by providing quick and emotion-appropriate responses to user inquiries.

[0266] System Configuration

[0267] This system consists of the following main components:

[0268] 1. User interface for receiving inquiries

[0269] 2. Terminals that receive and process data from the user interface.

[0270] 3. Servers that perform data processing and response generation.

[0271] 4. Means of analyzing user emotions using an emotion engine

[0272] 5. Communication means for returning the generated response to the user.

[0273] Specific steps

[0274] 1. User input:

[0275] A user enters a question in the chat window such as, "How does KOALA work?"

[0276] 2. Sending a request from the terminal to the server:

[0277] The terminal converts the user's input into JSON format and sends a POST request to the server. This request contains the user's question data.

[0278] 3. Server-side processing:

[0279] The server receives the request and tokenizes the user's input. The tokenized data is passed to a generative AI model (e.g., a large-scale language model) to generate an appropriate response.

[0280] 4. Analysis using an emotion engine:

[0281] Before the server generates a response using a generative AI model, it passes the user's input to an emotion engine for analysis. The emotion engine analyzes the user's emotional state in real time and incorporates the analysis results into the response generation.

[0282] 5. Response generation and transformation:

[0283] Based on the analysis results of the sentiment engine, the generative AI model generates a response text. The generated response text is converted into JSON format. The converted data has the following structure: "response": "KOALA uses an advanced analysis algorithm to process data in real time."

[0284] 6. Response transmission from the server to the terminal:

[0285] The server returns this JSON data to the terminal.

[0286] 7. Display of the response to the user:

[0287] The terminal displays the response data received from the server in the chat window. The user checks this response and, if they have further questions, enters them again in the chat window.

[0288] Specific example

[0289] When the user enters the question "What are the main functions of KOALA?", it operates as follows:

[0290] 1. The user enters a question into the terminal.

[0291] 2. The terminal sends the question data to the server.

[0292] 3. The server processes the data and analyzes the user's sentiment with the sentiment engine before using the generative AI model.

[0293] 4. The sentiment engine analyzes the user's sentiment state, and based on the results, the generative AI model generates a response.

[0294] 5. The server converts the generated response into JSON format and sends it to the terminal.

[0295] 6. The terminal displays the response in the chat window.

[0296] At this time, the following response is displayed to the user: "The main functions of KOALA are data analysis, real-time processing, and advanced visualization."

[0297] By introducing this system, the response to the user's inquiry will be emotion-responsive, and an improvement in customer satisfaction can be expected.

[0298] The following describes the processing flow.

[0299] Step 1:

[0300] The user enters a question in the chat window. Enter a specific question such as "What are the main functions of KOALA?"

[0301] Step 2:

[0302] The terminal receives the user's input and converts this input data into JSON format. The converted data has the following JSON structure:

[0303] json

[0304] {

[0305] "user_input": "What are the main functions of KOALA?"

[0306] }

[0307] [[ID=I45]]Step 3:

[0308] The terminal sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[0309] Step 4:

[0310] The server processes the POST request received from the terminal. At this time, the / chat endpoint of the Flask application receives the request.

[0311] Step 5:

[0312] The server extracts user input (user_input) from the received request data. Specifically, it extracts the data as follows:

[0313] python

[0314] data = request.json

[0315] user_input = data['user_input']

[0316] Step 6:

[0317] The server passes the extracted user input to the emotion engine for sentiment analysis. The sentiment engine analyzes the user's emotions as follows:

[0318] python

[0319] emotion = emotion_engine.analyze(user_input)

[0320] The emotion engine returns the analysis results. For example, it may determine that the user's input contains emotions such as "joy" or "anxiety."

[0321] Step 7:

[0322] The server performs a tokenization process based on the analysis results of the emotion engine. The tokenization process proceeds as follows:

[0323] python

[0324] inputs = tokenizer.encode(user_input, return_tensors='pt')

[0325] Step 8:

[0326] The server passes the tokenized data and sentiment analysis results to a generative AI model (large-scale language model) to generate response text. Specifically, the process is as follows:

[0327] python

[0328] outputs = model.generate(inputs, max_length=200, num_return_sequences=1, emotion=emotion)

[0329] response = tokenizer.decode(outputs[0], skip_special_tokens=True)

[0330] Step 9:

[0331] The server converts the generated response into JSON format. The converted JSON data has the following structure:

[0332] json

[0333] {

[0334] "Response": "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[0335] }

[0336] Step 10:

[0337] The server sends this JSON data back to the device.

[0338] Step 11:

[0339] The terminal analyzes the response data received from the server and displays it in the chat window. The user can review this response and, if they have further questions, enter them again in the chat window.

[0340] Through the steps described above, the system provides fast and emotionally responsive responses to user inquiries. This enables responses that take user emotions into consideration, which is expected to improve customer satisfaction.

[0341] (Example 2)

[0342] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0343] Traditional chat systems had the problem of lowering user satisfaction because they provided uniform responses without considering user emotions. Furthermore, the inconsistent quality of responses made it difficult to provide satisfactory support.

[0344] The specific processing performed by the specific 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 tokenizing user input, means for passing the tokenized data to sentiment analysis means for analysis, and means for passing the tokenized data to a generation AI model based on the sentiment analysis results to generate a response. This makes it possible to provide a response that corresponds to the user's emotions.

[0345] "User input" refers to the text or questions that users type into the chat window.

[0346] "Tokenization" refers to the process of dividing a text into units such as words or sentences.

[0347] "Sentiment analysis method" refers to a method that analyzes emotions from user input text and outputs the results.

[0348] A "generative AI model" refers to an algorithm that uses artificial intelligence to generate responses to user input.

[0349] "Response generation" refers to the process of generating an appropriate response based on user input.

[0350] "JSON format" is an abbreviation for JavaScript Object Notation, and refers to a format for representing data in a structured way.

[0351] This invention relates to a chat system that utilizes a generative AI model combined with an emotion engine that recognizes user emotions. This system significantly improves the efficiency and satisfaction of customer service by providing quick and emotion-appropriate responses to user inquiries.

[0352] System Configuration

[0353] This system consists of the following main components:

[0354] 1. User interface for receiving user input

[0355] 2. Terminals that receive and process data from the user interface.

[0356] 3. Servers that perform data processing and response generation.

[0357] 4. Means of analyzing user emotions using an emotion engine

[0358] 5. Communication means for returning the generated response to the user.

[0359] Program processing

[0360] The user accesses the system and enters a question or message into the chat window. For example, the user might type, "How does KOALA work?" The terminal converts this input into JSON format and sends a POST request to the server. This request contains the user's question data.

[0361] The server receives the request and tokenizes the user's input. The tokenized data is first passed to an emotion engine to analyze the user's emotional state. The emotion engine can use, for example, Microsoft® Azure® Text Analytics API. Based on the analysis results, the tokenized data is passed to a generative AI model (for example, a large-scale language model such as GPT-3) to generate an appropriate response.

[0362] The generated response text is converted back into JSON format and sent to the terminal. The terminal parses the received data and displays the response in the chat window. This allows the user to receive responses that are appropriate to their emotions.

[0363] Specific example

[0364] For example, if a user enters the question "What are the main functions of KOALA?", the system will behave as follows:

[0365] 1. The user enters the question into the device.

[0366] 2. The terminal sends the question data to the server.

[0367] 3. The server processes the data and analyzes the user's emotions using an emotion engine before using generative AI models.

[0368] 4. The emotion engine analyzes the user's emotional state, and based on the results, a generative AI model generates a response.

[0369] 5. The server converts the generated response into JSON format and sends it to the terminal.

[0370] 6. The device displays the response in the chat window.

[0371] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[0372] The introduction of this system will enable responses to user inquiries to become more emotionally responsive, which is expected to improve customer satisfaction.

[0373] Example of a prompt:

[0374] "The user is asking: 'How does KOALA work?' Its emotional state is neutral. Please provide an appropriate response to this."

[0375] This invention enables the generation of sophisticated responses that take emotions into consideration, and is expected to significantly improve the user experience.

[0376] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0377] Step 1:

[0378] The user enters a question into the chat window. This input is natural language text, such as "How does KOALA work?".

[0379] Input: User's question text

[0380] Output: Input completion event in the chat window

[0381] Step 2:

[0382] The terminal receives user input and converts it to JSON format. The JSON data will be in the following format:

[0383] json

[0384] {

[0385] "user_input": "How does KOALA work?"

[0386] }

[0387] Input: User's question text

[0388] Output: Question data in JSON format

[0389] Step 3:

[0390] The terminal sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[0391] Input: Question data in JSON format

[0392] Output: POST request to the server

[0393] Step 4:

[0394] The server receives the POST request, parses the JSON data, and extracts the user input portion.

[0395] Input: JSON data sent to the server

[0396] Output: Extracted user input text

[0397] Step 5:

[0398] The server tokenizes the extracted input text. For example, it might be split into "KOALA", "is", "how", "works", "is", and "is it?".

[0399] Input: User-input text

[0400] Output: Tokenized data

[0401] Step 6:

[0402] The server passes the tokenized data to the sentiment engine for sentiment analysis. The sentiment engine uses Microsoft Azure's Text Analytics API, among others. For example, it might analyze the sentiment as "Sentimental state: Neutral."

[0403] Input: Tokenized user input data

[0404] Output: Emotion analysis results

[0405] Step 7:

[0406] The server passes the tokenized data, based on the sentiment analysis results, to a generative AI model to generate a response. The generative AI model uses large-scale language models such as GPT-3. Here, a response such as "KOALA uses advanced analytical algorithms to process data in real time" is generated.

[0407] Input: Tokenized data and sentiment analysis results

[0408] Output: Generated response text

[0409] Step 8:

[0410] The server converts the generated response text into JSON format. The resulting data will look like this:

[0411] json

[0412] {

[0413] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[0414] }

[0415] Input: Generated response text

[0416] Output: Response data in JSON format

[0417] Step 9:

[0418] The server sends response data in JSON format to the terminal.

[0419] Input: Response data in JSON format

[0420] Output: Sending a response to the terminal

[0421] Step 10:

[0422] The terminal parses the JSON data received from the server and extracts the response text. The extracted response text is then displayed in the chat window.

[0423] Input: JSON data sent from the server

[0424] Output: Response text displayed in the chat window

[0425] This allows users to see appropriate responses tailored to their emotions in real time.

[0426] (Application Example 2)

[0427] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0428] Traditional chat systems have a problem in that they cannot flexibly change responses according to the user's emotions and situation, and therefore cannot sufficiently increase user satisfaction. In particular, e-commerce sites receive a wide range of customer inquiries, and quick and accurate responses are required, so providing responses that respond to emotions is effective. However, current systems do not have emotion analysis functions and can only return a uniform response.

[0429] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0430] In this invention, the server includes means for receiving user input, means for analyzing the user's emotions using an emotion analysis engine, means for generating a response using a generative AI model based on the analysis results, means for converting the generated response into JSON format, and means for returning the response to the user. This makes it possible to provide a response that is appropriate to the user's emotions.

[0431] "Means of receiving user input" refers to a function that receives messages and inquiries sent by users through an interface such as a chat window.

[0432] "Methods for tokenizing user input" refer to the process of dividing user-entered text into meaningful units to facilitate natural language processing.

[0433] "A means of passing tokenized data to a generative AI model to generate a response" refers to a function that inputs tokenized data into a generative AI model and generates an appropriate response for the user.

[0434] "Means for converting generated responses into JSON format" refers to a function that converts text responses generated by generative AI models into JSON format, facilitating data communication and display.

[0435] "Means of sending a response back to the user" refers to a function that sends a generated response from the server to the user's device and displays it to the user.

[0436] "Means of analyzing a user's emotions using an emotion analysis engine" refers to an engine and its functions for analyzing and understanding a user's emotional state from their text input.

[0437] "Means for adjusting responses based on analysis results" refers to a function that adjusts the content of responses generated by generative AI models based on the analysis results obtained by the emotion analysis engine.

[0438] This invention relates to a chat system for e-commerce sites that combines an emotion recognition engine and a generative AI model. This system enables the rapid and appropriate provision of responses tailored to the user's emotions. The specific configuration and processing methods of this system are described below.

[0439] System Configuration

[0440] This system consists of the following main components:

[0441] 1. Means for receiving user input:

[0442] The user enters text into the chat window via a smartphone app.

[0443] 2. Means of tokenizing user input:

[0444] Divide the text into meaningful units and convert them into a format that is easy to process.

[0445] 3. Means for analyzing user emotions using an emotion analysis engine:

[0446] The system analyzes the emotional state contained in the user's input. The software used includes the Hugging Face transformers library.

[0447] 4. Means of passing tokenized data to a generative AI model to generate a response:

[0448] Based on tokenized data and sentiment analysis results, a generative AI model generates an appropriate response. A large-scale language model is used here.

[0449] 5. Means for converting the generated response to JSON format:

[0450] The generated text response is converted to JSON format to improve communication efficiency.

[0451] 6. Means of returning the response to the user:

[0452] The server sends a response to the user's smartphone and displays it in the chat window.

[0453] Specific example

[0454] If a user enters the question "My order hasn't arrived yet. What's going on?" into the chat window of an online shopping site, the system will operate as follows:

[0455] 1. Receiving user input:

[0456] The device receives the text entered by the user.

[0457] 2. Emotion analysis:

[0458] The device passes the received text to an emotion analysis engine, which analyzes whether the user is expressing anxiety or anger.

[0459] 3. Response generation:

[0460] Based on the results of the emotion analysis, a generative AI model generates a response such as, "I apologize for the delay. I will check it immediately."

[0461] 4. Sending a response:

[0462] The generated response is converted to JSON format and sent from the server to the user's terminal.

[0463] 5. Display of response:

[0464] The user checks the response displayed in the chat window.

[0465] The software used includes the Hugging Face transformers library for emotion analysis and a generative AI model (large-scale language model). This enables appropriate responses based on the user's emotions, thereby improving customer satisfaction.

[0466] Example of a prompt:

[0467] "My order hasn't arrived yet. What's going on?"

[0468] This invention provides responses that take user emotions into consideration, significantly improving the quality of customer service on e-commerce sites.

[0469] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0470] Step 1:

[0471] The user types "My order hasn't arrived yet. What's going on?" into the chat window. This is the starting point for the entire system.

[0472] Step 2:

[0473] The terminal receives user input and records it as text data. This input is the user's text message, which serves as initial data for processing.

[0474] Step 3:

[0475] The terminal tokenizes the received text data. Tokenization is the process of dividing text into meaningful units (tokens). For example, it might be divided into "order," "but," "haven't," "arrived," "how," "is," "what," "is," "what," "is," "what," "is." This process makes the text data easier to analyze.

[0476] Step 4:

[0477] The tokenized text data is passed to an emotion analysis engine to analyze the user's emotions. The emotion analysis engine uses the Hugging Face transformers library to determine the user's emotional state (e.g., anxiety, anger). The output is the emotion classification result, with "anxiety" being output as an example.

[0478] Step 5:

[0479] Based on the sentiment analysis results, the server passes the tokenized data to a generative AI model to generate an appropriate response. The generative AI model uses a large-scale language model to generate a response such as, "I apologize for the delay. I will check on this immediately."

[0480] Step 6:

[0481] The generated response is converted to JSON format. The server converts the text response to JSON format in preparation for the next communication process. For example, it will be converted to a format like {"response": "Sorry for the wait. I will check it shortly."}.

[0482] Step 7:

[0483] The server sends data in JSON format to the terminal. The terminal receives the JSON data sent from the server and prepares for the next processing step.

[0484] Step 8:

[0485] The terminal parses the received JSON data and displays a response in the chat window. The user can then see the response displayed in the chat window, such as "Sorry for the wait. I will check it immediately," and ask further questions or take actions.

[0486] This series of processing steps allows users to receive appropriate responses that match their emotions, resulting in a high level of satisfaction.

[0487] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0488] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0489] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0490] [Second Embodiment]

[0491] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0492] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0493] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0494] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0495] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0496] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0497] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0498] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0499] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0500] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0501] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0502] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0503] As an embodiment of the present invention, the process of constructing a chat system using a generative AI model will be described. This system significantly improves the efficiency of customer service by providing quick and accurate responses to questions entered by users.

[0504] System Configuration

[0505] This system consists of the following main components:

[0506] 1. User interface for receiving inquiries

[0507] 2. Terminals that receive and process data from the user interface.

[0508] 3. Servers that perform data processing and response generation.

[0509] 4. Communication means for sending the generated response back to the user.

[0510] Specific steps

[0511] 1. User input:

[0512] A user enters a question in the chat window such as, "How does KOALA work?"

[0513] 2. Sending a request from the terminal to the server:

[0514] The terminal converts the user's input into JSON format and sends a POST request to the server. This request contains the user's question data.

[0515] 3. Server-side processing:

[0516] The server receives the request and tokenizes the user's input. The tokenized data is passed to a generative AI model (e.g., a large-scale language model) to generate an appropriate response.

[0517] 4. Response generation and transformation:

[0518] Convert the generated response text to JSON format. The converted data will have the following structure: "response": "KOALA processes the data in real time using advanced analytical algorithms."

[0519] 5. Sending a response from the server to the terminal:

[0520] The server sends this JSON data back to the device.

[0521] 6. Displaying responses to the user:

[0522] The terminal displays the response data received from the server in the chat window. The user can review this and continue asking further questions.

[0523] Specific example

[0524] If a user enters the question "What are the main functions of KOALA?", it will behave as follows:

[0525] 1. The user enters the question into the device.

[0526] 2. The terminal sends the question data to the server.

[0527] 3. The server processes the data and generates a response using a generative AI model.

[0528] 4. The server converts the generated response into JSON format and sends it to the terminal.

[0529] 5. The device displays the response in the chat window.

[0530] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[0531] The implementation of this system will dramatically improve the response speed to user inquiries, leading to increased productivity for sales representatives and reduced workload on the product. Furthermore, faster response times will improve customer satisfaction.

[0532] The following describes the processing flow.

[0533] Step 1:

[0534] The user types a question into the chat window. For example, they might type a specific question like, "How does KOALA work?"

[0535] Step 2:

[0536] The terminal receives user input and converts this input data into JSON format. The converted data has the following JSON structure:

[0537] json

[0538] {

[0539] "user_input": "How does KOALA work?"

[0540] }

[0541] Step 3:

[0542] The device sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[0543] Step 4:

[0544] The server processes the POST request received from the terminal. At this time, the / chat endpoint of the Flask application receives the request.

[0545] Step 5:

[0546] The server extracts user input (user_input) from the received request data. Specifically, it extracts the data as follows:

[0547] python

[0548] data = request.json

[0549] user_input = data['user_input']

[0550] Step 6:

[0551] The server tokenizes the extracted user input. For this purpose, it uses a tokenizer from a pre-loaded generative AI model.

[0552] python

[0553] inputs = tokenizer.encode(user_input, return_tensors='pt')

[0554] Step 7:

[0555] The server passes the tokenized data to a generative AI model (large-scale language model) to generate response text.

[0556] python

[0557] outputs = model.generate(inputs, max_length=200, num_return_sequences=1)

[0558] response = tokenizer.decode(outputs[0], skip_special_tokens=True)

[0559] Step 8:

[0560] The server converts the generated response into JSON format. The converted JSON data has the following structure:

[0561] json

[0562] {

[0563] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[0564] }

[0565] Step 9:

[0566] The server sends the generated response data back to the terminal. This response data is sent again in JSON format.

[0567] Step 10:

[0568] The terminal analyzes the response data received from the server and displays it in the chat window. The user reviews this response and, if they have further questions, enter them again in the chat window.

[0569] Through these steps, the system provides fast and accurate responses to user inquiries.

[0570] (Example 1)

[0571] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0572] In today's digital transformation, there is a demand for systems that can respond quickly and accurately to user inquiries. Conventional systems often process inquiries manually, which is time-consuming and labor-intensive. Furthermore, the quality of responses can vary depending on the person handling them, leading to decreased customer satisfaction. Therefore, this invention aims to solve these problems by providing a system that uses a generative AI model to deliver consistent, high-quality responses quickly.

[0573] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0574] In this invention, the server includes means for tokenizing user input and creating a prompt, means for passing the tokenized data to a generation AI model to generate a response, and means for converting the generated response into JSON format. This enables a rapid and accurate response to user inquiries.

[0575] A "user" refers to a person who uses the system to make an inquiry.

[0576] "Means for receiving input" refers to a device or software that has the function of receiving inquiries from users.

[0577] "JSON format" is an abbreviation for JavaScript Object Notation, and refers to a format that structures data using key-value pairs and represents it in an easy-to-read format.

[0578] "Tokenization" refers to the process of dividing input text into smaller units such as words and phrases.

[0579] A "prompt" refers to the initial conditions or question-style sentences that a generative AI model uses to receive input.

[0580] A "generative AI model" refers to a model that uses machine learning techniques to generate responses based on user input.

[0581] A "large-scale language model" refers to a model that is trained using a large amount of text data to understand and generate complex language patterns.

[0582] "Means for generating responses" refers to devices or software that create appropriate responses to user inquiries based on tokenized data.

[0583] "Means for sending back a response" refers to a device or software that has the function of sending back the response generated by the server to the user.

[0584] A "chat window" refers to an interface for users to input and display text-based inquiries and responses.

[0585] As an embodiment of the present invention, a method for constructing a chat system using a generative AI model will be described. This system significantly improves the efficiency of customer service by providing quick and accurate responses to questions entered by users.

[0586] System Configuration

[0587] This system consists of the following main components:

[0588] 1. User Interface: Receives inquiries from users.

[0589] 2. Terminal: Receives data from the user interface and sends it to the server.

[0590] 3. Server: Performs data processing and response generation.

[0591] 4. Communication method: The generated response is sent back to the user.

[0592] Hardware and software to be used

[0593] Hardware: Computers and smartphones used by users, and server machines on which servers are installed.

[0594] Software: Web pages and applications with user interfaces, software for sending requests used by devices, generative AI models running on servers (e.g., GPT-3), and Python libraries used for data processing (e.g., NLTK, spaCy, etc.).

[0595] System Operation Description

[0596] The system operates as follows:

[0597] 1. User input: The user enters their question into the chat window. In this example, they type, "How does KOALA work?"

[0598] 2. Terminal generation and transmission of requests: The terminal converts the user's input into JSON format and sends it to the server as a POST request.

[0599] 3. Receiving and processing requests by the server: The server receives the request and tokenizes the user's input. This process uses natural language processing libraries such as NLTK (Natural Language Toolkit).

[0600] 4. Response generation by generative AI model: The server inputs tokenized data into the generative AI model to generate an appropriate response. In this example, the response "KOALA processes data in real time using advanced analytical algorithms" is obtained.

[0601] 5. Server-side response conversion and transmission: The generated response is converted to JSON format and sent back to the terminal.

[0602] 6. Displaying responses via the terminal: The terminal displays the response received from the server in the chat window. This allows the user to see the generated response.

[0603] Specific example

[0604] For example, if a user enters the question "What are the main functions of KOALA?", the system will behave as follows:

[0605] 1. User input: The user types "What are the main functions of KOALA?" into the chat window.

[0606] 2. Terminal generation and sending of requests: The terminal converts the question into JSON format and sends it to the server.

[0607] 3. Server receives and processes the request: The server receives the query and performs tokenization.

[0608] 4. Response generation using a generative AI model: The server uses a generative AI model to generate a response that says, "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[0609] 5. Server conversion and transmission of response: The server converts the generated response into JSON format and sends it to the terminal.

[0610] 6. Display of responses by the device: The device displays the response in the chat window.

[0611] Example of a prompt

[0612] "What are the main functions of KOALA?"

[0613] This system enables quick and accurate responses to user inquiries, leading to increased productivity for sales representatives and improved customer satisfaction.

[0614] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0615] Step 1:

[0616] User input

[0617] The user enters a question into the chat window. In this example, they type, "How does KOALA work?". The user directly types the text on the keyboard, and it appears in the chat window. This input is then processed.

[0618] Specific operation: The user accesses a chat window in a web browser or application and types their inquiry using the keyboard. The user's input is displayed in the chat window in real time.

[0619] Step 2:

[0620] Request generation by the terminal

[0621] The terminal converts the user's input into JSON format and sends a POST request to the server. The input is the user's question text, "How does KOALA work?". Converting this to JSON format results in the following:

[0622] json

[0623] {

[0624] "Question": "How does KOALA work?"

[0625] }

[0626] Specific operation: Use JavaScript's XMLHttpRequest and fetch APIs to convert user input data into JSON format and send a POST request to the server.

[0627] Step 3:

[0628] Server receives request

[0629] The server receives a request in JSON format sent from the terminal. The input is data in JSON format.

[0630] json

[0631] {

[0632] "Question": "How does KOALA work?"

[0633] }

[0634] The server receives this data for analysis.

[0635] Specific operation: The server uses a web framework such as Flask or Django in Python to receive POST requests from the terminal and parse their contents.

[0636] Step 4:

[0637] Execute tokenization process

[0638] The server tokenizes the user's question received. The input is the user's question text. Tokenization divides this into units such as words and phrases. Natural language processing libraries such as NLTK and spaCy are used for this process. For example, it might be divided into "KOALA", "is", "how", "works", "of", "is it?".

[0639] Specific operation: The server uses libraries such as NLTK and spaCy to tokenize the user's input text.

[0640] Step 5:

[0641] Response generation using generative AI models

[0642] The server inputs tokenized data into a generative AI model, which then generates an appropriate response. The input is tokenized data. Based on this, the generative AI model (e.g., GPT-3) generates a response such as, "KOALA uses advanced analytical algorithms to process the data in real time."

[0643] Specific operation: The server uses the OpenAI API or similar tools to send prompts to the generated AI model, causing it to generate responses.

[0644] Step 6:

[0645] Response transformation

[0646] The server converts the generated response into JSON format. The input is the generated response text. When converted to JSON format, it looks like this:

[0647] json

[0648] {

[0649] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[0650] }

[0651] Specific operation: Use Python's standard library or other JSON libraries to convert the generated response into JSON format.

[0652] Step 7:

[0653] Sending a response

[0654] The server returns the converted JSON data to the terminal. The input is a response data in JSON format. This is then sent to the terminal.

[0655] Specific operation: The server uses features from Python's Flask or Django to return JSON data as an HTTP response.

[0656] Step 8:

[0657] Display of response

[0658] The terminal displays the response data received from the server in the chat window. The input is response data in JSON format.

[0659] json

[0660] {

[0661] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[0662] }

[0663] The device analyzes this information and displays it in the chat window in a format that is easy for the user to understand.

[0664] Specific operation: Using JavaScript, the response text is extracted from the received JSON data and inserted into an HTML element for display.

[0665] (Application Example 1)

[0666] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0667] In food delivery services, responding quickly and accurately to customer inquiries directly leads to improved customer satisfaction and operational efficiency. However, traditional systems often suffer from delayed responses or inappropriate answers depending on the nature of the inquiry. Furthermore, providing real-time information on delivery progress and menu item details is difficult. This results in increased customer dissatisfaction and a decline in service quality.

[0668] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0669] In this invention, the server includes means for receiving input from a user, means for tokenizing the user's input, means for passing the tokenized data to a generative AI model to generate a response, means for converting the generated response into JSON format, means for returning the response to the user, means for generating a response specific to food delivery services, and means for providing real-time information on the delivery progress and menu item details of food deliveries. This makes it possible to respond quickly and accurately to user inquiries and to provide real-time information on delivery progress and menu details.

[0670] "Means of receiving user input" refers to providing an interface for users of a food delivery service to enter inquiries or questions using a smartphone or other device.

[0671] "Methods for tokenizing user input" refer to processes that analyze the text entered by the user and divide it into words and phrases.

[0672] "Means of passing tokenized data to a generative AI model to generate a response" refers to the process by which a generative AI model generates an appropriate response based on tokenized input.

[0673] "Means for converting the generated response to JSON format" refers to the process of converting the generated text response into JSON, a common data exchange format.

[0674] "Means of sending the response back to the user" refers to sending the converted response data to the user's device so that the user can confirm the response.

[0675] "Means for generating responses specifically for food delivery services" refers to processing that provides professional and appropriate responses to questions and inquiries related to food delivery.

[0676] "A means of providing real-time responses regarding the progress of food delivery and details of menu items" refers to a system that immediately provides users with real-time information regarding the current progress of delivery and specific menu details.

[0677] Modes for carrying out the invention

[0678] Program processing

[0679] The system that embodies this application is a chat support system specifically designed for food delivery services, providing quick and accurate responses to user inquiries through a smartphone application.

[0680] Hardware and software to be used

[0681] Hardware: Cloud servers or on-premises servers

[0682] Software: Flask (Web framework), OpenAI API

[0683] Data processing and data calculation

[0684] 1. Receiving user input:

[0685] The server receives questions entered by users through a smartphone application. Users initiate an inquiry by typing their questions into the application's chat window.

[0686] 2. Tokenization of user input:

[0687] The server tokenizes the user's input. Tokenization is the process of analyzing the user's text and dividing it into words and phrases.

[0688] 3. Pass the tokenized data to a generative AI model to generate a response:

[0689] The tokenized data is input into a generative AI model (e.g., GPT-3) to generate an appropriate response. The generative AI model generates the most appropriate text response to the user's question.

[0690] 4. Converting the generated response to JSON format:

[0691] The generated response is converted to JSON format. JSON is a common format for data exchange and allows for proper formatting of the response.

[0692] 5. Sending the response back to the user:

[0693] The server sends the generated JSON response back to the user's smartphone application. The user can then view the response in the application's chat window.

[0694] 6. Generating responses specifically for food delivery services:

[0695] The generative AI model has the capability to generate responses specifically tailored to food delivery services. It quickly provides specific information to users in response to their questions, such as delivery progress and details of menu items.

[0696] 7. Real-time responses regarding delivery progress and menu item details:

[0697] The server provides real-time information on the progress of food deliveries and details of menu items. This allows users to quickly obtain information about the current delivery status and menu items.

[0698] Adding specific examples

[0699] As a concrete example, let's explain the behavior when a user enters the question, "How long will it take for my ordered pizza to arrive?"

[0700] 1. The user enters their question into the chat window of the smartphone application.

[0701] 2. The server receives this question and tokenizes the input text.

[0702] 3. Pass the tokenized data to the generative AI model (GPT-3) and use the following prompts to generate a response:

[0703] As a support bot for a food delivery service, answer user inquiries.

[0704] User: How long will it take for my pizza to arrive?

[0705] Bot:

[0706] 4. The generative AI model generates the response, "Pizza delivery usually takes 30-40 minutes, but the current delivery time can be specifically tracked based on GPS."

[0707] 5. Convert this response to JSON format and send it back to the user's smartphone application.

[0708] 6. Users can check the response on the application and obtain information about the delivery status.

[0709] In this application, users can receive quick and accurate answers to their inquiries, and real-time information on delivery progress and details of menu items can also be provided.

[0710] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0711] Program processing steps

[0712] Step 1:

[0713] The user enters their question into the chat window of the smartphone application. When the user types a message such as "How long will it take for my ordered pizza to arrive?" and presses the send button, the input data is sent to the server. At this stage, the input is the user's question text.

[0714] Step 2:

[0715] The server receives input text from the user and tokenizes it. Specifically, it divides the input text into words and phrases, converting the data into a format that is easy to analyze. In this example, the question text received as input, "How long will it take for my ordered pizza to arrive?", is output as tokenized data.

[0716] Step 3:

[0717] The server passes the tokenized data to a generative AI model, which then generates a response. The generative AI model (e.g., GPT-3) generates an appropriate response according to the format of the prompt. A concrete example of a prompt is as follows:

[0718] As a support bot for a food delivery service, answer user inquiries.

[0719] User: How long will it take for my pizza to arrive?

[0720] Bot:

[0721] Given tokenized data as input, a generative AI model generates a response such as, "Pizza delivery usually takes 30-40 minutes, but the current delivery time can be specifically tracked based on GPS." This is the output data.

[0722] Step 4:

[0723] The server converts the generated response into JSON format. This conversion process transforms the data into the standard JSON format. The generated response text is converted into JSON data like the following:

[0724] json

[0725] {

[0726] "Response": "Pizza delivery usually takes 30-40 minutes, but current delivery times can be tracked specifically based on GPS."

[0727] }

[0728] It receives the generated response text as input and outputs data in JSON format.

[0729] Step 5:

[0730] The server sends this JSON data back to the user's device. The device (smartphone application) receives this data and processes it to display it in the chat window. The output is receiving JSON data as input and displaying the response in the user's chat window.

[0731] Step 6:

[0732] The user checks the response displayed in the chat window of the smartphone application. This response allows the user to receive real-time information about the delivery progress and details of the food delivery menu. Specifically, the user reads and understands the displayed response. The input in this step is the response text displayed in the application, and the output is the user's understanding and satisfaction.

[0733] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0734] This invention relates to a chat system that utilizes a generative AI model combined with an emotion engine that recognizes user emotions. This system significantly improves the efficiency and satisfaction of customer service by providing quick and emotion-appropriate responses to user inquiries.

[0735] System Configuration

[0736] This system consists of the following main components:

[0737] 1. User interface for receiving inquiries

[0738] 2. Terminals that receive and process data from the user interface.

[0739] 3. Servers that perform data processing and response generation.

[0740] 4. Means of analyzing user emotions using an emotion engine

[0741] 5. Communication means for returning the generated response to the user.

[0742] Specific steps

[0743] 1. User input:

[0744] A user enters a question in the chat window such as, "How does KOALA work?"

[0745] 2. Sending a request from the terminal to the server:

[0746] The terminal converts the user's input into JSON format and sends a POST request to the server. This request contains the user's question data.

[0747] 3. Server-side processing:

[0748] The server receives the request and tokenizes the user's input. The tokenized data is passed to a generative AI model (e.g., a large-scale language model) to generate an appropriate response.

[0749] 4. Analysis using an emotion engine:

[0750] Before the server generates a response using a generative AI model, it passes the user's input to an emotion engine for analysis. The emotion engine analyzes the user's emotional state in real time and incorporates the analysis results into the response generation.

[0751] 5. Response generation and transformation:

[0752] Based on the analysis results of the emotion engine, a generative AI model generates response text. The generated response text is converted to JSON format. The converted data has the following structure: "response": "KOALA processes the data in real time using advanced analytical algorithms."

[0753] 6. Sending a response from the server to the terminal:

[0754] The server sends this JSON data back to the device.

[0755] 7. Displaying responses to the user:

[0756] The terminal displays the response data received from the server in the chat window. The user reviews this response and, if they have further questions, enter them again in the chat window.

[0757] Specific example

[0758] If a user enters the question "What are the main functions of KOALA?", it will behave as follows:

[0759] 1. The user enters the question into the device.

[0760] 2. The terminal sends the question data to the server.

[0761] 3. The server processes the data and analyzes the user's emotions using an emotion engine before using generative AI models.

[0762] 4. The emotion engine analyzes the user's emotional state, and based on the results, a generative AI model generates a response.

[0763] 5. The server converts the generated response into JSON format and sends it to the terminal.

[0764] 6. The device displays the response in the chat window.

[0765] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[0766] The introduction of this system will enable responses to user inquiries to become more emotionally responsive, which is expected to improve customer satisfaction.

[0767] The following describes the processing flow.

[0768] Step 1:

[0769] The user types a question into the chat window. For example, they might type a specific question like, "What are the main features of KOALA?"

[0770] Step 2:

[0771] The terminal receives user input and converts this input data into JSON format. The converted data has the following JSON structure:

[0772] json

[0773] {

[0774] "user_input": "What are the main functions of KOALA?"

[0775] }

[0776] Step 3:

[0777] The device sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[0778] Step 4:

[0779] The server processes the POST request received from the terminal. At this time, the / chat endpoint of the Flask application receives the request.

[0780] Step 5:

[0781] The server extracts user input (user_input) from the received request data. Specifically, it extracts the data as follows:

[0782] python

[0783] data = request.json

[0784] user_input = data['user_input']

[0785] Step 6:

[0786] The server passes the extracted user input to the emotion engine for sentiment analysis. The sentiment engine analyzes the user's emotions as follows:

[0787] python

[0788] emotion = emotion_engine.analyze(user_input)

[0789] The emotion engine returns the analysis results. For example, it may determine that the user's input contains emotions such as "joy" or "anxiety."

[0790] Step 7:

[0791] The server performs a tokenization process based on the analysis results of the emotion engine. The tokenization process proceeds as follows:

[0792] python

[0793] inputs = tokenizer.encode(user_input, return_tensors='pt')

[0794] Step 8:

[0795] The server passes the tokenized data and sentiment analysis results to a generative AI model (large-scale language model) to generate response text. Specifically, the process is as follows:

[0796] python

[0797] outputs = model.generate(inputs, max_length=200, num_return_sequences=1, emotion=emotion)

[0798] response = tokenizer.decode(outputs[0], skip_special_tokens=True)

[0799] Step 9:

[0800] The server converts the generated response into JSON format. The converted JSON data has the following structure:

[0801] json

[0802] {

[0803] "Response": "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[0804] }

[0805] Step 10:

[0806] The server sends this JSON data back to the device.

[0807] Step 11:

[0808] The terminal analyzes the response data received from the server and displays it in the chat window. The user can review this response and, if they have further questions, enter them again in the chat window.

[0809] Through the steps described above, the system provides fast and emotionally responsive responses to user inquiries. This enables responses that take user emotions into consideration, which is expected to improve customer satisfaction.

[0810] (Example 2)

[0811] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0812] Traditional chat systems had the problem of lowering user satisfaction because they provided uniform responses without considering user emotions. Furthermore, the inconsistent quality of responses made it difficult to provide satisfactory support.

[0813] The specific processing performed by the specific 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 tokenizing user input, means for passing the tokenized data to sentiment analysis means for analysis, and means for passing the tokenized data to a generation AI model based on the sentiment analysis results to generate a response. This makes it possible to provide a response that corresponds to the user's emotions.

[0814] "User input" refers to the text or questions that users type into the chat window.

[0815] "Tokenization" refers to the process of dividing a text into units such as words or sentences.

[0816] "Sentiment analysis method" refers to a method that analyzes emotions from user input text and outputs the results.

[0817] A "generative AI model" refers to an algorithm that uses artificial intelligence to generate responses to user input.

[0818] "Response generation" refers to the process of generating an appropriate response based on user input.

[0819] "JSON format" is an abbreviation for JavaScript Object Notation, and refers to a format for representing data in a structured way.

[0820] This invention relates to a chat system that utilizes a generative AI model combined with an emotion engine that recognizes user emotions. This system significantly improves the efficiency and satisfaction of customer service by providing quick and emotion-appropriate responses to user inquiries.

[0821] System Configuration

[0822] This system consists of the following main components:

[0823] 1. User interface for receiving user input

[0824] 2. Terminals that receive and process data from the user interface.

[0825] 3. Servers that perform data processing and response generation.

[0826] 4. Means of analyzing user emotions using an emotion engine

[0827] 5. Communication means for returning the generated response to the user.

[0828] Program processing

[0829] The user accesses the system and enters a question or message into the chat window. For example, the user might type, "How does KOALA work?" The terminal converts this input into JSON format and sends a POST request to the server. This request contains the user's question data.

[0830] The server receives the request and tokenizes the user's input. The tokenized data is first passed to an emotion engine to analyze the user's emotional state. The emotion engine can use, for example, the Microsoft Azure Text Analytics API. Based on the analysis results, the tokenized data is passed to a generative AI model (for example, a large-scale language model such as GPT-3) to generate an appropriate response.

[0831] The generated response text is converted back into JSON format and sent to the terminal. The terminal parses the received data and displays the response in the chat window. This allows the user to receive responses that are appropriate to their emotions.

[0832] Specific example

[0833] For example, if a user enters the question "What are the main functions of KOALA?", the system will behave as follows:

[0834] 1. The user enters the question into the device.

[0835] 2. The terminal sends the question data to the server.

[0836] 3. The server processes the data and analyzes the user's emotions using an emotion engine before using generative AI models.

[0837] 4. The emotion engine analyzes the user's emotional state, and based on the results, a generative AI model generates a response.

[0838] 5. The server converts the generated response into JSON format and sends it to the terminal.

[0839] 6. The device displays the response in the chat window.

[0840] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[0841] The introduction of this system will enable responses to user inquiries to become more emotionally responsive, which is expected to improve customer satisfaction.

[0842] Example of a prompt:

[0843] "The user is asking: 'How does KOALA work?' Its emotional state is neutral. Please provide an appropriate response to this."

[0844] This invention enables the generation of sophisticated responses that take emotions into consideration, and is expected to significantly improve the user experience.

[0845] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0846] Step 1:

[0847] The user enters a question into the chat window. This input is natural language text, such as "How does KOALA work?".

[0848] Input: User's question text

[0849] Output: Input completion event in the chat window

[0850] Step 2:

[0851] The terminal receives user input and converts it to JSON format. The JSON data will be in the following format:

[0852] json

[0853] {

[0854] "user_input": "How does KOALA work?"

[0855] }

[0856] Input: User's question text

[0857] Output: Question data in JSON format

[0858] Step 3:

[0859] The terminal sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[0860] Input: Question data in JSON format

[0861] Output: POST request to the server

[0862] Step 4:

[0863] The server receives the POST request, parses the JSON data, and extracts the user input portion.

[0864] Input: JSON data sent to the server

[0865] Output: Extracted user input text

[0866] Step 5:

[0867] The server tokenizes the extracted input text. For example, it might be split into "KOALA", "is", "how", "works", "is", and "is it?".

[0868] Input: User-input text

[0869] Output: Tokenized data

[0870] Step 6:

[0871] The server passes the tokenized data to the sentiment engine for sentiment analysis. The sentiment engine uses Microsoft Azure's Text Analytics API, among others. For example, it might analyze the sentiment as "Sentimental state: Neutral."

[0872] Input: Tokenized user input data

[0873] Output: Emotion analysis results

[0874] Step 7:

[0875] The server passes the tokenized data, based on the sentiment analysis results, to a generative AI model to generate a response. The generative AI model uses large-scale language models such as GPT-3. Here, a response such as "KOALA uses advanced analytical algorithms to process data in real time" is generated.

[0876] Input: Tokenized data and sentiment analysis results

[0877] Output: Generated response text

[0878] Step 8:

[0879] The server converts the generated response text into JSON format. The resulting data will look like this:

[0880] json

[0881] {

[0882] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[0883] }

[0884] Input: Generated response text

[0885] Output: Response data in JSON format

[0886] Step 9:

[0887] The server sends response data in JSON format to the terminal.

[0888] Input: Response data in JSON format

[0889] Output: Sending a response to the terminal

[0890] Step 10:

[0891] The terminal parses the JSON data received from the server and extracts the response text. The extracted response text is then displayed in the chat window.

[0892] Input: JSON data sent from the server

[0893] Output: Response text displayed in the chat window

[0894] This allows users to see appropriate responses tailored to their emotions in real time.

[0895] (Application Example 2)

[0896] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0897] Traditional chat systems have a problem in that they cannot flexibly change responses according to the user's emotions and situation, and therefore cannot sufficiently increase user satisfaction. In particular, e-commerce sites receive a wide range of customer inquiries, and quick and accurate responses are required, so providing responses that respond to emotions is effective. However, current systems do not have emotion analysis functions and can only return a uniform response.

[0898] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0899] In this invention, the server includes means for receiving user input, means for analyzing the user's emotions using an emotion analysis engine, means for generating a response using a generative AI model based on the analysis results, means for converting the generated response into JSON format, and means for returning the response to the user. This makes it possible to provide a response that is appropriate to the user's emotions.

[0900] "Means of receiving user input" refers to a function that receives messages and inquiries sent by users through an interface such as a chat window.

[0901] "Methods for tokenizing user input" refer to the process of dividing user-entered text into meaningful units to facilitate natural language processing.

[0902] "A means of passing tokenized data to a generative AI model to generate a response" refers to a function that inputs tokenized data into a generative AI model and generates an appropriate response for the user.

[0903] "Means for converting generated responses into JSON format" refers to a function that converts text responses generated by generative AI models into JSON format, facilitating data communication and display.

[0904] "Means of sending a response back to the user" refers to a function that sends a generated response from the server to the user's device and displays it to the user.

[0905] "Means of analyzing a user's emotions using an emotion analysis engine" refers to an engine and its functions for analyzing and understanding a user's emotional state from their text input.

[0906] "Means for adjusting responses based on analysis results" refers to a function that adjusts the content of responses generated by generative AI models based on the analysis results obtained by the emotion analysis engine.

[0907] This invention relates to a chat system for e-commerce sites that combines an emotion recognition engine and a generative AI model. This system enables the rapid and appropriate provision of responses tailored to the user's emotions. The specific configuration and processing methods of this system are described below.

[0908] System Configuration

[0909] This system consists of the following main components:

[0910] 1. Means for receiving user input:

[0911] The user enters text into the chat window via a smartphone app.

[0912] 2. Means of tokenizing user input:

[0913] Divide the text into meaningful units and convert them into a format that is easy to process.

[0914] 3. Means for analyzing user emotions using an emotion analysis engine:

[0915] The system analyzes the emotional state contained in the user's input. The software used includes the Hugging Face transformers library.

[0916] 4. Means of passing tokenized data to a generative AI model to generate a response:

[0917] Based on tokenized data and sentiment analysis results, a generative AI model generates an appropriate response. A large-scale language model is used here.

[0918] 5. Means for converting the generated response to JSON format:

[0919] The generated text response is converted to JSON format to improve communication efficiency.

[0920] 6. Means of returning the response to the user:

[0921] The server sends a response to the user's smartphone and displays it in the chat window.

[0922] Specific example

[0923] If a user enters the question "My order hasn't arrived yet. What's going on?" into the chat window of an online shopping site, the system will operate as follows:

[0924] 1. Receiving user input:

[0925] The device receives the text entered by the user.

[0926] 2. Emotion analysis:

[0927] The device passes the received text to an emotion analysis engine, which analyzes whether the user is expressing anxiety or anger.

[0928] 3. Response generation:

[0929] Based on the results of the emotion analysis, a generative AI model generates a response such as, "I apologize for the delay. I will check it immediately."

[0930] 4. Sending a response:

[0931] The generated response is converted to JSON format and sent from the server to the user's terminal.

[0932] 5. Display of response:

[0933] The user checks the response displayed in the chat window.

[0934] The software used includes the Hugging Face transformers library for emotion analysis and a generative AI model (large-scale language model). This enables appropriate responses based on the user's emotions, thereby improving customer satisfaction.

[0935] Example of a prompt:

[0936] "My order hasn't arrived yet. What's going on?"

[0937] This invention provides responses that take user emotions into consideration, significantly improving the quality of customer service on e-commerce sites.

[0938] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0939] Step 1:

[0940] The user types "My order hasn't arrived yet. What's going on?" into the chat window. This is the starting point for the entire system.

[0941] Step 2:

[0942] The terminal receives user input and records it as text data. This input is the user's text message, which serves as initial data for processing.

[0943] Step 3:

[0944] The terminal tokenizes the received text data. Tokenization is the process of dividing text into meaningful units (tokens). For example, it might be divided into "order," "but," "haven't," "arrived," "how," "is," "what," "is," "what," "is," "what," "is." This process makes the text data easier to analyze.

[0945] Step 4:

[0946] The tokenized text data is passed to an emotion analysis engine to analyze the user's emotions. The emotion analysis engine uses the Hugging Face transformers library to determine the user's emotional state (e.g., anxiety, anger). The output is the emotion classification result, with "anxiety" being output as an example.

[0947] Step 5:

[0948] Based on the sentiment analysis results, the server passes the tokenized data to a generative AI model to generate an appropriate response. The generative AI model uses a large-scale language model to generate a response such as, "I apologize for the delay. I will check on this immediately."

[0949] Step 6:

[0950] The generated response is converted to JSON format. The server converts the text response to JSON format in preparation for the next communication process. For example, it will be converted to a format like {"response": "Sorry for the wait. I will check it shortly."}.

[0951] Step 7:

[0952] The server sends data in JSON format to the terminal. The terminal receives the JSON data sent from the server and prepares for the next processing step.

[0953] Step 8:

[0954] The terminal parses the received JSON data and displays a response in the chat window. The user can then see the response displayed in the chat window, such as "Sorry for the wait. I will check it immediately," and ask further questions or take actions.

[0955] This series of processing steps allows users to receive appropriate responses that match their emotions, resulting in a high level of satisfaction.

[0956] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0957] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0958] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0959] [Third Embodiment]

[0960] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0961] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0962] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0963] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0964] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0965] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0966] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0967] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0968] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0969] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0970] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0971] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0972] As an embodiment of the present invention, the process of constructing a chat system using a generative AI model will be described. This system significantly improves the efficiency of customer service by providing quick and accurate responses to questions entered by users.

[0973] System Configuration

[0974] This system consists of the following main components:

[0975] 1. User interface for receiving inquiries

[0976] 2. Terminals that receive and process data from the user interface.

[0977] 3. Servers that perform data processing and response generation.

[0978] 4. Communication means for sending the generated response back to the user.

[0979] Specific steps

[0980] 1. User input:

[0981] A user enters a question in the chat window such as, "How does KOALA work?"

[0982] 2. Sending a request from the terminal to the server:

[0983] The terminal converts the user's input into JSON format and sends a POST request to the server. This request contains the user's question data.

[0984] 3. Server-side processing:

[0985] The server receives the request and tokenizes the user's input. The tokenized data is passed to a generative AI model (e.g., a large-scale language model) to generate an appropriate response.

[0986] 4. Response generation and transformation:

[0987] Convert the generated response text to JSON format. The converted data will have the following structure: "response": "KOALA processes the data in real time using advanced analytical algorithms."

[0988] 5. Sending a response from the server to the terminal:

[0989] The server sends this JSON data back to the device.

[0990] 6. Displaying responses to the user:

[0991] The terminal displays the response data received from the server in the chat window. The user can review this and continue asking further questions.

[0992] Specific example

[0993] If a user enters the question "What are the main functions of KOALA?", it will behave as follows:

[0994] 1. The user enters the question into the device.

[0995] 2. The terminal sends the question data to the server.

[0996] 3. The server processes the data and generates a response using a generative AI model.

[0997] 4. The server converts the generated response into JSON format and sends it to the terminal.

[0998] 5. The device displays the response in the chat window.

[0999] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1000] The implementation of this system will dramatically improve the response speed to user inquiries, leading to increased productivity for sales representatives and reduced workload on the product. Furthermore, faster response times will improve customer satisfaction.

[1001] The following describes the processing flow.

[1002] Step 1:

[1003] The user types a question into the chat window. For example, they might type a specific question like, "How does KOALA work?"

[1004] Step 2:

[1005] The terminal receives user input and converts this input data into JSON format. The converted data has the following JSON structure:

[1006] json

[1007] {

[1008] "user_input": "How does KOALA work?"

[1009] }

[1010] Step 3:

[1011] The device sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[1012] Step 4:

[1013] The server processes the POST request received from the terminal. At this time, the / chat endpoint of the Flask application receives the request.

[1014] Step 5:

[1015] The server extracts user input (user_input) from the received request data. Specifically, it extracts the data as follows:

[1016] python

[1017] data = request.json

[1018] user_input = data['user_input']

[1019] Step 6:

[1020] The server tokenizes the extracted user input. For this purpose, it uses a tokenizer from a pre-loaded generative AI model.

[1021] python

[1022] inputs = tokenizer.encode(user_input, return_tensors='pt')

[1023] Step 7:

[1024] The server passes the tokenized data to a generative AI model (large-scale language model) to generate response text.

[1025] python

[1026] outputs = model.generate(inputs, max_length=200, num_return_sequences=1)

[1027] response = tokenizer.decode(outputs[0], skip_special_tokens=True)

[1028] Step 8:

[1029] The server converts the generated response into JSON format. The converted JSON data has the following structure:

[1030] json

[1031] {

[1032] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[1033] }

[1034] Step 9:

[1035] The server sends the generated response data back to the terminal. This response data is sent again in JSON format.

[1036] Step 10:

[1037] The terminal analyzes the response data received from the server and displays it in the chat window. The user reviews this response and, if they have further questions, enter them again in the chat window.

[1038] Through these steps, the system provides fast and accurate responses to user inquiries.

[1039] (Example 1)

[1040] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1041] In today's digital transformation, there is a demand for systems that can respond quickly and accurately to user inquiries. Conventional systems often process inquiries manually, which is time-consuming and labor-intensive. Furthermore, the quality of responses can vary depending on the person handling them, leading to decreased customer satisfaction. Therefore, this invention aims to solve these problems by providing a system that uses a generative AI model to deliver consistent, high-quality responses quickly.

[1042] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1043] In this invention, the server includes means for tokenizing user input and creating a prompt, means for passing the tokenized data to a generation AI model to generate a response, and means for converting the generated response into JSON format. This enables a rapid and accurate response to user inquiries.

[1044] A "user" refers to a person who uses the system to make an inquiry.

[1045] "Means for receiving input" refers to a device or software that has the function of receiving inquiries from users.

[1046] "JSON format" is an abbreviation for JavaScript Object Notation, and refers to a format that structures data using key-value pairs and represents it in an easy-to-read format.

[1047] "Tokenization" refers to the process of dividing input text into smaller units such as words and phrases.

[1048] A "prompt" refers to the initial conditions or question-style sentences that a generative AI model uses to receive input.

[1049] A "generative AI model" refers to a model that uses machine learning techniques to generate responses based on user input.

[1050] A "large-scale language model" refers to a model that is trained using a large amount of text data to understand and generate complex language patterns.

[1051] "Means for generating responses" refers to devices or software that create appropriate responses to user inquiries based on tokenized data.

[1052] "Means for sending back a response" refers to a device or software that has the function of sending back the response generated by the server to the user.

[1053] A "chat window" refers to an interface for users to input and display text-based inquiries and responses.

[1054] As an embodiment of the present invention, a method for constructing a chat system using a generative AI model will be described. This system significantly improves the efficiency of customer service by providing quick and accurate responses to questions entered by users.

[1055] System Configuration

[1056] This system consists of the following main components:

[1057] 1. User Interface: Receives inquiries from users.

[1058] 2. Terminal: Receives data from the user interface and sends it to the server.

[1059] 3. Server: Performs data processing and response generation.

[1060] 4. Communication method: The generated response is sent back to the user.

[1061] Hardware and software to be used

[1062] Hardware: Computers and smartphones used by users, and server machines on which servers are installed.

[1063] Software: Web pages and applications with user interfaces, software for sending requests used by devices, generative AI models running on servers (e.g., GPT-3), and Python libraries used for data processing (e.g., NLTK, spaCy, etc.).

[1064] System Operation Description

[1065] The system operates as follows:

[1066] 1. User input: The user enters their question into the chat window. In this example, they type, "How does KOALA work?"

[1067] 2. Terminal generation and transmission of requests: The terminal converts the user's input into JSON format and sends it to the server as a POST request.

[1068] 3. Receiving and processing requests by the server: The server receives the request and tokenizes the user's input. This process uses natural language processing libraries such as NLTK (Natural Language Toolkit).

[1069] 4. Response generation by generative AI model: The server inputs tokenized data into the generative AI model to generate an appropriate response. In this example, the response "KOALA processes data in real time using advanced analytical algorithms" is obtained.

[1070] 5. Server-side response conversion and transmission: The generated response is converted to JSON format and sent back to the terminal.

[1071] 6. Displaying responses via the terminal: The terminal displays the response received from the server in the chat window. This allows the user to see the generated response.

[1072] Specific example

[1073] For example, if a user enters the question "What are the main functions of KOALA?", the system will behave as follows:

[1074] 1. User input: The user types "What are the main functions of KOALA?" into the chat window.

[1075] 2. Terminal generation and sending of requests: The terminal converts the question into JSON format and sends it to the server.

[1076] 3. Server receives and processes the request: The server receives the query and performs tokenization.

[1077] 4. Response generation using a generative AI model: The server uses a generative AI model to generate a response that says, "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1078] 5. Server conversion and transmission of response: The server converts the generated response into JSON format and sends it to the terminal.

[1079] 6. Display of responses by the device: The device displays the response in the chat window.

[1080] Example of a prompt

[1081] "What are the main functions of KOALA?"

[1082] This system enables quick and accurate responses to user inquiries, leading to increased productivity for sales representatives and improved customer satisfaction.

[1083] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1084] Step 1:

[1085] User input

[1086] The user enters a question into the chat window. In this example, they type, "How does KOALA work?". The user directly types the text on the keyboard, and it appears in the chat window. This input is then processed.

[1087] Specific operation: The user accesses a chat window in a web browser or application and types their inquiry using the keyboard. The user's input is displayed in the chat window in real time.

[1088] Step 2:

[1089] Request generation by the terminal

[1090] The terminal converts the user's input into JSON format and sends a POST request to the server. The input is the user's question text, "How does KOALA work?". Converting this to JSON format results in the following:

[1091] json

[1092] {

[1093] "Question": "How does KOALA work?"

[1094] }

[1095] Specific operation: Use JavaScript's XMLHttpRequest and fetch APIs to convert user input data into JSON format and send a POST request to the server.

[1096] Step 3:

[1097] Server receives request

[1098] The server receives a request in JSON format sent from the terminal. The input is data in JSON format.

[1099] json

[1100] {

[1101] "Question": "How does KOALA work?"

[1102] }

[1103] The server receives this data for analysis.

[1104] Specific operation: The server uses a web framework such as Flask or Django in Python to receive POST requests from the terminal and parse their contents.

[1105] Step 4:

[1106] Execute tokenization process

[1107] The server tokenizes the user's question received. The input is the user's question text. Tokenization divides this into units such as words and phrases. Natural language processing libraries such as NLTK and spaCy are used for this process. For example, it might be divided into "KOALA", "is", "how", "works", "of", "is it?".

[1108] Specific operation: The server uses libraries such as NLTK and spaCy to tokenize the user's input text.

[1109] Step 5:

[1110] Response generation using generative AI models

[1111] The server inputs tokenized data into a generative AI model, which then generates an appropriate response. The input is tokenized data. Based on this, the generative AI model (e.g., GPT-3) generates a response such as, "KOALA uses advanced analytical algorithms to process the data in real time."

[1112] Specific operation: The server uses the OpenAI API or similar tools to send prompts to the generated AI model, causing it to generate responses.

[1113] Step 6:

[1114] Response transformation

[1115] The server converts the generated response into JSON format. The input is the generated response text. When converted to JSON format, it looks like this:

[1116] json

[1117] {

[1118] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[1119] }

[1120] Specific operation: Use Python's standard library or other JSON libraries to convert the generated response into JSON format.

[1121] Step 7:

[1122] Sending a response

[1123] The server returns the converted JSON data to the terminal. The input is a response data in JSON format. This is then sent to the terminal.

[1124] Specific operation: The server uses features from Python's Flask or Django to return JSON data as an HTTP response.

[1125] Step 8:

[1126] Display of response

[1127] The terminal displays the response data received from the server in the chat window. The input is response data in JSON format.

[1128] json

[1129] {

[1130] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[1131] }

[1132] The device analyzes this information and displays it in the chat window in a format that is easy for the user to understand.

[1133] Specific operation: Using JavaScript, the response text is extracted from the received JSON data and inserted into an HTML element for display.

[1134] (Application Example 1)

[1135] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1136] In food delivery services, responding quickly and accurately to customer inquiries directly leads to improved customer satisfaction and operational efficiency. However, traditional systems often suffer from delayed responses or inappropriate answers depending on the nature of the inquiry. Furthermore, providing real-time information on delivery progress and menu item details is difficult. This results in increased customer dissatisfaction and a decline in service quality.

[1137] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1138] In this invention, the server includes means for receiving input from a user, means for tokenizing the user's input, means for passing the tokenized data to a generative AI model to generate a response, means for converting the generated response into JSON format, means for returning the response to the user, means for generating a response specific to food delivery services, and means for providing real-time information on the delivery progress and menu item details of food deliveries. This makes it possible to respond quickly and accurately to user inquiries and to provide real-time information on delivery progress and menu details.

[1139] "Means of receiving user input" refers to providing an interface for users of a food delivery service to enter inquiries or questions using a smartphone or other device.

[1140] "Methods for tokenizing user input" refer to processes that analyze the text entered by the user and divide it into words and phrases.

[1141] "Means of passing tokenized data to a generative AI model to generate a response" refers to the process by which a generative AI model generates an appropriate response based on tokenized input.

[1142] "Means for converting the generated response to JSON format" refers to the process of converting the generated text response into JSON, a common data exchange format.

[1143] "Means of sending the response back to the user" refers to sending the converted response data to the user's device so that the user can confirm the response.

[1144] "Means for generating responses specifically for food delivery services" refers to processing that provides professional and appropriate responses to questions and inquiries related to food delivery.

[1145] "A means of providing real-time responses regarding the progress of food delivery and details of menu items" refers to a system that immediately provides users with real-time information regarding the current progress of delivery and specific menu details.

[1146] Modes for carrying out the invention

[1147] Program processing

[1148] The system that embodies this application is a chat support system specifically designed for food delivery services, providing quick and accurate responses to user inquiries through a smartphone application.

[1149] Hardware and software to be used

[1150] Hardware: Cloud servers or on-premises servers

[1151] Software: Flask (Web framework), OpenAI API

[1152] Data processing and data calculation

[1153] 1. Receiving user input:

[1154] The server receives questions entered by users through a smartphone application. Users initiate an inquiry by typing their questions into the application's chat window.

[1155] 2. Tokenization of user input:

[1156] The server tokenizes the user's input. Tokenization is the process of analyzing the user's text and dividing it into words and phrases.

[1157] 3. Pass the tokenized data to a generative AI model to generate a response:

[1158] The tokenized data is input into a generative AI model (e.g., GPT-3) to generate an appropriate response. The generative AI model generates the most appropriate text response to the user's question.

[1159] 4. Converting the generated response to JSON format:

[1160] The generated response is converted to JSON format. JSON is a common format for data exchange and allows for proper formatting of the response.

[1161] 5. Sending the response back to the user:

[1162] The server sends the generated JSON response back to the user's smartphone application. The user can then view the response in the application's chat window.

[1163] 6. Generating responses specifically for food delivery services:

[1164] The generative AI model has the capability to generate responses specifically tailored to food delivery services. It quickly provides specific information to users in response to their questions, such as delivery progress and details of menu items.

[1165] 7. Real-time responses regarding delivery progress and menu item details:

[1166] The server provides real-time information on the progress of food deliveries and details of menu items. This allows users to quickly obtain information about the current delivery status and menu items.

[1167] Adding specific examples

[1168] As a concrete example, let's explain the behavior when a user enters the question, "How long will it take for my ordered pizza to arrive?"

[1169] 1. The user enters their question into the chat window of the smartphone application.

[1170] 2. The server receives this question and tokenizes the input text.

[1171] 3. Pass the tokenized data to the generative AI model (GPT-3) and use the following prompts to generate a response:

[1172] As a support bot for a food delivery service, answer user inquiries.

[1173] User: How long will it take for my pizza to arrive?

[1174] Bot:

[1175] 4. The generative AI model generates the response, "Pizza delivery usually takes 30-40 minutes, but the current delivery time can be specifically tracked based on GPS."

[1176] 5. Convert this response to JSON format and send it back to the user's smartphone application.

[1177] 6. Users can check the response on the application and obtain information about the delivery status.

[1178] In this application, users can receive quick and accurate answers to their inquiries, and real-time information on delivery progress and details of menu items can also be provided.

[1179] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1180] Program processing steps

[1181] Step 1:

[1182] The user enters their question into the chat window of the smartphone application. When the user types a message such as "How long will it take for my ordered pizza to arrive?" and presses the send button, the input data is sent to the server. At this stage, the input is the user's question text.

[1183] Step 2:

[1184] The server receives input text from the user and tokenizes it. Specifically, it divides the input text into words and phrases, converting the data into a format that is easy to analyze. In this example, the question text received as input, "How long will it take for my ordered pizza to arrive?", is output as tokenized data.

[1185] Step 3:

[1186] The server passes the tokenized data to a generative AI model, which then generates a response. The generative AI model (e.g., GPT-3) generates an appropriate response according to the format of the prompt. A concrete example of a prompt is as follows:

[1187] As a support bot for a food delivery service, answer user inquiries.

[1188] User: How long will it take for my pizza to arrive?

[1189] Bot:

[1190] Given tokenized data as input, a generative AI model generates a response such as, "Pizza delivery usually takes 30-40 minutes, but the current delivery time can be specifically tracked based on GPS." This is the output data.

[1191] Step 4:

[1192] The server converts the generated response into JSON format. This conversion process transforms the data into the standard JSON format. The generated response text is converted into JSON data like the following:

[1193] json

[1194] {

[1195] "Response": "Pizza delivery usually takes 30-40 minutes, but current delivery times can be tracked specifically based on GPS."

[1196] }

[1197] It receives the generated response text as input and outputs data in JSON format.

[1198] Step 5:

[1199] The server sends this JSON data back to the user's device. The device (smartphone application) receives this data and processes it to display it in the chat window. The output is receiving JSON data as input and displaying the response in the user's chat window.

[1200] Step 6:

[1201] The user checks the response displayed in the chat window of the smartphone application. This response allows the user to receive real-time information about the delivery progress and details of the food delivery menu. Specifically, the user reads and understands the displayed response. The input in this step is the response text displayed in the application, and the output is the user's understanding and satisfaction.

[1202] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1203] This invention relates to a chat system that utilizes a generative AI model combined with an emotion engine that recognizes user emotions. This system significantly improves the efficiency and satisfaction of customer service by providing quick and emotion-appropriate responses to user inquiries.

[1204] System Configuration

[1205] This system consists of the following main components:

[1206] 1. User interface for receiving inquiries

[1207] 2. Terminals that receive and process data from the user interface.

[1208] 3. Servers that perform data processing and response generation.

[1209] 4. Means of analyzing user emotions using an emotion engine

[1210] 5. Communication means for returning the generated response to the user.

[1211] Specific steps

[1212] 1. User input:

[1213] A user enters a question in the chat window such as, "How does KOALA work?"

[1214] 2. Sending a request from the terminal to the server:

[1215] The terminal converts the user's input into JSON format and sends a POST request to the server. This request contains the user's question data.

[1216] 3. Server-side processing:

[1217] The server receives the request and tokenizes the user's input. The tokenized data is passed to a generative AI model (e.g., a large-scale language model) to generate an appropriate response.

[1218] 4. Analysis using an emotion engine:

[1219] Before the server generates a response using a generative AI model, it passes the user's input to an emotion engine for analysis. The emotion engine analyzes the user's emotional state in real time and incorporates the analysis results into the response generation.

[1220] 5. Response generation and transformation:

[1221] Based on the analysis results of the emotion engine, a generative AI model generates response text. The generated response text is converted to JSON format. The converted data has the following structure: "response": "KOALA processes the data in real time using advanced analytical algorithms."

[1222] 6. Sending a response from the server to the terminal:

[1223] The server sends this JSON data back to the device.

[1224] 7. Displaying responses to the user:

[1225] The terminal displays the response data received from the server in the chat window. The user reviews this response and, if they have further questions, enter them again in the chat window.

[1226] Specific example

[1227] If a user enters the question "What are the main functions of KOALA?", it will behave as follows:

[1228] 1. The user enters the question into the device.

[1229] 2. The terminal sends the question data to the server.

[1230] 3. The server processes the data and analyzes the user's emotions using an emotion engine before using generative AI models.

[1231] 4. The emotion engine analyzes the user's emotional state, and based on the results, a generative AI model generates a response.

[1232] 5. The server converts the generated response into JSON format and sends it to the terminal.

[1233] 6. The device displays the response in the chat window.

[1234] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1235] The introduction of this system will enable responses to user inquiries to become more emotionally responsive, which is expected to improve customer satisfaction.

[1236] The following describes the processing flow.

[1237] Step 1:

[1238] The user types a question into the chat window. For example, they might type a specific question like, "What are the main features of KOALA?"

[1239] Step 2:

[1240] The terminal receives user input and converts this input data into JSON format. The converted data has the following JSON structure:

[1241] json

[1242] {

[1243] "user_input": "What are the main functions of KOALA?"

[1244] }

[1245] Step 3:

[1246] The device sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[1247] Step 4:

[1248] The server processes the POST request received from the terminal. At this time, the / chat endpoint of the Flask application receives the request.

[1249] Step 5:

[1250] The server extracts user input (user_input) from the received request data. Specifically, it extracts the data as follows:

[1251] python

[1252] data = request.json

[1253] user_input = data['user_input']

[1254] Step 6:

[1255] The server passes the extracted user input to the emotion engine for sentiment analysis. The sentiment engine analyzes the user's emotions as follows:

[1256] python

[1257] emotion = emotion_engine.analyze(user_input)

[1258] The emotion engine returns the analysis results. For example, it may determine that the user's input contains emotions such as "joy" or "anxiety."

[1259] Step 7:

[1260] The server performs a tokenization process based on the analysis results of the emotion engine. The tokenization process proceeds as follows:

[1261] python

[1262] inputs = tokenizer.encode(user_input, return_tensors='pt')

[1263] Step 8:

[1264] The server passes the tokenized data and sentiment analysis results to a generative AI model (large-scale language model) to generate response text. Specifically, the process is as follows:

[1265] python

[1266] outputs = model.generate(inputs, max_length=200, num_return_sequences=1, emotion=emotion)

[1267] response = tokenizer.decode(outputs[0], skip_special_tokens=True)

[1268] Step 9:

[1269] The server converts the generated response into JSON format. The converted JSON data has the following structure:

[1270] json

[1271] {

[1272] "Response": "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1273] }

[1274] Step 10:

[1275] The server sends this JSON data back to the device.

[1276] Step 11:

[1277] The terminal analyzes the response data received from the server and displays it in the chat window. The user can review this response and, if they have further questions, enter them again in the chat window.

[1278] Through the steps described above, the system provides fast and emotionally responsive responses to user inquiries. This enables responses that take user emotions into consideration, which is expected to improve customer satisfaction.

[1279] (Example 2)

[1280] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1281] Traditional chat systems had the problem of lowering user satisfaction because they provided uniform responses without considering user emotions. Furthermore, the inconsistent quality of responses made it difficult to provide satisfactory support.

[1282] The specific processing performed by the specific 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 tokenizing user input, means for passing the tokenized data to sentiment analysis means for analysis, and means for passing the tokenized data to a generation AI model based on the sentiment analysis results to generate a response. This makes it possible to provide a response that corresponds to the user's emotions.

[1283] "User input" refers to the text or questions that users type into the chat window.

[1284] "Tokenization" refers to the process of dividing a text into units such as words or sentences.

[1285] "Sentiment analysis method" refers to a method that analyzes emotions from user input text and outputs the results.

[1286] A "generative AI model" refers to an algorithm that uses artificial intelligence to generate responses to user input.

[1287] "Response generation" refers to the process of generating an appropriate response based on user input.

[1288] "JSON format" is an abbreviation for JavaScript Object Notation, and refers to a format for representing data in a structured way.

[1289] This invention relates to a chat system that utilizes a generative AI model combined with an emotion engine that recognizes user emotions. This system significantly improves the efficiency and satisfaction of customer service by providing quick and emotion-appropriate responses to user inquiries.

[1290] System Configuration

[1291] This system consists of the following main components:

[1292] 1. User interface for receiving user input

[1293] 2. Terminals that receive and process data from the user interface.

[1294] 3. Servers that perform data processing and response generation.

[1295] 4. Means of analyzing user emotions using an emotion engine

[1296] 5. Communication means for returning the generated response to the user.

[1297] Program processing

[1298] The user accesses the system and enters a question or message into the chat window. For example, the user might type, "How does KOALA work?" The terminal converts this input into JSON format and sends a POST request to the server. This request contains the user's question data.

[1299] The server receives the request and tokenizes the user's input. The tokenized data is first passed to an emotion engine to analyze the user's emotional state. The emotion engine can use, for example, the Microsoft Azure Text Analytics API. Based on the analysis results, the tokenized data is passed to a generative AI model (for example, a large-scale language model such as GPT-3) to generate an appropriate response.

[1300] The generated response text is converted back into JSON format and sent to the terminal. The terminal parses the received data and displays the response in the chat window. This allows the user to receive responses that are appropriate to their emotions.

[1301] Specific example

[1302] For example, if a user enters the question "What are the main functions of KOALA?", the system will behave as follows:

[1303] 1. The user enters the question into the device.

[1304] 2. The terminal sends the question data to the server.

[1305] 3. The server processes the data and analyzes the user's emotions using an emotion engine before using generative AI models.

[1306] 4. The emotion engine analyzes the user's emotional state, and based on the results, a generative AI model generates a response.

[1307] 5. The server converts the generated response into JSON format and sends it to the terminal.

[1308] 6. The device displays the response in the chat window.

[1309] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1310] The introduction of this system will enable responses to user inquiries to become more emotionally responsive, which is expected to improve customer satisfaction.

[1311] Example of a prompt:

[1312] "The user is asking: 'How does KOALA work?' Its emotional state is neutral. Please provide an appropriate response to this."

[1313] This invention enables the generation of sophisticated responses that take emotions into consideration, and is expected to significantly improve the user experience.

[1314] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1315] Step 1:

[1316] The user enters a question into the chat window. This input is natural language text, such as "How does KOALA work?".

[1317] Input: User's question text

[1318] Output: Input completion event in the chat window

[1319] Step 2:

[1320] The terminal receives user input and converts it to JSON format. The JSON data will be in the following format:

[1321] json

[1322] {

[1323] "user_input": "How does KOALA work?"

[1324] }

[1325] Input: User's question text

[1326] Output: Question data in JSON format

[1327] Step 3:

[1328] The terminal sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[1329] Input: Question data in JSON format

[1330] Output: POST request to the server

[1331] Step 4:

[1332] The server receives the POST request, parses the JSON data, and extracts the user input portion.

[1333] Input: JSON data sent to the server

[1334] Output: Extracted user input text

[1335] Step 5:

[1336] The server tokenizes the extracted input text. For example, it might be split into "KOALA", "is", "how", "works", "is", and "is it?".

[1337] Input: User-input text

[1338] Output: Tokenized data

[1339] Step 6:

[1340] The server passes the tokenized data to the sentiment engine for sentiment analysis. The sentiment engine uses Microsoft Azure's Text Analytics API, among others. For example, it might analyze the sentiment as "Sentimental state: Neutral."

[1341] Input: Tokenized user input data

[1342] Output: Emotion analysis results

[1343] Step 7:

[1344] The server passes the tokenized data, based on the sentiment analysis results, to a generative AI model to generate a response. The generative AI model uses large-scale language models such as GPT-3. Here, a response such as "KOALA uses advanced analytical algorithms to process data in real time" is generated.

[1345] Input: Tokenized data and sentiment analysis results

[1346] Output: Generated response text

[1347] Step 8:

[1348] The server converts the generated response text into JSON format. The resulting data will look like this:

[1349] json

[1350] {

[1351] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[1352] }

[1353] Input: Generated response text

[1354] Output: Response data in JSON format

[1355] Step 9:

[1356] The server sends response data in JSON format to the terminal.

[1357] Input: Response data in JSON format

[1358] Output: Sending a response to the terminal

[1359] Step 10:

[1360] The terminal parses the JSON data received from the server and extracts the response text. The extracted response text is then displayed in the chat window.

[1361] Input: JSON data sent from the server

[1362] Output: Response text displayed in the chat window

[1363] This allows users to see appropriate responses tailored to their emotions in real time.

[1364] (Application Example 2)

[1365] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1366] Traditional chat systems have a problem in that they cannot flexibly change responses according to the user's emotions and situation, and therefore cannot sufficiently increase user satisfaction. In particular, e-commerce sites receive a wide range of customer inquiries, and quick and accurate responses are required, so providing responses that respond to emotions is effective. However, current systems do not have emotion analysis functions and can only return a uniform response.

[1367] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1368] In this invention, the server includes means for receiving user input, means for analyzing the user's emotions using an emotion analysis engine, means for generating a response using a generative AI model based on the analysis results, means for converting the generated response into JSON format, and means for returning the response to the user. This makes it possible to provide a response that is appropriate to the user's emotions.

[1369] "Means of receiving user input" refers to a function that receives messages and inquiries sent by users through an interface such as a chat window.

[1370] "Methods for tokenizing user input" refer to the process of dividing user-entered text into meaningful units to facilitate natural language processing.

[1371] "A means of passing tokenized data to a generative AI model to generate a response" refers to a function that inputs tokenized data into a generative AI model and generates an appropriate response for the user.

[1372] "Means for converting generated responses into JSON format" refers to a function that converts text responses generated by generative AI models into JSON format, facilitating data communication and display.

[1373] "Means of sending a response back to the user" refers to a function that sends a generated response from the server to the user's device and displays it to the user.

[1374] "Means of analyzing a user's emotions using an emotion analysis engine" refers to an engine and its functions for analyzing and understanding a user's emotional state from their text input.

[1375] "Means for adjusting responses based on analysis results" refers to a function that adjusts the content of responses generated by generative AI models based on the analysis results obtained by the emotion analysis engine.

[1376] This invention relates to a chat system for e-commerce sites that combines an emotion recognition engine and a generative AI model. This system enables the rapid and appropriate provision of responses tailored to the user's emotions. The specific configuration and processing methods of this system are described below.

[1377] System Configuration

[1378] This system consists of the following main components:

[1379] 1. Means for receiving user input:

[1380] The user enters text into the chat window via a smartphone app.

[1381] 2. Means of tokenizing user input:

[1382] Divide the text into meaningful units and convert them into a format that is easy to process.

[1383] 3. Means for analyzing user emotions using an emotion analysis engine:

[1384] The system analyzes the emotional state contained in the user's input. The software used includes the Hugging Face transformers library.

[1385] 4. Means of passing tokenized data to a generative AI model to generate a response:

[1386] Based on tokenized data and sentiment analysis results, a generative AI model generates an appropriate response. A large-scale language model is used here.

[1387] 5. Means for converting the generated response to JSON format:

[1388] The generated text response is converted to JSON format to improve communication efficiency.

[1389] 6. Means of returning the response to the user:

[1390] The server sends a response to the user's smartphone and displays it in the chat window.

[1391] Specific example

[1392] If a user enters the question "My order hasn't arrived yet. What's going on?" into the chat window of an online shopping site, the system will operate as follows:

[1393] 1. Receiving user input:

[1394] The device receives the text entered by the user.

[1395] 2. Emotion analysis:

[1396] The device passes the received text to an emotion analysis engine, which analyzes whether the user is expressing anxiety or anger.

[1397] 3. Response generation:

[1398] Based on the results of the emotion analysis, a generative AI model generates a response such as, "I apologize for the delay. I will check it immediately."

[1399] 4. Sending a response:

[1400] The generated response is converted to JSON format and sent from the server to the user's terminal.

[1401] 5. Display of response:

[1402] The user checks the response displayed in the chat window.

[1403] The software used includes the Hugging Face transformers library for emotion analysis and a generative AI model (large-scale language model). This enables appropriate responses based on the user's emotions, thereby improving customer satisfaction.

[1404] Example of a prompt:

[1405] "My order hasn't arrived yet. What's going on?"

[1406] This invention provides responses that take user emotions into consideration, significantly improving the quality of customer service on e-commerce sites.

[1407] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1408] Step 1:

[1409] The user types "My order hasn't arrived yet. What's going on?" into the chat window. This is the starting point for the entire system.

[1410] Step 2:

[1411] The terminal receives user input and records it as text data. This input is the user's text message, which serves as initial data for processing.

[1412] Step 3:

[1413] The terminal tokenizes the received text data. Tokenization is the process of dividing text into meaningful units (tokens). For example, it might be divided into "order," "but," "haven't," "arrived," "how," "is," "what," "is," "what," "is," "what," "is." This process makes the text data easier to analyze.

[1414] Step 4:

[1415] The tokenized text data is passed to an emotion analysis engine to analyze the user's emotions. The emotion analysis engine uses the Hugging Face transformers library to determine the user's emotional state (e.g., anxiety, anger). The output is the emotion classification result, with "anxiety" being output as an example.

[1416] Step 5:

[1417] Based on the sentiment analysis results, the server passes the tokenized data to a generative AI model to generate an appropriate response. The generative AI model uses a large-scale language model to generate a response such as, "I apologize for the delay. I will check on this immediately."

[1418] Step 6:

[1419] The generated response is converted to JSON format. The server converts the text response to JSON format in preparation for the next communication process. For example, it will be converted to a format like {"response": "Sorry for the wait. I will check it shortly."}.

[1420] Step 7:

[1421] The server sends data in JSON format to the terminal. The terminal receives the JSON data sent from the server and prepares for the next processing step.

[1422] Step 8:

[1423] The terminal parses the received JSON data and displays a response in the chat window. The user can then see the response displayed in the chat window, such as "Sorry for the wait. I will check it immediately," and ask further questions or take actions.

[1424] This series of processing steps allows users to receive appropriate responses that match their emotions, resulting in a high level of satisfaction.

[1425] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1426] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1427] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[1428] [Fourth Embodiment]

[1429] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1430] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1431] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1432] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[1433] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1434] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1435] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1436] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[1437] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1438] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1439] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1440] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1441] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1442] As an embodiment of the present invention, the process of constructing a chat system using a generative AI model will be described. This system significantly improves the efficiency of customer service by providing quick and accurate responses to questions entered by users.

[1443] System Configuration

[1444] This system consists of the following main components:

[1445] 1. User interface for receiving inquiries

[1446] 2. Terminals that receive and process data from the user interface.

[1447] 3. Servers that perform data processing and response generation.

[1448] 4. Communication means for sending the generated response back to the user.

[1449] Specific steps

[1450] 1. User input:

[1451] A user enters a question in the chat window such as, "How does KOALA work?"

[1452] 2. Sending a request from the terminal to the server:

[1453] The terminal converts the user's input into JSON format and sends a POST request to the server. This request contains the user's question data.

[1454] 3. Server-side processing:

[1455] The server receives the request and tokenizes the user's input. The tokenized data is passed to a generative AI model (e.g., a large-scale language model) to generate an appropriate response.

[1456] 4. Response generation and transformation:

[1457] Convert the generated response text to JSON format. The converted data will have the following structure: "response": "KOALA processes the data in real time using advanced analytical algorithms."

[1458] 5. Sending a response from the server to the terminal:

[1459] The server sends this JSON data back to the device.

[1460] 6. Displaying responses to the user:

[1461] The terminal displays the response data received from the server in the chat window. The user can review this and continue asking further questions.

[1462] Specific example

[1463] If a user enters the question "What are the main functions of KOALA?", it will behave as follows:

[1464] 1. The user enters the question into the device.

[1465] 2. The terminal sends the question data to the server.

[1466] 3. The server processes the data and generates a response using a generative AI model.

[1467] 4. The server converts the generated response into JSON format and sends it to the terminal.

[1468] 5. The device displays the response in the chat window.

[1469] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1470] The implementation of this system will dramatically improve the response speed to user inquiries, leading to increased productivity for sales representatives and reduced workload on the product. Furthermore, faster response times will improve customer satisfaction.

[1471] The following describes the processing flow.

[1472] Step 1:

[1473] The user types a question into the chat window. For example, they might type a specific question like, "How does KOALA work?"

[1474] Step 2:

[1475] The terminal receives user input and converts this input data into JSON format. The converted data has the following JSON structure:

[1476] json

[1477] {

[1478] "user_input": "How does KOALA work?"

[1479] }

[1480] Step 3:

[1481] The device sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[1482] Step 4:

[1483] The server processes the POST request received from the terminal. At this time, the / chat endpoint of the Flask application receives the request.

[1484] Step 5:

[1485] The server extracts user input (user_input) from the received request data. Specifically, it extracts the data as follows:

[1486] python

[1487] data = request.json

[1488] user_input = data['user_input']

[1489] Step 6:

[1490] The server tokenizes the extracted user input. For this purpose, it uses a tokenizer from a pre-loaded generative AI model.

[1491] python

[1492] inputs = tokenizer.encode(user_input, return_tensors='pt')

[1493] Step 7:

[1494] The server passes the tokenized data to a generative AI model (large-scale language model) to generate response text.

[1495] python

[1496] outputs = model.generate(inputs, max_length=200, num_return_sequences=1)

[1497] response = tokenizer.decode(outputs[0], skip_special_tokens=True)

[1498] Step 8:

[1499] The server converts the generated response into JSON format. The converted JSON data has the following structure:

[1500] json

[1501] {

[1502] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[1503] }

[1504] Step 9:

[1505] The server sends the generated response data back to the terminal. This response data is sent again in JSON format.

[1506] Step 10:

[1507] The terminal analyzes the response data received from the server and displays it in the chat window. The user reviews this response and, if they have further questions, enter them again in the chat window.

[1508] Through these steps, the system provides fast and accurate responses to user inquiries.

[1509] (Example 1)

[1510] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1511] In today's digital transformation, there is a demand for systems that can respond quickly and accurately to user inquiries. Conventional systems often process inquiries manually, which is time-consuming and labor-intensive. Furthermore, the quality of responses can vary depending on the person handling them, leading to decreased customer satisfaction. Therefore, this invention aims to solve these problems by providing a system that uses a generative AI model to deliver consistent, high-quality responses quickly.

[1512] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1513] In this invention, the server includes means for tokenizing user input and creating a prompt, means for passing the tokenized data to a generation AI model to generate a response, and means for converting the generated response into JSON format. This enables a rapid and accurate response to user inquiries.

[1514] A "user" refers to a person who uses the system to make an inquiry.

[1515] "Means for receiving input" refers to a device or software that has the function of receiving inquiries from users.

[1516] "JSON format" is an abbreviation for JavaScript Object Notation, and refers to a format that structures data using key-value pairs and represents it in an easy-to-read format.

[1517] "Tokenization" refers to the process of dividing input text into smaller units such as words and phrases.

[1518] A "prompt" refers to the initial conditions or question-style sentences that a generative AI model uses to receive input.

[1519] A "generative AI model" refers to a model that uses machine learning techniques to generate responses based on user input.

[1520] A "large-scale language model" refers to a model that is trained using a large amount of text data to understand and generate complex language patterns.

[1521] "Means for generating responses" refers to devices or software that create appropriate responses to user inquiries based on tokenized data.

[1522] "Means for sending back a response" refers to a device or software that has the function of sending back the response generated by the server to the user.

[1523] A "chat window" refers to an interface for users to input and display text-based inquiries and responses.

[1524] As an embodiment of the present invention, a method for constructing a chat system using a generative AI model will be described. This system significantly improves the efficiency of customer service by providing quick and accurate responses to questions entered by users.

[1525] System Configuration

[1526] This system consists of the following main components:

[1527] 1. User Interface: Receives inquiries from users.

[1528] 2. Terminal: Receives data from the user interface and sends it to the server.

[1529] 3. Server: Performs data processing and response generation.

[1530] 4. Communication method: The generated response is sent back to the user.

[1531] Hardware and software to be used

[1532] Hardware: Computers and smartphones used by users, and server machines on which servers are installed.

[1533] Software: Web pages and applications with user interfaces, software for sending requests used by devices, generative AI models running on servers (e.g., GPT-3), and Python libraries used for data processing (e.g., NLTK, spaCy, etc.).

[1534] System Operation Description

[1535] The system operates as follows:

[1536] 1. User input: The user enters their question into the chat window. In this example, they type, "How does KOALA work?"

[1537] 2. Terminal generation and transmission of requests: The terminal converts the user's input into JSON format and sends it to the server as a POST request.

[1538] 3. Receiving and processing requests by the server: The server receives the request and tokenizes the user's input. This process uses natural language processing libraries such as NLTK (Natural Language Toolkit).

[1539] 4. Response generation by generative AI model: The server inputs tokenized data into the generative AI model to generate an appropriate response. In this example, the response "KOALA processes data in real time using advanced analytical algorithms" is obtained.

[1540] 5. Server-side response conversion and transmission: The generated response is converted to JSON format and sent back to the terminal.

[1541] 6. Displaying responses via the terminal: The terminal displays the response received from the server in the chat window. This allows the user to see the generated response.

[1542] Specific example

[1543] For example, if a user enters the question "What are the main functions of KOALA?", the system will behave as follows:

[1544] 1. User input: The user types "What are the main functions of KOALA?" into the chat window.

[1545] 2. Terminal generation and sending of requests: The terminal converts the question into JSON format and sends it to the server.

[1546] 3. Server receives and processes the request: The server receives the query and performs tokenization.

[1547] 4. Response generation using a generative AI model: The server uses a generative AI model to generate a response that says, "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1548] 5. Server conversion and transmission of response: The server converts the generated response into JSON format and sends it to the terminal.

[1549] 6. Display of responses by the device: The device displays the response in the chat window.

[1550] Example of a prompt

[1551] "What are the main functions of KOALA?"

[1552] This system enables quick and accurate responses to user inquiries, leading to increased productivity for sales representatives and improved customer satisfaction.

[1553] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1554] Step 1:

[1555] User input

[1556] The user enters a question into the chat window. In this example, they type, "How does KOALA work?". The user directly types the text on the keyboard, and it appears in the chat window. This input is then processed.

[1557] Specific operation: The user accesses a chat window in a web browser or application and types their inquiry using the keyboard. The user's input is displayed in the chat window in real time.

[1558] Step 2:

[1559] Request generation by the terminal

[1560] The terminal converts the user's input into JSON format and sends a POST request to the server. The input is the user's question text, "How does KOALA work?". Converting this to JSON format results in the following:

[1561] json

[1562] {

[1563] "Question": "How does KOALA work?"

[1564] }

[1565] Specific operation: Use JavaScript's XMLHttpRequest and fetch APIs to convert user input data into JSON format and send a POST request to the server.

[1566] Step 3:

[1567] Server receives request

[1568] The server receives a request in JSON format sent from the terminal. The input is data in JSON format.

[1569] json

[1570] {

[1571] "Question": "How does KOALA work?"

[1572] }

[1573] The server receives this data for analysis.

[1574] Specific operation: The server uses a web framework such as Flask or Django in Python to receive POST requests from the terminal and parse their contents.

[1575] Step 4:

[1576] Execute tokenization process

[1577] The server tokenizes the user's question received. The input is the user's question text. Tokenization divides this into units such as words and phrases. Natural language processing libraries such as NLTK and spaCy are used for this process. For example, it might be divided into "KOALA", "is", "how", "works", "of", "is it?".

[1578] Specific operation: The server uses libraries such as NLTK and spaCy to tokenize the user's input text.

[1579] Step 5:

[1580] Response generation using generative AI models

[1581] The server inputs tokenized data into a generative AI model, which then generates an appropriate response. The input is tokenized data. Based on this, the generative AI model (e.g., GPT-3) generates a response such as, "KOALA uses advanced analytical algorithms to process the data in real time."

[1582] Specific operation: The server uses the OpenAI API or similar tools to send prompts to the generated AI model, causing it to generate responses.

[1583] Step 6:

[1584] Response transformation

[1585] The server converts the generated response into JSON format. The input is the generated response text. When converted to JSON format, it looks like this:

[1586] json

[1587] {

[1588] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[1589] }

[1590] Specific operation: Use Python's standard library or other JSON libraries to convert the generated response into JSON format.

[1591] Step 7:

[1592] Sending a response

[1593] The server returns the converted JSON data to the terminal. The input is a response data in JSON format. This is then sent to the terminal.

[1594] Specific operation: The server uses features from Python's Flask or Django to return JSON data as an HTTP response.

[1595] Step 8:

[1596] Display of response

[1597] The terminal displays the response data received from the server in the chat window. The input is response data in JSON format.

[1598] json

[1599] {

[1600] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[1601] }

[1602] The device analyzes this information and displays it in the chat window in a format that is easy for the user to understand.

[1603] Specific operation: Using JavaScript, the response text is extracted from the received JSON data and inserted into an HTML element for display.

[1604] (Application Example 1)

[1605] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1606] In food delivery services, responding quickly and accurately to customer inquiries directly leads to improved customer satisfaction and operational efficiency. However, traditional systems often suffer from delayed responses or inappropriate answers depending on the nature of the inquiry. Furthermore, providing real-time information on delivery progress and menu item details is difficult. This results in increased customer dissatisfaction and a decline in service quality.

[1607] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1608] In this invention, the server includes means for receiving input from a user, means for tokenizing the user's input, means for passing the tokenized data to a generative AI model to generate a response, means for converting the generated response into JSON format, means for returning the response to the user, means for generating a response specific to food delivery services, and means for providing real-time information on the delivery progress and menu item details of food deliveries. This makes it possible to respond quickly and accurately to user inquiries and to provide real-time information on delivery progress and menu details.

[1609] "Means of receiving user input" refers to providing an interface for users of a food delivery service to enter inquiries or questions using a smartphone or other device.

[1610] "Methods for tokenizing user input" refer to processes that analyze the text entered by the user and divide it into words and phrases.

[1611] "Means of passing tokenized data to a generative AI model to generate a response" refers to the process by which a generative AI model generates an appropriate response based on tokenized input.

[1612] "Means for converting the generated response to JSON format" refers to the process of converting the generated text response into JSON, a common data exchange format.

[1613] "Means of sending the response back to the user" refers to sending the converted response data to the user's device so that the user can confirm the response.

[1614] "Means for generating responses specifically for food delivery services" refers to processing that provides professional and appropriate responses to questions and inquiries related to food delivery.

[1615] "A means of providing real-time responses regarding the progress of food delivery and details of menu items" refers to a system that immediately provides users with real-time information regarding the current progress of delivery and specific menu details.

[1616] Modes for carrying out the invention

[1617] Program processing

[1618] The system that embodies this application is a chat support system specifically designed for food delivery services, providing quick and accurate responses to user inquiries through a smartphone application.

[1619] Hardware and software to be used

[1620] Hardware: Cloud servers or on-premises servers

[1621] Software: Flask (Web framework), OpenAI API

[1622] Data processing and data calculation

[1623] 1. Receiving user input:

[1624] The server receives questions entered by users through a smartphone application. Users initiate an inquiry by typing their questions into the application's chat window.

[1625] 2. Tokenization of user input:

[1626] The server tokenizes the user's input. Tokenization is the process of analyzing the user's text and dividing it into words and phrases.

[1627] 3. Pass the tokenized data to a generative AI model to generate a response:

[1628] The tokenized data is input into a generative AI model (e.g., GPT-3) to generate an appropriate response. The generative AI model generates the most appropriate text response to the user's question.

[1629] 4. Converting the generated response to JSON format:

[1630] The generated response is converted to JSON format. JSON is a common format for data exchange and allows for proper formatting of the response.

[1631] 5. Sending the response back to the user:

[1632] The server sends the generated JSON response back to the user's smartphone application. The user can then view the response in the application's chat window.

[1633] 6. Generating responses specifically for food delivery services:

[1634] The generative AI model has the capability to generate responses specifically tailored to food delivery services. It quickly provides specific information to users in response to their questions, such as delivery progress and details of menu items.

[1635] 7. Real-time responses regarding delivery progress and menu item details:

[1636] The server provides real-time information on the progress of food deliveries and details of menu items. This allows users to quickly obtain information about the current delivery status and menu items.

[1637] Adding specific examples

[1638] As a concrete example, let's explain the behavior when a user enters the question, "How long will it take for my ordered pizza to arrive?"

[1639] 1. The user enters their question into the chat window of the smartphone application.

[1640] 2. The server receives this question and tokenizes the input text.

[1641] 3. Pass the tokenized data to the generative AI model (GPT-3) and use the following prompts to generate a response:

[1642] As a support bot for a food delivery service, answer user inquiries.

[1643] User: How long will it take for my pizza to arrive?

[1644] Bot:

[1645] 4. The generative AI model generates the response, "Pizza delivery usually takes 30-40 minutes, but the current delivery time can be specifically tracked based on GPS."

[1646] 5. Convert this response to JSON format and send it back to the user's smartphone application.

[1647] 6. Users can check the response on the application and obtain information about the delivery status.

[1648] In this application, users can receive quick and accurate answers to their inquiries, and real-time information on delivery progress and details of menu items can also be provided.

[1649] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1650] Program processing steps

[1651] Step 1:

[1652] The user enters their question into the chat window of the smartphone application. When the user types a message such as "How long will it take for my ordered pizza to arrive?" and presses the send button, the input data is sent to the server. At this stage, the input is the user's question text.

[1653] Step 2:

[1654] The server receives input text from the user and tokenizes it. Specifically, it divides the input text into words and phrases, converting the data into a format that is easy to analyze. In this example, the question text received as input, "How long will it take for my ordered pizza to arrive?", is output as tokenized data.

[1655] Step 3:

[1656] The server passes the tokenized data to a generative AI model, which then generates a response. The generative AI model (e.g., GPT-3) generates an appropriate response according to the format of the prompt. A concrete example of a prompt is as follows:

[1657] As a support bot for a food delivery service, answer user inquiries.

[1658] User: How long will it take for my pizza to arrive?

[1659] Bot:

[1660] Given tokenized data as input, a generative AI model generates a response such as, "Pizza delivery usually takes 30-40 minutes, but the current delivery time can be specifically tracked based on GPS." This is the output data.

[1661] Step 4:

[1662] The server converts the generated response into JSON format. This conversion process transforms the data into the standard JSON format. The generated response text is converted into JSON data like the following:

[1663] json

[1664] {

[1665] "Response": "Pizza delivery usually takes 30-40 minutes, but current delivery times can be tracked specifically based on GPS."

[1666] }

[1667] It receives the generated response text as input and outputs data in JSON format.

[1668] Step 5:

[1669] The server sends this JSON data back to the user's device. The device (smartphone application) receives this data and processes it to display it in the chat window. The output is receiving JSON data as input and displaying the response in the user's chat window.

[1670] Step 6:

[1671] The user checks the response displayed in the chat window of the smartphone application. This response allows the user to receive real-time information about the delivery progress and details of the food delivery menu. Specifically, the user reads and understands the displayed response. The input in this step is the response text displayed in the application, and the output is the user's understanding and satisfaction.

[1672] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1673] This invention relates to a chat system that utilizes a generative AI model combined with an emotion engine that recognizes user emotions. This system significantly improves the efficiency and satisfaction of customer service by providing quick and emotion-appropriate responses to user inquiries.

[1674] System Configuration

[1675] This system consists of the following main components:

[1676] 1. User interface for receiving inquiries

[1677] 2. Terminals that receive and process data from the user interface.

[1678] 3. Servers that perform data processing and response generation.

[1679] 4. Means of analyzing user emotions using an emotion engine

[1680] 5. Communication means for returning the generated response to the user.

[1681] Specific steps

[1682] 1. User input:

[1683] A user enters a question in the chat window such as, "How does KOALA work?"

[1684] 2. Sending a request from the terminal to the server:

[1685] The terminal converts the user's input into JSON format and sends a POST request to the server. This request contains the user's question data.

[1686] 3. Server-side processing:

[1687] The server receives the request and tokenizes the user's input. The tokenized data is passed to a generative AI model (e.g., a large-scale language model) to generate an appropriate response.

[1688] 4. Analysis using an emotion engine:

[1689] Before the server generates a response using a generative AI model, it passes the user's input to an emotion engine for analysis. The emotion engine analyzes the user's emotional state in real time and incorporates the analysis results into the response generation.

[1690] 5. Response generation and transformation:

[1691] Based on the analysis results of the emotion engine, a generative AI model generates response text. The generated response text is converted to JSON format. The converted data has the following structure: "response": "KOALA processes the data in real time using advanced analytical algorithms."

[1692] 6. Sending a response from the server to the terminal:

[1693] The server sends this JSON data back to the device.

[1694] 7. Displaying responses to the user:

[1695] The terminal displays the response data received from the server in the chat window. The user reviews this response and, if they have further questions, enter them again in the chat window.

[1696] Specific example

[1697] If a user enters the question "What are the main functions of KOALA?", it will behave as follows:

[1698] 1. The user enters the question into the device.

[1699] 2. The terminal sends the question data to the server.

[1700] 3. The server processes the data and analyzes the user's emotions using an emotion engine before using generative AI models.

[1701] 4. The emotion engine analyzes the user's emotional state, and based on the results, a generative AI model generates a response.

[1702] 5. The server converts the generated response into JSON format and sends it to the terminal.

[1703] 6. The device displays the response in the chat window.

[1704] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1705] The introduction of this system will enable responses to user inquiries to become more emotionally responsive, which is expected to improve customer satisfaction.

[1706] The following describes the processing flow.

[1707] Step 1:

[1708] The user types a question into the chat window. For example, they might type a specific question like, "What are the main features of KOALA?"

[1709] Step 2:

[1710] The terminal receives user input and converts this input data into JSON format. The converted data has the following JSON structure:

[1711] json

[1712] {

[1713] "user_input": "What are the main functions of KOALA?"

[1714] }

[1715] Step 3:

[1716] The device sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[1717] Step 4:

[1718] The server processes the POST request received from the terminal. At this time, the / chat endpoint of the Flask application receives the request.

[1719] Step 5:

[1720] The server extracts user input (user_input) from the received request data. Specifically, it extracts the data as follows:

[1721] python

[1722] data = request.json

[1723] user_input = data['user_input']

[1724] Step 6:

[1725] The server passes the extracted user input to the emotion engine for sentiment analysis. The sentiment engine analyzes the user's emotions as follows:

[1726] python

[1727] emotion = emotion_engine.analyze(user_input)

[1728] The emotion engine returns the analysis results. For example, it may determine that the user's input contains emotions such as "joy" or "anxiety."

[1729] Step 7:

[1730] The server performs a tokenization process based on the analysis results of the emotion engine. The tokenization process proceeds as follows:

[1731] python

[1732] inputs = tokenizer.encode(user_input, return_tensors='pt')

[1733] Step 8:

[1734] The server passes the tokenized data and sentiment analysis results to a generative AI model (large-scale language model) to generate response text. Specifically, the process is as follows:

[1735] python

[1736] outputs = model.generate(inputs, max_length=200, num_return_sequences=1, emotion=emotion)

[1737] response = tokenizer.decode(outputs[0], skip_special_tokens=True)

[1738] Step 9:

[1739] The server converts the generated response into JSON format. The converted JSON data has the following structure:

[1740] json

[1741] {

[1742] "Response": "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1743] }

[1744] Step 10:

[1745] The server sends this JSON data back to the device.

[1746] Step 11:

[1747] The terminal analyzes the response data received from the server and displays it in the chat window. The user can review this response and, if they have further questions, enter them again in the chat window.

[1748] Through the steps described above, the system provides fast and emotionally responsive responses to user inquiries. This enables responses that take user emotions into consideration, which is expected to improve customer satisfaction.

[1749] (Example 2)

[1750] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1751] Traditional chat systems had the problem of lowering user satisfaction because they provided uniform responses without considering user emotions. Furthermore, the inconsistent quality of responses made it difficult to provide satisfactory support.

[1752] The specific processing performed by the specific 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 tokenizing user input, means for passing the tokenized data to sentiment analysis means for analysis, and means for passing the tokenized data to a generation AI model based on the sentiment analysis results to generate a response. This makes it possible to provide a response that corresponds to the user's emotions.

[1753] "User input" refers to the text or questions that users type into the chat window.

[1754] "Tokenization" refers to the process of dividing a text into units such as words or sentences.

[1755] "Sentiment analysis method" refers to a method that analyzes emotions from user input text and outputs the results.

[1756] A "generative AI model" refers to an algorithm that uses artificial intelligence to generate responses to user input.

[1757] "Response generation" refers to the process of generating an appropriate response based on user input.

[1758] "JSON format" is an abbreviation for JavaScript Object Notation, and refers to a format for representing data in a structured way.

[1759] This invention relates to a chat system that utilizes a generative AI model combined with an emotion engine that recognizes user emotions. This system significantly improves the efficiency and satisfaction of customer service by providing quick and emotion-appropriate responses to user inquiries.

[1760] System Configuration

[1761] This system consists of the following main components:

[1762] 1. User interface for receiving user input

[1763] 2. Terminals that receive and process data from the user interface.

[1764] 3. Servers that perform data processing and response generation.

[1765] 4. Means of analyzing user emotions using an emotion engine

[1766] 5. Communication means for returning the generated response to the user.

[1767] Program processing

[1768] The user accesses the system and enters a question or message into the chat window. For example, the user might type, "How does KOALA work?" The terminal converts this input into JSON format and sends a POST request to the server. This request contains the user's question data.

[1769] The server receives the request and tokenizes the user's input. The tokenized data is first passed to an emotion engine to analyze the user's emotional state. The emotion engine can use, for example, the Microsoft Azure Text Analytics API. Based on the analysis results, the tokenized data is passed to a generative AI model (for example, a large-scale language model such as GPT-3) to generate an appropriate response.

[1770] The generated response text is converted back into JSON format and sent to the terminal. The terminal parses the received data and displays the response in the chat window. This allows the user to receive responses that are appropriate to their emotions.

[1771] Specific example

[1772] For example, if a user enters the question "What are the main functions of KOALA?", the system will behave as follows:

[1773] 1. The user enters the question into the device.

[1774] 2. The terminal sends the question data to the server.

[1775] 3. The server processes the data and analyzes the user's emotions using an emotion engine before using generative AI models.

[1776] 4. The emotion engine analyzes the user's emotional state, and based on the results, a generative AI model generates a response.

[1777] 5. The server converts the generated response into JSON format and sends it to the terminal.

[1778] 6. The device displays the response in the chat window.

[1779] At this point, the user will see the following response: "KOALA's main features include data analysis, real-time processing, and advanced visualization."

[1780] The introduction of this system will enable responses to user inquiries to become more emotionally responsive, which is expected to improve customer satisfaction.

[1781] Example of a prompt:

[1782] "The user is asking: 'How does KOALA work?' Its emotional state is neutral. Please provide an appropriate response to this."

[1783] This invention enables the generation of sophisticated responses that take emotions into consideration, and is expected to significantly improve the user experience.

[1784] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1785] Step 1:

[1786] The user enters a question into the chat window. This input is natural language text, such as "How does KOALA work?".

[1787] Input: User's question text

[1788] Output: Input completion event in the chat window

[1789] Step 2:

[1790] The terminal receives user input and converts it to JSON format. The JSON data will be in the following format:

[1791] json

[1792] {

[1793] "user_input": "How does KOALA work?"

[1794] }

[1795] Input: User's question text

[1796] Output: Question data in JSON format

[1797] Step 3:

[1798] The terminal sends the converted JSON data to the server as a POST request. This request contains the user's question data.

[1799] Input: Question data in JSON format

[1800] Output: POST request to the server

[1801] Step 4:

[1802] The server receives the POST request, parses the JSON data, and extracts the user input portion.

[1803] Input: JSON data sent to the server

[1804] Output: Extracted user input text

[1805] Step 5:

[1806] The server tokenizes the extracted input text. For example, it might be split into "KOALA", "is", "how", "works", "is", and "is it?".

[1807] Input: User-input text

[1808] Output: Tokenized data

[1809] Step 6:

[1810] The server passes the tokenized data to the sentiment engine for sentiment analysis. The sentiment engine uses Microsoft Azure's Text Analytics API, among others. For example, it might analyze the sentiment as "Sentimental state: Neutral."

[1811] Input: Tokenized user input data

[1812] Output: Emotion analysis results

[1813] Step 7:

[1814] The server passes the tokenized data, based on the sentiment analysis results, to a generative AI model to generate a response. The generative AI model uses large-scale language models such as GPT-3. Here, a response such as "KOALA uses advanced analytical algorithms to process data in real time" is generated.

[1815] Input: Tokenized data and sentiment analysis results

[1816] Output: Generated response text

[1817] Step 8:

[1818] The server converts the generated response text into JSON format. The resulting data will look like this:

[1819] json

[1820] {

[1821] "Response": "KOALA uses advanced analytical algorithms to process data in real time."

[1822] }

[1823] Input: Generated response text

[1824] Output: Response data in JSON format

[1825] Step 9:

[1826] The server sends response data in JSON format to the terminal.

[1827] Input: Response data in JSON format

[1828] Output: Sending a response to the terminal

[1829] Step 10:

[1830] The terminal parses the JSON data received from the server and extracts the response text. The extracted response text is then displayed in the chat window.

[1831] Input: JSON data sent from the server

[1832] Output: Response text displayed in the chat window

[1833] This allows users to see appropriate responses tailored to their emotions in real time.

[1834] (Application Example 2)

[1835] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1836] Traditional chat systems have a problem in that they cannot flexibly change responses according to the user's emotions and situation, and therefore cannot sufficiently increase user satisfaction. In particular, e-commerce sites receive a wide range of customer inquiries, and quick and accurate responses are required, so providing responses that respond to emotions is effective. However, current systems do not have emotion analysis functions and can only return a uniform response.

[1837] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1838] In this invention, the server includes means for receiving user input, means for analyzing the user's emotions using an emotion analysis engine, means for generating a response using a generative AI model based on the analysis results, means for converting the generated response into JSON format, and means for returning the response to the user. This makes it possible to provide a response that is appropriate to the user's emotions.

[1839] "Means of receiving user input" refers to a function that receives messages and inquiries sent by users through an interface such as a chat window.

[1840] "Methods for tokenizing user input" refer to the process of dividing user-entered text into meaningful units to facilitate natural language processing.

[1841] "A means of passing tokenized data to a generative AI model to generate a response" refers to a function that inputs tokenized data into a generative AI model and generates an appropriate response for the user.

[1842] "Means for converting generated responses into JSON format" refers to a function that converts text responses generated by generative AI models into JSON format, facilitating data communication and display.

[1843] "Means of sending a response back to the user" refers to a function that sends a generated response from the server to the user's device and displays it to the user.

[1844] "Means of analyzing a user's emotions using an emotion analysis engine" refers to an engine and its functions for analyzing and understanding a user's emotional state from their text input.

[1845] "Means for adjusting responses based on analysis results" refers to a function that adjusts the content of responses generated by generative AI models based on the analysis results obtained by the emotion analysis engine.

[1846] This invention relates to a chat system for e-commerce sites that combines an emotion recognition engine and a generative AI model. This system enables the rapid and appropriate provision of responses tailored to the user's emotions. The specific configuration and processing methods of this system are described below.

[1847] System Configuration

[1848] This system consists of the following main components:

[1849] 1. Means for receiving user input:

[1850] The user enters text into the chat window via a smartphone app.

[1851] 2. Means of tokenizing user input:

[1852] Divide the text into meaningful units and convert them into a format that is easy to process.

[1853] 3. Means for analyzing user emotions using an emotion analysis engine:

[1854] The system analyzes the emotional state contained in the user's input. The software used includes the Hugging Face transformers library.

[1855] 4. Means of passing tokenized data to a generative AI model to generate a response:

[1856] Based on tokenized data and sentiment analysis results, a generative AI model generates an appropriate response. A large-scale language model is used here.

[1857] 5. Means for converting the generated response to JSON format:

[1858] The generated text response is converted to JSON format to improve communication efficiency.

[1859] 6. Means of returning the response to the user:

[1860] The server sends a response to the user's smartphone and displays it in the chat window.

[1861] Specific example

[1862] If a user enters the question "My order hasn't arrived yet. What's going on?" into the chat window of an online shopping site, the system will operate as follows:

[1863] 1. Receiving user input:

[1864] The device receives the text entered by the user.

[1865] 2. Emotion analysis:

[1866] The device passes the received text to an emotion analysis engine, which analyzes whether the user is expressing anxiety or anger.

[1867] 3. Response generation:

[1868] Based on the results of the emotion analysis, a generative AI model generates a response such as, "I apologize for the delay. I will check it immediately."

[1869] 4. Sending a response:

[1870] The generated response is converted to JSON format and sent from the server to the user's terminal.

[1871] 5. Display of response:

[1872] The user checks the response displayed in the chat window.

[1873] The software used includes the Hugging Face transformers library for emotion analysis and a generative AI model (large-scale language model). This enables appropriate responses based on the user's emotions, thereby improving customer satisfaction.

[1874] Example of a prompt:

[1875] "My order hasn't arrived yet. What's going on?"

[1876] This invention provides responses that take user emotions into consideration, significantly improving the quality of customer service on e-commerce sites.

[1877] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1878] Step 1:

[1879] The user types "My order hasn't arrived yet. What's going on?" into the chat window. This is the starting point for the entire system.

[1880] Step 2:

[1881] The terminal receives user input and records it as text data. This input is the user's text message, which serves as initial data for processing.

[1882] Step 3:

[1883] The terminal tokenizes the received text data. Tokenization is the process of dividing text into meaningful units (tokens). For example, it might be divided into "order," "but," "haven't," "arrived," "how," "is," "what," "is," "what," "is," "what," "is." This process makes the text data easier to analyze.

[1884] Step 4:

[1885] The tokenized text data is passed to an emotion analysis engine to analyze the user's emotions. The emotion analysis engine uses the Hugging Face transformers library to determine the user's emotional state (e.g., anxiety, anger). The output is the emotion classification result, with "anxiety" being output as an example.

[1886] Step 5:

[1887] Based on the sentiment analysis results, the server passes the tokenized data to a generative AI model to generate an appropriate response. The generative AI model uses a large-scale language model to generate a response such as, "I apologize for the delay. I will check on this immediately."

[1888] Step 6:

[1889] The generated response is converted to JSON format. The server converts the text response to JSON format in preparation for the next communication process. For example, it will be converted to a format like {"response": "Sorry for the wait. I will check it shortly."}.

[1890] Step 7:

[1891] The server sends data in JSON format to the terminal. The terminal receives the JSON data sent from the server and prepares for the next processing step.

[1892] Step 8:

[1893] The terminal parses the received JSON data and displays a response in the chat window. The user can then see the response displayed in the chat window, such as "Sorry for the wait. I will check it immediately," and ask further questions or take actions.

[1894] This series of processing steps allows users to receive appropriate responses that match their emotions, resulting in a high level of satisfaction.

[1895] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1896] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1897] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[1898] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1899] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[1900] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[1901] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[1902] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[1903] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[1904] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[1905] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[1906] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[1907] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[1908] 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.

[1909] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[1910] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[1911] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[1912] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[1913] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[1914] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[1915] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted as being incorporated by reference.

[1916] The following is further disclosed regarding the embodiments described above.

[1917] (Claim 1)

[1918] A means of receiving input from the user,

[1919] A means of tokenizing user input,

[1920] A means of passing tokenized data to a generative AI model to generate a response,

[1921] A means of converting the generated response into JSON format,

[1922] A means of sending a response back to the user,

[1923] A system that includes this.

[1924] (Claim 2)

[1925] The system according to claim 1, wherein the generative AI model is a large-scale language model.

[1926] (Claim 3)

[1927] The system according to claim 1, further comprising means for displaying user input in a chat window.

[1928] "Example 1"

[1929] (Claim 1)

[1930] A means of receiving input from the user,

[1931] A means of converting user input into JSON format and sending it,

[1932] A means by which the server tokenizes user input and creates a prompt,

[1933] A means of passing tokenized data to a generating AI model to generate a response,

[1934] A means of converting the generated response into JSON format,

[1935] A means of sending a response back to the user,

[1936] A system that includes this.

[1937] (Claim 2)

[1938] The system according to claim 1, wherein the generative AI model is a large-scale language model.

[1939] (Claim 3)

[1940] The system according to claim 1, further comprising means for displaying user input in a chat window.

[1941] "Application Example 1"

[1942] (Claim 1)

[1943] A means of receiving input from the user,

[1944] A means of tokenizing user input,

[1945] A means of passing tokenized data to a generative AI model to generate a response,

[1946] A means of converting the generated response into JSON format,

[1947] A means of sending a response back to the user,

[1948] A means of generating responses specifically for food delivery services,

[1949] A means of providing real-time answers regarding the progress of food delivery and details of menu items,

[1950] A system that includes this.

[1951] (Claim 2)

[1952] The system according to claim 1, wherein the generative AI model is a large-scale language model.

[1953] (Claim 3)

[1954] The system according to claim 1, further comprising means for displaying user input in a chat window.

[1955] "Example 2 of combining an emotion engine"

[1956] (Claim 1)

[1957] A means of receiving input from the user,

[1958] A means of tokenizing user input,

[1959] A means of passing tokenized data to a sentiment analysis tool for analysis,

[1960] A means of generating a response by passing tokenized data, based on the results of sentiment analysis, to an AI model, and

[1961] A means of converting the generated response into JSON format,

[1962] A means of sending a response back to the user,

[1963] A system that includes this.

[1964] (Claim 2)

[1965] The system according to claim 1, wherein the generative AI model is a large-scale language model.

[1966] (Claim 3)

[1967] The system according to claim 1, further comprising means for displaying user input in a chat window.

[1968] "Application example 2 when combining with an emotional engine"

[1969] (Claim 1)

[1970] A means of receiving input from the user,

[1971] A means of tokenizing user input,

[1972] A means of passing tokenized data to a generative AI model to generate a response,

[1973] A means of converting the generated response into JSON format,

[1974] A means of sending a response back to the user,

[1975] A means of analyzing a user's emotions using an emotion analysis engine,

[1976] A means of adjusting the response based on the analysis results,

[1977] A system that includes this.

[1978] (Claim 2)

[1979] The system according to claim 1, wherein the generative AI model is a large-scale language model.

[1980] (Claim 3)

[1981] The system according to claim 1, further comprising means for displaying user input in a chat window. [Explanation of Symbols]

[1982] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. A means of receiving input from the user, A means of tokenizing user input, A means of passing tokenized data to a generative AI model to generate a response, A means of converting the generated response into JSON format, A means of sending a response back to the user, A system that includes this.

2. The system according to claim 1, wherein the generative AI model is a large-scale language model.

3. The system according to claim 1, further comprising means for displaying user input in a chat window.

Images