System

Abstract

A system is provided.SOLUTION: This system is provided with a means for analyzing the contents of an inquiry and classifying it into a designated category, a means for generating a reply sentence based on the analyzed result, a means for adding an attached file to the reply sentence and a means for transmitting a mail.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] In modern companies, the number of internal inquiries is increasing, requiring quick and efficient responses. Traditionally, responding to internal inquiries required significant human intervention, resulting in significant time and cost burdens. In particular, processes such as analyzing email content, generating replies, and organizing and sending attachments were time-consuming and reduced overall business efficiency. Furthermore, responses that take the sender's emotions into account were sometimes required, making maintaining and improving response quality a difficult challenge. [Means for solving the problem]

[0005] The present invention provides a system that analyzes the content of an inquiry and classifies it into a specified category, and further proposes a system that includes a system that generates a reply based on the analysis results, a system that adds an attachment to the reply, and a system that sends an email.

[0006] Specifically, it analyzes the content of received emails using natural language processing (NLP), extracts keywords and phrases from the email body, and has the means to infer the sender's sentiment. Based on the analysis results, it classifies the email into the corresponding category and automatically generates a reply. The generated reply also includes a means to provide a user interface that allows the user to edit, approve, or reject the reply. The process of automatically searching for attachments required for the reply and attaching them to the email is performed by RPA (Robotic Process Automation), and the email is finally sent, thereby streamlining the entire process and minimizing human intervention.

[0007] "Inquiry Content" refers to the entire information in the body, subject, and attachments of the email or message sent by the User.

[0008] "Analysis" refers to the process of analyzing the content of received emails and messages using natural language processing technology to identify intent and emotion.

[0009] "Classification" refers to sorting emails or messages into predefined categories based on the results of analysis.

[0010] "Reply text" refers to a reply text automatically generated by the system based on the results of analysis and classification.

[0011] An "attachment" is a file sent along with an email or message.

[0012] "Natural language processing (NLP)" refers to a broad range of technologies that enable computers to understand, interpret, and generate human language.

[0013] "Inferring emotions" refers to identifying the sender's emotional and psychological state from the content of a received email or message.

[0014] "User Interface" refers to the visual and functional interface through which a user interacts with and performs operations on a system.

[0015] "Robotic Process Automation (RPA)" refers to the technology of software robots used to streamline business processes.

[0016] "Sending" refers to the act of sending the email over the Internet to deliver the generated reply and any necessary attachments to the recipient.

[0017] "Means" refers to the methods or techniques used to achieve an end.

[0018] "User" means any individual or organization that uses the System to send email or take action on email received. [Brief explanation of the drawings]

[0019] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0020] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0021] First, the terms used in the following description will be explained.

[0022] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0023] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0024] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0025] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0026] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0027] [First embodiment]

[0028] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0029] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0030] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0031] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0032] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0033] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0034] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0035] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0036] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0037] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0038] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0039] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0040] This invention provides a system for automating and efficiently responding to inquiries. This system automates a series of processes: analyzing the content of an inquiry, classifying it into a specified category, generating a reply, adding an attachment, and sending an email. The main elements that make up this system are a server, a terminal, and a user.

[0041] 1. Content analysis and categorization by the server

[0042] server

[0043] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject line, and attachments, and to identify the intent of the specific inquiry and the sentiment of the sender. For this purpose, the server uses natural language processing (NLP) techniques to extract keywords and phrases from the email body and identify the sender's request and situation. Based on the analysis results, the server classifies the email into predefined categories (e.g., technical support, general inquiry, emergency response, etc.).

[0044] As a concrete example, when a server receives and analyzes an email that says, "Please send me the minutes of the meeting. It's urgent," the server will identify two main elements from the email's content: "Request for meeting minutes" and "Urgent," and categorize it based on those elements.

[0045] 2. Generating a Reply

[0046] server

[0047] The server generates an appropriate reply based on the results of category classification and sentiment analysis, using natural language generation (NLG) technology. The server stores different reply templates for each specified category, and automatically selects the appropriate template to generate a reply.

[0048] As a specific example, the server selects a template corresponding to a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0049] 3. Adding attachments to replies

[0050] Terminal

[0051] Robotic Process Automation (RPA) installed on the device automatically identifies the necessary attachments and adds them to the email. The device searches for the appropriate file in the company's file system or database, and once found, attaches it to the email.

[0052] As a concrete example, the device searches the corporate file system for a file labeled "meeting minutes," extracts it, and attaches it to an email.

[0053] 4. Replying to emails

[0054] Terminal

[0055] The RPA on the terminal checks the generated reply and attachments, and finally sends the email to the user. During this process, it checks whether the components of the email are complete and corrects any deficiencies.

[0056] As a specific example, the terminal sends an email reply to the user with the reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait." and the searched meeting minutes file.

[0057] Editing and Approval Features

[0058] User

[0059] It provides an interface that allows users to monitor this process behind the scenes and review, edit, and approve or reject the generated replies, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[0060] This system enables efficient responses to internal inquiries, saving time and costs. It also takes into consideration the quality of the email and the feelings of the sender, improving the overall quality of responses.

[0061] The processing flow will be explained below.

[0062] Step 1:

[0063] server

[0064] The server retrieves new emails from the mailbox and saves the subject, body, and attachments of the received email to a file for analysis.

[0065] Step 2:

[0066] server

[0067] The server analyzes the body, subject, and attachments of received emails using natural language processing (NLP) technology. Specifically, it extracts keywords and important phrases from the email body to identify the email's intent and request. In parallel, it performs sentiment analysis to identify the sender's psychological state and emotions.

[0068] Step 3:

[0069] server

[0070] Based on the analysis results, the server categorizes the email into a corresponding category, such as technical support, general inquiries, or emergency response, based on pre-defined rules.

[0071] Step 4:

[0072] server

[0073] The server uses the classified categories and the results of sentiment analysis to select an appropriate reply template. It then uses natural language generation (NLG) technology to generate a reply based on the selected template. At this stage, the reply text is enhanced to reflect the sender's sentiment.

[0074] Step 5:

[0075] Terminal

[0076] The device's RPA (Robotic Process Automation) searches for the required attachments by scanning the company's file system and databases and identifying the appropriate file based on the results of email analysis and the content of the reply.

[0077] Step 6:

[0078] Terminal

[0079] The RPA on the terminal adds the searched attachments to the email, so that the necessary documents are attached to the reply email according to the user's request.

[0080] Step 7:

[0081] Terminal

[0082] The RPA on the device checks all generated replies and attachments to ensure they are properly structured and checked for consistency and completeness.

[0083] Step 8:

[0084] Terminal

[0085] After the device performs a final check, it automatically sends an email containing the generated reply and attachment, completing the entire process of responding to the inquiry.

[0086] Step 9:

[0087] User

[0088] Users can review the generated replies through the system's user interface, and can edit, approve, or reject them as needed to ensure the quality of the final submission.

[0089] This series of processing flows enables effective and efficient responses to internal inquiries, reducing time and costs while also taking into consideration the quality of the email and the feelings of the sender.

[0090] Example 1

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

[0092] In today's business environment, there is a demand for fast and efficient response to inquiries. However, in many cases, manually analyzing, replying to, adding attachments to, and sending large volumes of emails is time-consuming and labor-intensive, making it difficult to maintain consistent response quality. Furthermore, due to insufficient sentiment analysis and automatic response generation, appropriate responses to senders may not be delivered promptly. Furthermore, since there is no human intervention possible for automatically generated replies, there is a risk of sending incorrect replies.

[0093] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0094] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category; means for generating a reply based on the analysis results using natural language generation technology; means for searching for necessary attachments from an internal company database, identifying them, and adding them to the reply; and means for checking the generated reply and attachments and sending the email. This enables automated response to inquiries, enabling fast and efficient responses. Furthermore, the use of natural language processing technology makes it possible to analyze the sender's sentiment and generate an appropriate response. Furthermore, by providing an interface that allows users to edit, approve, or reject the automatically generated reply, it is possible to ensure the quality of responses and reduce the risk of sending incorrect replies.

[0095] A "Contact Us" is a message sent by a user requesting specific information, support, or action.

[0096] A "category" is a criterion for classifying the contents of an inquiry into a specific group.

[0097] "Parsing" is the process of understanding the content of an email's body, subject, and attachments and extracting specific information.

[0098] "Natural language generation technology" refers to technology used by computers to generate human language, generating sentences based on specific templates or models.

[0099] A "reply" is a response message that is automatically generated in response to a received inquiry.

[0100] An "attachment" is an electronic file, such as a document or image, that is added to an email.

[0101] A "database" is a system for efficiently storing, managing, and retrieving structured information.

[0102] "Emotion inference" is the process of identifying the sender's emotions or mood from the content of a received email.

[0103] "User interface" refers to the platform or screen through which a user interacts with a system.

[0104] "Robotic Process Automation" (RPA) is a technology that mimics human actions to automate repetitive tasks such as data entry and file manipulation.

[0105] The "SMTP protocol" is a standard communication protocol for sending email over the Internet.

[0106] System Overview

[0107] This invention provides a system for automating and efficiently responding to inquiries. This system automates a series of processes: analyzing the content of an inquiry, classifying it into a specified category, generating a reply, adding an attachment, and sending an email. The main components are a server, a terminal, and a user.

[0108] 1. Content analysis and categorization by the server

[0109] server

[0110] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject, and attachments, and to identify the intent of the specific inquiry and the sender's sentiment. For this purpose, the server uses natural language processing (NLP) techniques such as "spaCy" and "NLTK." This allows it to extract keywords and phrases from the email body and identify the sender's request and situation.

[0111] To give a concrete example, when a server receives an email that says, "Please send me the meeting minutes. It's urgent," the server identifies two key elements from the email's content: "Request for meeting minutes" and "Urgent," and categorizes the email based on those elements.

[0112] 2. Generating a Reply

[0113] server

[0114] The server generates an appropriate reply based on the results of category classification and sentiment analysis, using natural language generation (NLG) technology such as GPT-3. The server stores different reply templates for each specified category, and automatically selects the appropriate template to generate a reply.

[0115] As a specific example, the server selects a template corresponding to a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0116] 3. Adding attachments to replies

[0117] Terminal

[0118] Robotic Process Automation (RPA) installed on the device automatically identifies the necessary attachments and adds them to the email. The device searches for the appropriate file in the company's file system or SQL database, and attaches it to the email once it finds it. RPA tools used include UiPath and Automation Anywhere.

[0119] As a concrete example, the device searches the corporate file system for a file labeled "meeting minutes," extracts it, and attaches it to an email.

[0120] 4. Replying to emails

[0121] Terminal

[0122] The RPA on the terminal checks the generated reply message and attachments, and finally sends the email to the user. It checks whether the email components are complete and corrects any errors. The SMTP protocol is used for sending.

[0123] As a specific example, the terminal sends an email reply to the user with the reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait." and the searched meeting minutes file.

[0124] 5. Editing and Approval Functions

[0125] User

[0126] The system provides an interface that allows users to review, edit, and approve or reject the generated responses using a web application built with a framework such as Django, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[0127] Examples of prompt statements

[0128] "Please tell me the process for analyzing the content of incoming emails, categorizing them into appropriate categories, and then generating and sending replies. Specific keywords are 'meeting minutes' and 'urgent'."

[0129] This system allows internal inquiries to be handled quickly and efficiently, saving time and costs, and provides high-quality responses that take the sender's feelings into consideration, improving the overall quality of responses.

[0130] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0131] Step 1: Receiving email

[0132] server

[0133] The server receives new emails from the mail server using the IMAP protocol. The input is the data of unread emails obtained from the mail server.

[0134] What it does: The server connects to the mail server using the IMAP library, retrieves unread emails from the user's mailbox, and stores them locally. The stored email data is used in the next analysis step.

[0135] Step 2: Content analysis

[0136] server

[0137] It analyzes the body, subject, attachments, etc. of received emails. The technology used is a natural language processing (NLP) library such as "spaCy" or "NLTK." The input is the raw received email data, and the output is the analysis results, extracting keywords and phrases.

[0138] What it does: The server runs a Python script to convert the email body, subject line, and attachments into text, then uses an NLP library to parse the text data and extract important keywords and phrases.

[0139] Step 3: Categorization

[0140] server

[0141] Based on the analyzed email content, a pre-trained AI model such as the BERT model is used to classify emails into predefined categories. The input is the analysis result, and the output is the category information.

[0142] What it does: The server calls the "categorization model" and provides it with the parsed keywords and phrases as input. The model classifies the email into the appropriate category (e.g., technical support, general inquiry, emergency response) and returns the result.

[0143] Step 4: Generate a reply

[0144] server

[0145] A reply sentence is generated based on the category classification. The technology used here is a generative AI model such as GPT-3. The input is category information and analysis results, and the output is the generated reply sentence.

[0146] Specific operation: The server calls the "reply generation model" and automatically generates a reply based on the template, category information, and analyzed content. The generated reply is used in the next step, adding an attachment.

[0147] Step 5: Find and add attachments

[0148] Terminal

[0149] Robotic Process Automation (RPA) is used to find the appropriate attachments and add them to emails. Tools used include UiPath and Automation Anywhere. The input is the generated reply and the parsed results, and the output is the reply with the attachments added.

[0150] What it does: The device searches the company's file system or SQL database to find the appropriate file, attaches it to an email, and passes it along with the reply to the next step.

[0151] Step 6: Sending an email

[0152] Terminal

[0153] The reply message and attachments generated using the RPA tool are checked, and the final email is sent to the user. The input is the reply message with the attachments added, and the output is the sending record.

[0154] Specific operation: The device will use the SMTP settings of the mail server to send the confirmed email, and will record the success or failure of the sending in a log, which can be viewed in the management interface.

[0155] Step 7: Review and Approval

[0156] User

[0157] It provides an interface for reviewing automatically generated replies and, if necessary, editing, approving, or rejecting them. The input is the generated reply, and the output is the approved or revised reply.

[0158] What it does: The user can review the reply through a web interface, make any necessary edits, and finally press the "Approve" button to send the email.

[0159] (Application example 1)

[0160] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0161] Traditionally, online shopping sites have mainly handled inquiries manually by customer support staff, which consumes a lot of resources and has issues with the speed and accuracy of responses. Furthermore, because the quality of responses depends on the quality of the staff, there is often inconsistency in the quality of responses. There was a need to solve these problems and realize efficient, high-quality responses using an automated system.

[0162] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0163] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply based on the analysis results, means for adding an attachment to the reply, means for sending an email, and means for operating as a smartphone application and providing a user-friendly interface. This allows for automated response to inquiries on online shopping sites, enabling fast and efficient responses. Furthermore, natural language processing technology can be used to infer the sender's emotions and generate high-quality replies using a generative AI model, thereby improving user satisfaction.

[0164] 1. "Means for analyzing the content of inquiries and classifying them into designated categories" refers to a system that uses natural language processing technology to analyze the body and subject of received emails and classify them into specific categories (e.g., technical support, general inquiries, emergency response, etc.).

[0165] 2. "Means for generating a reply based on the analysis results" refers to a mechanism that uses natural language generation technology to automatically generate an appropriate reply based on the analysis results.

[0166] 3. "Means for adding attachments to replies" refers to a system that searches for attachments required for replies from the company's file system or database and automatically attaches them to the email.

[0167] 4. "Means for sending email" means a mechanism that automatically sends an email containing the generated reply message and attachment to the user.

[0168] 5. "Means that operate as an application for smartphones and provide a user-friendly interface" refers to a mechanism that provides a user interface that allows users to intuitively and easily handle the inquiry response process through an application that operates on a smartphone.

[0169] 6. "Means of analyzing the content of received emails using natural language processing and inferring the sender's emotions" refers to a system that uses natural language processing technology to analyze the content of received emails and infer the sender's emotions and intentions.

[0170] 7. "Means of generating using a generative AI model" means a mechanism that uses generative AI technology (e.g., GPT-4) to automatically generate high-quality natural language generated replies.

[0171] 8. "Means for providing a user interface that allows the user to edit, approve, or reject the generated reply" refers to a mechanism that provides an interface that allows the user to review the generated reply and edit, approve, or reject it as necessary.

[0172] The present invention is a system for automating and efficiently handling inquiries on an online shopping site. This system is mainly composed of a server, terminals, and users.

[0173] Server-based content analysis and categorization

[0174] server

[0175] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject, and attachments, and to identify the intent of the specific inquiry and the sender's sentiment. For this purpose, the server uses natural language processing (NLP) technology. NLP technology uses Hugging Face Transformers to extract keywords and phrases from the email body and identify the sender's request and situation. Based on the analysis results, the server classifies the email into a predefined category (e.g., delivery delay, product inquiry, return, etc.). For example, if the server receives and analyzes an email that reads, "My item has not arrived yet. This is order number 12345," the server will identify the key element of "delivery delay" from the email's content and classify it into a category based on that.

[0176] Generate a reply

[0177] server

[0178] Based on the results of the analysis and sentiment analysis, the server generates an appropriate reply. This is done using natural language generation (NLG) technology, particularly OpenAI's GPT-4. The server maintains different reply templates for each specified category and selects the appropriate template to generate the reply. For example, the server selects the template for "Delivery Delay" and generates the reply, "We apologize for the delay in your order. We have checked the tracking information and can confirm that the item is currently being shipped." An example of a prompt to use is, "There has been a delay in the delivery of order number 12345. What should we do?"

[0179] Adding attachments to replies

[0180] Terminal

[0181] The RPA (Robotic Process Automation) installed on the terminal provides a mechanism to automatically identify necessary attachments and add them to emails. The terminal searches for the appropriate file from the company's file system or database (e.g., MySQL) and attaches it to the email once it finds it. For example, the terminal searches for a PDF file of tracking information labeled "Delivery Delay" from the company's file system, extracts it, and attaches it to the email.

[0182] Sending emails

[0183] Terminal

[0184] The RPA on the terminal checks the generated reply text and attachments, and finally sends the email to the user. During this process, it checks whether the components of the email are complete and corrects any deficiencies. For example, the terminal sends an email reply to the user with the reply text "We apologize for the delay in your order. We have checked the tracking information and confirmed that the shipping process is underway," along with a PDF file of the searched tracking information.

[0185] Editing and Approval Features

[0186] User

[0187] It provides an interface that allows users to monitor this process behind the scenes and review, edit, and approve or reject the generated replies, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[0188] This system significantly improves the efficiency of customer support inquiries on online shopping sites, saving time and costs. Furthermore, the use of natural language processing technology makes it possible to take into consideration the quality of the email and the sender's feelings, improving the overall quality of responses.

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

[0190] Step 1:

[0191] Retrieving incoming emails

[0192] The server acquires the received email from the mail server.

[0193] Input: Email data received from the email server

[0194] Output: A mail object for parsing

[0195] Specific operation: Connects to a mail server (e.g., IMAP server) and retrieves new emails. The retrieved emails are saved on the server as objects containing the body, subject, and attachments.

[0196] Step 2:

[0197] Email content analysis

[0198] The server analyzes the body and subject of the received email using natural language processing (NLP) techniques, specifically using Hugging Face Transformers to extract keywords and phrases.

[0199] Input: The retrieved email object

[0200] Output: Analysis results (keywords, category information)

[0201] What it does: The email body and subject are tokenized and input into the model. The model analyzes the context based on the tokens and extracts key keywords and sentiment.

[0202] Step 3:

[0203] Category Classification

[0204] The server classifies the email into predefined categories based on the analysis results.

[0205] Input: Analysis results (keywords, category information)

[0206] Output: Classification results (categories)

[0207] Specific behavior: The extracted keywords are compared with the pre-defined category classification rules to determine the most appropriate category. For example, if the keyword "Delivery Delay" is included, the email will be classified into the "Delivery Delay" category.

[0208] Step 4:

[0209] Generate a reply

[0210] The server generates a reply message based on the classified category using a generative AI model (e.g., GPT-4).

[0211] Input: Classification result (category)

[0212] Output: The generated reply

[0213] Specific operation: The category information is input into the generative AI model as a prompt sentence, and a reply sentence is generated. For example, if the category is classified as "Delivery Delay," the prompt will be "Please generate a reply sentence to address the delivery delay."

[0214] Step 5:

[0215] Searching for and adding attachments

[0216] The device will automatically search for any attachments needed for the reply and add them to the email.

[0217] Input: Generated reply, email object

[0218] Output: Reply email with attachment added

[0219] Specific operation: Search for the appropriate file from the company's file system or database (e.g., MySQL) and attach it to the reply email once found. For example, search for a "PDF file of order information" and attach it to the reply email.

[0220] Step 6:

[0221] Sending emails

[0222] The terminal checks the generated reply message and attachments, and finally sends the email to the user.

[0223] Input: Reply email with attachments added

[0224] Output: Email sent

[0225] Specific operation: Check all components of the reply email, confirm that there are no errors, and then send the email to the user via the email sending server (e.g., SMTP server).

[0226] Step 7:

[0227] Edit and Approve

[0228] The user has an interface to review, edit, and approve or reject the generated response.

[0229] Input: Generated reply, user interface

[0230] Output: Reply edited, approved, or rejected by the user

[0231] What happens: The user logs into the web or mobile interface, reviews the generated response, edits it as needed, and approves or rejects it. If approved, the next step is to send an email.

[0232] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0233] This invention relates to a system for automating and efficiently handling inquiries, and in particular provides a system that combines an emotion engine that recognizes user emotions. This system is composed of a server, terminals, and a mechanism that focuses on the user.

[0234] 1. Content analysis and emotion recognition by the server

[0235] server

[0236] The server first analyzes the received email. The body, subject, and attachments of the received email are analyzed using natural language processing (NLP) technology. Keywords and phrases are extracted from the email body to identify the specific intent of the inquiry.

[0237] Next, the emotion engine performs a sentiment analysis of the email. The emotion engine is equipped with algorithms that recognize the sender's emotions and evaluate their stress level and urgency. It also references past email history to learn the sender's emotional patterns, enabling more accurate analysis. This allows the server to obtain analysis results that take the sender's psychological state into account.

[0238] As a concrete example, when a server receives and analyzes an email that says, "Please send me the minutes of the meeting. It's urgent," the server identifies two elements: "request for meeting minutes" and "urgency," and evaluates the message as "high urgency" through its emotion engine.

[0239] 2. Categorization based on analysis results

[0240] server

[0241] Based on the results of the analysis and sentiment analysis, the server classifies the email into a corresponding category (e.g., technical support, general inquiries, emergency response, etc.).

[0242] 3. Generating a Reply

[0243] server

[0244] The server selects an appropriate reply template based on the results of category classification and sentiment analysis. It then uses natural language generation (NLG) technology to generate a reply based on the selected template. It adds expressions that take into account the sender's sentiment and urgency.

[0245] As a specific example, the server selects a template for a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0246] 4. Adding attachments to replies

[0247] Terminal

[0248] The robotic process automation (RPA) installed on the device automatically searches for required attachments. The device scans the company's file system and databases, and identifies and adds appropriate files based on the analysis results and the content of the reply.

[0249] As a specific example, the device searches for a file labeled "meeting minutes" and attaches it to an email.

[0250] 5. Sending emails

[0251] Terminal

[0252] The RPA on the terminal checks the contents of the generated reply and attachments, and finally sends the email. After confirming that there are no abnormalities, the properly constructed email is automatically sent.

[0253] As a specific example, the terminal sends an email reply to the user using the searched file of the meeting minutes, along with the generated reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0254] Editing and Approval Features

[0255] User

[0256] Users can review generated replies through the system's user interface and edit, approve, or reject them as needed, ensuring the quality of responses and the appropriateness of the final submission.

[0257] This system will streamline internal inquiries, saving time and costs, and also enable responses that take user sentiment into account, improving the overall quality of responses.

[0258] The processing flow will be explained below.

[0259] Step 1:

[0260] server

[0261] The server retrieves new emails from the mailbox, saving the email subject, body, and attachments for analysis.

[0262] Step 2:

[0263] server

[0264] The server analyzes the body, subject, and attachments of received emails using natural language processing (NLP) technology. Specifically, it extracts keywords and important phrases from the email body and identifies the specific intent of the inquiry.

[0265] Step 3:

[0266] server

[0267] The server uses an emotion engine to analyze the emotion of emails. The emotion engine analyzes the linguistic features of the email body and recognizes the sender's emotional state. Furthermore, it learns emotional patterns by referencing the sender's past email history, enabling more accurate emotion recognition.

[0268] Step 4:

[0269] server

[0270] The emotion engine uses the results of sentiment analysis to assess the urgency and importance of emails, for example, determining stress levels and whether a request is urgent.

[0271] Step 5:

[0272] server

[0273] Based on the results of the analysis and sentiment analysis, the server classifies the email into a corresponding category (e.g., technical support, general inquiries, emergency response, etc.).

[0274] Step 6:

[0275] server

[0276] The server selects an appropriate reply template based on the assigned category and the results of sentiment analysis. Based on the selected template, it generates a reply using natural language generation (NLG) technology. At this stage, sentiment-based expressions are added to the reply.

[0277] Step 7:

[0278] Terminal

[0279] RPA (Robotic Process Automation) installed on the terminal scans the company's file systems and databases and automatically searches for necessary attachments based on the analysis results and the content of the reply.

[0280] Step 8:

[0281] Terminal

[0282] The RPA on the terminal adds the retrieved attachments to the email, so that the necessary documents according to the user's request are attached to the reply email.

[0283] Step 9:

[0284] Terminal

[0285] The RPA on the terminal checks the generated reply and attachments to ensure that the content is consistent and there are no errors. After confirming that the configuration is correct, the email is sent.

[0286] Step 10:

[0287] User

[0288] Users review the generated replies through the system's user interface, editing, approving, or rejecting them as needed to ensure the quality of the final submission.

[0289] Step 11:

[0290] Terminal

[0291] The terminal then performs a final check of the reply text reviewed by the user and automatically sends it via email, completing the entire inquiry response process.

[0292] This process flow enables efficient and effective responses to internal inquiries, reducing time and costs, and providing high-quality responses that take into account the user's feelings and urgency.

[0293] Example 2

[0294] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0295] Currently, responding to customer inquiries requires manual email analysis, composing replies, and recognizing emotions, which requires a great deal of time and effort. In particular, it is difficult to respond while taking into account the sender's emotions and urgency, resulting in inconsistent response quality. This can lead to lower customer satisfaction and increases the time and cost required for responses.

[0296] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0297] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply based on the analysis result, means for adding an attachment to the reply, means for sending the email, means for recognizing the sender's emotions and evaluating the stress level and urgency, and means for learning emotional patterns using past communication history and improving analysis accuracy. This automates inquiry responses, improves response quality and efficiency, and increases customer satisfaction.

[0298] "Means for analyzing the content of inquiries and classifying them into designated categories" refers to a function for analyzing received emails and communication data and classifying them into appropriate categories (e.g., technical support, emergency response, etc.) based on their content.

[0299] "Means for generating replies based on analysis results" refers to a function that automatically creates replies based on the results of content analysis. These replies are generated using pre-prepared templates and generative AI models.

[0300] "Means for adding attachments to replies" refers to a function that automatically searches for and adds relevant files required for replies, including means for scanning the company's internal file systems and databases.

[0301] "Means for sending email" means a function for sending email, including the generated reply and any necessary attachments, to a specified recipient, typically using the SMTP protocol.

[0302] "Means for recognizing the sender's emotions and assessing their stress level and urgency" refers to a function that recognizes the sender's emotions from the content of received emails and communication data and assesses their stress level and urgency. It uses an emotion engine and natural language processing technology.

[0303] "Means for learning emotional patterns using past communication history to improve analysis accuracy" is a function that analyzes the history of emails and communication data accumulated in the past and learns the emotional patterns of the sender based on that data, thereby improving analysis accuracy.

[0304] "Means for providing a user interface for users to edit, approve, or reject" means providing a user interface with the functionality to allow users to review the generated reply text and edit, approve, or reject it as necessary.

[0305] This invention relates to a system for automating and efficiently handling inquiries, and in particular provides a system that combines an emotion engine that recognizes the sender's emotions. This system is composed of a mechanism that is centered around a server, a terminal, and a user.

[0306] Server-based content analysis and emotion recognition

[0307] server

[0308] The server first receives new emails from the email server. It extracts the body, subject, and attachments of the received emails and performs content analysis using a Python natural language processing (NLP) library (e.g., NLTK or spaCy). It extracts important keywords and phrases from the email body to identify the specific intent of the inquiry. For example, if an email is received saying, "Please send me the minutes of the meeting. It's urgent," the server identifies the keywords "meeting minutes" and "urgent."

[0309] The server then uses an emotion engine to analyze the sentiment of the received email. Specifically, it uses a sentiment analysis model using Hugging Face's Transformers library to evaluate the sender's emotions. For example, it recognizes that the sender is in a hurry from the phrase "I'm in a hurry" and evaluates the stress level and urgency.

[0310] Categorization based on analysis results

[0311] server

[0312] The server combines the results of the analysis and sentiment analysis and classifies the email into an appropriate category (e.g., technical support, general inquiry, emergency response, etc.). For example, an email rated as "high urgency" for "request for meeting minutes" would be classified into the "urgent response" category.

[0313] Generate a reply

[0314] server

[0315] The server selects the optimal reply template, taking into account the appropriate category and sentiment, and generates a reply based on that. This generation uses natural language generation (NLG) technology, leveraging a generative AI model such as OpenAI's GPT-3. For example, it could generate a reply such as, "The meeting minutes you requested will be sent to you shortly. We apologize for the wait."

[0316] Adding attachments to replies

[0317] Terminal

[0318] An RPA (e.g., UiPath or Blue Prism) installed on a terminal scans the company's file system and databases to automatically search for necessary attachments, such as files labeled "meeting minutes," and attach them to emails.

[0319] Sending emails

[0320] Terminal

[0321] The RPA on the terminal checks the generated reply message and the attached file, and after confirming that there are no abnormalities, sends the email using the SMTP protocol. Specifically, it sends a reply message saying, "The requested meeting minutes will be sent shortly. We apologize for the wait," with the meeting minutes file attached.

[0322] Editing and Approval Features

[0323] User

[0324] Users can review generated replies through the system's user interface and edit, approve, or reject them as needed, ensuring the quality of responses and the appropriateness of the final submission.

[0325] Specific examples and prompts for the generative AI model

[0326] Specific examples

[0327] 1. An email is received saying, "Please send me the minutes of the meeting. It's urgent." The server identifies the elements "request for meeting minutes" and "urgent" and evaluates it as "high urgency."

[0328] 2. The server classifies the request as an "urgent response" and generates a reply based on the corresponding reply template: "The meeting minutes you requested will be sent to you shortly. We apologize for the wait."

[0329] 3. The device searches for a file labeled "Meeting Minutes" and attaches it to the email.

[0330] 4. The device checks the email content and sends it after confirming that there are no abnormalities.

[0331] Prompt sentence for generative AI model

[0332] You received an email saying, "Please send me the minutes of the meeting. It's urgent." Use natural language generation technology to generate an appropriate reply to this email. The reply should include expressions that reflect the urgency of the request.

[0333] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0334] Step 1: Receiving and analyzing emails

[0335] server

[0336] The server receives new emails from the email server. It extracts the body, subject, and attachments of the received emails. It then uses this extracted data as input for analysis using a Python NLP library (e.g., NLTK or spaCy). Specifically, it extracts important keywords and phrases from the email body and identifies the specific intent of the inquiry. The output of this process is the email content with the identified keywords and phrases.

[0337] Input: Body, subject, and attachments of received emails

[0338] Output: Identified keywords and phrases

[0339] Step 2: Emotion Recognition

[0340] server

[0341] The server takes the keywords and phrases extracted in step 1 as input and analyzes the emotions using an emotion engine. It evaluates the sender's emotions using an emotion analysis model using Hugging Face's Transformers library. Specifically, it evaluates the sender's stress level and urgency. The output of this process is the sender's emotion evaluation result.

[0342] Input: Identified keywords and phrases

[0343] Output: Sender's sentiment evaluation result

[0344] Step 3: Categorization

[0345] server

[0346] The server combines the analysis results from step 1 with the emotion evaluation results from step 2 and classifies the email into an appropriate category. Specifically, it matches important keywords with the emotion evaluation results and classifies the email into categories such as "technical support," "general inquiries," and "emergency response." The output of this process is the email category information.

[0347] Input: Analysis results, sender's emotion evaluation results

[0348] Output: Email category information

[0349] Step 4: Generate a reply

[0350] server

[0351] The server uses the category information classified in step 3 as input to select the optimal reply template and generate a reply using natural language generation (NLG) technology. A generative AI model (e.g., OpenAI's GPT-3) is used to create a reply based on the template. The output of this process is the generated reply.

[0352] Input: Email category information

[0353] Output: The generated reply

[0354] Step 5: Add attachments

[0355] Terminal

[0356] An RPA (e.g., UiPath or Blue Prism) installed on a terminal scans the company's file system or database and automatically searches for the necessary attachments. For example, it searches for files labeled "meeting minutes" and attaches them to an email. The output of this process is a reply message with the attachment.

[0357] Input: Generated reply

[0358] Output: Reply with attachments

[0359] Step 6: Sending an email

[0360] Terminal

[0361] The RPA on the terminal checks the reply text with the attachment from step 5 and sends the email using the SMTP protocol. Specifically, after a final check of the reply text and attachment, it checks that there are no abnormalities and then sends the email. The output of this process is the sent email.

[0362] Input: Reply with attachment

[0363] Output: Sent email

[0364] Step 7: Editing and Approval Functions

[0365] User

[0366] The user reviews the reply generated in step 4 through the system's user interface, edits the reply if necessary, and finally approves or rejects it. This action checks the quality and appropriateness of the reply. The output of this process is a final reply approved by the user.

[0367] Input: Generated reply

[0368] Output: Final reply accepted by the user

[0369] (Application example 2)

[0370] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0371] When dealing with customers in brick-and-mortar stores, staff are required to respond appropriately to a variety of questions and requests. However, current systems make it difficult to respond appropriately by taking into account the customer's emotions, which can result in a decline in customer satisfaction. This also places a heavy burden on staff, making efficient operation difficult. In particular, current systems are insufficient when emotion recognition and rapid response are required.

[0372] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0373] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply, means for adding an attachment to the reply, means for analyzing text and recognizing emotions, means for inputting a question by voice and converting it into text, means for displaying the generated reply, and means for sending an email. This enables prompt and appropriate responses that take emotions into consideration when dealing with customers in a physical store.

[0374] An "inquiry" is a question or request made by a customer seeking information.

[0375] "Analysis" is the process of understanding the content based on input data and extracting classifications and meanings.

[0376] A "category" is a criterion for classifying content into a particular group or groups based on the analyzed content.

[0377] A "reply" is a response message generated in response to an inquiry.

[0378] An "attachment" is an additional data file that is added to a reply.

[0379] "Email" means a text message sent or received electronically.

[0380] A "means" is a method or device for achieving a specific purpose.

[0381] "Natural language processing" is a technology that uses computers to analyze, understand, and generate human language.

[0382] "Emotion recognition" is a technology that analyzes and judges emotional states from text and voice data.

[0383] "Text conversion" is a technology that converts voice data into text information.

[0384] "Display" refers to the visual presentation of analyzed or generated results.

[0385] "Smart glasses" are wearable devices that are capable of augmented reality and displaying information.

[0386] A "robot" is a mechanical device that is programmed to automatically perform specific tasks.

[0387] General Overview

[0388] This invention is a system for improving the efficiency of customer service in brick-and-mortar stores and responding in a way that takes into account the customer's emotions. This system includes a server and a smart device (smart glasses or a robot), and acquires and analyzes the customer's voice data to generate an appropriate reply and display the response.

[0389] Server Configuration

[0390] The server has the following main functions:

[0391] 1. Voice data to text conversion function

[0392] It receives voice data sent from smart devices and converts it into text using the Google Cloud Speech-to-Text API.

[0393] 2. Text analysis and emotion recognition

[0394] The converted text data is analyzed using a natural language processing (NLP) engine (e.g., spaCy or NLTK) to identify the intent of the inquiry, and an emotion engine (e.g., IBM Watson Tone Analyzer) to recognize the customer's emotions.

[0395] 3. Reply generation function

[0396] An appropriate reply is generated based on the analysis results and emotion recognition results, using a generative AI model (e.g., OpenAI GPT-3).

[0397] Smart Device Configuration

[0398] The smart devices (smart glasses and robots) have the following main functions:

[0399] 1. Audio data acquisition function

[0400] A function that allows customers to input questions and requests by voice and send them to the server.

[0401] 2. Display function

[0402] A function that displays text data (reply text) sent from the server, allowing staff to respond to customers quickly.

[0403] Specific examples

[0404] To illustrate, consider the following scenario:

[0405] 1. A customer asks a staff member wearing smart glasses, "I'd like to process a return."

[0406] 2. The staff member's smart glasses send this audio to a server, where it is converted into text using the Google Cloud Speech-to-Text API.

[0407] 3. The converted text is analyzed using spaCy and NLTK to identify the inquiry intent of "I would like to process a return." IBM Watson Tone Analyzer determines that the customer's emotions are calm.

[0408] 4. The server uses OpenAI GPT-3 to generate a reply saying, "Sorry, please go ahead and we'll help you process your return."

[0409] 5. This reply will be displayed on the staff member's smart glasses and the staff member will convey it to the customer.

[0410] Prompt Sentence Examples

[0411] Customer: "I'd like to process a return."

[0412] Smart Glasses: "Sorry, please come on over here and we'll help you process your return."

[0413] As described above, the present invention provides a system for quickly and efficiently handling customers in physical stores, and realizes high-quality service that takes into consideration the emotions of customers.

[0414] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0415] Step 1: Getting voice input

[0416] Subject: Terminal

[0417] How it works: Smart glasses or a robot are used to capture voice questions and requests from users.

[0418] Input: User's voice data.

[0419] Output: The audio data is sent to the server.

[0420] Step 2: Convert audio data to text

[0421] Subject: Server

[0422] Operation: Received voice data is converted into text data using the Google Cloud Speech-to-Text API.

[0423] Input: Audio data.

[0424] Output: Text data.

[0425] Step 3: Parsing the text

[0426] Subject: Server

[0427] How it works: Uses spaCy and NLTK to extract keywords and phrases from text data to identify query intent.

[0428] Input: Text data.

[0429] Output: Analysis result (query intent).

[0430] Step 4: Emotion Recognition

[0431] Subject: Server

[0432] How it works: Using IBM Watson Tone Analyzer, it recognizes the user's emotions from the converted text data and assesses stress levels and urgency.

[0433] Input: Text data.

[0434] Output: Emotion recognition results (stress level, urgency).

[0435] Step 5: Generate a reply

[0436] Subject: Server

[0437] How it works: Based on the analysis results and emotion recognition results, OpenAI GPT-3 is used to generate the optimal reply.

[0438] Input: Analysis results, emotion recognition results.

[0439] Output: The generated reply.

[0440] Step 6: View the reply

[0441] Subject: Terminal

[0442] How it works: The generated reply is displayed on smart glasses or a robot and conveyed to the user.

[0443] Input: The generated reply.

[0444] Output: Text to be displayed on smart glasses or robot.

[0445] For example, if a user asks, "I would like to process a return," the process will proceed as follows:

[0446] 1. The user asks a question by voice, and the smart glasses collect the voice data.

[0447] 2. The audio data is sent to the server and converted to text using the Google Cloud Speech-to-Text API.

[0448] 3. The converted text data was analyzed using spaCy and NLTK to identify the inquiry intent of "return procedure."

[0449] 4. IBM Watson Tone Analyzer recognizes emotions and determines that the user is calm.

[0450] 5. Using OpenAI GPT-3, we generated the following response: "Sorry, please go ahead and we'll help you process your return."

[0451] 6. The generated reply is displayed on the smart glasses and conveyed to the user.

[0452] This process enables fast and sensitive customer service in physical stores.

[0453] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

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

[0455] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0456] [Second embodiment]

[0457] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0458] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0459] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0460] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0461] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0462] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0463] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0464] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0465] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0466] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0467] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0468] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0469] This invention provides a system for automating and efficiently responding to inquiries. This system automates a series of processes: analyzing the content of an inquiry, classifying it into a specified category, generating a reply, adding an attachment, and sending an email. The main elements that make up this system are a server, a terminal, and a user.

[0470] 1. Content analysis and categorization by the server

[0471] server

[0472] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject line, and attachments, and to identify the intent of the specific inquiry and the sentiment of the sender. For this purpose, the server uses natural language processing (NLP) techniques to extract keywords and phrases from the email body and identify the sender's request and situation. Based on the analysis results, the server classifies the email into predefined categories (e.g., technical support, general inquiry, emergency response, etc.).

[0473] As a concrete example, when a server receives and analyzes an email that says, "Please send me the minutes of the meeting. It's urgent," the server will identify two main elements from the email's content: "Request for meeting minutes" and "Urgent," and categorize it based on those elements.

[0474] 2. Generating a Reply

[0475] server

[0476] The server generates an appropriate reply based on the results of category classification and sentiment analysis, using natural language generation (NLG) technology. The server stores different reply templates for each specified category, and automatically selects the appropriate template to generate a reply.

[0477] As a specific example, the server selects a template corresponding to a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0478] 3. Adding attachments to replies

[0479] Terminal

[0480] Robotic Process Automation (RPA) installed on the device automatically identifies the necessary attachments and adds them to the email. The device searches for the appropriate file in the company's file system or database, and once found, attaches it to the email.

[0481] As a concrete example, the device searches the corporate file system for a file labeled "meeting minutes," extracts it, and attaches it to an email.

[0482] 4. Replying to emails

[0483] Terminal

[0484] The RPA on the terminal checks the generated reply and attachments, and finally sends the email to the user. During this process, it checks whether the components of the email are complete and corrects any deficiencies.

[0485] As a specific example, the terminal sends an email reply to the user with the reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait." and the searched meeting minutes file.

[0486] Editing and Approval Features

[0487] User

[0488] It provides an interface that allows users to monitor this process behind the scenes and review, edit, and approve or reject the generated replies, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[0489] This system enables efficient responses to internal inquiries, saving time and costs. It also takes into consideration the quality of the email and the feelings of the sender, improving the overall quality of responses.

[0490] The processing flow will be explained below.

[0491] Step 1:

[0492] server

[0493] The server retrieves new emails from the mailbox and saves the subject, body, and attachments of the received email to a file for analysis.

[0494] Step 2:

[0495] server

[0496] The server analyzes the body, subject, and attachments of received emails using natural language processing (NLP) technology. Specifically, it extracts keywords and important phrases from the email body to identify the email's intent and request. In parallel, it performs sentiment analysis to identify the sender's psychological state and emotions.

[0497] Step 3:

[0498] server

[0499] Based on the analysis results, the server categorizes the email into a corresponding category, such as technical support, general inquiries, or emergency response, based on pre-defined rules.

[0500] Step 4:

[0501] server

[0502] The server uses the classified categories and the results of sentiment analysis to select an appropriate reply template. It then uses natural language generation (NLG) technology to generate a reply based on the selected template. At this stage, the reply text is enhanced to reflect the sender's sentiment.

[0503] Step 5:

[0504] Terminal

[0505] The device's RPA (Robotic Process Automation) searches for the required attachments by scanning the company's file system and databases and identifying the appropriate file based on the results of email analysis and the content of the reply.

[0506] Step 6:

[0507] Terminal

[0508] The RPA on the terminal adds the searched attachments to the email, so that the necessary documents are attached to the reply email according to the user's request.

[0509] Step 7:

[0510] Terminal

[0511] The RPA on the device checks all generated replies and attachments to ensure they are properly structured and checked for consistency and completeness.

[0512] Step 8:

[0513] Terminal

[0514] After the device performs a final check, it automatically sends an email containing the generated reply and attachment, completing the entire process of responding to the inquiry.

[0515] Step 9:

[0516] User

[0517] Users can review the generated replies through the system's user interface, and can edit, approve, or reject them as needed to ensure the quality of the final submission.

[0518] This series of processing flows enables effective and efficient responses to internal inquiries, reducing time and costs while also taking into consideration the quality of the email and the feelings of the sender.

[0519] Example 1

[0520] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0521] In today's business environment, there is a demand for fast and efficient response to inquiries. However, in many cases, manually analyzing, replying to, adding attachments to, and sending large volumes of emails is time-consuming and labor-intensive, making it difficult to maintain consistent response quality. Furthermore, due to insufficient sentiment analysis and automatic response generation, appropriate responses to senders may not be delivered promptly. Furthermore, since there is no human intervention possible for automatically generated replies, there is a risk of sending incorrect replies.

[0522] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0523] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category; means for generating a reply based on the analysis results using natural language generation technology; means for searching for necessary attachments from an internal company database, identifying them, and adding them to the reply; and means for checking the generated reply and attachments and sending the email. This enables automated response to inquiries, enabling fast and efficient responses. Furthermore, the use of natural language processing technology makes it possible to analyze the sender's sentiment and generate an appropriate response. Furthermore, by providing an interface that allows users to edit, approve, or reject the automatically generated reply, it is possible to ensure the quality of responses and reduce the risk of sending incorrect replies.

[0524] A "Contact Us" is a message sent by a user requesting specific information, support, or action.

[0525] A "category" is a criterion for classifying the contents of an inquiry into a specific group.

[0526] "Parsing" is the process of understanding the content of an email's body, subject, and attachments and extracting specific information.

[0527] "Natural language generation technology" refers to technology used by computers to generate human language, generating sentences based on specific templates or models.

[0528] A "reply" is a response message that is automatically generated in response to a received inquiry.

[0529] An "attachment" is an electronic file, such as a document or image, that is added to an email.

[0530] A "database" is a system for efficiently storing, managing, and retrieving structured information.

[0531] "Emotion inference" is the process of identifying the sender's emotions or mood from the content of a received email.

[0532] "User interface" refers to the platform or screen through which a user interacts with a system.

[0533] "Robotic Process Automation" (RPA) is a technology that mimics human actions to automate repetitive tasks such as data entry and file manipulation.

[0534] The "SMTP protocol" is a standard communication protocol for sending email over the Internet.

[0535] System Overview

[0536] This invention provides a system for automating and efficiently responding to inquiries. This system automates a series of processes: analyzing the content of an inquiry, classifying it into a specified category, generating a reply, adding an attachment, and sending an email. The main components are a server, a terminal, and a user.

[0537] 1. Content analysis and categorization by the server

[0538] server

[0539] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject, and attachments, and to identify the intent of the specific inquiry and the sender's sentiment. For this purpose, the server uses natural language processing (NLP) techniques such as "spaCy" and "NLTK." This allows it to extract keywords and phrases from the email body and identify the sender's request and situation.

[0540] To give a concrete example, when a server receives an email that says, "Please send me the meeting minutes. It's urgent," the server identifies two key elements from the email's content: "Request for meeting minutes" and "Urgent," and categorizes the email based on those elements.

[0541] 2. Generating a Reply

[0542] server

[0543] The server generates an appropriate reply based on the results of category classification and sentiment analysis, using natural language generation (NLG) technology such as GPT-3. The server stores different reply templates for each specified category, and automatically selects the appropriate template to generate a reply.

[0544] As a specific example, the server selects a template corresponding to a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0545] 3. Adding attachments to replies

[0546] Terminal

[0547] Robotic Process Automation (RPA) installed on the device automatically identifies the necessary attachments and adds them to the email. The device searches for the appropriate file in the company's file system or SQL database, and attaches it to the email once it finds it. RPA tools used include UiPath and Automation Anywhere.

[0548] As a concrete example, the device searches the corporate file system for a file labeled "meeting minutes," extracts it, and attaches it to an email.

[0549] 4. Replying to emails

[0550] Terminal

[0551] The RPA on the terminal checks the generated reply message and attachments, and finally sends the email to the user. It checks whether the email components are complete and corrects any errors. The SMTP protocol is used for sending.

[0552] As a specific example, the terminal sends an email reply to the user with the reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait." and the searched meeting minutes file.

[0553] 5. Editing and Approval Functions

[0554] User

[0555] The system provides an interface that allows users to review, edit, and approve or reject the generated responses using a web application built with a framework such as Django, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[0556] Examples of prompt statements

[0557] "Please tell me the process for analyzing the content of incoming emails, categorizing them into appropriate categories, and then generating and sending replies. Specific keywords are 'meeting minutes' and 'urgent'."

[0558] This system allows internal inquiries to be handled quickly and efficiently, saving time and costs, and provides high-quality responses that take the sender's feelings into consideration, improving the overall quality of responses.

[0559] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0560] Step 1: Receiving email

[0561] server

[0562] The server receives new emails from the mail server using the IMAP protocol. The input is the data of unread emails obtained from the mail server.

[0563] What it does: The server connects to the mail server using the IMAP library, retrieves unread emails from the user's mailbox, and stores them locally. The stored email data is used in the next analysis step.

[0564] Step 2: Content analysis

[0565] server

[0566] It analyzes the body, subject, attachments, etc. of received emails. The technology used is a natural language processing (NLP) library such as "spaCy" or "NLTK." The input is the raw received email data, and the output is the analysis results, extracting keywords and phrases.

[0567] What it does: The server runs a Python script to convert the email body, subject line, and attachments into text, then uses an NLP library to parse the text data and extract important keywords and phrases.

[0568] Step 3: Categorization

[0569] server

[0570] Based on the analyzed email content, a pre-trained AI model such as the BERT model is used to classify emails into predefined categories. The input is the analysis result, and the output is the category information.

[0571] What it does: The server calls the "categorization model" and provides it with the parsed keywords and phrases as input. The model classifies the email into the appropriate category (e.g., technical support, general inquiry, emergency response) and returns the result.

[0572] Step 4: Generate a reply

[0573] server

[0574] A reply sentence is generated based on the category classification. The technology used here is a generative AI model such as GPT-3. The input is category information and analysis results, and the output is the generated reply sentence.

[0575] Specific operation: The server calls the "reply generation model" and automatically generates a reply based on the template, category information, and analyzed content. The generated reply is used in the next step, adding an attachment.

[0576] Step 5: Find and add attachments

[0577] Terminal

[0578] Robotic Process Automation (RPA) is used to find the appropriate attachments and add them to emails. Tools used include UiPath and Automation Anywhere. The input is the generated reply and the parsed results, and the output is the reply with the attachments added.

[0579] What it does: The device searches the company's file system or SQL database to find the appropriate file, attaches it to an email, and passes it along with the reply to the next step.

[0580] Step 6: Sending an email

[0581] Terminal

[0582] The reply message and attachments generated using the RPA tool are checked, and the final email is sent to the user. The input is the reply message with the attachments added, and the output is the sending record.

[0583] Specific operation: The device will use the SMTP settings of the mail server to send the confirmed email, and will record the success or failure of the sending in a log, which can be viewed in the management interface.

[0584] Step 7: Review and Approval

[0585] User

[0586] It provides an interface for reviewing automatically generated replies and, if necessary, editing, approving, or rejecting them. The input is the generated reply, and the output is the approved or revised reply.

[0587] What it does: The user can review the reply through a web interface, make any necessary edits, and finally press the "Approve" button to send the email.

[0588] (Application example 1)

[0589] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0590] Traditionally, online shopping sites have mainly handled inquiries manually by customer support staff, which consumes a lot of resources and has issues with the speed and accuracy of responses. Furthermore, because the quality of responses depends on the quality of the staff, there is often inconsistency in the quality of responses. There was a need to solve these problems and realize efficient, high-quality responses using an automated system.

[0591] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0592] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply based on the analysis results, means for adding an attachment to the reply, means for sending an email, and means for operating as a smartphone application and providing a user-friendly interface. This allows for automated response to inquiries on online shopping sites, enabling fast and efficient responses. Furthermore, natural language processing technology can be used to infer the sender's emotions and generate high-quality replies using a generative AI model, thereby improving user satisfaction.

[0593] 1. "Means for analyzing the content of inquiries and classifying them into designated categories" refers to a system that uses natural language processing technology to analyze the body and subject of received emails and classify them into specific categories (e.g., technical support, general inquiries, emergency response, etc.).

[0594] 2. "Means for generating a reply based on the analysis results" refers to a mechanism that uses natural language generation technology to automatically generate an appropriate reply based on the analysis results.

[0595] 3. "Means for adding attachments to replies" refers to a system that searches for attachments required for replies from the company's file system or database and automatically attaches them to the email.

[0596] 4. "Means for sending email" means a mechanism that automatically sends an email containing the generated reply message and attachment to the user.

[0597] 5. "Means that operate as an application for smartphones and provide a user-friendly interface" refers to a mechanism that provides a user interface that allows users to intuitively and easily handle the inquiry response process through an application that operates on a smartphone.

[0598] 6. "Means of analyzing the content of received emails using natural language processing and inferring the sender's emotions" refers to a system that uses natural language processing technology to analyze the content of received emails and infer the sender's emotions and intentions.

[0599] 7. "Means of generating using a generative AI model" means a mechanism that uses generative AI technology (e.g., GPT-4) to automatically generate high-quality natural language generated replies.

[0600] 8. "Means for providing a user interface that allows the user to edit, approve, or reject the generated reply" refers to a mechanism that provides an interface that allows the user to review the generated reply and edit, approve, or reject it as necessary.

[0601] The present invention is a system for automating and efficiently handling inquiries on an online shopping site. This system is mainly composed of a server, terminals, and users.

[0602] Server-based content analysis and categorization

[0603] server

[0604] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject, and attachments, and to identify the intent of the specific inquiry and the sender's sentiment. For this purpose, the server uses natural language processing (NLP) technology. NLP technology uses Hugging Face Transformers to extract keywords and phrases from the email body and identify the sender's request and situation. Based on the analysis results, the server classifies the email into a predefined category (e.g., delivery delay, product inquiry, return, etc.). For example, if the server receives and analyzes an email that reads, "My item has not arrived yet. This is order number 12345," the server will identify the key element of "delivery delay" from the email's content and classify it into a category based on that.

[0605] Generate a reply

[0606] server

[0607] Based on the results of the analysis and sentiment analysis, the server generates an appropriate reply. This is done using natural language generation (NLG) technology, particularly OpenAI's GPT-4. The server maintains different reply templates for each specified category and selects the appropriate template to generate the reply. For example, the server selects the template for "Delivery Delay" and generates the reply, "We apologize for the delay in your order. We have checked the tracking information and can confirm that the item is currently being shipped." An example of a prompt to use is, "There has been a delay in the delivery of order number 12345. What should we do?"

[0608] Adding attachments to replies

[0609] Terminal

[0610] The RPA (Robotic Process Automation) installed on the terminal provides a mechanism to automatically identify necessary attachments and add them to emails. The terminal searches for the appropriate file from the company's file system or database (e.g., MySQL) and attaches it to the email once it finds it. For example, the terminal searches for a PDF file of tracking information labeled "Delivery Delay" from the company's file system, extracts it, and attaches it to the email.

[0611] Sending emails

[0612] Terminal

[0613] The RPA on the terminal checks the generated reply text and attachments, and finally sends the email to the user. During this process, it checks whether the components of the email are complete and corrects any deficiencies. For example, the terminal sends an email reply to the user with the reply text "We apologize for the delay in your order. We have checked the tracking information and confirmed that the shipping process is underway," along with a PDF file of the searched tracking information.

[0614] Editing and Approval Features

[0615] User

[0616] It provides an interface that allows users to monitor this process behind the scenes and review, edit, and approve or reject the generated replies, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[0617] This system significantly improves the efficiency of customer support inquiries on online shopping sites, saving time and costs. Furthermore, the use of natural language processing technology makes it possible to take into consideration the quality of the email and the sender's feelings, improving the overall quality of responses.

[0618] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0619] Step 1:

[0620] Retrieving incoming emails

[0621] The server acquires the received email from the mail server.

[0622] Input: Email data received from the email server

[0623] Output: A mail object for parsing

[0624] Specific operation: Connects to a mail server (e.g., IMAP server) and retrieves new emails. The retrieved emails are saved on the server as objects containing the body, subject, and attachments.

[0625] Step 2:

[0626] Email content analysis

[0627] The server analyzes the body and subject of the received email using natural language processing (NLP) techniques, specifically using Hugging Face Transformers to extract keywords and phrases.

[0628] Input: The retrieved email object

[0629] Output: Analysis results (keywords, category information)

[0630] What it does: The email body and subject are tokenized and input into the model. The model analyzes the context based on the tokens and extracts key keywords and sentiment.

[0631] Step 3:

[0632] Category Classification

[0633] The server classifies the email into predefined categories based on the analysis results.

[0634] Input: Analysis results (keywords, category information)

[0635] Output: Classification results (categories)

[0636] Specific behavior: The extracted keywords are compared with the pre-defined category classification rules to determine the most appropriate category. For example, if the keyword "Delivery Delay" is included, the email will be classified into the "Delivery Delay" category.

[0637] Step 4:

[0638] Generate a reply

[0639] The server generates a reply message based on the classified category using a generative AI model (e.g., GPT-4).

[0640] Input: Classification result (category)

[0641] Output: The generated reply

[0642] Specific operation: The category information is input into the generative AI model as a prompt sentence, and a reply sentence is generated. For example, if the category is classified as "Delivery Delay," the prompt will be "Please generate a reply sentence to address the delivery delay."

[0643] Step 5:

[0644] Searching for and adding attachments

[0645] The device will automatically search for any attachments needed for the reply and add them to the email.

[0646] Input: Generated reply, email object

[0647] Output: Reply email with attachment added

[0648] Specific operation: Search for the appropriate file from the company's file system or database (e.g., MySQL) and attach it to the reply email once found. For example, search for a "PDF file of order information" and attach it to the reply email.

[0649] Step 6:

[0650] Sending emails

[0651] The terminal checks the generated reply message and attachments, and finally sends the email to the user.

[0652] Input: Reply email with attachments added

[0653] Output: Email sent

[0654] Specific operation: Check all components of the reply email, confirm that there are no errors, and then send the email to the user via the email sending server (e.g., SMTP server).

[0655] Step 7:

[0656] Edit and Approve

[0657] The user has an interface to review, edit, and approve or reject the generated response.

[0658] Input: Generated reply, user interface

[0659] Output: Reply edited, approved, or rejected by the user

[0660] What happens: The user logs into the web or mobile interface, reviews the generated response, edits it as needed, and approves or rejects it. If approved, the next step is to send an email.

[0661] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0662] This invention relates to a system for automating and efficiently handling inquiries, and in particular provides a system that combines an emotion engine that recognizes user emotions. This system is composed of a server, terminals, and a mechanism that focuses on the user.

[0663] 1. Content analysis and emotion recognition by the server

[0664] server

[0665] The server first analyzes the received email. The body, subject, and attachments of the received email are analyzed using natural language processing (NLP) technology. Keywords and phrases are extracted from the email body to identify the specific intent of the inquiry.

[0666] Next, the emotion engine performs a sentiment analysis of the email. The emotion engine is equipped with algorithms that recognize the sender's emotions and evaluate their stress level and urgency. It also references past email history to learn the sender's emotional patterns, enabling more accurate analysis. This allows the server to obtain analysis results that take the sender's psychological state into account.

[0667] As a concrete example, when a server receives and analyzes an email that says, "Please send me the minutes of the meeting. It's urgent," the server identifies two elements: "request for meeting minutes" and "urgency," and evaluates the message as "high urgency" through its emotion engine.

[0668] 2. Categorization based on analysis results

[0669] server

[0670] Based on the results of the analysis and sentiment analysis, the server classifies the email into a corresponding category (e.g., technical support, general inquiries, emergency response, etc.).

[0671] 3. Generating a Reply

[0672] server

[0673] The server selects an appropriate reply template based on the results of category classification and sentiment analysis. It then uses natural language generation (NLG) technology to generate a reply based on the selected template. It adds expressions that take into account the sender's sentiment and urgency.

[0674] As a specific example, the server selects a template for a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0675] 4. Adding attachments to replies

[0676] Terminal

[0677] The robotic process automation (RPA) installed on the device automatically searches for required attachments. The device scans the company's file system and databases, and identifies and adds appropriate files based on the analysis results and the content of the reply.

[0678] As a specific example, the device searches for a file labeled "meeting minutes" and attaches it to an email.

[0679] 5. Sending emails

[0680] Terminal

[0681] The RPA on the terminal checks the contents of the generated reply and attachments, and finally sends the email. After confirming that there are no abnormalities, the properly constructed email is automatically sent.

[0682] As a specific example, the terminal sends an email reply to the user using the searched file of the meeting minutes, along with the generated reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0683] Editing and Approval Features

[0684] User

[0685] Users can review generated replies through the system's user interface and edit, approve, or reject them as needed, ensuring the quality of responses and the appropriateness of the final submission.

[0686] This system will streamline internal inquiries, saving time and costs, and also enable responses that take user sentiment into account, improving the overall quality of responses.

[0687] The processing flow will be explained below.

[0688] Step 1:

[0689] server

[0690] The server retrieves new emails from the mailbox, saving the email subject, body, and attachments for analysis.

[0691] Step 2:

[0692] server

[0693] The server analyzes the body, subject, and attachments of received emails using natural language processing (NLP) technology. Specifically, it extracts keywords and important phrases from the email body and identifies the specific intent of the inquiry.

[0694] Step 3:

[0695] server

[0696] The server uses an emotion engine to analyze the emotion of emails. The emotion engine analyzes the linguistic features of the email body and recognizes the sender's emotional state. Furthermore, it learns emotional patterns by referencing the sender's past email history, enabling more accurate emotion recognition.

[0697] Step 4:

[0698] server

[0699] The emotion engine uses the results of sentiment analysis to assess the urgency and importance of emails, for example, determining stress levels and whether a request is urgent.

[0700] Step 5:

[0701] server

[0702] Based on the results of the analysis and sentiment analysis, the server classifies the email into a corresponding category (e.g., technical support, general inquiries, emergency response, etc.).

[0703] Step 6:

[0704] server

[0705] The server selects an appropriate reply template based on the assigned category and the results of sentiment analysis. Based on the selected template, it generates a reply using natural language generation (NLG) technology. At this stage, sentiment-based expressions are added to the reply.

[0706] Step 7:

[0707] Terminal

[0708] RPA (Robotic Process Automation) installed on the terminal scans the company's file systems and databases and automatically searches for necessary attachments based on the analysis results and the content of the reply.

[0709] Step 8:

[0710] Terminal

[0711] The RPA on the terminal adds the retrieved attachments to the email, so that the necessary documents according to the user's request are attached to the reply email.

[0712] Step 9:

[0713] Terminal

[0714] The RPA on the terminal checks the generated reply and attachments to ensure that the content is consistent and there are no errors. After confirming that the configuration is correct, the email is sent.

[0715] Step 10:

[0716] User

[0717] Users review the generated replies through the system's user interface, editing, approving, or rejecting them as needed to ensure the quality of the final submission.

[0718] Step 11:

[0719] Terminal

[0720] The terminal then performs a final check of the reply text reviewed by the user and automatically sends it via email, completing the entire inquiry response process.

[0721] This process flow enables efficient and effective responses to internal inquiries, reducing time and costs, and providing high-quality responses that take into account the user's feelings and urgency.

[0722] Example 2

[0723] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0724] Currently, responding to customer inquiries requires manual email analysis, composing replies, and recognizing emotions, which requires a great deal of time and effort. In particular, it is difficult to respond while taking into account the sender's emotions and urgency, resulting in inconsistent response quality. This can lead to lower customer satisfaction and increases the time and cost required for responses.

[0725] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0726] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply based on the analysis result, means for adding an attachment to the reply, means for sending the email, means for recognizing the sender's emotions and evaluating the stress level and urgency, and means for learning emotional patterns using past communication history and improving analysis accuracy. This automates inquiry responses, improves response quality and efficiency, and increases customer satisfaction.

[0727] "Means for analyzing the content of inquiries and classifying them into designated categories" refers to a function for analyzing received emails and communication data and classifying them into appropriate categories (e.g., technical support, emergency response, etc.) based on their content.

[0728] "Means for generating replies based on analysis results" refers to a function that automatically creates replies based on the results of content analysis. These replies are generated using pre-prepared templates and generative AI models.

[0729] "Means for adding attachments to replies" refers to a function that automatically searches for and adds relevant files required for replies, including means for scanning the company's internal file systems and databases.

[0730] "Means for sending email" means a function for sending email, including the generated reply and any necessary attachments, to a specified recipient, typically using the SMTP protocol.

[0731] "Means for recognizing the sender's emotions and assessing their stress level and urgency" refers to a function that recognizes the sender's emotions from the content of received emails and communication data and assesses their stress level and urgency. It uses an emotion engine and natural language processing technology.

[0732] "Means for learning emotional patterns using past communication history to improve analysis accuracy" is a function that analyzes the history of emails and communication data accumulated in the past and learns the emotional patterns of the sender based on that data, thereby improving analysis accuracy.

[0733] "Means for providing a user interface for users to edit, approve, or reject" means providing a user interface with the functionality to allow users to review the generated reply text and edit, approve, or reject it as necessary.

[0734] This invention relates to a system for automating and efficiently handling inquiries, and in particular provides a system that combines an emotion engine that recognizes the sender's emotions. This system is composed of a mechanism that is centered around a server, a terminal, and a user.

[0735] Server-based content analysis and emotion recognition

[0736] server

[0737] The server first receives new emails from the email server. It extracts the body, subject, and attachments of the received emails and performs content analysis using a Python natural language processing (NLP) library (e.g., NLTK or spaCy). It extracts important keywords and phrases from the email body to identify the specific intent of the inquiry. For example, if an email is received saying, "Please send me the minutes of the meeting. It's urgent," the server identifies the keywords "meeting minutes" and "urgent."

[0738] The server then uses an emotion engine to analyze the sentiment of the received email. Specifically, it uses a sentiment analysis model using Hugging Face's Transformers library to evaluate the sender's emotions. For example, it recognizes that the sender is in a hurry from the phrase "I'm in a hurry" and evaluates the stress level and urgency.

[0739] Categorization based on analysis results

[0740] server

[0741] The server combines the results of the analysis and sentiment analysis and classifies the email into an appropriate category (e.g., technical support, general inquiry, emergency response, etc.). For example, an email rated as "high urgency" for "request for meeting minutes" would be classified into the "urgent response" category.

[0742] Generate a reply

[0743] server

[0744] The server selects the optimal reply template, taking into account the appropriate category and sentiment, and generates a reply based on that. This generation uses natural language generation (NLG) technology, leveraging a generative AI model such as OpenAI's GPT-3. For example, it could generate a reply such as, "The meeting minutes you requested will be sent to you shortly. We apologize for the wait."

[0745] Adding attachments to replies

[0746] Terminal

[0747] An RPA (e.g., UiPath or Blue Prism) installed on a terminal scans the company's file system and databases to automatically search for necessary attachments, such as files labeled "meeting minutes," and attach them to emails.

[0748] Sending emails

[0749] Terminal

[0750] The RPA on the terminal checks the generated reply message and the attached file, and after confirming that there are no abnormalities, sends the email using the SMTP protocol. Specifically, it sends a reply message saying, "The requested meeting minutes will be sent shortly. We apologize for the wait," with the meeting minutes file attached.

[0751] Editing and Approval Features

[0752] User

[0753] Users can review generated replies through the system's user interface and edit, approve, or reject them as needed, ensuring the quality of responses and the appropriateness of the final submission.

[0754] Specific examples and prompts for the generative AI model

[0755] Specific examples

[0756] 1. An email is received saying, "Please send me the minutes of the meeting. It's urgent." The server identifies the elements "request for meeting minutes" and "urgent" and evaluates it as "high urgency."

[0757] 2. The server classifies the request as an "urgent response" and generates a reply based on the corresponding reply template: "The meeting minutes you requested will be sent to you shortly. We apologize for the wait."

[0758] 3. The device searches for a file labeled "Meeting Minutes" and attaches it to the email.

[0759] 4. The device checks the email content and sends it after confirming that there are no abnormalities.

[0760] Prompt sentence for generative AI model

[0761] You received an email saying, "Please send me the minutes of the meeting. It's urgent." Use natural language generation technology to generate an appropriate reply to this email. The reply should include expressions that reflect the urgency of the request.

[0762] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0763] Step 1: Receiving and analyzing emails

[0764] server

[0765] The server receives new emails from the email server. It extracts the body, subject, and attachments of the received emails. It then uses this extracted data as input for analysis using a Python NLP library (e.g., NLTK or spaCy). Specifically, it extracts important keywords and phrases from the email body and identifies the specific intent of the inquiry. The output of this process is the email content with the identified keywords and phrases.

[0766] Input: Body, subject, and attachments of received emails

[0767] Output: Identified keywords and phrases

[0768] Step 2: Emotion Recognition

[0769] server

[0770] The server takes the keywords and phrases extracted in step 1 as input and analyzes the emotions using an emotion engine. It evaluates the sender's emotions using an emotion analysis model using Hugging Face's Transformers library. Specifically, it evaluates the sender's stress level and urgency. The output of this process is the sender's emotion evaluation result.

[0771] Input: Identified keywords and phrases

[0772] Output: Sender's sentiment evaluation result

[0773] Step 3: Categorization

[0774] server

[0775] The server combines the analysis results from step 1 with the emotion evaluation results from step 2 and classifies the email into an appropriate category. Specifically, it matches important keywords with the emotion evaluation results and classifies the email into categories such as "technical support," "general inquiries," and "emergency response." The output of this process is the email category information.

[0776] Input: Analysis results, sender's emotion evaluation results

[0777] Output: Email category information

[0778] Step 4: Generate a reply

[0779] server

[0780] The server uses the category information classified in step 3 as input to select the optimal reply template and generate a reply using natural language generation (NLG) technology. A generative AI model (e.g., OpenAI's GPT-3) is used to create a reply based on the template. The output of this process is the generated reply.

[0781] Input: Email category information

[0782] Output: The generated reply

[0783] Step 5: Add attachments

[0784] Terminal

[0785] An RPA (e.g., UiPath or Blue Prism) installed on a terminal scans the company's file system or database and automatically searches for the necessary attachments. For example, it searches for files labeled "meeting minutes" and attaches them to an email. The output of this process is a reply message with the attachment.

[0786] Input: Generated reply

[0787] Output: Reply with attachments

[0788] Step 6: Sending an email

[0789] Terminal

[0790] The RPA on the terminal checks the reply text with the attachment from step 5 and sends the email using the SMTP protocol. Specifically, after a final check of the reply text and attachment, it checks that there are no abnormalities and then sends the email. The output of this process is the sent email.

[0791] Input: Reply with attachment

[0792] Output: Sent email

[0793] Step 7: Editing and Approval Functions

[0794] User

[0795] The user reviews the reply generated in step 4 through the system's user interface, edits the reply if necessary, and finally approves or rejects it. This action checks the quality and appropriateness of the reply. The output of this process is a final reply approved by the user.

[0796] Input: Generated reply

[0797] Output: Final reply accepted by the user

[0798] (Application example 2)

[0799] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0800] When dealing with customers in brick-and-mortar stores, staff are required to respond appropriately to a variety of questions and requests. However, current systems make it difficult to respond appropriately by taking into account the customer's emotions, which can result in a decline in customer satisfaction. This also places a heavy burden on staff, making efficient operation difficult. In particular, current systems are insufficient when emotion recognition and rapid response are required.

[0801] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0802] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply, means for adding an attachment to the reply, means for analyzing text and recognizing emotions, means for inputting a question by voice and converting it into text, means for displaying the generated reply, and means for sending an email. This enables prompt and appropriate responses that take emotions into consideration when dealing with customers in a physical store.

[0803] An "inquiry" is a question or request made by a customer seeking information.

[0804] "Analysis" is the process of understanding the content based on input data and extracting classifications and meanings.

[0805] A "category" is a criterion for classifying content into a particular group or groups based on the analyzed content.

[0806] A "reply" is a response message generated in response to an inquiry.

[0807] An "attachment" is an additional data file that is added to a reply.

[0808] "Email" means a text message sent or received electronically.

[0809] A "means" is a method or device for achieving a specific purpose.

[0810] "Natural language processing" is a technology that uses computers to analyze, understand, and generate human language.

[0811] "Emotion recognition" is a technology that analyzes and judges emotional states from text and voice data.

[0812] "Text conversion" is a technology that converts voice data into text information.

[0813] "Display" refers to the visual presentation of analyzed or generated results.

[0814] "Smart glasses" are wearable devices that are capable of augmented reality and displaying information.

[0815] A "robot" is a mechanical device that is programmed to automatically perform specific tasks.

[0816] General Overview

[0817] This invention is a system for improving the efficiency of customer service in brick-and-mortar stores and responding in a way that takes into account the customer's emotions. This system includes a server and a smart device (smart glasses or a robot), and acquires and analyzes the customer's voice data to generate an appropriate reply and display the response.

[0818] Server Configuration

[0819] The server has the following main functions:

[0820] 1. Voice data to text conversion function

[0821] It receives voice data sent from smart devices and converts it into text using the Google Cloud Speech-to-Text API.

[0822] 2. Text analysis and emotion recognition

[0823] The converted text data is analyzed using a natural language processing (NLP) engine (e.g., spaCy or NLTK) to identify the intent of the inquiry, and an emotion engine (e.g., IBM Watson Tone Analyzer) to recognize the customer's emotions.

[0824] 3. Reply generation function

[0825] An appropriate reply is generated based on the analysis results and emotion recognition results, using a generative AI model (e.g., OpenAI GPT-3).

[0826] Smart Device Configuration

[0827] The smart devices (smart glasses and robots) have the following main functions:

[0828] 1. Audio data acquisition function

[0829] A function that allows customers to input questions and requests by voice and send them to the server.

[0830] 2. Display function

[0831] A function that displays text data (reply text) sent from the server, allowing staff to respond to customers quickly.

[0832] Specific examples

[0833] To illustrate, consider the following scenario:

[0834] 1. A customer asks a staff member wearing smart glasses, "I'd like to process a return."

[0835] 2. The staff member's smart glasses send this audio to a server, where it is converted into text using the Google Cloud Speech-to-Text API.

[0836] 3. The converted text is analyzed using spaCy and NLTK to identify the inquiry intent of "I would like to process a return." IBM Watson Tone Analyzer determines that the customer's emotions are calm.

[0837] 4. The server uses OpenAI GPT-3 to generate a reply saying, "Sorry, please go ahead and we'll help you process your return."

[0838] 5. This reply will be displayed on the staff member's smart glasses and the staff member will convey it to the customer.

[0839] Prompt Sentence Examples

[0840] Customer: "I'd like to process a return."

[0841] Smart Glasses: "Sorry, please come on over here and we'll help you process your return."

[0842] As described above, the present invention provides a system for quickly and efficiently handling customers in physical stores, and realizes high-quality service that takes into consideration the emotions of customers.

[0843] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0844] Step 1: Getting voice input

[0845] Subject: Terminal

[0846] How it works: Smart glasses or a robot are used to capture voice questions and requests from users.

[0847] Input: User's voice data.

[0848] Output: The audio data is sent to the server.

[0849] Step 2: Convert audio data to text

[0850] Subject: Server

[0851] Operation: Received voice data is converted into text data using the Google Cloud Speech-to-Text API.

[0852] Input: Audio data.

[0853] Output: Text data.

[0854] Step 3: Parsing the text

[0855] Subject: Server

[0856] How it works: Uses spaCy and NLTK to extract keywords and phrases from text data to identify query intent.

[0857] Input: Text data.

[0858] Output: Analysis result (query intent).

[0859] Step 4: Emotion Recognition

[0860] Subject: Server

[0861] How it works: Using IBM Watson Tone Analyzer, it recognizes the user's emotions from the converted text data and assesses stress levels and urgency.

[0862] Input: Text data.

[0863] Output: Emotion recognition results (stress level, urgency).

[0864] Step 5: Generate a reply

[0865] Subject: Server

[0866] How it works: Based on the analysis results and emotion recognition results, OpenAI GPT-3 is used to generate the optimal reply.

[0867] Input: Analysis results, emotion recognition results.

[0868] Output: The generated reply.

[0869] Step 6: View the reply

[0870] Subject: Terminal

[0871] How it works: The generated reply is displayed on smart glasses or a robot and conveyed to the user.

[0872] Input: The generated reply.

[0873] Output: Text to be displayed on smart glasses or robot.

[0874] For example, if a user asks, "I would like to process a return," the process will proceed as follows:

[0875] 1. The user asks a question by voice, and the smart glasses collect the voice data.

[0876] 2. The audio data is sent to the server and converted to text using the Google Cloud Speech-to-Text API.

[0877] 3. The converted text data was analyzed using spaCy and NLTK to identify the inquiry intent of "return procedure."

[0878] 4. IBM Watson Tone Analyzer recognizes emotions and determines that the user is calm.

[0879] 5. Using OpenAI GPT-3, we generated the following response: "Sorry, please go ahead and we'll help you process your return."

[0880] 6. The generated reply is displayed on the smart glasses and conveyed to the user.

[0881] This process enables fast and sensitive customer service in physical stores.

[0882] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0883] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0884] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0885] [Third embodiment]

[0886] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0887] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0888] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0889] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0890] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0891] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0892] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0893] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0894] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0895] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0896] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0897] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0898] This invention provides a system for automating and efficiently responding to inquiries. This system automates a series of processes: analyzing the content of an inquiry, classifying it into a specified category, generating a reply, adding an attachment, and sending an email. The main elements that make up this system are a server, a terminal, and a user.

[0899] 1. Content analysis and categorization by the server

[0900] server

[0901] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject line, and attachments, and to identify the intent of the specific inquiry and the sentiment of the sender. For this purpose, the server uses natural language processing (NLP) techniques to extract keywords and phrases from the email body and identify the sender's request and situation. Based on the analysis results, the server classifies the email into predefined categories (e.g., technical support, general inquiry, emergency response, etc.).

[0902] As a concrete example, when a server receives and analyzes an email that says, "Please send me the minutes of the meeting. It's urgent," the server will identify two main elements from the email's content: "Request for meeting minutes" and "Urgent," and categorize it based on those elements.

[0903] 2. Generating a Reply

[0904] server

[0905] The server generates an appropriate reply based on the results of category classification and sentiment analysis, using natural language generation (NLG) technology. The server stores different reply templates for each specified category, and automatically selects the appropriate template to generate a reply.

[0906] As a specific example, the server selects a template corresponding to a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0907] 3. Adding attachments to replies

[0908] Terminal

[0909] Robotic Process Automation (RPA) installed on the device automatically identifies the necessary attachments and adds them to the email. The device searches for the appropriate file in the company's file system or database, and once found, attaches it to the email.

[0910] As a concrete example, the device searches the corporate file system for a file labeled "meeting minutes," extracts it, and attaches it to an email.

[0911] 4. Replying to emails

[0912] Terminal

[0913] The RPA on the terminal checks the generated reply and attachments, and finally sends the email to the user. During this process, it checks whether the components of the email are complete and corrects any deficiencies.

[0914] As a specific example, the terminal sends an email reply to the user with the reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait." and the searched meeting minutes file.

[0915] Editing and Approval Features

[0916] User

[0917] It provides an interface that allows users to monitor this process behind the scenes and review, edit, and approve or reject the generated replies, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[0918] This system enables efficient responses to internal inquiries, saving time and costs. It also takes into consideration the quality of the email and the feelings of the sender, improving the overall quality of responses.

[0919] The processing flow will be explained below.

[0920] Step 1:

[0921] server

[0922] The server retrieves new emails from the mailbox and saves the subject, body, and attachments of the received email to a file for analysis.

[0923] Step 2:

[0924] server

[0925] The server analyzes the body, subject, and attachments of received emails using natural language processing (NLP) technology. Specifically, it extracts keywords and important phrases from the email body to identify the email's intent and request. In parallel, it performs sentiment analysis to identify the sender's psychological state and emotions.

[0926] Step 3:

[0927] server

[0928] Based on the analysis results, the server categorizes the email into a corresponding category, such as technical support, general inquiries, or emergency response, based on pre-defined rules.

[0929] Step 4:

[0930] server

[0931] The server uses the classified categories and the results of sentiment analysis to select an appropriate reply template. It then uses natural language generation (NLG) technology to generate a reply based on the selected template. At this stage, the reply text is enhanced to reflect the sender's sentiment.

[0932] Step 5:

[0933] Terminal

[0934] The device's RPA (Robotic Process Automation) searches for the required attachments by scanning the company's file system and databases and identifying the appropriate file based on the results of email analysis and the content of the reply.

[0935] Step 6:

[0936] Terminal

[0937] The RPA on the terminal adds the searched attachments to the email, so that the necessary documents are attached to the reply email according to the user's request.

[0938] Step 7:

[0939] Terminal

[0940] The RPA on the device checks all generated replies and attachments to ensure they are properly structured and checked for consistency and completeness.

[0941] Step 8:

[0942] Terminal

[0943] After the device performs a final check, it automatically sends an email containing the generated reply and attachment, completing the entire process of responding to the inquiry.

[0944] Step 9:

[0945] User

[0946] Users can review the generated replies through the system's user interface, and can edit, approve, or reject them as needed to ensure the quality of the final submission.

[0947] This series of processing flows enables effective and efficient responses to internal inquiries, reducing time and costs while also taking into consideration the quality of the email and the feelings of the sender.

[0948] Example 1

[0949] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0950] In today's business environment, there is a demand for fast and efficient response to inquiries. However, in many cases, manually analyzing, replying to, adding attachments to, and sending large volumes of emails is time-consuming and labor-intensive, making it difficult to maintain consistent response quality. Furthermore, due to insufficient sentiment analysis and automatic response generation, appropriate responses to senders may not be delivered promptly. Furthermore, since there is no human intervention possible for automatically generated replies, there is a risk of sending incorrect replies.

[0951] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0952] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category; means for generating a reply based on the analysis results using natural language generation technology; means for searching for necessary attachments from an internal company database, identifying them, and adding them to the reply; and means for checking the generated reply and attachments and sending the email. This enables automated response to inquiries, enabling fast and efficient responses. Furthermore, the use of natural language processing technology makes it possible to analyze the sender's sentiment and generate an appropriate response. Furthermore, by providing an interface that allows users to edit, approve, or reject the automatically generated reply, it is possible to ensure the quality of responses and reduce the risk of sending incorrect replies.

[0953] A "Contact Us" is a message sent by a user requesting specific information, support, or action.

[0954] A "category" is a criterion for classifying the contents of an inquiry into a specific group.

[0955] "Parsing" is the process of understanding the content of an email's body, subject, and attachments and extracting specific information.

[0956] "Natural language generation technology" refers to technology used by computers to generate human language, generating sentences based on specific templates or models.

[0957] A "reply" is a response message that is automatically generated in response to a received inquiry.

[0958] An "attachment" is an electronic file, such as a document or image, that is added to an email.

[0959] A "database" is a system for efficiently storing, managing, and retrieving structured information.

[0960] "Emotion inference" is the process of identifying the sender's emotions or mood from the content of a received email.

[0961] "User interface" refers to the platform or screen through which a user interacts with a system.

[0962] "Robotic Process Automation" (RPA) is a technology that mimics human actions to automate repetitive tasks such as data entry and file manipulation.

[0963] The "SMTP protocol" is a standard communication protocol for sending email over the Internet.

[0964] System Overview

[0965] This invention provides a system for automating and efficiently responding to inquiries. This system automates a series of processes: analyzing the content of an inquiry, classifying it into a specified category, generating a reply, adding an attachment, and sending an email. The main components are a server, a terminal, and a user.

[0966] 1. Content analysis and categorization by the server

[0967] server

[0968] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject, and attachments, and to identify the intent of the specific inquiry and the sender's sentiment. For this purpose, the server uses natural language processing (NLP) techniques such as "spaCy" and "NLTK." This allows it to extract keywords and phrases from the email body and identify the sender's request and situation.

[0969] To give a concrete example, when a server receives an email that says, "Please send me the meeting minutes. It's urgent," the server identifies two key elements from the email's content: "Request for meeting minutes" and "Urgent," and categorizes the email based on those elements.

[0970] 2. Generating a Reply

[0971] server

[0972] The server generates an appropriate reply based on the results of category classification and sentiment analysis, using natural language generation (NLG) technology such as GPT-3. The server stores different reply templates for each specified category, and automatically selects the appropriate template to generate a reply.

[0973] As a specific example, the server selects a template corresponding to a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[0974] 3. Adding attachments to replies

[0975] Terminal

[0976] Robotic Process Automation (RPA) installed on the device automatically identifies the necessary attachments and adds them to the email. The device searches for the appropriate file in the company's file system or SQL database, and attaches it to the email once it finds it. RPA tools used include UiPath and Automation Anywhere.

[0977] As a concrete example, the device searches the corporate file system for a file labeled "meeting minutes," extracts it, and attaches it to an email.

[0978] 4. Replying to emails

[0979] Terminal

[0980] The RPA on the terminal checks the generated reply message and attachments, and finally sends the email to the user. It checks whether the email components are complete and corrects any errors. The SMTP protocol is used for sending.

[0981] As a specific example, the terminal sends an email reply to the user with the reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait." and the searched meeting minutes file.

[0982] 5. Editing and Approval Functions

[0983] User

[0984] The system provides an interface that allows users to review, edit, and approve or reject the generated responses using a web application built with a framework such as Django, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[0985] Examples of prompt statements

[0986] "Please tell me the process for analyzing the content of incoming emails, categorizing them into appropriate categories, and then generating and sending replies. Specific keywords are 'meeting minutes' and 'urgent'."

[0987] This system allows internal inquiries to be handled quickly and efficiently, saving time and costs, and provides high-quality responses that take the sender's feelings into consideration, improving the overall quality of responses.

[0988] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0989] Step 1: Receiving email

[0990] server

[0991] The server receives new emails from the mail server using the IMAP protocol. The input is the data of unread emails obtained from the mail server.

[0992] What it does: The server connects to the mail server using the IMAP library, retrieves unread emails from the user's mailbox, and stores them locally. The stored email data is used in the next analysis step.

[0993] Step 2: Content analysis

[0994] server

[0995] It analyzes the body, subject, attachments, etc. of received emails. The technology used is a natural language processing (NLP) library such as "spaCy" or "NLTK." The input is the raw received email data, and the output is the analysis results, extracting keywords and phrases.

[0996] What it does: The server runs a Python script to convert the email body, subject line, and attachments into text, then uses an NLP library to parse the text data and extract important keywords and phrases.

[0997] Step 3: Categorization

[0998] server

[0999] Based on the analyzed email content, a pre-trained AI model such as the BERT model is used to classify emails into predefined categories. The input is the analysis result, and the output is the category information.

[1000] What it does: The server calls the "categorization model" and provides it with the parsed keywords and phrases as input. The model classifies the email into the appropriate category (e.g., technical support, general inquiry, emergency response) and returns the result.

[1001] Step 4: Generate a reply

[1002] server

[1003] A reply sentence is generated based on the category classification. The technology used here is a generative AI model such as GPT-3. The input is category information and analysis results, and the output is the generated reply sentence.

[1004] Specific operation: The server calls the "reply generation model" and automatically generates a reply based on the template, category information, and analyzed content. The generated reply is used in the next step, adding an attachment.

[1005] Step 5: Find and add attachments

[1006] Terminal

[1007] Robotic Process Automation (RPA) is used to find the appropriate attachments and add them to emails. Tools used include UiPath and Automation Anywhere. The input is the generated reply and the parsed results, and the output is the reply with the attachments added.

[1008] What it does: The device searches the company's file system or SQL database to find the appropriate file, attaches it to an email, and passes it along with the reply to the next step.

[1009] Step 6: Sending an email

[1010] Terminal

[1011] The reply message and attachments generated using the RPA tool are checked, and the final email is sent to the user. The input is the reply message with the attachments added, and the output is the sending record.

[1012] Specific operation: The device will use the SMTP settings of the mail server to send the confirmed email, and will record the success or failure of the sending in a log, which can be viewed in the management interface.

[1013] Step 7: Review and Approval

[1014] User

[1015] It provides an interface for reviewing automatically generated replies and, if necessary, editing, approving, or rejecting them. The input is the generated reply, and the output is the approved or revised reply.

[1016] What it does: The user can review the reply through a web interface, make any necessary edits, and finally press the "Approve" button to send the email.

[1017] (Application example 1)

[1018] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1019] Traditionally, online shopping sites have mainly handled inquiries manually by customer support staff, which consumes a lot of resources and has issues with the speed and accuracy of responses. Furthermore, because the quality of responses depends on the quality of the staff, there is often inconsistency in the quality of responses. There was a need to solve these problems and realize efficient, high-quality responses using an automated system.

[1020] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1021] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply based on the analysis results, means for adding an attachment to the reply, means for sending an email, and means for operating as a smartphone application and providing a user-friendly interface. This allows for automated response to inquiries on online shopping sites, enabling fast and efficient responses. Furthermore, natural language processing technology can be used to infer the sender's emotions and generate high-quality replies using a generative AI model, thereby improving user satisfaction.

[1022] 1. "Means for analyzing the content of inquiries and classifying them into designated categories" refers to a system that uses natural language processing technology to analyze the body and subject of received emails and classify them into specific categories (e.g., technical support, general inquiries, emergency response, etc.).

[1023] 2. "Means for generating a reply based on the analysis results" refers to a mechanism that uses natural language generation technology to automatically generate an appropriate reply based on the analysis results.

[1024] 3. "Means for adding attachments to replies" refers to a system that searches for attachments required for replies from the company's file system or database and automatically attaches them to the email.

[1025] 4. "Means for sending email" means a mechanism that automatically sends an email containing the generated reply message and attachment to the user.

[1026] 5. "Means that operate as an application for smartphones and provide a user-friendly interface" refers to a mechanism that provides a user interface that allows users to intuitively and easily handle the inquiry response process through an application that operates on a smartphone.

[1027] 6. "Means of analyzing the content of received emails using natural language processing and inferring the sender's emotions" refers to a system that uses natural language processing technology to analyze the content of received emails and infer the sender's emotions and intentions.

[1028] 7. "Means of generating using a generative AI model" means a mechanism that uses generative AI technology (e.g., GPT-4) to automatically generate high-quality natural language generated replies.

[1029] 8. "Means for providing a user interface that allows the user to edit, approve, or reject the generated reply" refers to a mechanism that provides an interface that allows the user to review the generated reply and edit, approve, or reject it as necessary.

[1030] The present invention is a system for automating and efficiently handling inquiries on an online shopping site. This system is mainly composed of a server, terminals, and users.

[1031] Server-based content analysis and categorization

[1032] server

[1033] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject, and attachments, and to identify the intent of the specific inquiry and the sender's sentiment. For this purpose, the server uses natural language processing (NLP) technology. NLP technology uses Hugging Face Transformers to extract keywords and phrases from the email body and identify the sender's request and situation. Based on the analysis results, the server classifies the email into a predefined category (e.g., delivery delay, product inquiry, return, etc.). For example, if the server receives and analyzes an email that reads, "My item has not arrived yet. This is order number 12345," the server will identify the key element of "delivery delay" from the email's content and classify it into a category based on that.

[1034] Generate a reply

[1035] server

[1036] Based on the results of the analysis and sentiment analysis, the server generates an appropriate reply. This is done using natural language generation (NLG) technology, particularly OpenAI's GPT-4. The server maintains different reply templates for each specified category and selects the appropriate template to generate the reply. For example, the server selects the template for "Delivery Delay" and generates the reply, "We apologize for the delay in your order. We have checked the tracking information and can confirm that the item is currently being shipped." An example of a prompt to use is, "There has been a delay in the delivery of order number 12345. What should we do?"

[1037] Adding attachments to replies

[1038] Terminal

[1039] The RPA (Robotic Process Automation) installed on the terminal provides a mechanism to automatically identify necessary attachments and add them to emails. The terminal searches for the appropriate file from the company's file system or database (e.g., MySQL) and attaches it to the email once it finds it. For example, the terminal searches for a PDF file of tracking information labeled "Delivery Delay" from the company's file system, extracts it, and attaches it to the email.

[1040] Sending emails

[1041] Terminal

[1042] The RPA on the terminal checks the generated reply text and attachments, and finally sends the email to the user. During this process, it checks whether the components of the email are complete and corrects any deficiencies. For example, the terminal sends an email reply to the user with the reply text "We apologize for the delay in your order. We have checked the tracking information and confirmed that the shipping process is underway," along with a PDF file of the searched tracking information.

[1043] Editing and Approval Features

[1044] User

[1045] It provides an interface that allows users to monitor this process behind the scenes and review, edit, and approve or reject the generated replies, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[1046] This system significantly improves the efficiency of customer support inquiries on online shopping sites, saving time and costs. Furthermore, the use of natural language processing technology makes it possible to take into consideration the quality of the email and the sender's feelings, improving the overall quality of responses.

[1047] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1048] Step 1:

[1049] Retrieving incoming emails

[1050] The server acquires the received email from the mail server.

[1051] Input: Email data received from the email server

[1052] Output: A mail object for parsing

[1053] Specific operation: Connects to a mail server (e.g., IMAP server) and retrieves new emails. The retrieved emails are saved on the server as objects containing the body, subject, and attachments.

[1054] Step 2:

[1055] Email content analysis

[1056] The server analyzes the body and subject of the received email using natural language processing (NLP) techniques, specifically using Hugging Face Transformers to extract keywords and phrases.

[1057] Input: The retrieved email object

[1058] Output: Analysis results (keywords, category information)

[1059] What it does: The email body and subject are tokenized and input into the model. The model analyzes the context based on the tokens and extracts key keywords and sentiment.

[1060] Step 3:

[1061] Category Classification

[1062] The server classifies the email into predefined categories based on the analysis results.

[1063] Input: Analysis results (keywords, category information)

[1064] Output: Classification results (categories)

[1065] Specific behavior: The extracted keywords are compared with the pre-defined category classification rules to determine the most appropriate category. For example, if the keyword "Delivery Delay" is included, the email will be classified into the "Delivery Delay" category.

[1066] Step 4:

[1067] Generate a reply

[1068] The server generates a reply message based on the classified category using a generative AI model (e.g., GPT-4).

[1069] Input: Classification result (category)

[1070] Output: The generated reply

[1071] Specific operation: The category information is input into the generative AI model as a prompt sentence, and a reply sentence is generated. For example, if the category is classified as "Delivery Delay," the prompt will be "Please generate a reply sentence to address the delivery delay."

[1072] Step 5:

[1073] Searching for and adding attachments

[1074] The device will automatically search for any attachments needed for the reply and add them to the email.

[1075] Input: Generated reply, email object

[1076] Output: Reply email with attachment added

[1077] Specific operation: Search for the appropriate file from the company's file system or database (e.g., MySQL) and attach it to the reply email once found. For example, search for a "PDF file of order information" and attach it to the reply email.

[1078] Step 6:

[1079] Sending emails

[1080] The terminal checks the generated reply message and attachments, and finally sends the email to the user.

[1081] Input: Reply email with attachments added

[1082] Output: Email sent

[1083] Specific operation: Check all components of the reply email, confirm that there are no errors, and then send the email to the user via the email sending server (e.g., SMTP server).

[1084] Step 7:

[1085] Edit and Approve

[1086] The user has an interface to review, edit, and approve or reject the generated response.

[1087] Input: Generated reply, user interface

[1088] Output: Reply edited, approved, or rejected by the user

[1089] What happens: The user logs into the web or mobile interface, reviews the generated response, edits it as needed, and approves or rejects it. If approved, the next step is to send an email.

[1090] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1091] This invention relates to a system for automating and efficiently handling inquiries, and in particular provides a system that combines an emotion engine that recognizes user emotions. This system is composed of a server, terminals, and a mechanism that focuses on the user.

[1092] 1. Content analysis and emotion recognition by the server

[1093] server

[1094] The server first analyzes the received email. The body, subject, and attachments of the received email are analyzed using natural language processing (NLP) technology. Keywords and phrases are extracted from the email body to identify the specific intent of the inquiry.

[1095] Next, the emotion engine performs a sentiment analysis of the email. The emotion engine is equipped with algorithms that recognize the sender's emotions and evaluate their stress level and urgency. It also references past email history to learn the sender's emotional patterns, enabling more accurate analysis. This allows the server to obtain analysis results that take the sender's psychological state into account.

[1096] As a concrete example, when a server receives and analyzes an email that says, "Please send me the minutes of the meeting. It's urgent," the server identifies two elements: "request for meeting minutes" and "urgency," and evaluates the message as "high urgency" through its emotion engine.

[1097] 2. Categorization based on analysis results

[1098] server

[1099] Based on the results of the analysis and sentiment analysis, the server classifies the email into a corresponding category (e.g., technical support, general inquiries, emergency response, etc.).

[1100] 3. Generating a Reply

[1101] server

[1102] The server selects an appropriate reply template based on the results of category classification and sentiment analysis. It then uses natural language generation (NLG) technology to generate a reply based on the selected template. It adds expressions that take into account the sender's sentiment and urgency.

[1103] As a specific example, the server selects a template for a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[1104] 4. Adding attachments to replies

[1105] Terminal

[1106] The robotic process automation (RPA) installed on the device automatically searches for required attachments. The device scans the company's file system and databases, and identifies and adds appropriate files based on the analysis results and the content of the reply.

[1107] As a specific example, the device searches for a file labeled "meeting minutes" and attaches it to an email.

[1108] 5. Sending emails

[1109] Terminal

[1110] The RPA on the terminal checks the contents of the generated reply and attachments, and finally sends the email. After confirming that there are no abnormalities, the properly constructed email is automatically sent.

[1111] As a specific example, the terminal sends an email reply to the user using the searched file of the meeting minutes, along with the generated reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[1112] Editing and Approval Features

[1113] User

[1114] Users can review generated replies through the system's user interface and edit, approve, or reject them as needed, ensuring the quality of responses and the appropriateness of the final submission.

[1115] This system will streamline internal inquiries, saving time and costs, and also enable responses that take user sentiment into account, improving the overall quality of responses.

[1116] The processing flow will be explained below.

[1117] Step 1:

[1118] server

[1119] The server retrieves new emails from the mailbox, saving the email subject, body, and attachments for analysis.

[1120] Step 2:

[1121] server

[1122] The server analyzes the body, subject, and attachments of received emails using natural language processing (NLP) technology. Specifically, it extracts keywords and important phrases from the email body and identifies the specific intent of the inquiry.

[1123] Step 3:

[1124] server

[1125] The server uses an emotion engine to analyze the emotion of emails. The emotion engine analyzes the linguistic features of the email body and recognizes the sender's emotional state. Furthermore, it learns emotional patterns by referencing the sender's past email history, enabling more accurate emotion recognition.

[1126] Step 4:

[1127] server

[1128] The emotion engine uses the results of sentiment analysis to assess the urgency and importance of emails, for example, determining stress levels and whether a request is urgent.

[1129] Step 5:

[1130] server

[1131] Based on the results of the analysis and sentiment analysis, the server classifies the email into a corresponding category (e.g., technical support, general inquiries, emergency response, etc.).

[1132] Step 6:

[1133] server

[1134] The server selects an appropriate reply template based on the assigned category and the results of sentiment analysis. Based on the selected template, it generates a reply using natural language generation (NLG) technology. At this stage, sentiment-based expressions are added to the reply.

[1135] Step 7:

[1136] Terminal

[1137] RPA (Robotic Process Automation) installed on the terminal scans the company's file systems and databases and automatically searches for necessary attachments based on the analysis results and the content of the reply.

[1138] Step 8:

[1139] Terminal

[1140] The RPA on the terminal adds the retrieved attachments to the email, so that the necessary documents according to the user's request are attached to the reply email.

[1141] Step 9:

[1142] Terminal

[1143] The RPA on the terminal checks the generated reply and attachments to ensure that the content is consistent and there are no errors. After confirming that the configuration is correct, the email is sent.

[1144] Step 10:

[1145] User

[1146] Users review the generated replies through the system's user interface, editing, approving, or rejecting them as needed to ensure the quality of the final submission.

[1147] Step 11:

[1148] Terminal

[1149] The terminal then performs a final check of the reply text reviewed by the user and automatically sends it via email, completing the entire inquiry response process.

[1150] This process flow enables efficient and effective responses to internal inquiries, reducing time and costs, and providing high-quality responses that take into account the user's feelings and urgency.

[1151] Example 2

[1152] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1153] Currently, responding to customer inquiries requires manual email analysis, composing replies, and recognizing emotions, which requires a great deal of time and effort. In particular, it is difficult to respond while taking into account the sender's emotions and urgency, resulting in inconsistent response quality. This can lead to lower customer satisfaction and increases the time and cost required for responses.

[1154] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1155] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply based on the analysis result, means for adding an attachment to the reply, means for sending the email, means for recognizing the sender's emotions and evaluating the stress level and urgency, and means for learning emotional patterns using past communication history and improving analysis accuracy. This automates inquiry responses, improves response quality and efficiency, and increases customer satisfaction.

[1156] "Means for analyzing the content of inquiries and classifying them into designated categories" refers to a function for analyzing received emails and communication data and classifying them into appropriate categories (e.g., technical support, emergency response, etc.) based on their content.

[1157] "Means for generating replies based on analysis results" refers to a function that automatically creates replies based on the results of content analysis. These replies are generated using pre-prepared templates and generative AI models.

[1158] "Means for adding attachments to replies" refers to a function that automatically searches for and adds relevant files required for replies, including means for scanning the company's internal file systems and databases.

[1159] "Means for sending email" means a function for sending email, including the generated reply and any necessary attachments, to a specified recipient, typically using the SMTP protocol.

[1160] "Means for recognizing the sender's emotions and assessing their stress level and urgency" refers to a function that recognizes the sender's emotions from the content of received emails and communication data and assesses their stress level and urgency. It uses an emotion engine and natural language processing technology.

[1161] "Means for learning emotional patterns using past communication history to improve analysis accuracy" is a function that analyzes the history of emails and communication data accumulated in the past and learns the emotional patterns of the sender based on that data, thereby improving analysis accuracy.

[1162] "Means for providing a user interface for users to edit, approve, or reject" means providing a user interface with the functionality to allow users to review the generated reply text and edit, approve, or reject it as necessary.

[1163] This invention relates to a system for automating and efficiently handling inquiries, and in particular provides a system that combines an emotion engine that recognizes the sender's emotions. This system is composed of a mechanism that is centered around a server, a terminal, and a user.

[1164] Server-based content analysis and emotion recognition

[1165] server

[1166] The server first receives new emails from the email server. It extracts the body, subject, and attachments of the received emails and performs content analysis using a Python natural language processing (NLP) library (e.g., NLTK or spaCy). It extracts important keywords and phrases from the email body to identify the specific intent of the inquiry. For example, if an email is received saying, "Please send me the minutes of the meeting. It's urgent," the server identifies the keywords "meeting minutes" and "urgent."

[1167] The server then uses an emotion engine to analyze the sentiment of the received email. Specifically, it uses a sentiment analysis model using Hugging Face's Transformers library to evaluate the sender's emotions. For example, it recognizes that the sender is in a hurry from the phrase "I'm in a hurry" and evaluates the stress level and urgency.

[1168] Categorization based on analysis results

[1169] server

[1170] The server combines the results of the analysis and sentiment analysis and classifies the email into an appropriate category (e.g., technical support, general inquiry, emergency response, etc.). For example, an email rated as "high urgency" for "request for meeting minutes" would be classified into the "urgent response" category.

[1171] Generate a reply

[1172] server

[1173] The server selects the optimal reply template, taking into account the appropriate category and sentiment, and generates a reply based on that. This generation uses natural language generation (NLG) technology, leveraging a generative AI model such as OpenAI's GPT-3. For example, it could generate a reply such as, "The meeting minutes you requested will be sent to you shortly. We apologize for the wait."

[1174] Adding attachments to replies

[1175] Terminal

[1176] An RPA (e.g., UiPath or Blue Prism) installed on a terminal scans the company's file system and databases to automatically search for necessary attachments, such as files labeled "meeting minutes," and attach them to emails.

[1177] Sending emails

[1178] Terminal

[1179] The RPA on the terminal checks the generated reply message and the attached file, and after confirming that there are no abnormalities, sends the email using the SMTP protocol. Specifically, it sends a reply message saying, "The requested meeting minutes will be sent shortly. We apologize for the wait," with the meeting minutes file attached.

[1180] Editing and Approval Features

[1181] User

[1182] Users can review generated replies through the system's user interface and edit, approve, or reject them as needed, ensuring the quality of responses and the appropriateness of the final submission.

[1183] Specific examples and prompts for the generative AI model

[1184] Specific examples

[1185] 1. An email is received saying, "Please send me the minutes of the meeting. It's urgent." The server identifies the elements "request for meeting minutes" and "urgent" and evaluates it as "high urgency."

[1186] 2. The server classifies the request as an "urgent response" and generates a reply based on the corresponding reply template: "The meeting minutes you requested will be sent to you shortly. We apologize for the wait."

[1187] 3. The device searches for a file labeled "Meeting Minutes" and attaches it to the email.

[1188] 4. The device checks the email content and sends it after confirming that there are no abnormalities.

[1189] Prompt sentence for generative AI model

[1190] You received an email saying, "Please send me the minutes of the meeting. It's urgent." Use natural language generation technology to generate an appropriate reply to this email. The reply should include expressions that reflect the urgency of the request.

[1191] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1192] Step 1: Receiving and analyzing emails

[1193] server

[1194] The server receives new emails from the email server. It extracts the body, subject, and attachments of the received emails. It then uses this extracted data as input for analysis using a Python NLP library (e.g., NLTK or spaCy). Specifically, it extracts important keywords and phrases from the email body and identifies the specific intent of the inquiry. The output of this process is the email content with the identified keywords and phrases.

[1195] Input: Body, subject, and attachments of received emails

[1196] Output: Identified keywords and phrases

[1197] Step 2: Emotion Recognition

[1198] server

[1199] The server takes the keywords and phrases extracted in step 1 as input and analyzes the emotions using an emotion engine. It evaluates the sender's emotions using an emotion analysis model using Hugging Face's Transformers library. Specifically, it evaluates the sender's stress level and urgency. The output of this process is the sender's emotion evaluation result.

[1200] Input: Identified keywords and phrases

[1201] Output: Sender's sentiment evaluation result

[1202] Step 3: Categorization

[1203] server

[1204] The server combines the analysis results from step 1 with the emotion evaluation results from step 2 and classifies the email into an appropriate category. Specifically, it matches important keywords with the emotion evaluation results and classifies the email into categories such as "technical support," "general inquiries," and "emergency response." The output of this process is the email category information.

[1205] Input: Analysis results, sender's emotion evaluation results

[1206] Output: Email category information

[1207] Step 4: Generate a reply

[1208] server

[1209] The server uses the category information classified in step 3 as input to select the optimal reply template and generate a reply using natural language generation (NLG) technology. A generative AI model (e.g., OpenAI's GPT-3) is used to create a reply based on the template. The output of this process is the generated reply.

[1210] Input: Email category information

[1211] Output: The generated reply

[1212] Step 5: Add attachments

[1213] Terminal

[1214] An RPA (e.g., UiPath or Blue Prism) installed on a terminal scans the company's file system or database and automatically searches for the necessary attachments. For example, it searches for files labeled "meeting minutes" and attaches them to an email. The output of this process is a reply message with the attachment.

[1215] Input: Generated reply

[1216] Output: Reply with attachments

[1217] Step 6: Sending an email

[1218] Terminal

[1219] The RPA on the terminal checks the reply text with the attachment from step 5 and sends the email using the SMTP protocol. Specifically, after a final check of the reply text and attachment, it checks that there are no abnormalities and then sends the email. The output of this process is the sent email.

[1220] Input: Reply with attachment

[1221] Output: Sent email

[1222] Step 7: Editing and Approval Functions

[1223] User

[1224] The user reviews the reply generated in step 4 through the system's user interface, edits the reply if necessary, and finally approves or rejects it. This action checks the quality and appropriateness of the reply. The output of this process is a final reply approved by the user.

[1225] Input: Generated reply

[1226] Output: Final reply accepted by the user

[1227] (Application example 2)

[1228] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1229] When dealing with customers in brick-and-mortar stores, staff are required to respond appropriately to a variety of questions and requests. However, current systems make it difficult to respond appropriately by taking into account the customer's emotions, which can result in a decline in customer satisfaction. This also places a heavy burden on staff, making efficient operation difficult. In particular, current systems are insufficient when emotion recognition and rapid response are required.

[1230] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1231] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply, means for adding an attachment to the reply, means for analyzing text and recognizing emotions, means for inputting a question by voice and converting it into text, means for displaying the generated reply, and means for sending an email. This enables prompt and appropriate responses that take emotions into consideration when dealing with customers in a physical store.

[1232] An "inquiry" is a question or request made by a customer seeking information.

[1233] "Analysis" is the process of understanding the content based on input data and extracting classifications and meanings.

[1234] A "category" is a criterion for classifying content into a particular group or groups based on the analyzed content.

[1235] A "reply" is a response message generated in response to an inquiry.

[1236] An "attachment" is an additional data file that is added to a reply.

[1237] "Email" means a text message sent or received electronically.

[1238] A "means" is a method or device for achieving a specific purpose.

[1239] "Natural language processing" is a technology that uses computers to analyze, understand, and generate human language.

[1240] "Emotion recognition" is a technology that analyzes and judges emotional states from text and voice data.

[1241] "Text conversion" is a technology that converts voice data into text information.

[1242] "Display" refers to the visual presentation of analyzed or generated results.

[1243] "Smart glasses" are wearable devices that are capable of augmented reality and displaying information.

[1244] A "robot" is a mechanical device that is programmed to automatically perform specific tasks.

[1245] General Overview

[1246] This invention is a system for improving the efficiency of customer service in brick-and-mortar stores and responding in a way that takes into account the customer's emotions. This system includes a server and a smart device (smart glasses or a robot), and acquires and analyzes the customer's voice data to generate an appropriate reply and display the response.

[1247] Server Configuration

[1248] The server has the following main functions:

[1249] 1. Voice data to text conversion function

[1250] It receives voice data sent from smart devices and converts it into text using the Google Cloud Speech-to-Text API.

[1251] 2. Text analysis and emotion recognition

[1252] The converted text data is analyzed using a natural language processing (NLP) engine (e.g., spaCy or NLTK) to identify the intent of the inquiry, and an emotion engine (e.g., IBM Watson Tone Analyzer) to recognize the customer's emotions.

[1253] 3. Reply generation function

[1254] An appropriate reply is generated based on the analysis results and emotion recognition results, using a generative AI model (e.g., OpenAI GPT-3).

[1255] Smart Device Configuration

[1256] The smart devices (smart glasses and robots) have the following main functions:

[1257] 1. Audio data acquisition function

[1258] A function that allows customers to input questions and requests by voice and send them to the server.

[1259] 2. Display function

[1260] A function that displays text data (reply text) sent from the server, allowing staff to respond to customers quickly.

[1261] Specific examples

[1262] To illustrate, consider the following scenario:

[1263] 1. A customer asks a staff member wearing smart glasses, "I'd like to process a return."

[1264] 2. The staff member's smart glasses send this audio to a server, where it is converted into text using the Google Cloud Speech-to-Text API.

[1265] 3. The converted text is analyzed using spaCy and NLTK to identify the inquiry intent of "I would like to process a return." IBM Watson Tone Analyzer determines that the customer's emotions are calm.

[1266] 4. The server uses OpenAI GPT-3 to generate a reply saying, "Sorry, please go ahead and we'll help you process your return."

[1267] 5. This reply will be displayed on the staff member's smart glasses and the staff member will convey it to the customer.

[1268] Prompt Sentence Examples

[1269] Customer: "I'd like to process a return."

[1270] Smart Glasses: "Sorry, please come on over here and we'll help you process your return."

[1271] As described above, the present invention provides a system for quickly and efficiently handling customers in physical stores, and realizes high-quality service that takes into consideration the emotions of customers.

[1272] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1273] Step 1: Getting voice input

[1274] Subject: Terminal

[1275] How it works: Smart glasses or a robot are used to capture voice questions and requests from users.

[1276] Input: User's voice data.

[1277] Output: The audio data is sent to the server.

[1278] Step 2: Convert audio data to text

[1279] Subject: Server

[1280] Operation: Received voice data is converted into text data using the Google Cloud Speech-to-Text API.

[1281] Input: Audio data.

[1282] Output: Text data.

[1283] Step 3: Parsing the text

[1284] Subject: Server

[1285] How it works: Uses spaCy and NLTK to extract keywords and phrases from text data to identify query intent.

[1286] Input: Text data.

[1287] Output: Analysis result (query intent).

[1288] Step 4: Emotion Recognition

[1289] Subject: Server

[1290] How it works: Using IBM Watson Tone Analyzer, it recognizes the user's emotions from the converted text data and assesses stress levels and urgency.

[1291] Input: Text data.

[1292] Output: Emotion recognition results (stress level, urgency).

[1293] Step 5: Generate a reply

[1294] Subject: Server

[1295] How it works: Based on the analysis results and emotion recognition results, OpenAI GPT-3 is used to generate the optimal reply.

[1296] Input: Analysis results, emotion recognition results.

[1297] Output: The generated reply.

[1298] Step 6: View the reply

[1299] Subject: Terminal

[1300] How it works: The generated reply is displayed on smart glasses or a robot and conveyed to the user.

[1301] Input: The generated reply.

[1302] Output: Text to be displayed on smart glasses or robot.

[1303] For example, if a user asks, "I would like to process a return," the process will proceed as follows:

[1304] 1. The user asks a question by voice, and the smart glasses collect the voice data.

[1305] 2. The audio data is sent to the server and converted to text using the Google Cloud Speech-to-Text API.

[1306] 3. The converted text data was analyzed using spaCy and NLTK to identify the inquiry intent of "return procedure."

[1307] 4. IBM Watson Tone Analyzer recognizes emotions and determines that the user is calm.

[1308] 5. Using OpenAI GPT-3, we generated the following response: "Sorry, please go ahead and we'll help you process your return."

[1309] 6. The generated reply is displayed on the smart glasses and conveyed to the user.

[1310] This process enables fast and sensitive customer service in physical stores.

[1311] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1312] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1313] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1314] [Fourth embodiment]

[1315] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1316] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1317] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1318] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1319] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1320] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1321] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1322] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1323] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1324] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1325] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1326] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1327] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1328] This invention provides a system for automating and efficiently responding to inquiries. This system automates a series of processes: analyzing the content of an inquiry, classifying it into a specified category, generating a reply, adding an attachment, and sending an email. The main elements that make up this system are a server, a terminal, and a user.

[1329] 1. Content analysis and categorization by the server

[1330] server

[1331] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject line, and attachments, and to identify the intent of the specific inquiry and the sentiment of the sender. For this purpose, the server uses natural language processing (NLP) techniques to extract keywords and phrases from the email body and identify the sender's request and situation. Based on the analysis results, the server classifies the email into predefined categories (e.g., technical support, general inquiry, emergency response, etc.).

[1332] As a concrete example, when a server receives and analyzes an email that says, "Please send me the minutes of the meeting. It's urgent," the server will identify two main elements from the email's content: "Request for meeting minutes" and "Urgent," and categorize it based on those elements.

[1333] 2. Generating a Reply

[1334] server

[1335] The server generates an appropriate reply based on the results of category classification and sentiment analysis, using natural language generation (NLG) technology. The server stores different reply templates for each specified category, and automatically selects the appropriate template to generate a reply.

[1336] As a specific example, the server selects a template corresponding to a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[1337] 3. Adding attachments to replies

[1338] Terminal

[1339] Robotic Process Automation (RPA) installed on the device automatically identifies the necessary attachments and adds them to the email. The device searches for the appropriate file in the company's file system or database, and once found, attaches it to the email.

[1340] As a concrete example, the device searches the corporate file system for a file labeled "meeting minutes," extracts it, and attaches it to an email.

[1341] 4. Replying to emails

[1342] Terminal

[1343] The RPA on the terminal checks the generated reply and attachments, and finally sends the email to the user. During this process, it checks whether the components of the email are complete and corrects any deficiencies.

[1344] As a specific example, the terminal sends an email reply to the user with the reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait." and the searched meeting minutes file.

[1345] Editing and Approval Features

[1346] User

[1347] It provides an interface that allows users to monitor this process behind the scenes and review, edit, and approve or reject the generated replies, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[1348] This system enables efficient responses to internal inquiries, saving time and costs. It also takes into consideration the quality of the email and the feelings of the sender, improving the overall quality of responses.

[1349] The processing flow will be explained below.

[1350] Step 1:

[1351] server

[1352] The server retrieves new emails from the mailbox and saves the subject, body, and attachments of the received email to a file for analysis.

[1353] Step 2:

[1354] server

[1355] The server analyzes the body, subject, and attachments of received emails using natural language processing (NLP) technology. Specifically, it extracts keywords and important phrases from the email body to identify the email's intent and request. In parallel, it performs sentiment analysis to identify the sender's psychological state and emotions.

[1356] Step 3:

[1357] server

[1358] Based on the analysis results, the server categorizes the email into a corresponding category, such as technical support, general inquiries, or emergency response, based on pre-defined rules.

[1359] Step 4:

[1360] server

[1361] The server uses the classified categories and the results of sentiment analysis to select an appropriate reply template. It then uses natural language generation (NLG) technology to generate a reply based on the selected template. At this stage, the reply text is enhanced to reflect the sender's sentiment.

[1362] Step 5:

[1363] Terminal

[1364] The device's RPA (Robotic Process Automation) searches for the required attachments by scanning the company's file system and databases and identifying the appropriate file based on the results of email analysis and the content of the reply.

[1365] Step 6:

[1366] Terminal

[1367] The RPA on the terminal adds the searched attachments to the email, so that the necessary documents are attached to the reply email according to the user's request.

[1368] Step 7:

[1369] Terminal

[1370] The RPA on the device checks all generated replies and attachments to ensure they are properly structured and checked for consistency and completeness.

[1371] Step 8:

[1372] Terminal

[1373] After the device performs a final check, it automatically sends an email containing the generated reply and attachment, completing the entire process of responding to the inquiry.

[1374] Step 9:

[1375] User

[1376] Users can review the generated replies through the system's user interface, and can edit, approve, or reject them as needed to ensure the quality of the final submission.

[1377] This series of processing flows enables effective and efficient responses to internal inquiries, reducing time and costs while also taking into consideration the quality of the email and the feelings of the sender.

[1378] Example 1

[1379] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1380] In today's business environment, there is a demand for fast and efficient response to inquiries. However, in many cases, manually analyzing, replying to, adding attachments to, and sending large volumes of emails is time-consuming and labor-intensive, making it difficult to maintain consistent response quality. Furthermore, due to insufficient sentiment analysis and automatic response generation, appropriate responses to senders may not be delivered promptly. Furthermore, since there is no human intervention possible for automatically generated replies, there is a risk of sending incorrect replies.

[1381] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1382] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category; means for generating a reply based on the analysis results using natural language generation technology; means for searching for necessary attachments from an internal company database, identifying them, and adding them to the reply; and means for checking the generated reply and attachments and sending the email. This enables automated response to inquiries, enabling fast and efficient responses. Furthermore, the use of natural language processing technology makes it possible to analyze the sender's sentiment and generate an appropriate response. Furthermore, by providing an interface that allows users to edit, approve, or reject the automatically generated reply, it is possible to ensure the quality of responses and reduce the risk of sending incorrect replies.

[1383] A "Contact Us" is a message sent by a user requesting specific information, support, or action.

[1384] A "category" is a criterion for classifying the contents of an inquiry into a specific group.

[1385] "Parsing" is the process of understanding the content of an email's body, subject, and attachments and extracting specific information.

[1386] "Natural language generation technology" refers to technology used by computers to generate human language, generating sentences based on specific templates or models.

[1387] A "reply" is a response message that is automatically generated in response to a received inquiry.

[1388] An "attachment" is an electronic file, such as a document or image, that is added to an email.

[1389] A "database" is a system for efficiently storing, managing, and retrieving structured information.

[1390] "Emotion inference" is the process of identifying the sender's emotions or mood from the content of a received email.

[1391] "User interface" refers to the platform or screen through which a user interacts with a system.

[1392] "Robotic Process Automation" (RPA) is a technology that mimics human actions to automate repetitive tasks such as data entry and file manipulation.

[1393] The "SMTP protocol" is a standard communication protocol for sending email over the Internet.

[1394] System Overview

[1395] This invention provides a system for automating and efficiently responding to inquiries. This system automates a series of processes: analyzing the content of an inquiry, classifying it into a specified category, generating a reply, adding an attachment, and sending an email. The main components are a server, a terminal, and a user.

[1396] 1. Content analysis and categorization by the server

[1397] server

[1398] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject, and attachments, and to identify the intent of the specific inquiry and the sender's sentiment. For this purpose, the server uses natural language processing (NLP) techniques such as "spaCy" and "NLTK." This allows it to extract keywords and phrases from the email body and identify the sender's request and situation.

[1399] To give a concrete example, when a server receives an email that says, "Please send me the meeting minutes. It's urgent," the server identifies two key elements from the email's content: "Request for meeting minutes" and "Urgent," and categorizes the email based on those elements.

[1400] 2. Generating a Reply

[1401] server

[1402] The server generates an appropriate reply based on the results of category classification and sentiment analysis, using natural language generation (NLG) technology such as GPT-3. The server stores different reply templates for each specified category, and automatically selects the appropriate template to generate a reply.

[1403] As a specific example, the server selects a template corresponding to a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[1404] 3. Adding attachments to replies

[1405] Terminal

[1406] Robotic Process Automation (RPA) installed on the device automatically identifies the necessary attachments and adds them to the email. The device searches for the appropriate file in the company's file system or SQL database, and attaches it to the email once it finds it. RPA tools used include UiPath and Automation Anywhere.

[1407] As a concrete example, the device searches the corporate file system for a file labeled "meeting minutes," extracts it, and attaches it to an email.

[1408] 4. Replying to emails

[1409] Terminal

[1410] The RPA on the terminal checks the generated reply message and attachments, and finally sends the email to the user. It checks whether the email components are complete and corrects any errors. The SMTP protocol is used for sending.

[1411] As a specific example, the terminal sends an email reply to the user with the reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait." and the searched meeting minutes file.

[1412] 5. Editing and Approval Functions

[1413] User

[1414] The system provides an interface that allows users to review, edit, and approve or reject the generated responses using a web application built with a framework such as Django, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[1415] Examples of prompt statements

[1416] "Please tell me the process for analyzing the content of incoming emails, categorizing them into appropriate categories, and then generating and sending replies. Specific keywords are 'meeting minutes' and 'urgent'."

[1417] This system allows internal inquiries to be handled quickly and efficiently, saving time and costs, and provides high-quality responses that take the sender's feelings into consideration, improving the overall quality of responses.

[1418] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1419] Step 1: Receiving email

[1420] server

[1421] The server receives new emails from the mail server using the IMAP protocol. The input is the data of unread emails obtained from the mail server.

[1422] What it does: The server connects to the mail server using the IMAP library, retrieves unread emails from the user's mailbox, and stores them locally. The stored email data is used in the next analysis step.

[1423] Step 2: Content analysis

[1424] server

[1425] It analyzes the body, subject, attachments, etc. of received emails. The technology used is a natural language processing (NLP) library such as "spaCy" or "NLTK." The input is the raw received email data, and the output is the analysis results, extracting keywords and phrases.

[1426] What it does: The server runs a Python script to convert the email body, subject line, and attachments into text, then uses an NLP library to parse the text data and extract important keywords and phrases.

[1427] Step 3: Categorization

[1428] server

[1429] Based on the analyzed email content, a pre-trained AI model such as the BERT model is used to classify emails into predefined categories. The input is the analysis result, and the output is the category information.

[1430] What it does: The server calls the "categorization model" and provides it with the parsed keywords and phrases as input. The model classifies the email into the appropriate category (e.g., technical support, general inquiry, emergency response) and returns the result.

[1431] Step 4: Generate a reply

[1432] server

[1433] A reply sentence is generated based on the category classification. The technology used here is a generative AI model such as GPT-3. The input is category information and analysis results, and the output is the generated reply sentence.

[1434] Specific operation: The server calls the "reply generation model" and automatically generates a reply based on the template, category information, and analyzed content. The generated reply is used in the next step, adding an attachment.

[1435] Step 5: Find and add attachments

[1436] Terminal

[1437] Robotic Process Automation (RPA) is used to find the appropriate attachments and add them to emails. Tools used include UiPath and Automation Anywhere. The input is the generated reply and the parsed results, and the output is the reply with the attachments added.

[1438] What it does: The device searches the company's file system or SQL database to find the appropriate file, attaches it to an email, and passes it along with the reply to the next step.

[1439] Step 6: Sending an email

[1440] Terminal

[1441] The reply message and attachments generated using the RPA tool are checked, and the final email is sent to the user. The input is the reply message with the attachments added, and the output is the sending record.

[1442] Specific operation: The device will use the SMTP settings of the mail server to send the confirmed email, and will record the success or failure of the sending in a log, which can be viewed in the management interface.

[1443] Step 7: Review and Approval

[1444] User

[1445] It provides an interface for reviewing automatically generated replies and, if necessary, editing, approving, or rejecting them. The input is the generated reply, and the output is the approved or revised reply.

[1446] What it does: The user can review the reply through a web interface, make any necessary edits, and finally press the "Approve" button to send the email.

[1447] (Application example 1)

[1448] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1449] Traditionally, online shopping sites have mainly handled inquiries manually by customer support staff, which consumes a lot of resources and has issues with the speed and accuracy of responses. Furthermore, because the quality of responses depends on the quality of the staff, there is often inconsistency in the quality of responses. There was a need to solve these problems and realize efficient, high-quality responses using an automated system.

[1450] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1451] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply based on the analysis results, means for adding an attachment to the reply, means for sending an email, and means for operating as a smartphone application and providing a user-friendly interface. This allows for automated response to inquiries on online shopping sites, enabling fast and efficient responses. Furthermore, natural language processing technology can be used to infer the sender's emotions and generate high-quality replies using a generative AI model, thereby improving user satisfaction.

[1452] 1. "Means for analyzing the content of inquiries and classifying them into designated categories" refers to a system that uses natural language processing technology to analyze the body and subject of received emails and classify them into specific categories (e.g., technical support, general inquiries, emergency response, etc.).

[1453] 2. "Means for generating a reply based on the analysis results" refers to a mechanism that uses natural language generation technology to automatically generate an appropriate reply based on the analysis results.

[1454] 3. "Means for adding attachments to replies" refers to a system that searches for attachments required for replies from the company's file system or database and automatically attaches them to the email.

[1455] 4. "Means for sending email" means a mechanism that automatically sends an email containing the generated reply message and attachment to the user.

[1456] 5. "Means that operate as an application for smartphones and provide a user-friendly interface" refers to a mechanism that provides a user interface that allows users to intuitively and easily handle the inquiry response process through an application that operates on a smartphone.

[1457] 6. "Means of analyzing the content of received emails using natural language processing and inferring the sender's emotions" refers to a system that uses natural language processing technology to analyze the content of received emails and infer the sender's emotions and intentions.

[1458] 7. "Means of generating using a generative AI model" means a mechanism that uses generative AI technology (e.g., GPT-4) to automatically generate high-quality natural language generated replies.

[1459] 8. "Means for providing a user interface that allows the user to edit, approve, or reject the generated reply" refers to a mechanism that provides an interface that allows the user to review the generated reply and edit, approve, or reject it as necessary.

[1460] The present invention is a system for automating and efficiently handling inquiries on an online shopping site. This system is mainly composed of a server, terminals, and users.

[1461] Server-based content analysis and categorization

[1462] server

[1463] The server first analyzes the received email. The main purpose of the analysis is to understand the content of the email body, subject, and attachments, and to identify the intent of the specific inquiry and the sender's sentiment. For this purpose, the server uses natural language processing (NLP) technology. NLP technology uses Hugging Face Transformers to extract keywords and phrases from the email body and identify the sender's request and situation. Based on the analysis results, the server classifies the email into a predefined category (e.g., delivery delay, product inquiry, return, etc.). For example, if the server receives and analyzes an email that reads, "My item has not arrived yet. This is order number 12345," the server will identify the key element of "delivery delay" from the email's content and classify it into a category based on that.

[1464] Generate a reply

[1465] server

[1466] Based on the results of the analysis and sentiment analysis, the server generates an appropriate reply. This is done using natural language generation (NLG) technology, particularly OpenAI's GPT-4. The server maintains different reply templates for each specified category and selects the appropriate template to generate the reply. For example, the server selects the template for "Delivery Delay" and generates the reply, "We apologize for the delay in your order. We have checked the tracking information and can confirm that the item is currently being shipped." An example of a prompt to use is, "There has been a delay in the delivery of order number 12345. What should we do?"

[1467] Adding attachments to replies

[1468] Terminal

[1469] The RPA (Robotic Process Automation) installed on the terminal provides a mechanism to automatically identify necessary attachments and add them to emails. The terminal searches for the appropriate file from the company's file system or database (e.g., MySQL) and attaches it to the email once it finds it. For example, the terminal searches for a PDF file of tracking information labeled "Delivery Delay" from the company's file system, extracts it, and attaches it to the email.

[1470] Sending emails

[1471] Terminal

[1472] The RPA on the terminal checks the generated reply text and attachments, and finally sends the email to the user. During this process, it checks whether the components of the email are complete and corrects any deficiencies. For example, the terminal sends an email reply to the user with the reply text "We apologize for the delay in your order. We have checked the tracking information and confirmed that the shipping process is underway," along with a PDF file of the searched tracking information.

[1473] Editing and Approval Features

[1474] User

[1475] It provides an interface that allows users to monitor this process behind the scenes and review, edit, and approve or reject the generated replies, allowing users to ensure the quality of the auto-generated responses and make manual adjustments if necessary.

[1476] This system significantly improves the efficiency of customer support inquiries on online shopping sites, saving time and costs. Furthermore, the use of natural language processing technology makes it possible to take into consideration the quality of the email and the sender's feelings, improving the overall quality of responses.

[1477] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1478] Step 1:

[1479] Retrieving incoming emails

[1480] The server acquires the received email from the mail server.

[1481] Input: Email data received from the email server

[1482] Output: A mail object for parsing

[1483] Specific operation: Connects to a mail server (e.g., IMAP server) and retrieves new emails. The retrieved emails are saved on the server as objects containing the body, subject, and attachments.

[1484] Step 2:

[1485] Email content analysis

[1486] The server analyzes the body and subject of the received email using natural language processing (NLP) techniques, specifically using Hugging Face Transformers to extract keywords and phrases.

[1487] Input: The retrieved email object

[1488] Output: Analysis results (keywords, category information)

[1489] What it does: The email body and subject are tokenized and input into the model. The model analyzes the context based on the tokens and extracts key keywords and sentiment.

[1490] Step 3:

[1491] Category Classification

[1492] The server classifies the email into predefined categories based on the analysis results.

[1493] Input: Analysis results (keywords, category information)

[1494] Output: Classification results (categories)

[1495] Specific behavior: The extracted keywords are compared with the pre-defined category classification rules to determine the most appropriate category. For example, if the keyword "Delivery Delay" is included, the email will be classified into the "Delivery Delay" category.

[1496] Step 4:

[1497] Generate a reply

[1498] The server generates a reply message based on the classified category using a generative AI model (e.g., GPT-4).

[1499] Input: Classification result (category)

[1500] Output: The generated reply

[1501] Specific operation: The category information is input into the generative AI model as a prompt sentence, and a reply sentence is generated. For example, if the category is classified as "Delivery Delay," the prompt will be "Please generate a reply sentence to address the delivery delay."

[1502] Step 5:

[1503] Searching for and adding attachments

[1504] The device will automatically search for any attachments needed for the reply and add them to the email.

[1505] Input: Generated reply, email object

[1506] Output: Reply email with attachment added

[1507] Specific operation: Search for the appropriate file from the company's file system or database (e.g., MySQL) and attach it to the reply email once found. For example, search for a "PDF file of order information" and attach it to the reply email.

[1508] Step 6:

[1509] Sending emails

[1510] The terminal checks the generated reply message and attachments, and finally sends the email to the user.

[1511] Input: Reply email with attachments added

[1512] Output: Email sent

[1513] Specific operation: Check all components of the reply email, confirm that there are no errors, and then send the email to the user via the email sending server (e.g., SMTP server).

[1514] Step 7:

[1515] Edit and Approve

[1516] The user has an interface to review, edit, and approve or reject the generated response.

[1517] Input: Generated reply, user interface

[1518] Output: Reply edited, approved, or rejected by the user

[1519] What happens: The user logs into the web or mobile interface, reviews the generated response, edits it as needed, and approves or rejects it. If approved, the next step is to send an email.

[1520] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1521] This invention relates to a system for automating and efficiently handling inquiries, and in particular provides a system that combines an emotion engine that recognizes user emotions. This system is composed of a server, terminals, and a mechanism that focuses on the user.

[1522] 1. Content analysis and emotion recognition by the server

[1523] server

[1524] The server first analyzes the received email. The body, subject, and attachments of the received email are analyzed using natural language processing (NLP) technology. Keywords and phrases are extracted from the email body to identify the specific intent of the inquiry.

[1525] Next, the emotion engine performs a sentiment analysis of the email. The emotion engine is equipped with algorithms that recognize the sender's emotions and evaluate their stress level and urgency. It also references past email history to learn the sender's emotional patterns, enabling more accurate analysis. This allows the server to obtain analysis results that take the sender's psychological state into account.

[1526] As a concrete example, when a server receives and analyzes an email that says, "Please send me the minutes of the meeting. It's urgent," the server identifies two elements: "request for meeting minutes" and "urgency," and evaluates the message as "high urgency" through its emotion engine.

[1527] 2. Categorization based on analysis results

[1528] server

[1529] Based on the results of the analysis and sentiment analysis, the server classifies the email into a corresponding category (e.g., technical support, general inquiries, emergency response, etc.).

[1530] 3. Generating a Reply

[1531] server

[1532] The server selects an appropriate reply template based on the results of category classification and sentiment analysis. It then uses natural language generation (NLG) technology to generate a reply based on the selected template. It adds expressions that take into account the sender's sentiment and urgency.

[1533] As a specific example, the server selects a template for a "request for meeting minutes" and generates a reply message that reads, "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[1534] 4. Adding attachments to replies

[1535] Terminal

[1536] The robotic process automation (RPA) installed on the device automatically searches for required attachments. The device scans the company's file system and databases, and identifies and adds appropriate files based on the analysis results and the content of the reply.

[1537] As a specific example, the device searches for a file labeled "meeting minutes" and attaches it to an email.

[1538] 5. Sending emails

[1539] Terminal

[1540] The RPA on the terminal checks the contents of the generated reply and attachments, and finally sends the email. After confirming that there are no abnormalities, the properly constructed email is automatically sent.

[1541] As a specific example, the terminal sends an email reply to the user using the searched file of the meeting minutes, along with the generated reply message "The requested meeting minutes will be sent to you shortly. We apologize for the wait."

[1542] Editing and Approval Features

[1543] User

[1544] Users can review generated replies through the system's user interface and edit, approve, or reject them as needed, ensuring the quality of responses and the appropriateness of the final submission.

[1545] This system will streamline internal inquiries, saving time and costs, and also enable responses that take user sentiment into account, improving the overall quality of responses.

[1546] The processing flow will be explained below.

[1547] Step 1:

[1548] server

[1549] The server retrieves new emails from the mailbox, saving the email subject, body, and attachments for analysis.

[1550] Step 2:

[1551] server

[1552] The server analyzes the body, subject, and attachments of received emails using natural language processing (NLP) technology. Specifically, it extracts keywords and important phrases from the email body and identifies the specific intent of the inquiry.

[1553] Step 3:

[1554] server

[1555] The server uses an emotion engine to analyze the emotion of emails. The emotion engine analyzes the linguistic features of the email body and recognizes the sender's emotional state. Furthermore, it learns emotional patterns by referencing the sender's past email history, enabling more accurate emotion recognition.

[1556] Step 4:

[1557] server

[1558] The emotion engine uses the results of sentiment analysis to assess the urgency and importance of emails, for example, determining stress levels and whether a request is urgent.

[1559] Step 5:

[1560] server

[1561] Based on the results of the analysis and sentiment analysis, the server classifies the email into a corresponding category (e.g., technical support, general inquiries, emergency response, etc.).

[1562] Step 6:

[1563] server

[1564] The server selects an appropriate reply template based on the assigned category and the results of sentiment analysis. Based on the selected template, it generates a reply using natural language generation (NLG) technology. At this stage, sentiment-based expressions are added to the reply.

[1565] Step 7:

[1566] Terminal

[1567] RPA (Robotic Process Automation) installed on the terminal scans the company's file systems and databases and automatically searches for necessary attachments based on the analysis results and the content of the reply.

[1568] Step 8:

[1569] Terminal

[1570] The RPA on the terminal adds the retrieved attachments to the email, so that the necessary documents according to the user's request are attached to the reply email.

[1571] Step 9:

[1572] Terminal

[1573] The RPA on the terminal checks the generated reply and attachments to ensure that the content is consistent and there are no errors. After confirming that the configuration is correct, the email is sent.

[1574] Step 10:

[1575] User

[1576] Users review the generated replies through the system's user interface, editing, approving, or rejecting them as needed to ensure the quality of the final submission.

[1577] Step 11:

[1578] Terminal

[1579] The terminal then performs a final check of the reply text reviewed by the user and automatically sends it via email, completing the entire inquiry response process.

[1580] This process flow enables efficient and effective responses to internal inquiries, reducing time and costs, and providing high-quality responses that take into account the user's feelings and urgency.

[1581] Example 2

[1582] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1583] Currently, responding to customer inquiries requires manual email analysis, composing replies, and recognizing emotions, which requires a great deal of time and effort. In particular, it is difficult to respond while taking into account the sender's emotions and urgency, resulting in inconsistent response quality. This can lead to lower customer satisfaction and increases the time and cost required for responses.

[1584] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1585] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply based on the analysis result, means for adding an attachment to the reply, means for sending the email, means for recognizing the sender's emotions and evaluating the stress level and urgency, and means for learning emotional patterns using past communication history and improving analysis accuracy. This automates inquiry responses, improves response quality and efficiency, and increases customer satisfaction.

[1586] "Means for analyzing the content of inquiries and classifying them into designated categories" refers to a function for analyzing received emails and communication data and classifying them into appropriate categories (e.g., technical support, emergency response, etc.) based on their content.

[1587] "Means for generating replies based on analysis results" refers to a function that automatically creates replies based on the results of content analysis. These replies are generated using pre-prepared templates and generative AI models.

[1588] "Means for adding attachments to replies" refers to a function that automatically searches for and adds relevant files required for replies, including means for scanning the company's internal file systems and databases.

[1589] "Means for sending email" means a function for sending email, including the generated reply and any necessary attachments, to a specified recipient, typically using the SMTP protocol.

[1590] "Means for recognizing the sender's emotions and assessing their stress level and urgency" refers to a function that recognizes the sender's emotions from the content of received emails and communication data and assesses their stress level and urgency. It uses an emotion engine and natural language processing technology.

[1591] "Means for learning emotional patterns using past communication history to improve analysis accuracy" is a function that analyzes the history of emails and communication data accumulated in the past and learns the emotional patterns of the sender based on that data, thereby improving analysis accuracy.

[1592] "Means for providing a user interface for users to edit, approve, or reject" means providing a user interface with the functionality to allow users to review the generated reply text and edit, approve, or reject it as necessary.

[1593] This invention relates to a system for automating and efficiently handling inquiries, and in particular provides a system that combines an emotion engine that recognizes the sender's emotions. This system is composed of a mechanism that is centered around a server, a terminal, and a user.

[1594] Server-based content analysis and emotion recognition

[1595] server

[1596] The server first receives new emails from the email server. It extracts the body, subject, and attachments of the received emails and performs content analysis using a Python natural language processing (NLP) library (e.g., NLTK or spaCy). It extracts important keywords and phrases from the email body to identify the specific intent of the inquiry. For example, if an email is received saying, "Please send me the minutes of the meeting. It's urgent," the server identifies the keywords "meeting minutes" and "urgent."

[1597] The server then uses an emotion engine to analyze the sentiment of the received email. Specifically, it uses a sentiment analysis model using Hugging Face's Transformers library to evaluate the sender's emotions. For example, it recognizes that the sender is in a hurry from the phrase "I'm in a hurry" and evaluates the stress level and urgency.

[1598] Categorization based on analysis results

[1599] server

[1600] The server combines the results of the analysis and sentiment analysis and classifies the email into an appropriate category (e.g., technical support, general inquiry, emergency response, etc.). For example, an email rated as "high urgency" for "request for meeting minutes" would be classified into the "urgent response" category.

[1601] Generate a reply

[1602] server

[1603] The server selects the optimal reply template, taking into account the appropriate category and sentiment, and generates a reply based on that. This generation uses natural language generation (NLG) technology, leveraging a generative AI model such as OpenAI's GPT-3. For example, it could generate a reply such as, "The meeting minutes you requested will be sent to you shortly. We apologize for the wait."

[1604] Adding attachments to replies

[1605] Terminal

[1606] An RPA (e.g., UiPath or Blue Prism) installed on a terminal scans the company's file system and databases to automatically search for necessary attachments, such as files labeled "meeting minutes," and attach them to emails.

[1607] Sending emails

[1608] Terminal

[1609] The RPA on the terminal checks the generated reply message and the attached file, and after confirming that there are no abnormalities, sends the email using the SMTP protocol. Specifically, it sends a reply message saying, "The requested meeting minutes will be sent shortly. We apologize for the wait," with the meeting minutes file attached.

[1610] Editing and Approval Features

[1611] User

[1612] Users can review generated replies through the system's user interface and edit, approve, or reject them as needed, ensuring the quality of responses and the appropriateness of the final submission.

[1613] Specific examples and prompts for the generative AI model

[1614] Specific examples

[1615] 1. An email is received saying, "Please send me the minutes of the meeting. It's urgent." The server identifies the elements "request for meeting minutes" and "urgent" and evaluates it as "high urgency."

[1616] 2. The server classifies the request as an "urgent response" and generates a reply based on the corresponding reply template: "The meeting minutes you requested will be sent to you shortly. We apologize for the wait."

[1617] 3. The device searches for a file labeled "Meeting Minutes" and attaches it to the email.

[1618] 4. The device checks the email content and sends it after confirming that there are no abnormalities.

[1619] Prompt sentence for generative AI model

[1620] You received an email saying, "Please send me the minutes of the meeting. It's urgent." Use natural language generation technology to generate an appropriate reply to this email. The reply should include expressions that reflect the urgency of the request.

[1621] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1622] Step 1: Receiving and analyzing emails

[1623] server

[1624] The server receives new emails from the email server. It extracts the body, subject, and attachments of the received emails. It then uses this extracted data as input for analysis using a Python NLP library (e.g., NLTK or spaCy). Specifically, it extracts important keywords and phrases from the email body and identifies the specific intent of the inquiry. The output of this process is the email content with the identified keywords and phrases.

[1625] Input: Body, subject, and attachments of received emails

[1626] Output: Identified keywords and phrases

[1627] Step 2: Emotion Recognition

[1628] server

[1629] The server takes the keywords and phrases extracted in step 1 as input and analyzes the emotions using an emotion engine. It evaluates the sender's emotions using an emotion analysis model using Hugging Face's Transformers library. Specifically, it evaluates the sender's stress level and urgency. The output of this process is the sender's emotion evaluation result.

[1630] Input: Identified keywords and phrases

[1631] Output: Sender's sentiment evaluation result

[1632] Step 3: Categorization

[1633] server

[1634] The server combines the analysis results from step 1 with the emotion evaluation results from step 2 and classifies the email into an appropriate category. Specifically, it matches important keywords with the emotion evaluation results and classifies the email into categories such as "technical support," "general inquiries," and "emergency response." The output of this process is the email category information.

[1635] Input: Analysis results, sender's emotion evaluation results

[1636] Output: Email category information

[1637] Step 4: Generate a reply

[1638] server

[1639] The server uses the category information classified in step 3 as input to select the optimal reply template and generate a reply using natural language generation (NLG) technology. A generative AI model (e.g., OpenAI's GPT-3) is used to create a reply based on the template. The output of this process is the generated reply.

[1640] Input: Email category information

[1641] Output: The generated reply

[1642] Step 5: Add attachments

[1643] Terminal

[1644] An RPA (e.g., UiPath or Blue Prism) installed on a terminal scans the company's file system or database and automatically searches for the necessary attachments. For example, it searches for files labeled "meeting minutes" and attaches them to an email. The output of this process is a reply message with the attachment.

[1645] Input: Generated reply

[1646] Output: Reply with attachments

[1647] Step 6: Sending an email

[1648] Terminal

[1649] The RPA on the terminal checks the reply text with the attachment from step 5 and sends the email using the SMTP protocol. Specifically, after a final check of the reply text and attachment, it checks that there are no abnormalities and then sends the email. The output of this process is the sent email.

[1650] Input: Reply with attachment

[1651] Output: Sent email

[1652] Step 7: Editing and Approval Functions

[1653] User

[1654] The user reviews the reply generated in step 4 through the system's user interface, edits the reply if necessary, and finally approves or rejects it. This action checks the quality and appropriateness of the reply. The output of this process is a final reply approved by the user.

[1655] Input: Generated reply

[1656] Output: Final reply accepted by the user

[1657] (Application example 2)

[1658] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1659] When dealing with customers in brick-and-mortar stores, staff are required to respond appropriately to a variety of questions and requests. However, current systems make it difficult to respond appropriately by taking into account the customer's emotions, which can result in a decline in customer satisfaction. This also places a heavy burden on staff, making efficient operation difficult. In particular, current systems are insufficient when emotion recognition and rapid response are required.

[1660] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1661] In this invention, the server includes means for analyzing the content of the inquiry and classifying it into a specified category, means for generating a reply, means for adding an attachment to the reply, means for analyzing text and recognizing emotions, means for inputting a question by voice and converting it into text, means for displaying the generated reply, and means for sending an email. This enables prompt and appropriate responses that take emotions into consideration when dealing with customers in a physical store.

[1662] An "inquiry" is a question or request made by a customer seeking information.

[1663] "Analysis" is the process of understanding the content based on input data and extracting classifications and meanings.

[1664] A "category" is a criterion for classifying content into a particular group or groups based on the analyzed content.

[1665] A "reply" is a response message generated in response to an inquiry.

[1666] An "attachment" is an additional data file that is added to a reply.

[1667] "Email" means a text message sent or received electronically.

[1668] A "means" is a method or device for achieving a specific purpose.

[1669] "Natural language processing" is a technology that uses computers to analyze, understand, and generate human language.

[1670] "Emotion recognition" is a technology that analyzes and judges emotional states from text and voice data.

[1671] "Text conversion" is a technology that converts voice data into text information.

[1672] "Display" refers to the visual presentation of analyzed or generated results.

[1673] "Smart glasses" are wearable devices that are capable of augmented reality and displaying information.

[1674] A "robot" is a mechanical device that is programmed to automatically perform specific tasks.

[1675] General Overview

[1676] This invention is a system for improving the efficiency of customer service in brick-and-mortar stores and responding in a way that takes into account the customer's emotions. This system includes a server and a smart device (smart glasses or a robot), and acquires and analyzes the customer's voice data to generate an appropriate reply and display the response.

[1677] Server Configuration

[1678] The server has the following main functions:

[1679] 1. Voice data to text conversion function

[1680] It receives voice data sent from smart devices and converts it into text using the Google Cloud Speech-to-Text API.

[1681] 2. Text analysis and emotion recognition

[1682] The converted text data is analyzed using a natural language processing (NLP) engine (e.g., spaCy or NLTK) to identify the intent of the inquiry, and an emotion engine (e.g., IBM Watson Tone Analyzer) to recognize the customer's emotions.

[1683] 3. Reply generation function

[1684] An appropriate reply is generated based on the analysis results and emotion recognition results, using a generative AI model (e.g., OpenAI GPT-3).

[1685] Smart Device Configuration

[1686] The smart devices (smart glasses and robots) have the following main functions:

[1687] 1. Audio data acquisition function

[1688] A function that allows customers to input questions and requests by voice and send them to the server.

[1689] 2. Display function

[1690] A function that displays text data (reply text) sent from the server, allowing staff to respond to customers quickly.

[1691] Specific examples

[1692] To illustrate, consider the following scenario:

[1693] 1. A customer asks a staff member wearing smart glasses, "I'd like to process a return."

[1694] 2. The staff member's smart glasses send this audio to a server, where it is converted into text using the Google Cloud Speech-to-Text API.

[1695] 3. The converted text is analyzed using spaCy and NLTK to identify the inquiry intent of "I would like to process a return." IBM Watson Tone Analyzer determines that the customer's emotions are calm.

[1696] 4. The server uses OpenAI GPT-3 to generate a reply saying, "Sorry, please go ahead and we'll help you process your return."

[1697] 5. This reply will be displayed on the staff member's smart glasses and the staff member will convey it to the customer.

[1698] Prompt Sentence Examples

[1699] Customer: "I'd like to process a return."

[1700] Smart Glasses: "Sorry, please come on over here and we'll help you process your return."

[1701] As described above, the present invention provides a system for quickly and efficiently handling customers in physical stores, and realizes high-quality service that takes into consideration the emotions of customers.

[1702] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1703] Step 1: Getting voice input

[1704] Subject: Terminal

[1705] How it works: Smart glasses or a robot are used to capture voice questions and requests from users.

[1706] Input: User's voice data.

[1707] Output: The audio data is sent to the server.

[1708] Step 2: Convert audio data to text

[1709] Subject: Server

[1710] Operation: Received voice data is converted into text data using the Google Cloud Speech-to-Text API.

[1711] Input: Audio data.

[1712] Output: Text data.

[1713] Step 3: Parsing the text

[1714] Subject: Server

[1715] How it works: Uses spaCy and NLTK to extract keywords and phrases from text data to identify query intent.

[1716] Input: Text data.

[1717] Output: Analysis result (query intent).

[1718] Step 4: Emotion Recognition

[1719] Subject: Server

[1720] How it works: Using IBM Watson Tone Analyzer, it recognizes the user's emotions from the converted text data and assesses stress levels and urgency.

[1721] Input: Text data.

[1722] Output: Emotion recognition results (stress level, urgency).

[1723] Step 5: Generate a reply

[1724] Subject: Server

[1725] How it works: Based on the analysis results and emotion recognition results, OpenAI GPT-3 is used to generate the optimal reply.

[1726] Input: Analysis results, emotion recognition results.

[1727] Output: The generated reply.

[1728] Step 6: View the reply

[1729] Subject: Terminal

[1730] How it works: The generated reply is displayed on smart glasses or a robot and conveyed to the user.

[1731] Input: The generated reply.

[1732] Output: Text to be displayed on smart glasses or robot.

[1733] For example, if a user asks, "I would like to process a return," the process will proceed as follows:

[1734] 1. The user asks a question by voice, and the smart glasses collect the voice data.

[1735] 2. The audio data is sent to the server and converted to text using the Google Cloud Speech-to-Text API.

[1736] 3. The converted text data was analyzed using spaCy and NLTK to identify the inquiry intent of "return procedure."

[1737] 4. IBM Watson Tone Analyzer recognizes emotions and determines that the user is calm.

[1738] 5. Using OpenAI GPT-3, we generated the following response: "Sorry, please go ahead and we'll help you process your return."

[1739] 6. The generated reply is displayed on the smart glasses and conveyed to the user.

[1740] This process enables fast and sensitive customer service in physical stores.

[1741] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1742] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1743] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1744] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1745] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1746] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1747] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1748] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1749] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1750] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1751] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1752] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1753] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1754] 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.

[1755] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1756] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1757] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1758] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1759] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1760] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1761] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1762] The following is further disclosed regarding the above embodiment.

[1763] (Claim 1)

[1764] A means for analyzing the content of inquiries and classifying them into designated categories;

[1765] A means for generating a reply message based on the analysis result;

[1766] A way to add attachments to replies,

[1767] A means for sending emails;

[1768] A system including:

[1769] (Claim 2)

[1770] 10. The system according to claim 1, further comprising means for analyzing the content of a received email using natural language processing to infer the sentiment of the sender.

[1771] (Claim 3)

[1772] 10. The system of claim 1, further comprising: means for providing a user interface for a user to edit, approve, or reject the generated reply.

[1773] "Example 1"

[1774] (Claim 1)

[1775] A means for analyzing the content of inquiries and classifying them into designated categories;

[1776] A means for generating a reply sentence based on the analysis result using natural language generation technology;

[1777] A means to search for and identify required attachments from a corporate database and add them to the reply;

[1778] A way to view the generated reply and attachments and send the email;

[1779] A system including:

[1780] (Claim 2)

[1781] 10. The system according to claim 1, further comprising means for analyzing the content of a received email using natural language processing technology and inferring the sentiment of the sender.

[1782] (Claim 3)

[1783] 10. The system of claim 1, further comprising: means for providing an interface for a user to edit, approve, or reject the generated reply.

[1784] "Application Example 1"

[1785] (Claim 1)

[1786] A means for analyzing the content of inquiries and classifying them into designated categories;

[1787] A means for generating a reply message based on the analysis result;

[1788] A way to add attachments to replies,

[1789] A means for sending emails;

[1790] means for operating as an application for a smartphone and providing a user-friendly interface;

[1791] A system including:

[1792] (Claim 2)

[1793] 10. The system according to claim 1, further comprising means for analyzing the content of a received email using natural language processing to infer the sentiment of the sender.

[1794] (Claim 3)

[1795] 10. The system of claim 1, further comprising: means for generating the generated reply sentence using a generative AI model.

[1796] (Claim 4)

[1797] 10. The system of claim 1, further comprising: means for providing a user interface for a user to edit, approve, or reject the generated reply.

[1798] "Example 2: Combining Emotion Engines"

[1799] (Claim 1)

[1800] A means for analyzing the content of inquiries and classifying them into designated categories;

[1801] A means for generating a reply message based on the analysis result;

[1802] A way to add attachments to replies,

[1803] A means for sending emails;

[1804] A means of recognizing the sender's emotions and assessing their stress level and urgency;

[1805] A means for learning emotion patterns using past communication history and improving analysis accuracy;

[1806] A system including:

[1807] (Claim 2)

[1808] 10. The system of claim 1, further comprising means for analyzing received communications using natural language processing to infer sentiment of the sender.

[1809] (Claim 3)

[1810] 10. The system of claim 1, further comprising: means for providing a user interface for a user to edit, approve, or reject the generated reply.

[1811] "Application example 2 when combining emotion engines"

[1812] (Claim 1)

[1813] A means for analyzing the content of inquiries and classifying them into designated categories;

[1814] A means for generating a reply message based on the analysis result;

[1815] A way to add attachments to replies,

[1816] A means for sending emails;

[1817] A way to input questions by voice and convert them into text;

[1818] A means of analyzing text and recognizing emotions;

[1819] Includes a means to display the generated reply

[1820] system.

[1821] (Claim 2)

[1822] 10. The system according to claim 1, further comprising means for analyzing the content of a received email using natural language processing to infer the sentiment of the sender.

[1823] (Claim 3)

[1824] 10. The system of claim 1, further comprising: means for providing a user interface for a user to edit, approve, or reject the generated reply.

[1825] (Claim 4)

[1826] The system of claim 1, wherein the display means is smart glasses or a robot. [Explanation of symbols]

[1827] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A means for analyzing the content of inquiries and classifying them into designated categories; A means for generating a reply message based on the analysis result; A way to add attachments to replies, A means for sending emails; A system including:

2. 2. The system according to claim 1, further comprising means for analyzing the content of a received email by natural language processing and inferring the feelings of the sender.

3. The system of claim 1 , further comprising: means for providing a user interface for a user to edit, approve, or reject the generated reply.

Images