system

Abstract

We provide the system. [Solution] A means for receiving voice complaints from customers, A means of recording the received audio, Methods for transcribing recorded audio data, A means of analyzing the transcribed text to understand the content of the complaint, A means of generating an appropriate apology response based on the content of the complaint, A means of providing a generated apology response, A system that includes this.

Description

Technical Field

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

Background Art

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

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Claim handling at a customer center is labor-intensive and burdensome for staff. Also, if claims calls or claims / inquiry emails cannot be handled quickly and appropriately, there is a risk of a decrease in customer satisfaction and deterioration of the corporate image. Furthermore, in claim handling, there is a problem that it is difficult to maintain consistent high-quality handling because the quality of handling varies from staff to staff.

Means for Solving the Problems

[0005] The present invention solves the above problem by providing a system that includes means for receiving voice complaints from customers, recording the voice and transcribing it, analyzing the transcribed text to understand the content of the complaint, and generating an appropriate apology response based on the content of the complaint and providing that apology response. Furthermore, the present invention provides a system that includes means for receiving complaints and inquiries via email, analyzing their content to understand the content of the complaints and inquiries, and generating and providing an appropriate reply. The present invention can efficiently provide high-quality and consistent responses by analyzing transcribed text and email content using a natural language processing model and automatically generating reply content and apology responses based on the results.

[0006] "Voice complaints" refer to complaints that customers communicate by voice, such as over the phone.

[0007] "Means of recording" refers to a function or device for electronically storing audio data.

[0008] "Methods for transcription" refer to functions or systems for converting audio data into text data.

[0009] "Means of analysis" refer to algorithms and software used to analyze text data and understand its content.

[0010] "Complaint details" refer to the specific key points of the customer's dissatisfaction or problems.

[0011] An "apology response" is a response that includes an apology in response to a customer complaint.

[0012] "Means of providing" refers to functions and interfaces for displaying the generated apology responses and reply content to the user.

[0013] "Email content" refers to the information contained in the body of an email containing a complaint or inquiry.

[0014] A "natural language processing model" is a machine learning model or algorithm for analyzing text data to understand its meaning.

[0015] The "reply content" is the content of the answer provided in response to a customer's email.

[0016] The "means for automatic generation" is a function or system that automatically creates reply content or apology responses based on the analysis results.

Brief Description of Drawings

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

Mode for Carrying Out the Invention

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

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

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

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

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

[0023] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0024] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0025] [First Embodiment]

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

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

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

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

[0030] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0031] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0032] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

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

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

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

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

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

[0038] This invention relates to a system for streamlining complaint handling in customer service centers. The system aims to analyze the content of complaint calls and complaint / inquiry emails, and automatically generate and provide appropriate apologies and replies. The main components of this system include speech recognition, natural language processing (NLP), and automated response generation functions.

[0039] Telephone-based complaint handling system

[0040] Receiving and recording complaint calls

[0041] When a customer service operator receives a complaint call, the operator's PC or telephone system records the audio. The recorded audio data is transmitted to a server in real time.

[0042] Speech recognition and transcription

[0043] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This allows the content of the claim to be obtained as text.

[0044] Text analysis

[0045] The server analyzes the transcribed text using a natural language processing (NLP) model to understand the complaint. This analysis extracts the specific points of what the customer is unhappy about.

[0046] Generating an apology response

[0047] Based on the understood complaint, the server automatically generates an appropriate apology response. For example, in response to a complaint that "delivery is delayed," a response such as "We are very sorry for the delay, sir / madam. We are currently checking the delivery status, so please wait a moment." will be generated.

[0048] Providing a response

[0049] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can use this response as a reference to proceed with handling the complaint.

[0050] Complaint and inquiry email handling system

[0051] Receiving and analyzing emails

[0052] When a customer service staff member (a user) receives a complaint or inquiry email, the terminal sends the email body to the server. The server uses an NLP model to analyze the email's content. Here, it understands what specific problem or inquiry the email represents.

[0053] Generating a reply

[0054] Based on the analysis results, the server automatically generates an appropriate response. For example, in response to a complaint that "the product was damaged," a response such as "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product." will be generated.

[0055] Providing a reply

[0056] The generated reply is sent to the device and displayed in the staff member's email client. The staff member can review the reply, edit it as needed, and then send back the email.

[0057] Specific example

[0058] Examples of complaint calls

[0059] User: The operator answers the phone.

[0060] Terminal: Records the call content and sends the audio data to the server.

[0061] Server: Performs transcription using a speech recognition API.

[0062] Server: Analyzes text using an NLP model to understand the content of the claim.

[0063] Server: Generates an appropriate apology response and sends it to the terminal.

[0064] User: Use the generated response as a reference to handle the complaint.

[0065] Example of a complaint email

[0066] User: Staff receive the email.

[0067] Terminal: Sends the email body to the server.

[0068] Server: Analyzes the content using an NLP model to understand the claim.

[0069] Server: Generates an appropriate reply and sends it to the terminal.

[0070] User: Review the generated reply before sending your email.

[0071] In this way, this system improves the efficiency of customer support operations and enhances customer satisfaction. Furthermore, it reduces variations in the quality of service provided by individual staff members, enabling the delivery of consistently high-quality support.

[0072] The following describes the processing flow.

[0073] Handling complaint calls

[0074] Step 1:

[0075] A user (a customer service operator) receives a complaint call.

[0076] An operator receives a call from a customer.

[0077] Step 2:

[0078] The terminal (operator's PC or telephone system) records the call.

[0079] The call is initiated, and the audio data is recorded in real time.

[0080] Step 3:

[0081] The device transmits the recorded audio data to the server in real time.

[0082] The recorded audio is compressed and sent to the server.

[0083] Step 4:

[0084] The server sends the received audio data to a speech recognition API for transcription.

[0085] The speech data is converted into text data using a speech recognition API.

[0086] Step 5:

[0087] The server sends the transcribed text to a natural language processing (NLP) model for analysis.

[0088] We use an NLP model to extract the types of complaints and key keywords.

[0089] Step 6:

[0090] The server generates an appropriate apology response based on the analysis results.

[0091] For each claim, a response is generated using a pre-prepared template or an AI model.

[0092] Step 7:

[0093] The server generates an apology response, sends it to the terminal, and displays it to the user.

[0094] An appropriate apology response is displayed on the operator's monitor.

[0095] Step 8:

[0096] We will handle the complaint based on the apology response displayed to the user.

[0097] The operator will then proceed with handling the actual complaint based on the displayed response.

[0098] Handling complaints and inquiry emails

[0099] Step 1:

[0100] The user (customer center staff) receives complaint and inquiry emails.

[0101] Check your email client for new emails.

[0102] Step 2:

[0103] The device sends the email content to the server.

[0104] The email body and subject are forwarded to the server via the API.

[0105] Step 3:

[0106] The server analyzes the email content using a natural language processing (NLP) model.

[0107] The content of emails is analyzed to extract specific details of complaints and inquiries.

[0108] Step 4:

[0109] The server automatically generates an appropriate response based on the analysis results.

[0110] The response content is generated from an AI model or template.

[0111] Step 5:

[0112] The server generates a reply and sends it to the terminal for the user to see.

[0113] The reply will appear in the staff member's email client.

[0114] Step 6:

[0115] The user reviews the displayed reply and edits it as needed.

[0116] Staff will review the reply and make any necessary corrections.

[0117] Step 7:

[0118] The user sends an email with the revised reply.

[0119] Send the confirmed reply to the customer.

[0120] Through these specific processing steps, the present invention can significantly improve the efficiency of customer support and reduce the burden on staff.

[0121] (Example 1)

[0122] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0123] In customer service settings, a tremendous amount of time and effort is spent handling complaints and inquiries. This can lead to inconsistencies in service quality among staff members, potentially lowering customer satisfaction. Furthermore, delays in situations where a quick response is required can further increase customer dissatisfaction. Therefore, there is a need to handle complaints and inquiries efficiently and effectively, and to achieve consistent, high-quality service.

[0124] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0125] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voices, and means for transcribing the recorded voice data. This allows the content of the complaints to be efficiently converted into text data, making it possible to analyze them.

[0126] The server also includes means for analyzing the transcribed text and understanding the content of the claim, means for generating an appropriate apology response based on the content of the claim, and means for providing the generated apology response. This enables the rapid generation of appropriate responses to claims and reduces variability in response quality.

[0127] Furthermore, the server includes means for transmitting recorded audio data in real time, means for converting audio to text using a speech recognition API, means for analyzing text using a natural language processing model, means for automatically generating apology responses using a generative AI model, means for analyzing email content, and means for generating reply content. This enables consistent, high-quality, and efficient complaint handling in both voice calls and email correspondence.

[0128] A "customer" is someone who receives a service and makes complaints or inquiries to the service provider.

[0129] "Voice complaints" refer to expressions of dissatisfaction or complaints made by customers using voice communication methods such as telephone calls.

[0130] "Means of recording" refers to devices and software used to record audio data in digital format.

[0131] "Methods for transcription" refer to technologies and devices used to convert audio data into text data.

[0132] "Means of analyzing text" refers to technologies and devices that use natural language processing techniques to understand text data and analyze its content.

[0133] "Means for generating apology responses" refers to technologies or devices that automatically create appropriate apology statements based on the content of a complaint.

[0134] "Means of providing generated apology responses" refers to devices or software that display and make available the generated apology messages to operators and staff.

[0135] "Means of sending to a server in real time" refers to technologies and devices for instantly transferring recorded audio data to a server.

[0136] A "speech recognition API" refers to an application programming interface (API) provided as an external service for converting speech data into text data.

[0137] A "natural language processing model" refers to a machine learning model designed to understand and analyze human language.

[0138] A "generative AI model" refers to an artificial intelligence (AI) model that generates text based on given input data.

[0139] "Complaint and inquiry emails" refer to complaints or questions that customers send via email.

[0140] "Means of analyzing email content" refers to technologies and devices that analyze the body of a received email and understand its content.

[0141] "Means for generating reply content" refers to technologies or devices that automatically create appropriate reply messages based on the content of an email.

[0142] This invention is a system for streamlining complaint and inquiry handling in customer service. This system enables customers to receive prompt, consistent, and high-quality responses to complaints and inquiries. Specific embodiments for implementing this invention are described below.

[0143] Receiving and recording complaint calls

[0144] When a user makes a complaint call, the operator's PC or the telephone system records the audio. The recorded audio data is transmitted to a server in real time. The recording function is implemented using digital recording devices or software integrated into the telephone system.

[0145] Speech recognition and transcription

[0146] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This speech recognition process uses an external speech recognition service, such as Google Cloud Speech-to-Text API. Based on the speech recognition results returned by the API, the call content is obtained as text data.

[0147] Text analysis

[0148] The server feeds the acquired text data into a natural language processing (NLP) model to analyze the complaint content. Advanced NLP models such as GPT-4 (registered trademark) are used for this process. The NLP model analysis clearly extracts customer dissatisfaction and requests. For example, specific complaints such as "delivery is delayed" are identified.

[0149] Generating an apology response

[0150] The server generates an appropriate apology response based on the analysis results. A generative AI model is used to create the best possible apology based on the input text data. For example, a response such as, "We sincerely apologize for the delay, customer. We are currently checking the delivery status, so please wait a moment," might be generated.

[0151] Providing a response

[0152] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can then use the displayed apology as a reference to proceed with customer service.

[0153] Receiving and analyzing complaint and inquiry emails

[0154] When a user receives a complaint or inquiry email, the terminal sends the email body to the server. The server analyzes the email content using an NLP model. Here too, natural language processing models such as GPT-4 are used. The specific problems and inquiries in the email are clarified.

[0155] Generating a reply

[0156] The server generates an appropriate reply based on the analysis results. A generation AI model is used to generate an email reply such as, "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product."

[0157] Providing a reply

[0158] The generated reply is sent to the device and displayed in the staff member's email client. The staff member reviews the displayed reply, edits it as needed, and then replies to the email.

[0159] Specific example

[0160] Examples of complaint calls

[0161] User: The operator calls and says, "Your delivery is delayed."

[0162] Terminal: Records call content and sends it to the server in real time.

[0163] Server: Uses a speech recognition API to convert speech data into text.

[0164] Server: Uses an NLP model to analyze text data and understand the content of the claims.

[0165] Server: Generates an appropriate apology response and sends it to the terminal.

[0166] User: Use the generated apology response as a reference to proceed with handling the complaint.

[0167] Example of a complaint email

[0168] User: Staff receive an email stating that "the product was damaged."

[0169] Terminal: Sends the email body to the server.

[0170] Server: Uses an NLP model to analyze email content and understand the nature of the complaint.

[0171] Server: Generates an appropriate reply and sends it to the terminal.

[0172] User: Review the generated reply and send a reply email.

[0173] Examples of specific prompt messages include, "Please generate an apology letter to handle a delivery delay complaint," or "Please create an email regarding the replacement of a damaged product."

[0174] In this way, the system improves the efficiency of customer support operations and enhances customer satisfaction. Furthermore, it reduces variations in the quality of service provided by individual staff members, enabling the delivery of consistently high-quality support.

[0175] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0176] Step 1: Receiving and recording complaint calls

[0177] A user receives a complaint call. Here, the user is a customer service operator. The terminal (the operator's PC or telephone system) records the call. Specifically, when the operator answers the call, the telephone system starts recording simultaneously with the start of the call. The recorded audio data is transmitted to the server in real time. The input is the customer's voice call, and the output is the recorded audio data.

[0178] Step 2: Speech recognition and transcription

[0179] The server sends the received audio data to a speech recognition API. For example, the Google Cloud Speech-to-Text API is used here. The server receives the API's processing result and converts the audio into text data. The input is audio data, and the output is text data. Specifically, the server sends the audio data to the speech recognition API and stores the resulting string data.

[0180] Step 3: Text Analysis

[0181] The server inputs the acquired text data into a natural language processing (NLP) model, which then analyzes the text. For example, the GPT-4 NLP model is used. The server then analyzes the results returned by the NLP model to understand the specific content of the claim. The input is the transcribed text data, and the output is the analysis result (the key points of the claim). Specifically, key points such as "delivery is delayed" and "order number 12345" are extracted.

[0182] Step 4: Generating an apology response

[0183] The server generates an appropriate apology response based on the analysis results from the NLP model. A generative AI model is used for this purpose. For example, it might generate a response such as, "We sincerely apologize for the delay. We are currently checking the delivery status, so please wait a moment." The input is the analysis results, and the output is the generated apology response. Specifically, a prompt for generating the apology response is input to the generative AI model, and the resulting apology message is retrieved.

[0184] Step 5: Provide a response

[0185] The server generates an apology response and sends it to the terminal. The terminal displays the received apology response on the operator's monitor. This allows the operator to refer to the generated apology and proceed with customer service. The input is the generated apology response, and the output is the text displayed on the operator's monitor. Specifically, the terminal executes the process of displaying the apology response received from the server on its screen.

[0186] Step 6: Receiving and analyzing complaint and inquiry emails

[0187] A user receives a complaint or inquiry email. Here, the user is a staff member at a customer service center. The terminal sends the received email body to the server. The server analyzes the email content using an NLP model. Specifically, it uses GPT-4 or similar to input the email text and analyzes its content. The input is the email body, and the output is the analysis result (the main points of the complaint or inquiry).

[0188] Step 7: Generating the reply content

[0189] The server generates an appropriate response based on the analysis results. A generative AI model is used for this generation. For example, a response such as "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product." might be generated. The input is the analysis results, and the output is the generated response. Specifically, the server prompts the generative AI model based on the analysis results and retrieves the generated response.

[0190] Step 8: Provide a reply

[0191] The server generates a reply and sends it to the terminal. The terminal displays this reply in the staff member's email client. This allows the staff member to review the generated reply, edit it as needed, and send the email. The input is the generated reply, and the output is the text displayed in the staff member's email client. Specifically, the terminal executes the process of displaying the reply received from the server in the email client.

[0192] (Application Example 1)

[0193] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0194] Traditional customer support systems face the challenge of responding quickly and appropriately to customer complaints and inquiries. Similarly, timely and accurate responses are required for interactive complaints and inquiries from viewers. This can easily lead to decreased customer satisfaction and inconsistent support quality.

[0195] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0196] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text and understanding the content of the complaint, means for generating an appropriate apology response based on the content of the complaint, means for providing the generated apology response, and means for automatically analyzing complaints and inquiries from viewers and automatically generating appropriate responses, including voice recognition means, natural language processing means, and response generation means. This enables a rapid, consistent, and high-quality response to complaints and inquiries from customers and viewers.

[0197] "Customer" refers to anyone who purchases or uses a product or service.

[0198] "Viewer" refers to anyone who watches or listens to videos, music, or other content provided through content distribution services.

[0199] A "complaint" refers to a customer or viewer reporting dissatisfaction or problems with a product or service.

[0200] "Inquiry" refers to a question or request from a customer or viewer seeking information or support.

[0201] "Speech recognition means" refers to systems and devices that analyze speech as digital data and convert it into text data.

[0202] "Natural language processing methods" refer to technologies and algorithms used to analyze and understand human language using computers.

[0203] "Response generation means" refers to algorithms or systems that automatically generate appropriate responses or answers based on analyzed claims or inquiries.

[0204] "Recording means" refers to devices or systems for saving received audio in digital format.

[0205] "Transcription methods" refer to systems and software used to convert recorded audio data into text data.

[0206] An "apology response" refers to a statement expressing regret for a customer's or viewer's complaint and outlining actions taken to resolve the issue.

[0207] "Automated analysis" refers to the process of using computers to analyze data and understand its content and intent.

[0208] This invention is a system for providing a quick and accurate response to complaints and inquiries from viewers and customers. This system consists of the following main components:

[0209] 1. Speech Recognition Method: Receive and digitally store voice complaints from customers. This involves using APIs or software for speech recognition (e.g., Google Speech Recognition API).

[0210] 2. Natural Language Processing Methods: Analyze the content of received audio data and complaint / inquiry emails to understand the intentions of customers and viewers. Specifically, utilize natural language processing models (e.g., Hugging Face's transformers library).

[0211] 3. Response generation means: Based on the analyzed complaint content, appropriate apology responses and reply content are automatically generated. This enables a quick and appropriate response.

[0212] 4. Recording method: The voice of the complaint is recorded and sent to the server. A recording device or telephone system is responsible for this role.

[0213] 5. Transcription method: Convert the recorded audio data into text data. This is done using a speech recognition API.

[0214] Server Role

[0215] The server plays a central role in processing the received audio and text data. First, it uses speech recognition to convert the audio into text. Then, it uses natural language processing to analyze the text and understand what the customer or audience wants. Finally, it uses response generation to automatically generate appropriate replies or apologies and send them to the terminal.

[0216] Terminal role

[0217] The terminal receives data sent from the server and provides it to the user. It also plays a role in recording the voice of the complaint and sending it to the server. Furthermore, it displays the generated response or reply content, which the user can use as a reference when handling the complaint.

[0218] User roles

[0219] Users are operators or staff who handle complaints from viewers and customers. They are required to use the generated responses and reply content as a reference to take appropriate action.

[0220] Specific example

[0221] For example, if a viewer sends a complaint such as "the video keeps cutting out" via voice message within the app, the following response is automatically generated:

[0222] Viewer: "The video keeps cutting out."

[0223] App: "We apologize for the issue you are experiencing. We are currently working on a solution."

[0224] In this way, this system can respond quickly and appropriately to viewer and customer complaints and inquiries, contributing to improved customer satisfaction.

[0225] Example of a prompt

[0226] "The video keeps cutting out."

[0227] "No sound."

[0228] "My subscription is not being renewed."

[0229] This allows for accurate support to be provided for specific complaints and inquiries.

[0230] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0231] Step 1:

[0232] The terminal receives audio complaints from viewers or customers. The received audio complaints are recorded digitally through the terminal's built-in microphone or an externally connected microphone.

[0233] Input: Audience or customer voice complaint

[0234] Output: Digital audio data

[0235] Step 2:

[0236] The terminal sends the recorded audio data to the server. The transmitted audio data is converted to an appropriate format on the server side and made ready for processing by speech recognition software.

[0237] Input: Digital audio data

[0238] Output: Audio data sent to the server

[0239] Step 3:

[0240] The server converts the audio data into text data using speech recognition technology. This process utilizes tools such as the Google Speech Recognition API to convert speech to text.

[0241] Input: Audio data sent to the server

[0242] Output: Transcribed text data

[0243] Step 4:

[0244] The server analyzes the transcribed text using natural language processing techniques. It uses libraries such as Hugging Face's transformers to understand the claims in the text and analyze the intent of the customer or audience.

[0245] Input: Transcribed text data

[0246] Output: Analysis results of the claim content

[0247] Step 5:

[0248] The server generates appropriate apology responses and reply content based on the analysis results. In this process, a generative AI model is used to automatically generate response texts by providing appropriate prompt sentences as input.

[0249] Input: Analysis results of the claim

[0250] Output: Automated apology response or reply content

[0251] Step 6:

[0252] The server sends the generated apology or reply to the terminal. The terminal displays this reply on the operator's or staff's screen, allowing them to use it as a reference when taking action.

[0253] Input: Automated apology or reply content

[0254] Output: Response sent to the terminal

[0255] Step 7:

[0256] Users review the generated apology or reply displayed on their device and use it to respond to viewers or customers. They can also edit the reply content at their discretion if necessary.

[0257] Input: The generated apology or reply displayed on the device.

[0258] Output: Edited response content and response to the audience or customer

[0259] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0260] This invention relates to a system for improving the efficiency of complaint handling and enhancing customer satisfaction in customer centers. This system not only analyzes the content of complaint calls and complaint / inquiry emails, but also incorporates an emotion engine to recognize the user's emotions, enabling it to automatically generate more appropriate apology responses and reply messages.

[0261] A telephone complaint handling system incorporating an emotion engine.

[0262] Receiving and recording complaint calls

[0263] When a customer service operator receives a complaint call, the operator's PC or telephone system records the audio. The recorded audio data is transmitted to a server in real time.

[0264] Speech recognition and transcription

[0265] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This allows the content of the claim to be obtained as text.

[0266] Text and sentiment analysis

[0267] The server analyzes the transcribed text using a natural language processing (NLP) model to understand the complaint. During this process, the server also uses an emotion engine to recognize the user's emotions, for example, whether the customer is angry or sad.

[0268] Generating an apology response

[0269] Based on the understood complaint and emotions, the server automatically generates an appropriate apology response. For example, if a complaint about "delayed delivery" and the emotion of "anger" are recognized, an emphatic apology such as "We sincerely apologize. We will immediately investigate the issue and take appropriate action" will be generated for the customer.

[0270] Providing a response

[0271] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can use this response as a reference to proceed with handling the complaint.

[0272] Inquiry email response system with integrated emotion engine

[0273] Receiving and analyzing emails

[0274] When a customer service staff member (a user) receives a complaint or inquiry email, the terminal sends the email body to the server. The server uses an NLP model to analyze the email's content. Here, it understands what specific problem or inquiry the email represents.

[0275] Analysis of email content and emotions

[0276] The server then uses an emotion engine to recognize the user's emotions from the email content. For example, it extracts emotions such as "dissatisfaction" or "disappointment" from the email body.

[0277] Generating a reply

[0278] Based on the analysis results and emotions, the server automatically generates appropriate reply content. For example, when a claim of "the product was broken" and an emotion of "disappointment" are recognized, a response such as "We apologize for the inconvenience. We will arrange a replacement immediately." is generated.

[0279] Provision of Reply

[0280] The generated reply content is sent to the terminal and displayed on the email client of the staff who is the user. The staff can check the reply content, edit it if necessary, and then reply to the email.

[0281] Specific Example

[0282] Example of Claim Call

[0283] User: The operator answers the phone.

[0284] Terminal: Records the call content and sends the voice data to the server.

[0285] Server: Performs speech recognition using the API to convert the voice to text.

[0286] Server: Performs text and emotion analysis using the NLP model to understand the claim content and emotion.

[0287] Server: Generates an appropriate apology reply and sends it to the terminal.

[0288] User: Handles the claim based on the generated reply.

[0289] Example of Claim Email

[0290] User: The staff receives the email.

[0291] Terminal: Sends the email body to the server.

[0292] Server: Analyzes content and emotion using an NLP model to understand the content and emotions of the complaint.

[0293] Server: Generates an appropriate reply and sends it to the terminal.

[0294] User: Review the generated reply before sending your email.

[0295] These specific processing steps enable this system to contribute to improving the efficiency of customer support and customer satisfaction. The incorporation of an emotion engine allows for more appropriate responses that take customer emotions into consideration.

[0296] The following describes the processing flow.

[0297] A telephone complaint handling system incorporating an emotion engine.

[0298] Step 1:

[0299] A user (a customer service operator) receives a complaint call.

[0300] The operator answers the phone using the usual procedure.

[0301] Step 2:

[0302] The terminal (operator's PC or telephone system) records the call.

[0303] When a call is initiated, the audio is automatically recorded.

[0304] Step 3:

[0305] The device transmits the recorded audio data to the server in real time.

[0306] The recorded audio data is divided into packets and sent to the server.

[0307] Step 4:

[0308] The server sends the received voice data to the speech recognition API for speech-to-text conversion.

[0309] The voice data is converted into text format through the speech recognition API.

[0310] Step 5:

[0311] The server sends the speech-to-text converted text to the natural language processing (NLP) model for content analysis.

[0312] The NLP model extracts claim content and important keywords.

[0313] Step 6:

[0314] The server uses the sentiment engine to recognize the user's sentiment from the speech-to-text converted text.

[0315] The sentiment engine analyzes the sentiment expressions in the text to identify the customer's sentiment (e.g., anger, sadness, confusion, etc.).

[0316] Step 7:

[0317] Based on the analysis results and the recognized sentiment, the server generates an appropriate apology response.

[0318] The response is created using a template or AI model according to the claim content and the customer's sentiment. For example, when the customer is angry about a late delivery, a strong apology like "We sincerely apologize. We will check immediately and take action." is created.

[0319] Step 8:

[0320] The server sends the generated apology response to the terminal for display to the user.

[0321] The generated apology response is displayed on the operator's monitor.

[0322] Step 9:

[0323] We will handle the complaint based on the apology response displayed to the user.

[0324] The operator uses the displayed response to proceed with the actual complaint handling. Responses can also be customized as needed.

[0325] Inquiry email response system with integrated emotion engine

[0326] Step 1:

[0327] The user (customer center staff) receives complaint and inquiry emails.

[0328] Check for new emails through your email client.

[0329] Step 2:

[0330] The device sends the email content to the server.

[0331] The email body and subject are transferred to the server via the API.

[0332] Step 3:

[0333] The server analyzes the email content using a natural language processing (NLP) model.

[0334] The NLP model analyzes the content and topic of the email.

[0335] Step 4:

[0336] The server uses an emotion engine to recognize the user's emotions from the email content.

[0337] The emotion engine analyzes emotional expressions within emails to identify the customer's emotions (e.g., dissatisfaction, disappointment, confusion, etc.).

[0338] Step 5:

[0339] The server automatically generates an appropriate response based on the analysis results and recognized emotions.

[0340] The system generates response messages using templates or AI models tailored to the nature of the complaint and the customer's feelings. For example, if a customer says, "I'm disappointed because the product is broken," a response such as, "We are very sorry for the inconvenience. We will arrange for a replacement immediately," might be generated.

[0341] Step 6:

[0342] The server generates a reply and sends it to the terminal for the user to see.

[0343] The generated reply will be displayed in the staff member's email client.

[0344] Step 7:

[0345] The user reviews the displayed reply and edits it as needed.

[0346] Staff will review the reply and make any necessary corrections.

[0347] Step 8:

[0348] The user sends an email with the revised reply.

[0349] Send the finalized response to the customer to complete the process.

[0350] As described above, by incorporating an emotion engine, the quality of complaint and inquiry handling can be improved, and responses that take customer emotions into consideration can be made. This system can promote the efficiency of customer support and improve customer satisfaction.

[0351] (Example 2)

[0352] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0353] Handling customer complaints in modern customer centers is time-consuming and labor-intensive. Furthermore, the quality of service provided by operators varies, making it difficult to consistently improve customer satisfaction. In particular, it is crucial to accurately understand customer emotions and respond with appropriate and courteous care. A method is needed to address these challenges efficiently and effectively.

[0354] The identification processing performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text to understand the content of the complaint, means for analyzing the transcribed text using a natural language processing model and an emotion analysis engine to understand the customer's emotions, means for generating an appropriate apology response based on the content of the complaint and the customer's emotions, and means for providing the generated apology response. This makes it possible to accurately grasp the content of the customer's complaint and emotions, and to quickly generate and provide an appropriate response accordingly. Furthermore, it is possible to reduce the burden on operators and ensure consistency in response quality, thereby improving customer satisfaction.

[0355] "Means for receiving customer voice complaints" refers to a function for electronically receiving complaints made by customers via telephone.

[0356] "Means for recording received audio" refers to a function for saving received audio data to a recording medium.

[0357] "Methods for transcribing recorded audio data" refers to functions that utilize speech recognition technology to convert recorded audio data into text data.

[0358] "Means for analyzing transcribed text and understanding the content of a claim" refers to a function that uses natural language processing technology to analyze text data obtained through transcription and recognize the content contained therein.

[0359] "A means of understanding customer emotions by analyzing transcribed text using natural language processing models and sentiment analysis engines" refers to a function that uses specific algorithms to analyze the linguistic structure and emotional aspects within the text and grasp the customer's emotional state.

[0360] "A means of generating an appropriate apology response based on the content of the complaint and the customer's feelings" refers to a function that automatically generates the optimal response text corresponding to the specific content of the recognized complaint and the customer's feelings.

[0361] "Means of providing a generated apology response" refers to a function for displaying the generated response to the operator or sending it directly to the customer.

[0362] "Means for receiving complaints and inquiries via email" refers to a function for receiving complaints and inquiries sent via email.

[0363] "A means of analyzing the content of received emails to understand complaints, inquiries, and customer sentiment" refers to a function that uses natural language processing and sentiment analysis technologies to analyze the text of received emails and grasp their content and customer sentiment.

[0364] "A means of generating appropriate response content based on complaints, inquiries, and customer sentiment" refers to a function that automatically generates the optimal response based on recognized email content and customer sentiment.

[0365] "A means of automatically generating response content and apology responses based on the content of the complaint and the customer's feelings using a generative AI model" refers to a function that uses an artificial intelligence model to generate appropriate responses to complaints and inquiries from specific prompt sentences.

[0366] "A means of instructing a generation AI model to provide an appropriate response using a prompt statement" refers to a function that provides a specific input (prompt) to the AI ​​model and generates an appropriate output based on that input.

[0367] This invention relates to a system for improving the efficiency of complaint handling and enhancing customer satisfaction in customer centers. Specifically, it analyzes both voice and email complaints and automatically generates appropriate apologies and replies.

[0368] System Configuration

[0369] This system consists of the following main elements:

[0370] 1. Users: Customer center operators and staff will be the primary users of the system.

[0371] 2. Terminals: This includes PCs, telephone systems, and email clients used by operators and staff.

[0372] 3. Server: This server runs the speech recognition API, natural language processing (NLP) models, sentiment analysis engine, and generative AI models.

[0373] Hardware and software to be used

[0374] Speech Recognition API: Use the Google Cloud Speech-to-Text API to convert speech data into text.

[0375] Natural Language Processing Models: Text data is analyzed using NLP models such as BERT and GPT-3(registered trademark).

[0376] Sentiment analysis engine: Uses IBM Watson® Tone Analyzer and other tools to analyze emotions from text.

[0377] Generative AI model: Uses GPT-3 and other models to generate appropriate apology responses and reply content.

[0378] Processing flow

[0379] Regarding the customer complaint phone handling system:

[0380] When a user receives a call, the device records the audio and sends it to the server in real time.

[0381] The server sends the audio data to the Google Cloud Speech-to-Text API, where it is converted into text data.

[0382] The server sends text data to an NLP model to analyze the claim details.

[0383] The server then uses an emotion analysis engine to analyze the customer's emotions.

[0384] Based on the nature of the complaint and the emotional state, the server uses a generative AI model to generate an appropriate apology response.

[0385] The generated apology response is sent to the device and displayed on the user's monitor.

[0386] Regarding the complaint email handling system:

[0387] When a user receives an email, the email body is sent from the device to the server.

[0388] The server uses an NLP model to analyze the email content and understand the complaint and emotions behind it.

[0389] Based on the analysis results, the server uses a generative AI model to generate appropriate responses.

[0390] The generated reply is sent to the device and displayed in the user's email client.

[0391] Examples of specific cases and prompt statements

[0392] For example, if an operator receives a complaint about a delivery delay:

[0393] User: The operator answers the phone.

[0394] Terminal: Records the call content and sends the audio data to the server.

[0395] Server: Performs transcription using a speech recognition API (e.g., Google Cloud Speech-to-Text).

[0396] Server: Uses NLP models to analyze text and sentiment to understand the content and emotions of complaints (e.g., BERT and IBM Watson Tone Analyzer).

[0397] Server: Generates an appropriate apology response and sends it to the terminal.

[0398] User: Use the generated response as a reference to handle the complaint.

[0399] Examples of prompt statements:

[0400] "We've received a complaint about a delivery delay. The customer is angry. Please generate an appropriate apology letter."

[0401] "We received an email inquiry about a damaged product. The customer is disappointed. Please generate an appropriate reply."

[0402] This system streamlines customer service complaint handling and enables appropriate responses that take customer emotions into consideration. This, in turn, leads to improved customer satisfaction.

[0403] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0404] Customer complaint phone handling system

[0405] Processing steps

[0406] Step 1:

[0407] A customer service operator, acting as the user, receives a complaint call. When the operator presses the "Start Call" button, the terminal begins recording the conversation. The recorded audio data is transmitted to the server in real time. The input is the call audio, and the output is the audio data stored on the server.

[0408] Step 2:

[0409] The server sends the received audio data to the Google Cloud Speech-to-Text API, where it is converted into text data. In this step, audio data is input and text data is generated as output. Specifically, the server sends an API request and receives a response.

[0410] Step 3:

[0411] The server sends the text data converted from the speech to a natural language processing (NLP) model such as BERT or GPT-3 to analyze the claim content. The input for this step is the transcribed text, and the analyzed claim content is output. Specifically, the server sends an API request to the NLP model and receives the analysis results.

[0412] Step 4:

[0413] The server then sends the analyzed text to a sentiment analysis engine, such as IBM Watson Tone Analyzer, to analyze the customer's emotions. In this step, the text containing the complaint is input, and the sentiment analysis results are obtained as output. The specific operation involves sending requests to the sentiment analysis API and receiving responses.

[0414] Step 5:

[0415] The server sends prompt messages to a generative AI model (e.g., GPT-3) based on the understood complaint content and sentiment state, generating an appropriate apology response. The input data is the complaint content and sentiment analysis results, and the output is the generated apology response. Specifically, the server defines appropriate prompt messages for the AI ​​model and sends an API request.

[0416] Step 6:

[0417] The generated apology response is sent to the terminal and displayed on the operator's monitor. In this step, the generated response text is input, and the displayed monitor is output. Specifically, a UI update request is made to display the response text.

[0418] Complaint email handling system

[0419] Processing steps

[0420] Step 1:

[0421] When a customer service staff member (a user) receives a complaint or inquiry email, the email body is sent from the terminal to the server. The input is the received email, and the output is the email data stored on the server. Specifically, the email client sends the email body via the API.

[0422] Step 2:

[0423] The server uses an NLP model to analyze the email content. The email body is input, and the complaint and inquiry details are output. Specifically, an API request is sent to analyze the text data, and the analysis results are received.

[0424] Step 3:

[0425] The server uses an emotion analysis engine to understand customer emotions during the process of analyzing email content and inquiry details. The input data is the email body, and the output data is the emotion analysis result. Specifically, it sends a request to the emotion analysis API and receives a response.

[0426] Step 4:

[0427] The server uses a generative AI model to generate an appropriate response based on the analyzed content and emotions. In this step, the complaint and customer emotions are used as input, and the output is the generated response. Specifically, a prompt is sent to the generative AI model to generate an appropriate response.

[0428] Step 5:

[0429] The generated reply is sent to the terminal and displayed in the staff member's email client. The input data is the generated reply text, and the output is the displayed email client. Specifically, a UI update request is executed to display the reply text.

[0430] Through the processing steps described above, this system contributes to improving the efficiency of customer support and enhancing customer satisfaction. By understanding customer emotions and responding appropriately and quickly, customer satisfaction can be increased.

[0431] (Application Example 2)

[0432] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0433] In modern customer service centers, responding quickly and appropriately to customer complaints and inquiries is a crucial challenge. However, especially when complaints are emotional, it is often difficult for staff to accurately understand those emotions and respond appropriately. Therefore, there is a need for systems that can improve customer satisfaction and increase the efficiency of complaint handling.

[0434] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0435] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text to understand the content of the complaint and the customer's feelings, means for generating an appropriate apology response based on the content of the complaint and the customer's feelings, and means for providing the generated apology response. This enables appropriate responses that take into account the customer's feelings.

[0436] "Means of understanding customer emotions" refers to the process of analyzing and recognizing the emotions behind customer words, such as anger, disappointment, or satisfaction, using natural language processing technology.

[0437] A "means for generating apology responses" is a system that automatically generates appropriate and empathetic apologies and responses based on the analyzed complaint content and customer emotions.

[0438] "Methods for generating response content" refers to a process that automatically constructs appropriate responses to customers, taking into account complaints, inquiries, and related emotions.

[0439] A "natural language processing model" is a machine learning model that analyzes text data and uses that information to understand meaning and intent.

[0440] A "generative AI model" is an artificial intelligence model that automatically generates text or answers based on given prompts or input data.

[0441] A "prompt" is an instruction or input sentence used by a generative AI model to generate a response, and it serves as a guideline for providing appropriate output based on its content.

[0442] This invention is a system that supports customer service in physical stores. Its purpose is to enable store staff to respond quickly and accurately to customer complaints and inquiries. A specific example of the system is described below.

[0443] Hardware and software to be used

[0444] The system uses the following hardware and software.

[0445] Hardware: Smartphones, smart glasses, head-mounted displays

[0446] Software: Python, TENSORFLOW®, Transformers package, SpeechRecognition package

[0447] System Configuration

[0448] 1. Means of receiving voice complaints from customers:

[0449] Staff members wear smartphones, smart glasses, or head-mounted displays to receive customer complaints and inquiries.

[0450] 2. Means for recording received audio:

[0451] The received audio data is recorded on the staff member's terminal and sent to the server in real time.

[0452] 3. Methods for transcribing recorded audio data:

[0453] The server converts the received audio data into text data using the SpeechRecognition package. This allows the content of complaints and inquiries to be obtained as text.

[0454] 4. Means for analyzing transcribed text to understand the content of the complaint and the customer's feelings:

[0455] The server analyzes the converted text data using the Transformers package and a natural language processing model to understand the content of the complaint, while simultaneously recognizing the customer's emotions using an emotion engine.

[0456] 5. Means for generating an appropriate apology response based on the nature of the complaint and the customer's feelings:

[0457] Based on the analyzed complaint details and customer emotions, the server uses a generative AI model to generate an appropriate apology response. For example, if the complaint is "the product was broken" and the customer's emotion is recognized as "anger," an apology response such as "We sincerely apologize. We will arrange for a replacement immediately" will be generated.

[0458] 6. Means of providing the generated apology response:

[0459] The generated apology response is sent from the server to the staff member's terminal and displayed on their smartphone, smart glasses, or head-mounted display. The staff member uses this response as a reference when dealing with the customer.

[0460] Example of a prompt

[0461] Examples of prompt statements that can actually be used include the following:

[0462] "Analyze the customer's emotions from the following text: 'The product was broken. Please do something about it.'"

[0463] Specific example

[0464] Examples of customer service:

[0465] When a staff member receives a customer complaint using smart glasses, the audio is recorded by the smart glasses and sent to a server in real time. The server converts the received audio into text and analyzes the emotions. Then, an appropriate apology response is displayed on the staff member's glasses. The staff member uses the displayed apology response to address the customer and resolve the issue quickly.

[0466] Thus, the present invention aims to improve customer satisfaction in physical stores by enabling appropriate and prompt responses that take customer emotions into consideration.

[0467] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0468] Step 1:

[0469] Receive voice complaints from customers.

[0470] Input: A customer makes a verbal complaint to a store staff member.

[0471] Specific operation: The store staff member's smartphone, smart glasses, or head-mounted display receives audio in real time.

[0472] Output: Received audio data.

[0473] Step 2:

[0474] Record the received audio.

[0475] Input: Audio data received in Step 1.

[0476] Specific operation: The device (smartphone, smart glasses, head-mounted display) records audio and sends the recorded data to the server.

[0477] Output: Recorded audio data.

[0478] Step 3:

[0479] Transcribing recorded audio data into text.

[0480] Input: Audio data recorded in Step 2.

[0481] Specific operation: The server uses a speech recognition API to convert the audio data into text data.

[0482] Output: Transcripted text data.

[0483] Step 4:

[0484] The transcribed text is analyzed to understand the content of the complaint and the customer's feelings.

[0485] Input: Text data obtained in Step 3.

[0486] Specific operation: The server uses a natural language processing (NLP) model to analyze the content of the complaint from the text data. At the same time, it uses an emotion engine to recognize the customer's emotions.

[0487] Output: Analyzed complaint details and perceived customer sentiment.

[0488] Step 5:

[0489] Based on the nature of the complaint and the customer's feelings, generate an appropriate apology response.

[0490] Input: Complaint details and customer sentiment analyzed in Step 4.

[0491] Specific operation: The server uses an AI model to generate an appropriate apology response based on the entered complaint content and emotions. For example, if the complaint is "the product was broken" and the emotion "anger" is recognized, an apology response such as "We sincerely apologize. We will arrange for a replacement immediately." will be generated.

[0492] Output: Generated apology response.

[0493] Step 6:

[0494] Provide a generated apology response.

[0495] Input: The apology response generated in Step 5.

[0496] Specific operation: The server sends the generated apology response to the staff member's device (smartphone, smart glasses, or head-mounted display). The staff member can then use the apology response as a reference when dealing with the customer.

[0497] Output: The apology response displayed on the terminal.

[0498] This allows the system to provide quick and appropriate responses in physical stores, taking customer emotions into consideration, thereby improving customer satisfaction.

[0499] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0500] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0501] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0502] [Second Embodiment]

[0503] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0504] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0505] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0506] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0507] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0508] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0509] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0510] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0511] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0512] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0513] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0514] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0515] This invention relates to a system for streamlining complaint handling in customer service centers. The system aims to analyze the content of complaint calls and complaint / inquiry emails, and automatically generate and provide appropriate apologies and replies. The main components of this system include speech recognition, natural language processing (NLP), and automated response generation functions.

[0516] Telephone-based complaint handling system

[0517] Receiving and recording complaint calls

[0518] When a customer service operator receives a complaint call, the operator's PC or telephone system records the audio. The recorded audio data is transmitted to a server in real time.

[0519] Speech recognition and transcription

[0520] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This allows the content of the claim to be obtained as text.

[0521] Text analysis

[0522] The server analyzes the transcribed text using a natural language processing (NLP) model to understand the complaint. This analysis extracts the specific points of what the customer is unhappy about.

[0523] Generating an apology response

[0524] Based on the understood complaint, the server automatically generates an appropriate apology response. For example, in response to a complaint that "delivery is delayed," a response such as "We are very sorry for the delay, sir / madam. We are currently checking the delivery status, so please wait a moment." will be generated.

[0525] Providing a response

[0526] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can use this response as a reference to proceed with handling the complaint.

[0527] Complaint and inquiry email handling system

[0528] Receiving and analyzing emails

[0529] When a customer service staff member (a user) receives a complaint or inquiry email, the terminal sends the email body to the server. The server uses an NLP model to analyze the email's content. Here, it understands what specific problem or inquiry the email represents.

[0530] Generating a reply

[0531] Based on the analysis results, the server automatically generates an appropriate response. For example, in response to a complaint that "the product was damaged," a response such as "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product." will be generated.

[0532] Providing a reply

[0533] The generated reply is sent to the device and displayed in the staff member's email client. The staff member can review the reply, edit it as needed, and then send back the email.

[0534] Specific example

[0535] Examples of complaint calls

[0536] User: The operator answers the phone.

[0537] Terminal: Records the call content and sends the audio data to the server.

[0538] Server: Performs transcription using a speech recognition API.

[0539] Server: Analyzes text using an NLP model to understand the content of the claim.

[0540] Server: Generates an appropriate apology response and sends it to the terminal.

[0541] User: Use the generated response as a reference to handle the complaint.

[0542] Example of a complaint email

[0543] User: Staff receive the email.

[0544] Terminal: Sends the email body to the server.

[0545] Server: Analyzes the content using an NLP model to understand the claim.

[0546] Server: Generates an appropriate reply and sends it to the terminal.

[0547] User: Review the generated reply before sending your email.

[0548] In this way, this system improves the efficiency of customer support operations and enhances customer satisfaction. Furthermore, it reduces variations in the quality of service provided by individual staff members, enabling the delivery of consistently high-quality support.

[0549] The following describes the processing flow.

[0550] Handling complaint calls

[0551] Step 1:

[0552] A user (a customer service operator) receives a complaint call.

[0553] An operator receives a call from a customer.

[0554] Step 2:

[0555] The terminal (operator's PC or telephone system) records the call.

[0556] The call is initiated, and the audio data is recorded in real time.

[0557] Step 3:

[0558] The device transmits the recorded audio data to the server in real time.

[0559] The recorded audio is compressed and sent to the server.

[0560] Step 4:

[0561] The server sends the received audio data to a speech recognition API for transcription.

[0562] The speech data is converted into text data using a speech recognition API.

[0563] Step 5:

[0564] The server sends the transcribed text to a natural language processing (NLP) model for analysis.

[0565] We use an NLP model to extract the types of complaints and key keywords.

[0566] Step 6:

[0567] The server generates an appropriate apology response based on the analysis results.

[0568] For each claim, a response is generated using a pre-prepared template or an AI model.

[0569] Step 7:

[0570] The server generates an apology response, sends it to the terminal, and displays it to the user.

[0571] An appropriate apology response is displayed on the operator's monitor.

[0572] Step 8:

[0573] We will handle the complaint based on the apology response displayed to the user.

[0574] The operator will then proceed with handling the actual complaint based on the displayed response.

[0575] Handling complaints and inquiry emails

[0576] Step 1:

[0577] The user (customer center staff) receives complaint and inquiry emails.

[0578] Check your email client for new emails.

[0579] Step 2:

[0580] The device sends the email content to the server.

[0581] The email body and subject are forwarded to the server via the API.

[0582] Step 3:

[0583] The server analyzes the email content using a natural language processing (NLP) model.

[0584] The content of emails is analyzed to extract specific details of complaints and inquiries.

[0585] Step 4:

[0586] The server automatically generates an appropriate response based on the analysis results.

[0587] The response content is generated from an AI model or template.

[0588] Step 5:

[0589] The server generates a reply and sends it to the terminal for the user to see.

[0590] The reply will appear in the staff member's email client.

[0591] Step 6:

[0592] The user reviews the displayed reply and edits it as needed.

[0593] Staff will review the reply and make any necessary corrections.

[0594] Step 7:

[0595] The user sends an email with the revised reply.

[0596] Send the confirmed reply to the customer.

[0597] Through these specific processing steps, the present invention can significantly improve the efficiency of customer support and reduce the burden on staff.

[0598] (Example 1)

[0599] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0600] In customer service settings, a tremendous amount of time and effort is spent handling complaints and inquiries. This can lead to inconsistencies in service quality among staff members, potentially lowering customer satisfaction. Furthermore, delays in situations where a quick response is required can further increase customer dissatisfaction. Therefore, there is a need to handle complaints and inquiries efficiently and effectively, and to achieve consistent, high-quality service.

[0601] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0602] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voices, and means for transcribing the recorded voice data. This allows the content of the complaints to be efficiently converted into text data, making it possible to analyze them.

[0603] The server also includes means for analyzing the transcribed text and understanding the content of the claim, means for generating an appropriate apology response based on the content of the claim, and means for providing the generated apology response. This enables the rapid generation of appropriate responses to claims and reduces variability in response quality.

[0604] Furthermore, the server includes means for transmitting recorded audio data in real time, means for converting audio to text using a speech recognition API, means for analyzing text using a natural language processing model, means for automatically generating apology responses using a generative AI model, means for analyzing email content, and means for generating reply content. This enables consistent, high-quality, and efficient complaint handling in both voice calls and email correspondence.

[0605] A "customer" is someone who receives a service and makes complaints or inquiries to the service provider.

[0606] "Voice complaints" refer to expressions of dissatisfaction or complaints made by customers using voice communication methods such as telephone calls.

[0607] "Means of recording" refers to devices and software used to record audio data in digital format.

[0608] "Methods for transcription" refer to technologies and devices used to convert audio data into text data.

[0609] "Means of analyzing text" refers to technologies and devices that use natural language processing techniques to understand text data and analyze its content.

[0610] "Means for generating apology responses" refers to technologies or devices that automatically create appropriate apology statements based on the content of a complaint.

[0611] "Means of providing generated apology responses" refers to devices or software that display and make available the generated apology messages to operators and staff.

[0612] "Means of sending to a server in real time" refers to technologies and devices for instantly transferring recorded audio data to a server.

[0613] A "speech recognition API" refers to an application programming interface (API) provided as an external service for converting speech data into text data.

[0614] A "natural language processing model" refers to a machine learning model designed to understand and analyze human language.

[0615] A "generative AI model" refers to an artificial intelligence (AI) model that generates text based on given input data.

[0616] "Complaint and inquiry emails" refer to complaints or questions that customers send via email.

[0617] "Means of analyzing email content" refers to technologies and devices that analyze the body of a received email and understand its content.

[0618] "Means for generating reply content" refers to technologies or devices that automatically create appropriate reply messages based on the content of an email.

[0619] This invention is a system for streamlining complaint and inquiry handling in customer service. This system enables customers to receive prompt, consistent, and high-quality responses to complaints and inquiries. Specific embodiments for implementing this invention are described below.

[0620] Receiving and recording complaint calls

[0621] When a user makes a complaint call, the operator's PC or the telephone system records the audio. The recorded audio data is transmitted to a server in real time. The recording function is implemented using digital recording devices or software integrated into the telephone system.

[0622] Speech recognition and transcription

[0623] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This speech recognition process uses an external speech recognition service, such as the Google Cloud Speech-to-Text API. Based on the speech recognition results returned by the API, the call content is retrieved as text data.

[0624] Text analysis

[0625] The server feeds the acquired text data into a natural language processing (NLP) model to analyze the complaint content. Advanced NLP models such as GPT-4 are used for this process. The NLP model analysis clearly extracts customer dissatisfaction and requests. For example, specific complaints such as "delivery is delayed" are identified.

[0626] Generating an apology response

[0627] The server generates an appropriate apology response based on the analysis results. A generative AI model is used to create the best possible apology based on the input text data. For example, a response such as, "We sincerely apologize for the delay, customer. We are currently checking the delivery status, so please wait a moment," might be generated.

[0628] Providing a response

[0629] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can then use the displayed apology as a reference to proceed with customer service.

[0630] Receiving and analyzing complaint and inquiry emails

[0631] When a user receives a complaint or inquiry email, the terminal sends the email body to the server. The server analyzes the email content using an NLP model. Here too, natural language processing models such as GPT-4 are used. The specific problems and inquiries in the email are clarified.

[0632] Generating a reply

[0633] The server generates an appropriate reply based on the analysis results. A generation AI model is used to generate an email reply such as, "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product."

[0634] Providing a reply

[0635] The generated reply is sent to the device and displayed in the staff member's email client. The staff member reviews the displayed reply, edits it as needed, and then replies to the email.

[0636] Specific example

[0637] Examples of complaint calls

[0638] User: The operator calls and says, "Your delivery is delayed."

[0639] Terminal: Records call content and sends it to the server in real time.

[0640] Server: Uses a speech recognition API to convert speech data into text.

[0641] Server: Uses an NLP model to analyze text data and understand the content of the claims.

[0642] Server: Generates an appropriate apology response and sends it to the terminal.

[0643] User: Use the generated apology response as a reference to proceed with handling the complaint.

[0644] Example of a complaint email

[0645] User: Staff receive an email stating that "the product was damaged."

[0646] Terminal: Sends the email body to the server.

[0647] Server: Uses an NLP model to analyze email content and understand the nature of the complaint.

[0648] Server: Generates an appropriate reply and sends it to the terminal.

[0649] User: Review the generated reply and send a reply email.

[0650] Examples of specific prompt messages include, "Please generate an apology letter to handle a delivery delay complaint," or "Please create an email regarding the replacement of a damaged product."

[0651] In this way, the system improves the efficiency of customer support operations and enhances customer satisfaction. Furthermore, it reduces variations in the quality of service provided by individual staff members, enabling the delivery of consistently high-quality support.

[0652] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0653] Step 1: Receiving and recording complaint calls

[0654] A user receives a complaint call. Here, the user is a customer service operator. The terminal (the operator's PC or telephone system) records the call. Specifically, when the operator answers the call, the telephone system starts recording simultaneously with the start of the call. The recorded audio data is transmitted to the server in real time. The input is the customer's voice call, and the output is the recorded audio data.

[0655] Step 2: Speech recognition and transcription

[0656] The server sends the received audio data to a speech recognition API. For example, the Google Cloud Speech-to-Text API is used here. The server receives the API's processing result and converts the audio into text data. The input is audio data, and the output is text data. Specifically, the server sends the audio data to the speech recognition API and stores the resulting string data.

[0657] Step 3: Text Analysis

[0658] The server inputs the acquired text data into a natural language processing (NLP) model, which then analyzes the text. For example, the GPT-4 NLP model is used. The server then analyzes the results returned by the NLP model to understand the specific content of the claim. The input is the transcribed text data, and the output is the analysis result (the key points of the claim). Specifically, key points such as "delivery is delayed" and "order number 12345" are extracted.

[0659] Step 4: Generating an apology response

[0660] The server generates an appropriate apology response based on the analysis results from the NLP model. A generative AI model is used for this purpose. For example, it might generate a response such as, "We sincerely apologize for the delay. We are currently checking the delivery status, so please wait a moment." The input is the analysis results, and the output is the generated apology response. Specifically, a prompt for generating the apology response is input to the generative AI model, and the resulting apology message is retrieved.

[0661] Step 5: Provide a response

[0662] The server generates an apology response and sends it to the terminal. The terminal displays the received apology response on the operator's monitor. This allows the operator to refer to the generated apology and proceed with customer service. The input is the generated apology response, and the output is the text displayed on the operator's monitor. Specifically, the terminal executes the process of displaying the apology response received from the server on its screen.

[0663] Step 6: Receiving and analyzing complaint and inquiry emails

[0664] A user receives a complaint or inquiry email. Here, the user is a staff member at a customer service center. The terminal sends the received email body to the server. The server analyzes the email content using an NLP model. Specifically, it uses GPT-4 or similar to input the email text and analyzes its content. The input is the email body, and the output is the analysis result (the main points of the complaint or inquiry).

[0665] Step 7: Generating the reply content

[0666] The server generates an appropriate response based on the analysis results. A generative AI model is used for this generation. For example, a response such as "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product." might be generated. The input is the analysis results, and the output is the generated response. Specifically, the server prompts the generative AI model based on the analysis results and retrieves the generated response.

[0667] Step 8: Provide a reply

[0668] The server generates a reply and sends it to the terminal. The terminal displays this reply in the staff member's email client. This allows the staff member to review the generated reply, edit it as needed, and send the email. The input is the generated reply, and the output is the text displayed in the staff member's email client. Specifically, the terminal executes the process of displaying the reply received from the server in the email client.

[0669] (Application Example 1)

[0670] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0671] Traditional customer support systems face the challenge of responding quickly and appropriately to customer complaints and inquiries. Similarly, timely and accurate responses are required for interactive complaints and inquiries from viewers. This can easily lead to decreased customer satisfaction and inconsistent support quality.

[0672] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0673] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text and understanding the content of the complaint, means for generating an appropriate apology response based on the content of the complaint, means for providing the generated apology response, and means for automatically analyzing complaints and inquiries from viewers and automatically generating appropriate responses, including voice recognition means, natural language processing means, and response generation means. This enables a rapid, consistent, and high-quality response to complaints and inquiries from customers and viewers.

[0674] "Customer" refers to anyone who purchases or uses a product or service.

[0675] "Viewer" refers to anyone who watches or listens to videos, music, or other content provided through content distribution services.

[0676] A "complaint" refers to a customer or viewer reporting dissatisfaction or problems with a product or service.

[0677] "Inquiry" refers to a question or request from a customer or viewer seeking information or support.

[0678] "Speech recognition means" refers to systems and devices that analyze speech as digital data and convert it into text data.

[0679] "Natural language processing methods" refer to technologies and algorithms used to analyze and understand human language using computers.

[0680] "Response generation means" refers to algorithms or systems that automatically generate appropriate responses or answers based on analyzed claims or inquiries.

[0681] "Recording means" refers to devices or systems for saving received audio in digital format.

[0682] "Transcription methods" refer to systems and software used to convert recorded audio data into text data.

[0683] An "apology response" refers to a statement expressing regret for a customer's or viewer's complaint and outlining actions taken to resolve the issue.

[0684] "Automated analysis" refers to the process of using computers to analyze data and understand its content and intent.

[0685] This invention is a system for providing a quick and accurate response to complaints and inquiries from viewers and customers. This system consists of the following main components:

[0686] 1. Speech Recognition Method: Receive and digitally store voice complaints from customers. This involves using APIs or software for speech recognition (e.g., Google Speech Recognition API).

[0687] 2. Natural Language Processing Methods: Analyze the content of received audio data and complaint / inquiry emails to understand the intentions of customers and viewers. Specifically, utilize natural language processing models (e.g., Hugging Face's transformers library).

[0688] 3. Response generation means: Based on the analyzed complaint content, appropriate apology responses and reply content are automatically generated. This enables a quick and appropriate response.

[0689] 4. Recording method: The voice of the complaint is recorded and sent to the server. A recording device or telephone system is responsible for this role.

[0690] 5. Transcription method: Convert the recorded audio data into text data. This is done using a speech recognition API.

[0691] Server Role

[0692] The server plays a central role in processing the received audio and text data. First, it uses speech recognition to convert the audio into text. Then, it uses natural language processing to analyze the text and understand what the customer or audience wants. Finally, it uses response generation to automatically generate appropriate replies or apologies and send them to the terminal.

[0693] Terminal role

[0694] The terminal receives data sent from the server and provides it to the user. It also plays a role in recording the voice of the complaint and sending it to the server. Furthermore, it displays the generated response or reply content, which the user can use as a reference when handling the complaint.

[0695] User roles

[0696] Users are operators or staff who handle complaints from viewers and customers. They are required to use the generated responses and reply content as a reference to take appropriate action.

[0697] Specific example

[0698] For example, if a viewer sends a complaint such as "the video keeps cutting out" via voice message within the app, the following response is automatically generated:

[0699] Viewer: "The video keeps cutting out."

[0700] App: "We apologize for the issue you are experiencing. We are currently working on a solution."

[0701] In this way, this system can respond quickly and appropriately to viewer and customer complaints and inquiries, contributing to improved customer satisfaction.

[0702] Example of a prompt

[0703] "The video keeps cutting out."

[0704] "No sound."

[0705] "My subscription is not being renewed."

[0706] This allows for accurate support to be provided for specific complaints and inquiries.

[0707] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0708] Step 1:

[0709] The terminal receives audio complaints from viewers or customers. The received audio complaints are recorded digitally through the terminal's built-in microphone or an externally connected microphone.

[0710] Input: Audience or customer voice complaint

[0711] Output: Digital audio data

[0712] Step 2:

[0713] The terminal sends the recorded audio data to the server. The transmitted audio data is converted to an appropriate format on the server side and made ready for processing by speech recognition software.

[0714] Input: Digital audio data

[0715] Output: Audio data sent to the server

[0716] Step 3:

[0717] The server converts the audio data into text data using speech recognition technology. This process utilizes tools such as the Google Speech Recognition API to convert speech to text.

[0718] Input: Audio data sent to the server

[0719] Output: Transcribed text data

[0720] Step 4:

[0721] The server analyzes the transcribed text using natural language processing techniques. It uses libraries such as Hugging Face's transformers to understand the claims in the text and analyze the intent of the customer or audience.

[0722] Input: Transcribed text data

[0723] Output: Analysis results of the claim content

[0724] Step 5:

[0725] The server generates appropriate apology responses and reply content based on the analysis results. In this process, a generative AI model is used to automatically generate response texts by providing appropriate prompt sentences as input.

[0726] Input: Analysis results of the claim

[0727] Output: Automated apology response or reply content

[0728] Step 6:

[0729] The server sends the generated apology or reply to the terminal. The terminal displays this reply on the operator's or staff's screen, allowing them to use it as a reference when taking action.

[0730] Input: Automated apology or reply content

[0731] Output: Response sent to the terminal

[0732] Step 7:

[0733] Users review the generated apology or reply displayed on their device and use it to respond to viewers or customers. They can also edit the reply content at their discretion if necessary.

[0734] Input: The generated apology or reply displayed on the device.

[0735] Output: Edited response content and response to the audience or customer

[0736] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0737] This invention relates to a system for improving the efficiency of complaint handling and enhancing customer satisfaction in customer centers. This system not only analyzes the content of complaint calls and complaint / inquiry emails, but also incorporates an emotion engine to recognize the user's emotions, enabling it to automatically generate more appropriate apology responses and reply messages.

[0738] A telephone complaint handling system incorporating an emotion engine.

[0739] Receiving and recording complaint calls

[0740] When a customer service operator receives a complaint call, the operator's PC or telephone system records the audio. The recorded audio data is transmitted to a server in real time.

[0741] Speech recognition and transcription

[0742] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This allows the content of the claim to be obtained as text.

[0743] Text and sentiment analysis

[0744] The server analyzes the transcribed text using a natural language processing (NLP) model to understand the complaint. During this process, the server also uses an emotion engine to recognize the user's emotions, for example, whether the customer is angry or sad.

[0745] Generating an apology response

[0746] Based on the understood complaint and emotions, the server automatically generates an appropriate apology response. For example, if a complaint about "delayed delivery" and the emotion of "anger" are recognized, an emphatic apology such as "We sincerely apologize. We will immediately investigate the issue and take appropriate action" will be generated for the customer.

[0747] Providing a response

[0748] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can use this response as a reference to proceed with handling the complaint.

[0749] Inquiry email response system with integrated emotion engine

[0750] Receiving and analyzing emails

[0751] When a customer service staff member (a user) receives a complaint or inquiry email, the terminal sends the email body to the server. The server uses an NLP model to analyze the email's content. Here, it understands what specific problem or inquiry the email represents.

[0752] Analysis of email content and emotions

[0753] The server then uses an emotion engine to recognize the user's emotions from the email content. For example, it extracts emotions such as "dissatisfaction" or "disappointment" from the email body.

[0754] Generating a reply

[0755] Based on the analysis results and emotions, the server automatically generates an appropriate response. For example, if a complaint that "the product was broken" and the emotion of "disappointment" are recognized, a response such as "We are very sorry for the inconvenience. We will arrange for a replacement immediately" will be generated.

[0756] Providing a reply

[0757] The generated reply is sent to the device and displayed in the staff member's email client. The staff member can review the reply, edit it as needed, and then send back the email.

[0758] Specific example

[0759] Examples of complaint calls

[0760] User: The operator answers the phone.

[0761] Terminal: Records the call content and sends the audio data to the server.

[0762] Server: Performs transcription using a speech recognition API.

[0763] Server: Uses an NLP model to analyze text and sentiment to understand the content and emotions behind the complaint.

[0764] Server: Generates an appropriate apology response and sends it to the terminal.

[0765] User: Use the generated response as a reference to handle the complaint.

[0766] Example of a complaint email

[0767] User: Staff receive the email.

[0768] Terminal: Sends the email body to the server.

[0769] Server: Analyzes content and emotion using an NLP model to understand the content and emotions of the complaint.

[0770] Server: Generates an appropriate reply and sends it to the terminal.

[0771] User: Review the generated reply before sending your email.

[0772] These specific processing steps enable this system to contribute to improving the efficiency of customer support and customer satisfaction. The incorporation of an emotion engine allows for more appropriate responses that take customer emotions into consideration.

[0773] The following describes the processing flow.

[0774] A telephone complaint handling system incorporating an emotion engine.

[0775] Step 1:

[0776] A user (a customer service operator) receives a complaint call.

[0777] The operator answers the phone using the usual procedure.

[0778] Step 2:

[0779] The terminal (operator's PC or telephone system) records the call.

[0780] When a call is initiated, the audio is automatically recorded.

[0781] Step 3:

[0782] The device transmits the recorded audio data to the server in real time.

[0783] The recorded audio data is divided into packets and sent to the server.

[0784] Step 4:

[0785] The server sends the received audio data to a speech recognition API for transcription.

[0786] The audio data is converted to text format via a speech recognition API.

[0787] Step 5:

[0788] The server sends the transcribed text to a natural language processing (NLP) model for analysis.

[0789] The NLP model extracts the content of the complaint and important keywords.

[0790] Step 6:

[0791] The server uses an emotion engine to recognize the user's emotions from the transcribed text.

[0792] The emotion engine analyzes emotional expressions within the text to identify the customer's emotions (e.g., anger, sadness, confusion, etc.).

[0793] Step 7:

[0794] The server generates an appropriate apology response based on the analysis results and the recognized emotions.

[0795] Responses are generated using templates or AI models tailored to the nature of the complaint and the customer's emotions. For example, if a customer is "angry because their delivery is delayed," an emphatic apology such as "We sincerely apologize. We will investigate and address the issue immediately" will be generated.

[0796] Step 8:

[0797] The server generates an apology response, sends it to the terminal, and displays it to the user.

[0798] The generated apology response is displayed on the operator's monitor.

[0799] Step 9:

[0800] We will handle the complaint based on the apology response displayed to the user.

[0801] The operator uses the displayed response to proceed with the actual complaint handling. Responses can also be customized as needed.

[0802] Inquiry email response system with integrated emotion engine

[0803] Step 1:

[0804] The user (customer center staff) receives complaint and inquiry emails.

[0805] Check for new emails through your email client.

[0806] Step 2:

[0807] The device sends the email content to the server.

[0808] The email body and subject are transferred to the server via the API.

[0809] Step 3:

[0810] The server analyzes the email content using a natural language processing (NLP) model.

[0811] The NLP model analyzes the content and topic of the email.

[0812] Step 4:

[0813] The server uses an emotion engine to recognize the user's emotions from the email content.

[0814] The emotion engine analyzes emotional expressions within emails to identify the customer's emotions (e.g., dissatisfaction, disappointment, confusion, etc.).

[0815] Step 5:

[0816] The server automatically generates an appropriate response based on the analysis results and recognized emotions.

[0817] The system generates response messages using templates or AI models tailored to the nature of the complaint and the customer's feelings. For example, if a customer says, "I'm disappointed because the product is broken," a response such as, "We are very sorry for the inconvenience. We will arrange for a replacement immediately," might be generated.

[0818] Step 6:

[0819] The server generates a reply and sends it to the terminal for the user to see.

[0820] The generated reply will be displayed in the staff member's email client.

[0821] Step 7:

[0822] The user reviews the displayed reply and edits it as needed.

[0823] Staff will review the reply and make any necessary corrections.

[0824] Step 8:

[0825] The user sends an email with the revised reply.

[0826] Send the finalized response to the customer to complete the process.

[0827] As described above, by incorporating an emotion engine, the quality of complaint and inquiry handling can be improved, and responses that take customer emotions into consideration can be made. This system can promote the efficiency of customer support and improve customer satisfaction.

[0828] (Example 2)

[0829] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0830] Handling customer complaints in modern customer centers is time-consuming and labor-intensive. Furthermore, the quality of service provided by operators varies, making it difficult to consistently improve customer satisfaction. In particular, it is crucial to accurately understand customer emotions and respond with appropriate and courteous care. A method is needed to address these challenges efficiently and effectively.

[0831] The identification processing performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text to understand the content of the complaint, means for analyzing the transcribed text using a natural language processing model and an emotion analysis engine to understand the customer's emotions, means for generating an appropriate apology response based on the content of the complaint and the customer's emotions, and means for providing the generated apology response. This makes it possible to accurately grasp the content of the customer's complaint and emotions, and to quickly generate and provide an appropriate response accordingly. Furthermore, it is possible to reduce the burden on operators and ensure consistency in response quality, thereby improving customer satisfaction.

[0832] "Means for receiving customer voice complaints" refers to a function for electronically receiving complaints made by customers via telephone.

[0833] "Means for recording received audio" refers to a function for saving received audio data to a recording medium.

[0834] "Methods for transcribing recorded audio data" refers to functions that utilize speech recognition technology to convert recorded audio data into text data.

[0835] "Means for analyzing transcribed text and understanding the content of a claim" refers to a function that uses natural language processing technology to analyze text data obtained through transcription and recognize the content contained therein.

[0836] "A means of understanding customer emotions by analyzing transcribed text using natural language processing models and sentiment analysis engines" refers to a function that uses specific algorithms to analyze the linguistic structure and emotional aspects within the text and grasp the customer's emotional state.

[0837] "A means of generating an appropriate apology response based on the content of the complaint and the customer's feelings" refers to a function that automatically generates the optimal response text corresponding to the specific content of the recognized complaint and the customer's feelings.

[0838] "Means of providing a generated apology response" refers to a function for displaying the generated response to the operator or sending it directly to the customer.

[0839] "Means for receiving complaints and inquiries via email" refers to a function for receiving complaints and inquiries sent via email.

[0840] "A means of analyzing the content of received emails to understand complaints, inquiries, and customer sentiment" refers to a function that uses natural language processing and sentiment analysis technologies to analyze the text of received emails and grasp their content and customer sentiment.

[0841] "A means of generating appropriate response content based on complaints, inquiries, and customer sentiment" refers to a function that automatically generates the optimal response based on recognized email content and customer sentiment.

[0842] "A means of automatically generating response content and apology responses based on the content of the complaint and the customer's feelings using a generative AI model" refers to a function that uses an artificial intelligence model to generate appropriate responses to complaints and inquiries from specific prompt sentences.

[0843] "A means of instructing a generation AI model to provide an appropriate response using a prompt statement" refers to a function that provides a specific input (prompt) to the AI ​​model and generates an appropriate output based on that input.

[0844] This invention relates to a system for improving the efficiency of complaint handling and enhancing customer satisfaction in customer centers. Specifically, it analyzes both voice and email complaints and automatically generates appropriate apologies and replies.

[0845] System Configuration

[0846] This system consists of the following main elements:

[0847] 1. Users: Customer center operators and staff will be the primary users of the system.

[0848] 2. Terminals: This includes PCs, telephone systems, and email clients used by operators and staff.

[0849] 3. Server: This server runs the speech recognition API, natural language processing (NLP) models, sentiment analysis engine, and generative AI models.

[0850] Hardware and software to be used

[0851] Speech Recognition API: Use the Google Cloud Speech-to-Text API to convert speech data into text.

[0852] Natural Language Processing Models: Text data is analyzed using NLP models such as BERT and GPT-3.

[0853] Sentiment analysis engine: Analyzes emotions from text using tools such as IBM Watson Tone Analyzer.

[0854] Generative AI model: Uses GPT-3 and other models to generate appropriate apology responses and reply content.

[0855] Processing flow

[0856] Regarding the customer complaint phone handling system:

[0857] When a user receives a call, the device records the audio and sends it to the server in real time.

[0858] The server sends the audio data to the Google Cloud Speech-to-Text API, where it is converted into text data.

[0859] The server sends text data to an NLP model to analyze the claim details.

[0860] The server then uses an emotion analysis engine to analyze the customer's emotions.

[0861] Based on the nature of the complaint and the emotional state, the server uses a generative AI model to generate an appropriate apology response.

[0862] The generated apology response is sent to the device and displayed on the user's monitor.

[0863] Regarding the complaint email handling system:

[0864] When a user receives an email, the email body is sent from the device to the server.

[0865] The server uses an NLP model to analyze the email content and understand the complaint and emotions behind it.

[0866] Based on the analysis results, the server uses a generative AI model to generate appropriate responses.

[0867] The generated reply is sent to the device and displayed in the user's email client.

[0868] Examples of specific cases and prompt statements

[0869] For example, if an operator receives a complaint about a delivery delay:

[0870] User: The operator answers the phone.

[0871] Terminal: Records the call content and sends the audio data to the server.

[0872] Server: Performs transcription using a speech recognition API (e.g., Google Cloud Speech-to-Text).

[0873] Server: Uses NLP models to analyze text and sentiment to understand the content and emotions of complaints (e.g., BERT and IBM Watson Tone Analyzer).

[0874] Server: Generates an appropriate apology response and sends it to the terminal.

[0875] User: Use the generated response as a reference to handle the complaint.

[0876] Examples of prompt statements:

[0877] "We've received a complaint about a delivery delay. The customer is angry. Please generate an appropriate apology letter."

[0878] "We received an email inquiry about a damaged product. The customer is disappointed. Please generate an appropriate reply."

[0879] This system streamlines customer service complaint handling and enables appropriate responses that take customer emotions into consideration. This, in turn, leads to improved customer satisfaction.

[0880] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0881] Customer complaint phone handling system

[0882] Processing steps

[0883] Step 1:

[0884] A customer service operator, acting as the user, receives a complaint call. When the operator presses the "Start Call" button, the terminal begins recording the conversation. The recorded audio data is transmitted to the server in real time. The input is the call audio, and the output is the audio data stored on the server.

[0885] Step 2:

[0886] The server sends the received audio data to the Google Cloud Speech-to-Text API, where it is converted into text data. In this step, audio data is input and text data is generated as output. Specifically, the server sends an API request and receives a response.

[0887] Step 3:

[0888] The server sends the text data converted from the speech to a natural language processing (NLP) model such as BERT or GPT-3 to analyze the claim content. The input for this step is the transcribed text, and the analyzed claim content is output. Specifically, the server sends an API request to the NLP model and receives the analysis results.

[0889] Step 4:

[0890] The server then sends the analyzed text to a sentiment analysis engine, such as IBM Watson Tone Analyzer, to analyze the customer's emotions. In this step, the text containing the complaint is input, and the sentiment analysis results are obtained as output. The specific operation involves sending requests to the sentiment analysis API and receiving responses.

[0891] Step 5:

[0892] The server sends prompt messages to a generative AI model (e.g., GPT-3) based on the understood complaint content and sentiment state, generating an appropriate apology response. The input data is the complaint content and sentiment analysis results, and the output is the generated apology response. Specifically, the server defines appropriate prompt messages for the AI ​​model and sends an API request.

[0893] Step 6:

[0894] The generated apology response is sent to the terminal and displayed on the operator's monitor. In this step, the generated response text is input, and the displayed monitor is output. Specifically, a UI update request is made to display the response text.

[0895] Complaint email handling system

[0896] Processing steps

[0897] Step 1:

[0898] When a customer service staff member (a user) receives a complaint or inquiry email, the email body is sent from the terminal to the server. The input is the received email, and the output is the email data stored on the server. Specifically, the email client sends the email body via the API.

[0899] Step 2:

[0900] The server uses an NLP model to analyze the email content. The email body is input, and the complaint and inquiry details are output. Specifically, an API request is sent to analyze the text data, and the analysis results are received.

[0901] Step 3:

[0902] The server uses an emotion analysis engine to understand customer emotions during the process of analyzing email content and inquiry details. The input data is the email body, and the output data is the emotion analysis result. Specifically, it sends a request to the emotion analysis API and receives a response.

[0903] Step 4:

[0904] The server uses a generative AI model to generate an appropriate response based on the analyzed content and emotions. In this step, the complaint and customer emotions are used as input, and the output is the generated response. Specifically, a prompt is sent to the generative AI model to generate an appropriate response.

[0905] Step 5:

[0906] The generated reply is sent to the terminal and displayed in the staff member's email client. The input data is the generated reply text, and the output is the displayed email client. Specifically, a UI update request is executed to display the reply text.

[0907] Through the processing steps described above, this system contributes to improving the efficiency of customer support and enhancing customer satisfaction. By understanding customer emotions and responding appropriately and quickly, customer satisfaction can be increased.

[0908] (Application Example 2)

[0909] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0910] In modern customer service centers, responding quickly and appropriately to customer complaints and inquiries is a crucial challenge. However, especially when complaints are emotional, it is often difficult for staff to accurately understand those emotions and respond appropriately. Therefore, there is a need for systems that can improve customer satisfaction and increase the efficiency of complaint handling.

[0911] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0912] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text to understand the content of the complaint and the customer's feelings, means for generating an appropriate apology response based on the content of the complaint and the customer's feelings, and means for providing the generated apology response. This enables appropriate responses that take into account the customer's feelings.

[0913] "Means of understanding customer emotions" refers to the process of analyzing and recognizing the emotions behind customer words, such as anger, disappointment, or satisfaction, using natural language processing technology.

[0914] A "means for generating apology responses" is a system that automatically generates appropriate and empathetic apologies and responses based on the analyzed complaint content and customer emotions.

[0915] "Methods for generating response content" refers to a process that automatically constructs appropriate responses to customers, taking into account complaints, inquiries, and related emotions.

[0916] A "natural language processing model" is a machine learning model that analyzes text data and uses that information to understand meaning and intent.

[0917] A "generative AI model" is an artificial intelligence model that automatically generates text or answers based on given prompts or input data.

[0918] A "prompt" is an instruction or input sentence used by a generative AI model to generate a response, and it serves as a guideline for providing appropriate output based on its content.

[0919] This invention is a system that supports customer service in physical stores. Its purpose is to enable store staff to respond quickly and accurately to customer complaints and inquiries. A specific example of the system is described below.

[0920] Hardware and software to be used

[0921] The system uses the following hardware and software.

[0922] Hardware: Smartphones, smart glasses, head-mounted displays

[0923] Software: Python, TensorFlow, Transformers package, SpeechRecognition package

[0924] System Configuration

[0925] 1. Means of receiving voice complaints from customers:

[0926] Staff members wear smartphones, smart glasses, or head-mounted displays to receive customer complaints and inquiries.

[0927] 2. Means for recording received audio:

[0928] The received audio data is recorded on the staff member's terminal and sent to the server in real time.

[0929] 3. Methods for transcribing recorded audio data:

[0930] The server converts the received audio data into text data using the SpeechRecognition package. This allows the content of complaints and inquiries to be obtained as text.

[0931] 4. Means for analyzing transcribed text to understand the content of the complaint and the customer's feelings:

[0932] The server analyzes the converted text data using the Transformers package and a natural language processing model to understand the content of the complaint, while simultaneously recognizing the customer's emotions using an emotion engine.

[0933] 5. Means for generating an appropriate apology response based on the nature of the complaint and the customer's feelings:

[0934] Based on the analyzed complaint details and customer emotions, the server uses a generative AI model to generate an appropriate apology response. For example, if the complaint is "the product was broken" and the customer's emotion is recognized as "anger," an apology response such as "We sincerely apologize. We will arrange for a replacement immediately" will be generated.

[0935] 6. Means of providing the generated apology response:

[0936] The generated apology response is sent from the server to the staff member's terminal and displayed on their smartphone, smart glasses, or head-mounted display. The staff member uses this response as a reference when dealing with the customer.

[0937] Example of a prompt

[0938] Examples of prompt statements that can actually be used include the following:

[0939] "Analyze the customer's emotions from the following text: 'The product was broken. Please do something about it.'"

[0940] Specific example

[0941] Examples of customer service:

[0942] When a staff member receives a customer complaint using smart glasses, the audio is recorded by the smart glasses and sent to a server in real time. The server converts the received audio into text and analyzes the emotions. Then, an appropriate apology response is displayed on the staff member's glasses. The staff member uses the displayed apology response to address the customer and resolve the issue quickly.

[0943] Thus, the present invention aims to improve customer satisfaction in physical stores by enabling appropriate and prompt responses that take customer emotions into consideration.

[0944] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0945] Step 1:

[0946] Receive voice complaints from customers.

[0947] Input: A customer makes a verbal complaint to a store staff member.

[0948] Specific operation: The store staff member's smartphone, smart glasses, or head-mounted display receives audio in real time.

[0949] Output: Received audio data.

[0950] Step 2:

[0951] Record the received audio.

[0952] Input: Audio data received in Step 1.

[0953] Specific operation: The device (smartphone, smart glasses, head-mounted display) records audio and sends the recorded data to the server.

[0954] Output: Recorded audio data.

[0955] Step 3:

[0956] Transcribe the recorded audio data into text.

[0957] Input: Audio data recorded in Step 2.

[0958] Specific operation: The server uses a speech recognition API to convert the audio data into text data.

[0959] Output: Transcripted text data.

[0960] Step 4:

[0961] The transcribed text is analyzed to understand the content of the complaint and the customer's feelings.

[0962] Input: Text data obtained in Step 3.

[0963] Specific operation: The server uses a natural language processing (NLP) model to analyze the content of the complaint from the text data. At the same time, it uses an emotion engine to recognize the customer's emotions.

[0964] Output: Analyzed complaint details and perceived customer sentiment.

[0965] Step 5:

[0966] Based on the nature of the complaint and the customer's feelings, generate an appropriate apology response.

[0967] Input: Complaint details and customer sentiment analyzed in Step 4.

[0968] Specific operation: The server uses an AI model to generate an appropriate apology response based on the entered complaint content and emotions. For example, if the complaint is "the product was broken" and the emotion "anger" is recognized, an apology response such as "We sincerely apologize. We will arrange for a replacement immediately." will be generated.

[0969] Output: Generated apology response.

[0970] Step 6:

[0971] Provide a generated apology response.

[0972] Input: The apology response generated in Step 5.

[0973] Specific operation: The server sends the generated apology response to the staff member's device (smartphone, smart glasses, or head-mounted display). The staff member can then use the apology response as a reference when dealing with the customer.

[0974] Output: The apology response displayed on the terminal.

[0975] This allows the system to provide quick and appropriate responses in physical stores, taking customer emotions into consideration, thereby improving customer satisfaction.

[0976] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0977] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0978] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0979] [Third Embodiment]

[0980] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0981] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0982] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0983] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0984] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0985] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0986] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0987] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0988] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0989] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0990] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0991] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0992] This invention relates to a system for streamlining complaint handling in customer service centers. The system aims to analyze the content of complaint calls and complaint / inquiry emails, and automatically generate and provide appropriate apologies and replies. The main components of this system include speech recognition, natural language processing (NLP), and automated response generation functions.

[0993] Telephone-based complaint handling system

[0994] Receiving and recording complaint calls

[0995] When a customer service operator receives a complaint call, the operator's PC or telephone system records the audio. The recorded audio data is transmitted to a server in real time.

[0996] Speech recognition and transcription

[0997] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This allows the content of the claim to be obtained as text.

[0998] Text analysis

[0999] The server analyzes the transcribed text using a natural language processing (NLP) model to understand the complaint. This analysis extracts the specific points of what the customer is unhappy about.

[1000] Generating an apology response

[1001] Based on the understood complaint, the server automatically generates an appropriate apology response. For example, in response to a complaint that "delivery is delayed," a response such as "We are very sorry for the delay, sir / madam. We are currently checking the delivery status, so please wait a moment." will be generated.

[1002] Providing a response

[1003] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can use this response as a reference to proceed with handling the complaint.

[1004] Complaint and inquiry email handling system

[1005] Receiving and analyzing emails

[1006] When a customer service staff member (a user) receives a complaint or inquiry email, the terminal sends the email body to the server. The server uses an NLP model to analyze the email's content. Here, it understands what specific problem or inquiry the email represents.

[1007] Generating a reply

[1008] Based on the analysis results, the server automatically generates an appropriate response. For example, in response to a complaint that "the product was damaged," a response such as "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product." will be generated.

[1009] Providing a reply

[1010] The generated reply is sent to the device and displayed in the staff member's email client. The staff member can review the reply, edit it as needed, and then send back the email.

[1011] Specific example

[1012] Examples of complaint calls

[1013] User: The operator answers the phone.

[1014] Terminal: Records the call content and sends the audio data to the server.

[1015] Server: Performs transcription using a speech recognition API.

[1016] Server: Analyzes text using an NLP model to understand the content of the claim.

[1017] Server: Generates an appropriate apology response and sends it to the terminal.

[1018] User: Use the generated response as a reference to handle the complaint.

[1019] Example of a complaint email

[1020] User: Staff receive the email.

[1021] Terminal: Sends the email body to the server.

[1022] Server: Analyzes the content using an NLP model to understand the claim.

[1023] Server: Generates an appropriate reply and sends it to the terminal.

[1024] User: Review the generated reply before sending your email.

[1025] In this way, this system improves the efficiency of customer support operations and enhances customer satisfaction. Furthermore, it reduces variations in the quality of service provided by individual staff members, enabling the delivery of consistently high-quality support.

[1026] The following describes the processing flow.

[1027] Handling complaint calls

[1028] Step 1:

[1029] A user (a customer service operator) receives a complaint call.

[1030] An operator receives a call from a customer.

[1031] Step 2:

[1032] The terminal (operator's PC or telephone system) records the call.

[1033] The call is initiated, and the audio data is recorded in real time.

[1034] Step 3:

[1035] The device transmits the recorded audio data to the server in real time.

[1036] The recorded audio is compressed and sent to the server.

[1037] Step 4:

[1038] The server sends the received audio data to a speech recognition API for transcription.

[1039] The speech data is converted into text data using a speech recognition API.

[1040] Step 5:

[1041] The server sends the transcribed text to a natural language processing (NLP) model for analysis.

[1042] We use an NLP model to extract the types of complaints and key keywords.

[1043] Step 6:

[1044] The server generates an appropriate apology response based on the analysis results.

[1045] For each claim, a response is generated using a pre-prepared template or an AI model.

[1046] Step 7:

[1047] The server generates an apology response, sends it to the terminal, and displays it to the user.

[1048] An appropriate apology response is displayed on the operator's monitor.

[1049] Step 8:

[1050] We will handle the complaint based on the apology response displayed to the user.

[1051] The operator will then proceed with handling the actual complaint based on the displayed response.

[1052] Handling complaints and inquiry emails

[1053] Step 1:

[1054] The user (customer center staff) receives complaint and inquiry emails.

[1055] Check your email client for new emails.

[1056] Step 2:

[1057] The device sends the email content to the server.

[1058] The email body and subject are forwarded to the server via the API.

[1059] Step 3:

[1060] The server analyzes the email content using a natural language processing (NLP) model.

[1061] The content of emails is analyzed to extract specific details of complaints and inquiries.

[1062] Step 4:

[1063] The server automatically generates an appropriate response based on the analysis results.

[1064] The response content is generated from an AI model or template.

[1065] Step 5:

[1066] The server generates a reply and sends it to the terminal for the user to see.

[1067] The reply will appear in the staff member's email client.

[1068] Step 6:

[1069] The user reviews the displayed reply and edits it as needed.

[1070] Staff will review the reply and make any necessary corrections.

[1071] Step 7:

[1072] The user sends an email with the revised reply.

[1073] Send the confirmed reply to the customer.

[1074] Through these specific processing steps, the present invention can significantly improve the efficiency of customer support and reduce the burden on staff.

[1075] (Example 1)

[1076] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1077] In customer service settings, a tremendous amount of time and effort is spent handling complaints and inquiries. This can lead to inconsistencies in service quality among staff members, potentially lowering customer satisfaction. Furthermore, delays in situations where a quick response is required can further increase customer dissatisfaction. Therefore, there is a need to handle complaints and inquiries efficiently and effectively, and to achieve consistent, high-quality service.

[1078] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1079] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voices, and means for transcribing the recorded voice data. This allows the content of the complaints to be efficiently converted into text data, making it possible to analyze them.

[1080] The server also includes means for analyzing the transcribed text and understanding the content of the claim, means for generating an appropriate apology response based on the content of the claim, and means for providing the generated apology response. This enables the rapid generation of appropriate responses to claims and reduces variability in response quality.

[1081] Furthermore, the server includes means for transmitting recorded audio data in real time, means for converting audio to text using a speech recognition API, means for analyzing text using a natural language processing model, means for automatically generating apology responses using a generative AI model, means for analyzing email content, and means for generating reply content. This enables consistent, high-quality, and efficient complaint handling in both voice calls and email correspondence.

[1082] A "customer" is someone who receives a service and makes complaints or inquiries to the service provider.

[1083] "Voice complaints" refer to expressions of dissatisfaction or complaints made by customers using voice communication methods such as telephone calls.

[1084] "Means of recording" refers to devices and software used to record audio data in digital format.

[1085] "Methods for transcription" refer to technologies and devices used to convert audio data into text data.

[1086] "Means of analyzing text" refers to technologies and devices that use natural language processing techniques to understand text data and analyze its content.

[1087] "Means for generating apology responses" refers to technologies or devices that automatically create appropriate apology statements based on the content of a complaint.

[1088] "Means of providing generated apology responses" refers to devices or software that display and make available the generated apology messages to operators and staff.

[1089] "Means of sending to a server in real time" refers to technologies and devices for instantly transferring recorded audio data to a server.

[1090] A "speech recognition API" refers to an application programming interface (API) provided as an external service for converting speech data into text data.

[1091] A "natural language processing model" refers to a machine learning model designed to understand and analyze human language.

[1092] A "generative AI model" refers to an artificial intelligence (AI) model that generates text based on given input data.

[1093] "Complaint and inquiry emails" refer to complaints or questions that customers send via email.

[1094] "Means of analyzing email content" refers to technologies and devices that analyze the body of a received email and understand its content.

[1095] "Means for generating reply content" refers to technologies or devices that automatically create appropriate reply messages based on the content of an email.

[1096] This invention is a system for streamlining complaint and inquiry handling in customer service. This system enables customers to receive prompt, consistent, and high-quality responses to complaints and inquiries. Specific embodiments for implementing this invention are described below.

[1097] Receiving and recording complaint calls

[1098] When a user makes a complaint call, the operator's PC or the telephone system records the audio. The recorded audio data is transmitted to a server in real time. The recording function is implemented using digital recording devices or software integrated into the telephone system.

[1099] Speech recognition and transcription

[1100] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This speech recognition process uses an external speech recognition service, such as the Google Cloud Speech-to-Text API. Based on the speech recognition results returned by the API, the call content is retrieved as text data.

[1101] Text analysis

[1102] The server feeds the acquired text data into a natural language processing (NLP) model to analyze the complaint content. Advanced NLP models such as GPT-4 are used for this process. The NLP model analysis clearly extracts customer dissatisfaction and requests. For example, specific complaints such as "delivery is delayed" are identified.

[1103] Generating an apology response

[1104] The server generates an appropriate apology response based on the analysis results. A generative AI model is used to create the best possible apology based on the input text data. For example, a response such as, "We sincerely apologize for the delay, customer. We are currently checking the delivery status, so please wait a moment," might be generated.

[1105] Providing a response

[1106] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can then use the displayed apology as a reference to proceed with customer service.

[1107] Receiving and analyzing complaint and inquiry emails

[1108] When a user receives a complaint or inquiry email, the terminal sends the email body to the server. The server analyzes the email content using an NLP model. Here too, natural language processing models such as GPT-4 are used. The specific problems and inquiries in the email are clarified.

[1109] Generating a reply

[1110] The server generates an appropriate reply based on the analysis results. A generation AI model is used to generate an email reply such as, "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product."

[1111] Providing a reply

[1112] The generated reply is sent to the device and displayed in the staff member's email client. The staff member reviews the displayed reply, edits it as needed, and then replies to the email.

[1113] Specific example

[1114] Examples of complaint calls

[1115] User: The operator calls and says, "Your delivery is delayed."

[1116] Terminal: Records call content and sends it to the server in real time.

[1117] Server: Uses a speech recognition API to convert speech data into text.

[1118] Server: Uses an NLP model to analyze text data and understand the content of the claims.

[1119] Server: Generates an appropriate apology response and sends it to the terminal.

[1120] User: Use the generated apology response as a reference to proceed with handling the complaint.

[1121] Example of a complaint email

[1122] User: Staff receive an email stating that "the product was damaged."

[1123] Terminal: Sends the email body to the server.

[1124] Server: Uses an NLP model to analyze email content and understand the nature of the complaint.

[1125] Server: Generates an appropriate reply and sends it to the terminal.

[1126] User: Review the generated reply and send a reply email.

[1127] Examples of specific prompt messages include, "Please generate an apology letter to handle a delivery delay complaint," or "Please create an email regarding the replacement of a damaged product."

[1128] In this way, the system improves the efficiency of customer support operations and enhances customer satisfaction. Furthermore, it reduces variations in the quality of service provided by individual staff members, enabling the delivery of consistently high-quality support.

[1129] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1130] Step 1: Receiving and recording complaint calls

[1131] A user receives a complaint call. Here, the user is a customer service operator. The terminal (the operator's PC or telephone system) records the call. Specifically, when the operator answers the call, the telephone system starts recording simultaneously with the start of the call. The recorded audio data is transmitted to the server in real time. The input is the customer's voice call, and the output is the recorded audio data.

[1132] Step 2: Speech recognition and transcription

[1133] The server sends the received audio data to a speech recognition API. For example, the Google Cloud Speech-to-Text API is used here. The server receives the API's processing result and converts the audio into text data. The input is audio data, and the output is text data. Specifically, the server sends the audio data to the speech recognition API and stores the resulting string data.

[1134] Step 3: Text Analysis

[1135] The server inputs the acquired text data into a natural language processing (NLP) model, which then analyzes the text. For example, the GPT-4 NLP model is used. The server then analyzes the results returned by the NLP model to understand the specific content of the claim. The input is the transcribed text data, and the output is the analysis result (the key points of the claim). Specifically, key points such as "delivery is delayed" and "order number 12345" are extracted.

[1136] Step 4: Generating an apology response

[1137] The server generates an appropriate apology response based on the analysis results from the NLP model. A generative AI model is used for this purpose. For example, it might generate a response such as, "We sincerely apologize for the delay. We are currently checking the delivery status, so please wait a moment." The input is the analysis results, and the output is the generated apology response. Specifically, a prompt for generating the apology response is input to the generative AI model, and the resulting apology message is retrieved.

[1138] Step 5: Provide a response

[1139] The server generates an apology response and sends it to the terminal. The terminal displays the received apology response on the operator's monitor. This allows the operator to refer to the generated apology and proceed with customer service. The input is the generated apology response, and the output is the text displayed on the operator's monitor. Specifically, the terminal executes the process of displaying the apology response received from the server on its screen.

[1140] Step 6: Receiving and analyzing complaint and inquiry emails

[1141] A user receives a complaint or inquiry email. Here, the user is a staff member at a customer service center. The terminal sends the received email body to the server. The server analyzes the email content using an NLP model. Specifically, it uses GPT-4 or similar to input the email text and analyzes its content. The input is the email body, and the output is the analysis result (the main points of the complaint or inquiry).

[1142] Step 7: Generating the reply content

[1143] The server generates an appropriate response based on the analysis results. A generative AI model is used for this generation. For example, a response such as "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product." might be generated. The input is the analysis results, and the output is the generated response. Specifically, the server prompts the generative AI model based on the analysis results and retrieves the generated response.

[1144] Step 8: Provide a reply

[1145] The server generates a reply and sends it to the terminal. The terminal displays this reply in the staff member's email client. This allows the staff member to review the generated reply, edit it as needed, and send the email. The input is the generated reply, and the output is the text displayed in the staff member's email client. Specifically, the terminal executes the process of displaying the reply received from the server in the email client.

[1146] (Application Example 1)

[1147] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1148] Traditional customer support systems face the challenge of responding quickly and appropriately to customer complaints and inquiries. Similarly, timely and accurate responses are required for interactive complaints and inquiries from viewers. This can easily lead to decreased customer satisfaction and inconsistent support quality.

[1149] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1150] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text and understanding the content of the complaint, means for generating an appropriate apology response based on the content of the complaint, means for providing the generated apology response, and means for automatically analyzing complaints and inquiries from viewers and automatically generating appropriate responses, including voice recognition means, natural language processing means, and response generation means. This enables a rapid, consistent, and high-quality response to complaints and inquiries from customers and viewers.

[1151] "Customer" refers to anyone who purchases or uses a product or service.

[1152] "Viewer" refers to anyone who watches or listens to videos, music, or other content provided through content distribution services.

[1153] A "complaint" refers to a customer or viewer reporting dissatisfaction or problems with a product or service.

[1154] "Inquiry" refers to a question or request from a customer or viewer seeking information or support.

[1155] "Speech recognition means" refers to systems and devices that analyze speech as digital data and convert it into text data.

[1156] "Natural language processing methods" refer to technologies and algorithms used to analyze and understand human language using computers.

[1157] "Response generation means" refers to algorithms or systems that automatically generate appropriate responses or answers based on analyzed claims or inquiries.

[1158] "Recording means" refers to devices or systems for saving received audio in digital format.

[1159] "Transcription methods" refer to systems and software used to convert recorded audio data into text data.

[1160] An "apology response" refers to a statement expressing regret for a customer's or viewer's complaint and outlining actions taken to resolve the issue.

[1161] "Automated analysis" refers to the process of using computers to analyze data and understand its content and intent.

[1162] This invention is a system for providing a quick and accurate response to complaints and inquiries from viewers and customers. This system consists of the following main components:

[1163] 1. Speech Recognition Method: Receive and digitally store voice complaints from customers. This involves using APIs or software for speech recognition (e.g., Google Speech Recognition API).

[1164] 2. Natural Language Processing Methods: Analyze the content of received audio data and complaint / inquiry emails to understand the intentions of customers and viewers. Specifically, utilize natural language processing models (e.g., Hugging Face's transformers library).

[1165] 3. Response generation means: Based on the analyzed complaint content, appropriate apology responses and reply content are automatically generated. This enables a quick and appropriate response.

[1166] 4. Recording method: The voice of the complaint is recorded and sent to the server. A recording device or telephone system is responsible for this role.

[1167] 5. Transcription method: Convert the recorded audio data into text data. This is done using a speech recognition API.

[1168] Server Role

[1169] The server plays a central role in processing the received audio and text data. First, it uses speech recognition to convert the audio into text. Then, it uses natural language processing to analyze the text and understand what the customer or audience wants. Finally, it uses response generation to automatically generate appropriate replies or apologies and send them to the terminal.

[1170] Terminal role

[1171] The terminal receives data sent from the server and provides it to the user. It also plays a role in recording the voice of the complaint and sending it to the server. Furthermore, it displays the generated response or reply content, which the user can use as a reference when handling the complaint.

[1172] User roles

[1173] Users are operators or staff who handle complaints from viewers and customers. They are required to use the generated responses and reply content as a reference to take appropriate action.

[1174] Specific example

[1175] For example, if a viewer sends a complaint such as "the video keeps cutting out" via voice message within the app, the following response is automatically generated:

[1176] Viewer: "The video keeps cutting out."

[1177] App: "We apologize for the issue you are experiencing. We are currently working on a solution."

[1178] In this way, this system can respond quickly and appropriately to viewer and customer complaints and inquiries, contributing to improved customer satisfaction.

[1179] Example of a prompt

[1180] "The video keeps cutting out."

[1181] "No sound."

[1182] "My subscription is not being renewed."

[1183] This allows for accurate support to be provided for specific complaints and inquiries.

[1184] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1185] Step 1:

[1186] The terminal receives audio complaints from viewers or customers. The received audio complaints are recorded digitally through the terminal's built-in microphone or an externally connected microphone.

[1187] Input: Audience or customer voice complaint

[1188] Output: Digital audio data

[1189] Step 2:

[1190] The terminal sends the recorded audio data to the server. The transmitted audio data is converted to an appropriate format on the server side and made ready for processing by speech recognition software.

[1191] Input: Digital audio data

[1192] Output: Audio data sent to the server

[1193] Step 3:

[1194] The server converts the audio data into text data using speech recognition technology. This process utilizes tools such as the Google Speech Recognition API to convert speech to text.

[1195] Input: Audio data sent to the server

[1196] Output: Transcribed text data

[1197] Step 4:

[1198] The server analyzes the transcribed text using natural language processing techniques. It uses libraries such as Hugging Face's transformers to understand the claims in the text and analyze the intent of the customer or audience.

[1199] Input: Transcribed text data

[1200] Output: Analysis results of the claim content

[1201] Step 5:

[1202] The server generates appropriate apology responses and reply content based on the analysis results. In this process, a generative AI model is used to automatically generate response texts by providing appropriate prompt sentences as input.

[1203] Input: Analysis results of the claim

[1204] Output: Automated apology response or reply content

[1205] Step 6:

[1206] The server sends the generated apology or reply to the terminal. The terminal displays this reply on the operator's or staff's screen, allowing them to use it as a reference when taking action.

[1207] Input: Automated apology or reply content

[1208] Output: Response sent to the terminal

[1209] Step 7:

[1210] Users review the generated apology or reply displayed on their device and use it to respond to viewers or customers. They can also edit the reply content at their discretion if necessary.

[1211] Input: The generated apology or reply displayed on the device.

[1212] Output: Edited response content and response to the audience or customer

[1213] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1214] This invention relates to a system for improving the efficiency of complaint handling and enhancing customer satisfaction in customer centers. This system not only analyzes the content of complaint calls and complaint / inquiry emails, but also incorporates an emotion engine to recognize the user's emotions, enabling it to automatically generate more appropriate apology responses and reply messages.

[1215] A telephone complaint handling system incorporating an emotion engine.

[1216] Receiving and recording complaint calls

[1217] When a customer service operator receives a complaint call, the operator's PC or telephone system records the audio. The recorded audio data is transmitted to a server in real time.

[1218] Speech recognition and transcription

[1219] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This allows the content of the claim to be obtained as text.

[1220] Text and sentiment analysis

[1221] The server analyzes the transcribed text using a natural language processing (NLP) model to understand the complaint. During this process, the server also uses an emotion engine to recognize the user's emotions, for example, whether the customer is angry or sad.

[1222] Generating an apology response

[1223] Based on the understood complaint and emotions, the server automatically generates an appropriate apology response. For example, if a complaint about "delayed delivery" and the emotion of "anger" are recognized, an emphatic apology such as "We sincerely apologize. We will immediately investigate the issue and take appropriate action" will be generated for the customer.

[1224] Providing a response

[1225] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can use this response as a reference to proceed with handling the complaint.

[1226] Inquiry email response system with integrated emotion engine

[1227] Receiving and analyzing emails

[1228] When a customer service staff member (a user) receives a complaint or inquiry email, the terminal sends the email body to the server. The server uses an NLP model to analyze the email's content. Here, it understands what specific problem or inquiry the email represents.

[1229] Analysis of email content and emotions

[1230] The server then uses an emotion engine to recognize the user's emotions from the email content. For example, it extracts emotions such as "dissatisfaction" or "disappointment" from the email body.

[1231] Generating a reply

[1232] Based on the analysis results and emotions, the server automatically generates an appropriate response. For example, if a complaint that "the product was broken" and the emotion of "disappointment" are recognized, a response such as "We are very sorry for the inconvenience. We will arrange for a replacement immediately" will be generated.

[1233] Providing a reply

[1234] The generated reply is sent to the device and displayed in the staff member's email client. The staff member can review the reply, edit it as needed, and then send back the email.

[1235] Specific example

[1236] Examples of complaint calls

[1237] User: The operator answers the phone.

[1238] Terminal: Records the call content and sends the audio data to the server.

[1239] Server: Performs transcription using a speech recognition API.

[1240] Server: Uses an NLP model to analyze text and sentiment to understand the content and emotions behind the complaint.

[1241] Server: Generates an appropriate apology response and sends it to the terminal.

[1242] User: Use the generated response as a reference to handle the complaint.

[1243] Example of a complaint email

[1244] User: Staff receive the email.

[1245] Terminal: Sends the email body to the server.

[1246] Server: Analyzes content and emotion using an NLP model to understand the content and emotions of the complaint.

[1247] Server: Generates an appropriate reply and sends it to the terminal.

[1248] User: Review the generated reply before sending your email.

[1249] These specific processing steps enable this system to contribute to improving the efficiency of customer support and customer satisfaction. The incorporation of an emotion engine allows for more appropriate responses that take customer emotions into consideration.

[1250] The following describes the processing flow.

[1251] A telephone complaint handling system incorporating an emotion engine.

[1252] Step 1:

[1253] A user (a customer service operator) receives a complaint call.

[1254] The operator answers the phone using the usual procedure.

[1255] Step 2:

[1256] The terminal (operator's PC or telephone system) records the call.

[1257] When a call is initiated, the audio is automatically recorded.

[1258] Step 3:

[1259] The device transmits the recorded audio data to the server in real time.

[1260] The recorded audio data is divided into packets and sent to the server.

[1261] Step 4:

[1262] The server sends the received audio data to a speech recognition API for transcription.

[1263] The audio data is converted to text format via a speech recognition API.

[1264] Step 5:

[1265] The server sends the transcribed text to a natural language processing (NLP) model for analysis.

[1266] The NLP model extracts the content of the complaint and important keywords.

[1267] Step 6:

[1268] The server uses an emotion engine to recognize the user's emotions from the transcribed text.

[1269] The emotion engine analyzes emotional expressions within the text to identify the customer's emotions (e.g., anger, sadness, confusion, etc.).

[1270] Step 7:

[1271] The server generates an appropriate apology response based on the analysis results and the recognized emotions.

[1272] Responses are generated using templates or AI models tailored to the nature of the complaint and the customer's emotions. For example, if a customer is "angry because their delivery is delayed," an emphatic apology such as "We sincerely apologize. We will investigate and address the issue immediately" will be generated.

[1273] Step 8:

[1274] The server generates an apology response, sends it to the terminal, and displays it to the user.

[1275] The generated apology response is displayed on the operator's monitor.

[1276] Step 9:

[1277] We will handle the complaint based on the apology response displayed to the user.

[1278] The operator uses the displayed response to proceed with the actual complaint handling. Responses can also be customized as needed.

[1279] Inquiry email response system with integrated emotion engine

[1280] Step 1:

[1281] The user (customer center staff) receives complaint and inquiry emails.

[1282] Check for new emails through your email client.

[1283] Step 2:

[1284] The device sends the email content to the server.

[1285] The email body and subject are transferred to the server via the API.

[1286] Step 3:

[1287] The server analyzes the email content using a natural language processing (NLP) model.

[1288] The NLP model analyzes the content and topic of the email.

[1289] Step 4:

[1290] The server uses an emotion engine to recognize the user's emotions from the email content.

[1291] The emotion engine analyzes emotional expressions within emails to identify the customer's emotions (e.g., dissatisfaction, disappointment, confusion, etc.).

[1292] Step 5:

[1293] The server automatically generates an appropriate response based on the analysis results and recognized emotions.

[1294] The system generates response messages using templates or AI models tailored to the nature of the complaint and the customer's feelings. For example, if a customer says, "I'm disappointed because the product is broken," a response such as, "We are very sorry for the inconvenience. We will arrange for a replacement immediately," might be generated.

[1295] Step 6:

[1296] The server generates a reply and sends it to the terminal for the user to see.

[1297] The generated reply will be displayed in the staff member's email client.

[1298] Step 7:

[1299] The user reviews the displayed reply and edits it as needed.

[1300] Staff will review the reply and make any necessary corrections.

[1301] Step 8:

[1302] The user sends an email with the revised reply.

[1303] Send the finalized response to the customer to complete the process.

[1304] As described above, by incorporating an emotion engine, the quality of complaint and inquiry handling can be improved, and responses that take customer emotions into consideration can be made. This system can promote the efficiency of customer support and improve customer satisfaction.

[1305] (Example 2)

[1306] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1307] Handling customer complaints in modern customer centers is time-consuming and labor-intensive. Furthermore, the quality of service provided by operators varies, making it difficult to consistently improve customer satisfaction. In particular, it is crucial to accurately understand customer emotions and respond with appropriate and courteous care. A method is needed to address these challenges efficiently and effectively.

[1308] The identification processing performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text to understand the content of the complaint, means for analyzing the transcribed text using a natural language processing model and an emotion analysis engine to understand the customer's emotions, means for generating an appropriate apology response based on the content of the complaint and the customer's emotions, and means for providing the generated apology response. This makes it possible to accurately grasp the content of the customer's complaint and emotions, and to quickly generate and provide an appropriate response accordingly. Furthermore, it is possible to reduce the burden on operators and ensure consistency in response quality, thereby improving customer satisfaction.

[1309] "Means for receiving customer voice complaints" refers to a function for electronically receiving complaints made by customers via telephone.

[1310] "Means for recording received audio" refers to a function for saving received audio data to a recording medium.

[1311] "Methods for transcribing recorded audio data" refers to functions that utilize speech recognition technology to convert recorded audio data into text data.

[1312] "Means for analyzing transcribed text and understanding the content of a claim" refers to a function that uses natural language processing technology to analyze text data obtained through transcription and recognize the content contained therein.

[1313] "A means of understanding customer emotions by analyzing transcribed text using natural language processing models and sentiment analysis engines" refers to a function that uses specific algorithms to analyze the linguistic structure and emotional aspects within the text and grasp the customer's emotional state.

[1314] "A means of generating an appropriate apology response based on the content of the complaint and the customer's feelings" refers to a function that automatically generates the optimal response text corresponding to the specific content of the recognized complaint and the customer's feelings.

[1315] "Means of providing a generated apology response" refers to a function for displaying the generated response to the operator or sending it directly to the customer.

[1316] "Means for receiving complaints and inquiries via email" refers to a function for receiving complaints and inquiries sent via email.

[1317] "A means of analyzing the content of received emails to understand complaints, inquiries, and customer sentiment" refers to a function that uses natural language processing and sentiment analysis technologies to analyze the text of received emails and grasp their content and customer sentiment.

[1318] "A means of generating appropriate response content based on complaints, inquiries, and customer sentiment" refers to a function that automatically generates the optimal response based on recognized email content and customer sentiment.

[1319] "A means of automatically generating response content and apology responses based on the content of the complaint and the customer's feelings using a generative AI model" refers to a function that uses an artificial intelligence model to generate appropriate responses to complaints and inquiries from specific prompt sentences.

[1320] "A means of instructing a generation AI model to provide an appropriate response using a prompt statement" refers to a function that provides a specific input (prompt) to the AI ​​model and generates an appropriate output based on that input.

[1321] This invention relates to a system for improving the efficiency of complaint handling and enhancing customer satisfaction in customer centers. Specifically, it analyzes both voice and email complaints and automatically generates appropriate apologies and replies.

[1322] System Configuration

[1323] This system consists of the following main elements:

[1324] 1. Users: Customer center operators and staff will be the primary users of the system.

[1325] 2. Terminals: This includes PCs, telephone systems, and email clients used by operators and staff.

[1326] 3. Server: This server runs the speech recognition API, natural language processing (NLP) models, sentiment analysis engine, and generative AI models.

[1327] Hardware and software to be used

[1328] Speech Recognition API: Use the Google Cloud Speech-to-Text API to convert speech data into text.

[1329] Natural Language Processing Models: Text data is analyzed using NLP models such as BERT and GPT-3.

[1330] Sentiment analysis engine: Analyzes emotions from text using tools such as IBM Watson Tone Analyzer.

[1331] Generative AI model: Uses GPT-3 and other models to generate appropriate apology responses and reply content.

[1332] Processing flow

[1333] Regarding the customer complaint phone handling system:

[1334] When a user receives a call, the device records the audio and sends it to the server in real time.

[1335] The server sends the audio data to the Google Cloud Speech-to-Text API, where it is converted into text data.

[1336] The server sends text data to an NLP model to analyze the claim details.

[1337] The server then uses an emotion analysis engine to analyze the customer's emotions.

[1338] Based on the nature of the complaint and the emotional state, the server uses a generative AI model to generate an appropriate apology response.

[1339] The generated apology response is sent to the device and displayed on the user's monitor.

[1340] Regarding the complaint email handling system:

[1341] When a user receives an email, the email body is sent from the device to the server.

[1342] The server uses an NLP model to analyze the email content and understand the complaint and emotions behind it.

[1343] Based on the analysis results, the server uses a generative AI model to generate appropriate responses.

[1344] The generated reply is sent to the device and displayed in the user's email client.

[1345] Examples of specific cases and prompt statements

[1346] For example, if an operator receives a complaint about a delivery delay:

[1347] User: The operator answers the phone.

[1348] Terminal: Records the call content and sends the audio data to the server.

[1349] Server: Performs transcription using a speech recognition API (e.g., Google Cloud Speech-to-Text).

[1350] Server: Uses NLP models to analyze text and sentiment to understand the content and emotions of complaints (e.g., BERT and IBM Watson Tone Analyzer).

[1351] Server: Generates an appropriate apology response and sends it to the terminal.

[1352] User: Use the generated response as a reference to handle the complaint.

[1353] Examples of prompt statements:

[1354] "We've received a complaint about a delivery delay. The customer is angry. Please generate an appropriate apology letter."

[1355] "We received an email inquiry about a damaged product. The customer is disappointed. Please generate an appropriate reply."

[1356] This system streamlines customer service complaint handling and enables appropriate responses that take customer emotions into consideration. This, in turn, leads to improved customer satisfaction.

[1357] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1358] Customer complaint phone handling system

[1359] Processing steps

[1360] Step 1:

[1361] A customer service operator, acting as the user, receives a complaint call. When the operator presses the "Start Call" button, the terminal begins recording the conversation. The recorded audio data is transmitted to the server in real time. The input is the call audio, and the output is the audio data stored on the server.

[1362] Step 2:

[1363] The server sends the received audio data to the Google Cloud Speech-to-Text API, where it is converted into text data. In this step, audio data is input and text data is generated as output. Specifically, the server sends an API request and receives a response.

[1364] Step 3:

[1365] The server sends the text data converted from the speech to a natural language processing (NLP) model such as BERT or GPT-3 to analyze the claim content. The input for this step is the transcribed text, and the analyzed claim content is output. Specifically, the server sends an API request to the NLP model and receives the analysis results.

[1366] Step 4:

[1367] The server then sends the analyzed text to a sentiment analysis engine, such as IBM Watson Tone Analyzer, to analyze the customer's emotions. In this step, the text containing the complaint is input, and the sentiment analysis results are obtained as output. The specific operation involves sending requests to the sentiment analysis API and receiving responses.

[1368] Step 5:

[1369] The server sends prompt messages to a generative AI model (e.g., GPT-3) based on the understood complaint content and sentiment state, generating an appropriate apology response. The input data is the complaint content and sentiment analysis results, and the output is the generated apology response. Specifically, the server defines appropriate prompt messages for the AI ​​model and sends an API request.

[1370] Step 6:

[1371] The generated apology response is sent to the terminal and displayed on the operator's monitor. In this step, the generated response text is input, and the displayed monitor is output. Specifically, a UI update request is made to display the response text.

[1372] Complaint email handling system

[1373] Processing steps

[1374] Step 1:

[1375] When a customer service staff member (a user) receives a complaint or inquiry email, the email body is sent from the terminal to the server. The input is the received email, and the output is the email data stored on the server. Specifically, the email client sends the email body via the API.

[1376] Step 2:

[1377] The server uses an NLP model to analyze the email content. The email body is input, and the complaint and inquiry details are output. Specifically, an API request is sent to analyze the text data, and the analysis results are received.

[1378] Step 3:

[1379] The server uses an emotion analysis engine to understand customer emotions during the process of analyzing email content and inquiry details. The input data is the email body, and the output data is the emotion analysis result. Specifically, it sends a request to the emotion analysis API and receives a response.

[1380] Step 4:

[1381] The server uses a generative AI model to generate an appropriate response based on the analyzed content and emotions. In this step, the complaint and customer emotions are used as input, and the output is the generated response. Specifically, a prompt is sent to the generative AI model to generate an appropriate response.

[1382] Step 5:

[1383] The generated reply is sent to the terminal and displayed in the staff member's email client. The input data is the generated reply text, and the output is the displayed email client. Specifically, a UI update request is executed to display the reply text.

[1384] Through the processing steps described above, this system contributes to improving the efficiency of customer support and enhancing customer satisfaction. By understanding customer emotions and responding appropriately and quickly, customer satisfaction can be increased.

[1385] (Application Example 2)

[1386] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[1387] In modern customer service centers, responding quickly and appropriately to customer complaints and inquiries is a crucial challenge. However, especially when complaints are emotional, it is often difficult for staff to accurately understand those emotions and respond appropriately. Therefore, there is a need for systems that can improve customer satisfaction and increase the efficiency of complaint handling.

[1388] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1389] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text to understand the content of the complaint and the customer's feelings, means for generating an appropriate apology response based on the content of the complaint and the customer's feelings, and means for providing the generated apology response. This enables appropriate responses that take into account the customer's feelings.

[1390] "Means of understanding customer emotions" refers to the process of analyzing and recognizing the emotions behind customer words, such as anger, disappointment, or satisfaction, using natural language processing technology.

[1391] A "means for generating apology responses" is a system that automatically generates appropriate and empathetic apologies and responses based on the analyzed complaint content and customer emotions.

[1392] "Methods for generating response content" refers to a process that automatically constructs appropriate responses to customers, taking into account complaints, inquiries, and related emotions.

[1393] A "natural language processing model" is a machine learning model that analyzes text data and uses that information to understand meaning and intent.

[1394] A "generative AI model" is an artificial intelligence model that automatically generates text or answers based on given prompts or input data.

[1395] A "prompt" is an instruction or input sentence used by a generative AI model to generate a response, and it serves as a guideline for providing appropriate output based on its content.

[1396] This invention is a system that supports customer service in physical stores. Its purpose is to enable store staff to respond quickly and accurately to customer complaints and inquiries. A specific example of the system is described below.

[1397] Hardware and software to be used

[1398] The system uses the following hardware and software.

[1399] Hardware: Smartphones, smart glasses, head-mounted displays

[1400] Software: Python, TensorFlow, Transformers package, SpeechRecognition package

[1401] System Configuration

[1402] 1. Means of receiving voice complaints from customers:

[1403] Staff members wear smartphones, smart glasses, or head-mounted displays to receive customer complaints and inquiries.

[1404] 2. Means for recording received audio:

[1405] The received audio data is recorded on the staff member's terminal and sent to the server in real time.

[1406] 3. Methods for transcribing recorded audio data:

[1407] The server converts the received audio data into text data using the SpeechRecognition package. This allows the content of complaints and inquiries to be obtained as text.

[1408] 4. Means for analyzing transcribed text to understand the content of the complaint and the customer's feelings:

[1409] The server analyzes the converted text data using the Transformers package and a natural language processing model to understand the content of the complaint, while simultaneously recognizing the customer's emotions using an emotion engine.

[1410] 5. Means for generating an appropriate apology response based on the nature of the complaint and the customer's feelings:

[1411] Based on the analyzed complaint details and customer emotions, the server uses a generative AI model to generate an appropriate apology response. For example, if the complaint is "the product was broken" and the customer's emotion is recognized as "anger," an apology response such as "We sincerely apologize. We will arrange for a replacement immediately" will be generated.

[1412] 6. Means of providing the generated apology response:

[1413] The generated apology response is sent from the server to the staff member's terminal and displayed on their smartphone, smart glasses, or head-mounted display. The staff member uses this response as a reference when dealing with the customer.

[1414] Example of a prompt

[1415] Examples of prompt statements that can actually be used include the following:

[1416] "Analyze the customer's emotions from the following text: 'The product was broken. Please do something about it.'"

[1417] Specific example

[1418] Examples of customer service:

[1419] When a staff member receives a customer complaint using smart glasses, the audio is recorded by the smart glasses and sent to a server in real time. The server converts the received audio into text and analyzes the emotions. Then, an appropriate apology response is displayed on the staff member's glasses. The staff member uses the displayed apology response to address the customer and resolve the issue quickly.

[1420] Thus, the present invention aims to improve customer satisfaction in physical stores by enabling appropriate and prompt responses that take customer emotions into consideration.

[1421] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1422] Step 1:

[1423] Receive voice complaints from customers.

[1424] Input: A customer makes a verbal complaint to a store staff member.

[1425] Specific operation: The store staff member's smartphone, smart glasses, or head-mounted display receives audio in real time.

[1426] Output: Received audio data.

[1427] Step 2:

[1428] Record the received audio.

[1429] Input: Audio data received in Step 1.

[1430] Specific operation: The device (smartphone, smart glasses, head-mounted display) records audio and sends the recorded data to the server.

[1431] Output: Recorded audio data.

[1432] Step 3:

[1433] Transcribe the recorded audio data into text.

[1434] Input: Audio data recorded in Step 2.

[1435] Specific operation: The server uses a speech recognition API to convert the audio data into text data.

[1436] Output: Transcripted text data.

[1437] Step 4:

[1438] The transcribed text is analyzed to understand the content of the complaint and the customer's feelings.

[1439] Input: Text data obtained in Step 3.

[1440] Specific operation: The server uses a natural language processing (NLP) model to analyze the content of the complaint from the text data. At the same time, it uses an emotion engine to recognize the customer's emotions.

[1441] Output: Analyzed complaint details and perceived customer sentiment.

[1442] Step 5:

[1443] Based on the nature of the complaint and the customer's feelings, generate an appropriate apology response.

[1444] Input: Complaint details and customer sentiment analyzed in Step 4.

[1445] Specific operation: The server uses an AI model to generate an appropriate apology response based on the entered complaint content and emotions. For example, if the complaint is "the product was broken" and the emotion "anger" is recognized, an apology response such as "We sincerely apologize. We will arrange for a replacement immediately." will be generated.

[1446] Output: Generated apology response.

[1447] Step 6:

[1448] Provide a generated apology response.

[1449] Input: The apology response generated in Step 5.

[1450] Specific operation: The server sends the generated apology response to the staff member's device (smartphone, smart glasses, or head-mounted display). The staff member can then use the apology response as a reference when dealing with the customer.

[1451] Output: The apology response displayed on the terminal.

[1452] This allows the system to provide quick and appropriate responses in physical stores, taking customer emotions into consideration, thereby improving customer satisfaction.

[1453] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1454] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1455] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[1456] [Fourth Embodiment]

[1457] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1458] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1459] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1460] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[1461] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1462] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1463] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1464] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[1465] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1466] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1467] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1468] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1469] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1470] This invention relates to a system for streamlining complaint handling in customer service centers. The system aims to analyze the content of complaint calls and complaint / inquiry emails, and automatically generate and provide appropriate apologies and replies. The main components of this system include speech recognition, natural language processing (NLP), and automated response generation functions.

[1471] Telephone-based complaint handling system

[1472] Receiving and recording complaint calls

[1473] When a customer service operator receives a complaint call, the operator's PC or telephone system records the audio. The recorded audio data is transmitted to a server in real time.

[1474] Speech recognition and transcription

[1475] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This allows the content of the claim to be obtained as text.

[1476] Text analysis

[1477] The server analyzes the transcribed text using a natural language processing (NLP) model to understand the complaint. This analysis extracts the specific points of what the customer is unhappy about.

[1478] Generating an apology response

[1479] Based on the understood complaint, the server automatically generates an appropriate apology response. For example, in response to a complaint that "delivery is delayed," a response such as "We are very sorry for the delay, sir / madam. We are currently checking the delivery status, so please wait a moment." will be generated.

[1480] Providing a response

[1481] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can use this response as a reference to proceed with handling the complaint.

[1482] Complaint and inquiry email handling system

[1483] Receiving and analyzing emails

[1484] When a customer service staff member (a user) receives a complaint or inquiry email, the terminal sends the email body to the server. The server uses an NLP model to analyze the email's content. Here, it understands what specific problem or inquiry the email represents.

[1485] Generating a reply

[1486] Based on the analysis results, the server automatically generates an appropriate response. For example, in response to a complaint that "the product was damaged," a response such as "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product." will be generated.

[1487] Providing a reply

[1488] The generated reply is sent to the device and displayed in the staff member's email client. The staff member can review the reply, edit it as needed, and then send back the email.

[1489] Specific example

[1490] Examples of complaint calls

[1491] User: The operator answers the phone.

[1492] Terminal: Records the call content and sends the audio data to the server.

[1493] Server: Performs transcription using a speech recognition API.

[1494] Server: Analyzes text using an NLP model to understand the content of the claim.

[1495] Server: Generates an appropriate apology response and sends it to the terminal.

[1496] User: Use the generated response as a reference to handle the complaint.

[1497] Example of a complaint email

[1498] User: Staff receive the email.

[1499] Terminal: Sends the email body to the server.

[1500] Server: Analyzes the content using an NLP model to understand the claim.

[1501] Server: Generates an appropriate reply and sends it to the terminal.

[1502] User: Review the generated reply before sending your email.

[1503] In this way, this system improves the efficiency of customer support operations and enhances customer satisfaction. Furthermore, it reduces variations in the quality of service provided by individual staff members, enabling the delivery of consistently high-quality support.

[1504] The following describes the processing flow.

[1505] Handling complaint calls

[1506] Step 1:

[1507] A user (a customer service operator) receives a complaint call.

[1508] An operator receives a call from a customer.

[1509] Step 2:

[1510] The terminal (operator's PC or telephone system) records the call.

[1511] The call is initiated, and the audio data is recorded in real time.

[1512] Step 3:

[1513] The device transmits the recorded audio data to the server in real time.

[1514] The recorded audio is compressed and sent to the server.

[1515] Step 4:

[1516] The server sends the received audio data to a speech recognition API for transcription.

[1517] The speech data is converted into text data using a speech recognition API.

[1518] Step 5:

[1519] The server sends the transcribed text to a natural language processing (NLP) model for analysis.

[1520] We use an NLP model to extract the types of complaints and key keywords.

[1521] Step 6:

[1522] The server generates an appropriate apology response based on the analysis results.

[1523] For each claim, a response is generated using a pre-prepared template or an AI model.

[1524] Step 7:

[1525] The server generates an apology response, sends it to the terminal, and displays it to the user.

[1526] An appropriate apology response is displayed on the operator's monitor.

[1527] Step 8:

[1528] We will handle the complaint based on the apology response displayed to the user.

[1529] The operator will then proceed with handling the actual complaint based on the displayed response.

[1530] Handling complaints and inquiry emails

[1531] Step 1:

[1532] The user (customer center staff) receives complaint and inquiry emails.

[1533] Check your email client for new emails.

[1534] Step 2:

[1535] The device sends the email content to the server.

[1536] The email body and subject are forwarded to the server via the API.

[1537] Step 3:

[1538] The server analyzes the email content using a natural language processing (NLP) model.

[1539] The content of emails is analyzed to extract specific details of complaints and inquiries.

[1540] Step 4:

[1541] The server automatically generates an appropriate response based on the analysis results.

[1542] The response content is generated from an AI model or template.

[1543] Step 5:

[1544] The server generates a reply and sends it to the terminal for the user to see.

[1545] The reply will appear in the staff member's email client.

[1546] Step 6:

[1547] The user reviews the displayed reply and edits it as needed.

[1548] Staff will review the reply and make any necessary corrections.

[1549] Step 7:

[1550] The user sends an email with the revised reply.

[1551] Send the confirmed reply to the customer.

[1552] Through these specific processing steps, the present invention can significantly improve the efficiency of customer support and reduce the burden on staff.

[1553] (Example 1)

[1554] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1555] In customer service settings, a tremendous amount of time and effort is spent handling complaints and inquiries. This can lead to inconsistencies in service quality among staff members, potentially lowering customer satisfaction. Furthermore, delays in situations where a quick response is required can further increase customer dissatisfaction. Therefore, there is a need to handle complaints and inquiries efficiently and effectively, and to achieve consistent, high-quality service.

[1556] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1557] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voices, and means for transcribing the recorded voice data. This allows the content of the complaints to be efficiently converted into text data, making it possible to analyze them.

[1558] The server also includes means for analyzing the transcribed text and understanding the content of the claim, means for generating an appropriate apology response based on the content of the claim, and means for providing the generated apology response. This enables the rapid generation of appropriate responses to claims and reduces variability in response quality.

[1559] Furthermore, the server includes means for transmitting recorded audio data in real time, means for converting audio to text using a speech recognition API, means for analyzing text using a natural language processing model, means for automatically generating apology responses using a generative AI model, means for analyzing email content, and means for generating reply content. This enables consistent, high-quality, and efficient complaint handling in both voice calls and email correspondence.

[1560] A "customer" is someone who receives a service and makes complaints or inquiries to the service provider.

[1561] "Voice complaints" refer to expressions of dissatisfaction or complaints made by customers using voice communication methods such as telephone calls.

[1562] "Means of recording" refers to devices and software used to record audio data in digital format.

[1563] "Methods for transcription" refer to technologies and devices used to convert audio data into text data.

[1564] "Means of analyzing text" refers to technologies and devices that use natural language processing techniques to understand text data and analyze its content.

[1565] "Means for generating apology responses" refers to technologies or devices that automatically create appropriate apology statements based on the content of a complaint.

[1566] "Means of providing generated apology responses" refers to devices or software that display and make available the generated apology messages to operators and staff.

[1567] "Means of sending to a server in real time" refers to technologies and devices for instantly transferring recorded audio data to a server.

[1568] A "speech recognition API" refers to an application programming interface (API) provided as an external service for converting speech data into text data.

[1569] A "natural language processing model" refers to a machine learning model designed to understand and analyze human language.

[1570] A "generative AI model" refers to an artificial intelligence (AI) model that generates text based on given input data.

[1571] "Complaint and inquiry emails" refer to complaints or questions that customers send via email.

[1572] "Means of analyzing email content" refers to technologies and devices that analyze the body of a received email and understand its content.

[1573] "Means for generating reply content" refers to technologies or devices that automatically create appropriate reply messages based on the content of an email.

[1574] This invention is a system for streamlining complaint and inquiry handling in customer service. This system enables customers to receive prompt, consistent, and high-quality responses to complaints and inquiries. Specific embodiments for implementing this invention are described below.

[1575] Receiving and recording complaint calls

[1576] When a user makes a complaint call, the operator's PC or the telephone system records the audio. The recorded audio data is transmitted to a server in real time. The recording function is implemented using digital recording devices or software integrated into the telephone system.

[1577] Speech recognition and transcription

[1578] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This speech recognition process uses an external speech recognition service, such as the Google Cloud Speech-to-Text API. Based on the speech recognition results returned by the API, the call content is retrieved as text data.

[1579] Text analysis

[1580] The server feeds the acquired text data into a natural language processing (NLP) model to analyze the complaint content. Advanced NLP models such as GPT-4 are used for this process. The NLP model analysis clearly extracts customer dissatisfaction and requests. For example, specific complaints such as "delivery is delayed" are identified.

[1581] Generating an apology response

[1582] The server generates an appropriate apology response based on the analysis results. A generative AI model is used to create the best possible apology based on the input text data. For example, a response such as, "We sincerely apologize for the delay, customer. We are currently checking the delivery status, so please wait a moment," might be generated.

[1583] Providing a response

[1584] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can then use the displayed apology as a reference to proceed with customer service.

[1585] Receiving and analyzing complaint and inquiry emails

[1586] When a user receives a complaint or inquiry email, the terminal sends the email body to the server. The server analyzes the email content using an NLP model. Here too, natural language processing models such as GPT-4 are used. The specific problems and inquiries in the email are clarified.

[1587] Generating a reply

[1588] The server generates an appropriate reply based on the analysis results. A generation AI model is used to generate an email reply such as, "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product."

[1589] Providing a reply

[1590] The generated reply is sent to the device and displayed in the staff member's email client. The staff member reviews the displayed reply, edits it as needed, and then replies to the email.

[1591] Specific example

[1592] Examples of complaint calls

[1593] User: The operator calls and says, "Your delivery is delayed."

[1594] Terminal: Records call content and sends it to the server in real time.

[1595] Server: Uses a speech recognition API to convert speech data into text.

[1596] Server: Uses an NLP model to analyze text data and understand the content of the claims.

[1597] Server: Generates an appropriate apology response and sends it to the terminal.

[1598] User: Use the generated apology response as a reference to proceed with handling the complaint.

[1599] Example of a complaint email

[1600] User: Staff receive an email stating that "the product was damaged."

[1601] Terminal: Sends the email body to the server.

[1602] Server: Uses an NLP model to analyze email content and understand the nature of the complaint.

[1603] Server: Generates an appropriate reply and sends it to the terminal.

[1604] User: Review the generated reply and send a reply email.

[1605] Examples of specific prompt messages include, "Please generate an apology letter to handle a delivery delay complaint," or "Please create an email regarding the replacement of a damaged product."

[1606] In this way, the system improves the efficiency of customer support operations and enhances customer satisfaction. Furthermore, it reduces variations in the quality of service provided by individual staff members, enabling the delivery of consistently high-quality support.

[1607] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1608] Step 1: Receiving and recording complaint calls

[1609] A user receives a complaint call. Here, the user is a customer service operator. The terminal (the operator's PC or telephone system) records the call. Specifically, when the operator answers the call, the telephone system starts recording simultaneously with the start of the call. The recorded audio data is transmitted to the server in real time. The input is the customer's voice call, and the output is the recorded audio data.

[1610] Step 2: Speech recognition and transcription

[1611] The server sends the received audio data to a speech recognition API. For example, the Google Cloud Speech-to-Text API is used here. The server receives the API's processing result and converts the audio into text data. The input is audio data, and the output is text data. Specifically, the server sends the audio data to the speech recognition API and stores the resulting string data.

[1612] Step 3: Text Analysis

[1613] The server inputs the acquired text data into a natural language processing (NLP) model, which then analyzes the text. For example, the GPT-4 NLP model is used. The server then analyzes the results returned by the NLP model to understand the specific content of the claim. The input is the transcribed text data, and the output is the analysis result (the key points of the claim). Specifically, key points such as "delivery is delayed" and "order number 12345" are extracted.

[1614] Step 4: Generating an apology response

[1615] The server generates an appropriate apology response based on the analysis results from the NLP model. A generative AI model is used for this purpose. For example, it might generate a response such as, "We sincerely apologize for the delay. We are currently checking the delivery status, so please wait a moment." The input is the analysis results, and the output is the generated apology response. Specifically, a prompt for generating the apology response is input to the generative AI model, and the resulting apology message is retrieved.

[1616] Step 5: Provide a response

[1617] The server generates an apology response and sends it to the terminal. The terminal displays the received apology response on the operator's monitor. This allows the operator to refer to the generated apology and proceed with customer service. The input is the generated apology response, and the output is the text displayed on the operator's monitor. Specifically, the terminal executes the process of displaying the apology response received from the server on its screen.

[1618] Step 6: Receiving and analyzing complaint and inquiry emails

[1619] A user receives a complaint or inquiry email. Here, the user is a staff member at a customer service center. The terminal sends the received email body to the server. The server analyzes the email content using an NLP model. Specifically, it uses GPT-4 or similar to input the email text and analyzes its content. The input is the email body, and the output is the analysis result (the main points of the complaint or inquiry).

[1620] Step 7: Generating the reply content

[1621] The server generates an appropriate response based on the analysis results. A generative AI model is used for this generation. For example, a response such as "We sincerely apologize for the inconvenience caused. We will promptly replace the damaged product." might be generated. The input is the analysis results, and the output is the generated response. Specifically, the server prompts the generative AI model based on the analysis results and retrieves the generated response.

[1622] Step 8: Provide a reply

[1623] The server generates a reply and sends it to the terminal. The terminal displays this reply in the staff member's email client. This allows the staff member to review the generated reply, edit it as needed, and send the email. The input is the generated reply, and the output is the text displayed in the staff member's email client. Specifically, the terminal executes the process of displaying the reply received from the server in the email client.

[1624] (Application Example 1)

[1625] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1626] Traditional customer support systems face the challenge of responding quickly and appropriately to customer complaints and inquiries. Similarly, timely and accurate responses are required for interactive complaints and inquiries from viewers. This can easily lead to decreased customer satisfaction and inconsistent support quality.

[1627] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1628] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text and understanding the content of the complaint, means for generating an appropriate apology response based on the content of the complaint, means for providing the generated apology response, and means for automatically analyzing complaints and inquiries from viewers and automatically generating appropriate responses, including voice recognition means, natural language processing means, and response generation means. This enables a rapid, consistent, and high-quality response to complaints and inquiries from customers and viewers.

[1629] "Customer" refers to anyone who purchases or uses a product or service.

[1630] "Viewer" refers to anyone who watches or listens to videos, music, or other content provided through content distribution services.

[1631] A "complaint" refers to a customer or viewer reporting dissatisfaction or problems with a product or service.

[1632] "Inquiry" refers to a question or request from a customer or viewer seeking information or support.

[1633] "Speech recognition means" refers to systems and devices that analyze speech as digital data and convert it into text data.

[1634] "Natural language processing methods" refer to technologies and algorithms used to analyze and understand human language using computers.

[1635] "Response generation means" refers to algorithms or systems that automatically generate appropriate responses or answers based on analyzed claims or inquiries.

[1636] "Recording means" refers to devices or systems for saving received audio in digital format.

[1637] "Transcription methods" refer to systems and software used to convert recorded audio data into text data.

[1638] An "apology response" refers to a statement expressing regret for a customer's or viewer's complaint and outlining actions taken to resolve the issue.

[1639] "Automated analysis" refers to the process of using computers to analyze data and understand its content and intent.

[1640] This invention is a system for providing a quick and accurate response to complaints and inquiries from viewers and customers. This system consists of the following main components:

[1641] 1. Speech Recognition Method: Receive and digitally store voice complaints from customers. This involves using APIs or software for speech recognition (e.g., Google Speech Recognition API).

[1642] 2. Natural Language Processing Methods: Analyze the content of received audio data and complaint / inquiry emails to understand the intentions of customers and viewers. Specifically, utilize natural language processing models (e.g., Hugging Face's transformers library).

[1643] 3. Response generation means: Based on the analyzed complaint content, appropriate apology responses and reply content are automatically generated. This enables a quick and appropriate response.

[1644] 4. Recording method: The voice of the complaint is recorded and sent to the server. A recording device or telephone system is responsible for this role.

[1645] 5. Transcription method: Convert the recorded audio data into text data. This is done using a speech recognition API.

[1646] Server Role

[1647] The server plays a central role in processing the received audio and text data. First, it uses speech recognition to convert the audio into text. Then, it uses natural language processing to analyze the text and understand what the customer or audience wants. Finally, it uses response generation to automatically generate appropriate replies or apologies and send them to the terminal.

[1648] Terminal role

[1649] The terminal receives data sent from the server and provides it to the user. It also plays a role in recording the voice of the complaint and sending it to the server. Furthermore, it displays the generated response or reply content, which the user can use as a reference when handling the complaint.

[1650] User roles

[1651] Users are operators or staff who handle complaints from viewers and customers. They are required to use the generated responses and reply content as a reference to take appropriate action.

[1652] Specific example

[1653] For example, if a viewer sends a complaint such as "the video keeps cutting out" via voice message within the app, the following response is automatically generated:

[1654] Viewer: "The video keeps cutting out."

[1655] App: "We apologize for the issue you are experiencing. We are currently working on a solution."

[1656] In this way, this system can respond quickly and appropriately to viewer and customer complaints and inquiries, contributing to improved customer satisfaction.

[1657] Example of a prompt

[1658] "The video keeps cutting out."

[1659] "No sound."

[1660] "My subscription is not being renewed."

[1661] This allows for accurate support to be provided for specific complaints and inquiries.

[1662] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1663] Step 1:

[1664] The terminal receives audio complaints from viewers or customers. The received audio complaints are recorded digitally through the terminal's built-in microphone or an externally connected microphone.

[1665] Input: Audience or customer voice complaint

[1666] Output: Digital audio data

[1667] Step 2:

[1668] The terminal sends the recorded audio data to the server. The transmitted audio data is converted to an appropriate format on the server side and made ready for processing by speech recognition software.

[1669] Input: Digital audio data

[1670] Output: Audio data sent to the server

[1671] Step 3:

[1672] The server converts the audio data into text data using speech recognition technology. This process utilizes tools such as the Google Speech Recognition API to convert speech to text.

[1673] Input: Audio data sent to the server

[1674] Output: Transcribed text data

[1675] Step 4:

[1676] The server analyzes the transcribed text using natural language processing techniques. It uses libraries such as Hugging Face's transformers to understand the claims in the text and analyze the intent of the customer or audience.

[1677] Input: Transcribed text data

[1678] Output: Analysis results of the claim content

[1679] Step 5:

[1680] The server generates appropriate apology responses and reply content based on the analysis results. In this process, a generative AI model is used to automatically generate response texts by providing appropriate prompt sentences as input.

[1681] Input: Analysis results of the claim

[1682] Output: Automated apology response or reply content

[1683] Step 6:

[1684] The server sends the generated apology or reply to the terminal. The terminal displays this reply on the operator's or staff's screen, allowing them to use it as a reference when taking action.

[1685] Input: Automated apology or reply content

[1686] Output: Response sent to the terminal

[1687] Step 7:

[1688] Users review the generated apology or reply displayed on their device and use it to respond to viewers or customers. They can also edit the reply content at their discretion if necessary.

[1689] Input: The generated apology or reply displayed on the device.

[1690] Output: Edited response content and response to the audience or customer

[1691] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1692] This invention relates to a system for improving the efficiency of complaint handling and enhancing customer satisfaction in customer centers. This system not only analyzes the content of complaint calls and complaint / inquiry emails, but also incorporates an emotion engine to recognize the user's emotions, enabling it to automatically generate more appropriate apology responses and reply messages.

[1693] A telephone complaint handling system incorporating an emotion engine.

[1694] Receiving and recording complaint calls

[1695] When a customer service operator receives a complaint call, the operator's PC or telephone system records the audio. The recorded audio data is transmitted to a server in real time.

[1696] Speech recognition and transcription

[1697] The server sends the received audio data to a speech recognition API, which converts the audio into text data. This allows the content of the claim to be obtained as text.

[1698] Text and sentiment analysis

[1699] The server analyzes the transcribed text using a natural language processing (NLP) model to understand the complaint. During this process, the server also uses an emotion engine to recognize the user's emotions, for example, whether the customer is angry or sad.

[1700] Generating an apology response

[1701] Based on the understood complaint and emotions, the server automatically generates an appropriate apology response. For example, if a complaint about "delayed delivery" and the emotion of "anger" are recognized, an emphatic apology such as "We sincerely apologize. We will immediately investigate the issue and take appropriate action" will be generated for the customer.

[1702] Providing a response

[1703] The generated apology response is sent to the terminal and displayed on the operator's monitor. The operator can use this response as a reference to proceed with handling the complaint.

[1704] Inquiry email response system with integrated emotion engine

[1705] Receiving and analyzing emails

[1706] When a customer service staff member (a user) receives a complaint or inquiry email, the terminal sends the email body to the server. The server uses an NLP model to analyze the email's content. Here, it understands what specific problem or inquiry the email represents.

[1707] Analysis of email content and emotions

[1708] The server then uses an emotion engine to recognize the user's emotions from the email content. For example, it extracts emotions such as "dissatisfaction" or "disappointment" from the email body.

[1709] Generating a reply

[1710] Based on the analysis results and emotions, the server automatically generates an appropriate response. For example, if a complaint that "the product was broken" and the emotion of "disappointment" are recognized, a response such as "We are very sorry for the inconvenience. We will arrange for a replacement immediately" will be generated.

[1711] Providing a reply

[1712] The generated reply is sent to the device and displayed in the staff member's email client. The staff member can review the reply, edit it as needed, and then send back the email.

[1713] Specific example

[1714] Examples of complaint calls

[1715] User: The operator answers the phone.

[1716] Terminal: Records the call content and sends the audio data to the server.

[1717] Server: Performs transcription using a speech recognition API.

[1718] Server: Uses an NLP model to analyze text and sentiment to understand the content and emotions behind the complaint.

[1719] Server: Generates an appropriate apology response and sends it to the terminal.

[1720] User: Use the generated response as a reference to handle the complaint.

[1721] Example of a complaint email

[1722] User: Staff receive the email.

[1723] Terminal: Sends the email body to the server.

[1724] Server: Analyzes content and emotion using an NLP model to understand the content and emotions of the complaint.

[1725] Server: Generates an appropriate reply and sends it to the terminal.

[1726] User: Review the generated reply before sending your email.

[1727] These specific processing steps enable this system to contribute to improving the efficiency of customer support and customer satisfaction. The incorporation of an emotion engine allows for more appropriate responses that take customer emotions into consideration.

[1728] The following describes the processing flow.

[1729] A telephone complaint handling system incorporating an emotion engine.

[1730] Step 1:

[1731] A user (a customer service operator) receives a complaint call.

[1732] The operator answers the phone using the usual procedure.

[1733] Step 2:

[1734] The terminal (operator's PC or telephone system) records the call.

[1735] When a call is initiated, the audio is automatically recorded.

[1736] Step 3:

[1737] The device transmits the recorded audio data to the server in real time.

[1738] The recorded audio data is divided into packets and sent to the server.

[1739] Step 4:

[1740] The server sends the received audio data to a speech recognition API for transcription.

[1741] The audio data is converted to text format via a speech recognition API.

[1742] Step 5:

[1743] The server sends the transcribed text to a natural language processing (NLP) model for analysis.

[1744] The NLP model extracts the content of the complaint and important keywords.

[1745] Step 6:

[1746] The server uses an emotion engine to recognize the user's emotions from the transcribed text.

[1747] The emotion engine analyzes emotional expressions within the text to identify the customer's emotions (e.g., anger, sadness, confusion, etc.).

[1748] Step 7:

[1749] The server generates an appropriate apology response based on the analysis results and the recognized emotions.

[1750] Responses are generated using templates or AI models tailored to the nature of the complaint and the customer's emotions. For example, if a customer is "angry because their delivery is delayed," an emphatic apology such as "We sincerely apologize. We will investigate and address the issue immediately" will be generated.

[1751] Step 8:

[1752] The server generates an apology response, sends it to the terminal, and displays it to the user.

[1753] The generated apology response is displayed on the operator's monitor.

[1754] Step 9:

[1755] We will handle the complaint based on the apology response displayed to the user.

[1756] The operator uses the displayed response to proceed with the actual complaint handling. Responses can also be customized as needed.

[1757] Inquiry email response system with integrated emotion engine

[1758] Step 1:

[1759] The user (customer center staff) receives complaint and inquiry emails.

[1760] Check for new emails through your email client.

[1761] Step 2:

[1762] The device sends the email content to the server.

[1763] The email body and subject are transferred to the server via the API.

[1764] Step 3:

[1765] The server analyzes the email content using a natural language processing (NLP) model.

[1766] The NLP model analyzes the content and topic of the email.

[1767] Step 4:

[1768] The server uses an emotion engine to recognize the user's emotions from the email content.

[1769] The emotion engine analyzes emotional expressions within emails to identify the customer's emotions (e.g., dissatisfaction, disappointment, confusion, etc.).

[1770] Step 5:

[1771] The server automatically generates an appropriate response based on the analysis results and recognized emotions.

[1772] The system generates response messages using templates or AI models tailored to the nature of the complaint and the customer's feelings. For example, if a customer says, "I'm disappointed because the product is broken," a response such as, "We are very sorry for the inconvenience. We will arrange for a replacement immediately," might be generated.

[1773] Step 6:

[1774] The server generates a reply and sends it to the terminal for the user to see.

[1775] The generated reply will be displayed in the staff member's email client.

[1776] Step 7:

[1777] The user reviews the displayed reply and edits it as needed.

[1778] Staff will review the reply and make any necessary corrections.

[1779] Step 8:

[1780] The user sends an email with the revised reply.

[1781] Send the finalized response to the customer to complete the process.

[1782] As described above, by incorporating an emotion engine, the quality of complaint and inquiry handling can be improved, and responses that take customer emotions into consideration can be made. This system can promote the efficiency of customer support and improve customer satisfaction.

[1783] (Example 2)

[1784] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1785] Handling customer complaints in modern customer centers is time-consuming and labor-intensive. Furthermore, the quality of service provided by operators varies, making it difficult to consistently improve customer satisfaction. In particular, it is crucial to accurately understand customer emotions and respond with appropriate and courteous care. A method is needed to address these challenges efficiently and effectively.

[1786] The identification processing performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text to understand the content of the complaint, means for analyzing the transcribed text using a natural language processing model and an emotion analysis engine to understand the customer's emotions, means for generating an appropriate apology response based on the content of the complaint and the customer's emotions, and means for providing the generated apology response. This makes it possible to accurately grasp the content of the customer's complaint and emotions, and to quickly generate and provide an appropriate response accordingly. Furthermore, it is possible to reduce the burden on operators and ensure consistency in response quality, thereby improving customer satisfaction.

[1787] "Means for receiving customer voice complaints" refers to a function for electronically receiving complaints made by customers via telephone.

[1788] "Means for recording received audio" refers to a function for saving received audio data to a recording medium.

[1789] "Methods for transcribing recorded audio data" refers to functions that utilize speech recognition technology to convert recorded audio data into text data.

[1790] "Means for analyzing transcribed text and understanding the content of a claim" refers to a function that uses natural language processing technology to analyze text data obtained through transcription and recognize the content contained therein.

[1791] "A means of understanding customer emotions by analyzing transcribed text using natural language processing models and sentiment analysis engines" refers to a function that uses specific algorithms to analyze the linguistic structure and emotional aspects within the text and grasp the customer's emotional state.

[1792] "A means of generating an appropriate apology response based on the content of the complaint and the customer's feelings" refers to a function that automatically generates the optimal response text corresponding to the specific content of the recognized complaint and the customer's feelings.

[1793] "Means of providing a generated apology response" refers to a function for displaying the generated response to the operator or sending it directly to the customer.

[1794] "Means for receiving complaints and inquiries via email" refers to a function for receiving complaints and inquiries sent via email.

[1795] "A means of analyzing the content of received emails to understand complaints, inquiries, and customer sentiment" refers to a function that uses natural language processing and sentiment analysis technologies to analyze the text of received emails and grasp their content and customer sentiment.

[1796] "A means of generating appropriate response content based on complaints, inquiries, and customer sentiment" refers to a function that automatically generates the optimal response based on recognized email content and customer sentiment.

[1797] "A means of automatically generating response content and apology responses based on the content of the complaint and the customer's feelings using a generative AI model" refers to a function that uses an artificial intelligence model to generate appropriate responses to complaints and inquiries from specific prompt sentences.

[1798] "A means of instructing a generation AI model to provide an appropriate response using a prompt statement" refers to a function that provides a specific input (prompt) to the AI ​​model and generates an appropriate output based on that input.

[1799] This invention relates to a system for improving the efficiency of complaint handling and enhancing customer satisfaction in customer centers. Specifically, it analyzes both voice and email complaints and automatically generates appropriate apologies and replies.

[1800] System Configuration

[1801] This system consists of the following main elements:

[1802] 1. Users: Customer center operators and staff will be the primary users of the system.

[1803] 2. Terminals: This includes PCs, telephone systems, and email clients used by operators and staff.

[1804] 3. Server: This server runs the speech recognition API, natural language processing (NLP) models, sentiment analysis engine, and generative AI models.

[1805] Hardware and software to be used

[1806] Speech Recognition API: Use the Google Cloud Speech-to-Text API to convert speech data into text.

[1807] Natural Language Processing Models: Text data is analyzed using NLP models such as BERT and GPT-3.

[1808] Sentiment analysis engine: Analyzes emotions from text using tools such as IBM Watson Tone Analyzer.

[1809] Generative AI model: Uses GPT-3 and other models to generate appropriate apology responses and reply content.

[1810] Processing flow

[1811] Regarding the customer complaint phone handling system:

[1812] When a user receives a call, the device records the audio and sends it to the server in real time.

[1813] The server sends the audio data to the Google Cloud Speech-to-Text API, where it is converted into text data.

[1814] The server sends text data to an NLP model to analyze the claim details.

[1815] The server then uses an emotion analysis engine to analyze the customer's emotions.

[1816] Based on the nature of the complaint and the emotional state, the server uses a generative AI model to generate an appropriate apology response.

[1817] The generated apology response is sent to the device and displayed on the user's monitor.

[1818] Regarding the complaint email handling system:

[1819] When a user receives an email, the email body is sent from the device to the server.

[1820] The server uses an NLP model to analyze the email content and understand the complaint and emotions behind it.

[1821] Based on the analysis results, the server uses a generative AI model to generate appropriate responses.

[1822] The generated reply is sent to the device and displayed in the user's email client.

[1823] Examples of specific cases and prompt statements

[1824] For example, if an operator receives a complaint about a delivery delay:

[1825] User: The operator answers the phone.

[1826] Terminal: Records the call content and sends the audio data to the server.

[1827] Server: Performs transcription using a speech recognition API (e.g., Google Cloud Speech-to-Text).

[1828] Server: Uses NLP models to analyze text and sentiment to understand the content and emotions of complaints (e.g., BERT and IBM Watson Tone Analyzer).

[1829] Server: Generates an appropriate apology response and sends it to the terminal.

[1830] User: Use the generated response as a reference to handle the complaint.

[1831] Examples of prompt statements:

[1832] "We've received a complaint about a delivery delay. The customer is angry. Please generate an appropriate apology letter."

[1833] "We received an email inquiry about a damaged product. The customer is disappointed. Please generate an appropriate reply."

[1834] This system streamlines customer service complaint handling and enables appropriate responses that take customer emotions into consideration. This, in turn, leads to improved customer satisfaction.

[1835] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1836] Customer complaint phone handling system

[1837] Processing steps

[1838] Step 1:

[1839] A customer service operator, acting as the user, receives a complaint call. When the operator presses the "Start Call" button, the terminal begins recording the conversation. The recorded audio data is transmitted to the server in real time. The input is the call audio, and the output is the audio data stored on the server.

[1840] Step 2:

[1841] The server sends the received audio data to the Google Cloud Speech-to-Text API, where it is converted into text data. In this step, audio data is input and text data is generated as output. Specifically, the server sends an API request and receives a response.

[1842] Step 3:

[1843] The server sends the text data converted from the speech to a natural language processing (NLP) model such as BERT or GPT-3 to analyze the claim content. The input for this step is the transcribed text, and the analyzed claim content is output. Specifically, the server sends an API request to the NLP model and receives the analysis results.

[1844] Step 4:

[1845] The server then sends the analyzed text to a sentiment analysis engine, such as IBM Watson Tone Analyzer, to analyze the customer's emotions. In this step, the text containing the complaint is input, and the sentiment analysis results are obtained as output. The specific operation involves sending requests to the sentiment analysis API and receiving responses.

[1846] Step 5:

[1847] The server sends prompt messages to a generative AI model (e.g., GPT-3) based on the understood complaint content and sentiment state, generating an appropriate apology response. The input data is the complaint content and sentiment analysis results, and the output is the generated apology response. Specifically, the server defines appropriate prompt messages for the AI ​​model and sends an API request.

[1848] Step 6:

[1849] The generated apology response is sent to the terminal and displayed on the operator's monitor. In this step, the generated response text is input, and the displayed monitor is output. Specifically, a UI update request is made to display the response text.

[1850] Complaint email handling system

[1851] Processing steps

[1852] Step 1:

[1853] When a customer service staff member (a user) receives a complaint or inquiry email, the email body is sent from the terminal to the server. The input is the received email, and the output is the email data stored on the server. Specifically, the email client sends the email body via the API.

[1854] Step 2:

[1855] The server uses an NLP model to analyze the email content. The email body is input, and the complaint and inquiry details are output. Specifically, an API request is sent to analyze the text data, and the analysis results are received.

[1856] Step 3:

[1857] The server uses an emotion analysis engine to understand customer emotions during the process of analyzing email content and inquiry details. The input data is the email body, and the output data is the emotion analysis result. Specifically, it sends a request to the emotion analysis API and receives a response.

[1858] Step 4:

[1859] The server uses a generative AI model to generate an appropriate response based on the analyzed content and emotions. In this step, the complaint and customer emotions are used as input, and the output is the generated response. Specifically, a prompt is sent to the generative AI model to generate an appropriate response.

[1860] Step 5:

[1861] The generated reply is sent to the terminal and displayed in the staff member's email client. The input data is the generated reply text, and the output is the displayed email client. Specifically, a UI update request is executed to display the reply text.

[1862] Through the processing steps described above, this system contributes to improving the efficiency of customer support and enhancing customer satisfaction. By understanding customer emotions and responding appropriately and quickly, customer satisfaction can be increased.

[1863] (Application Example 2)

[1864] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1865] In modern customer service centers, responding quickly and appropriately to customer complaints and inquiries is a crucial challenge. However, especially when complaints are emotional, it is often difficult for staff to accurately understand those emotions and respond appropriately. Therefore, there is a need for systems that can improve customer satisfaction and increase the efficiency of complaint handling.

[1866] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1867] In this invention, the server includes means for receiving voice complaints from customers, means for recording the received voice, means for transcribing the recorded voice data, means for analyzing the transcribed text to understand the content of the complaint and the customer's feelings, means for generating an appropriate apology response based on the content of the complaint and the customer's feelings, and means for providing the generated apology response. This enables appropriate responses that take into account the customer's feelings.

[1868] "Means of understanding customer emotions" refers to the process of analyzing and recognizing the emotions behind customer words, such as anger, disappointment, or satisfaction, using natural language processing technology.

[1869] A "means for generating apology responses" is a system that automatically generates appropriate and empathetic apologies and responses based on the analyzed complaint content and customer emotions.

[1870] "Methods for generating response content" refers to a process that automatically constructs appropriate responses to customers, taking into account complaints, inquiries, and related emotions.

[1871] A "natural language processing model" is a machine learning model that analyzes text data and uses that information to understand meaning and intent.

[1872] A "generative AI model" is an artificial intelligence model that automatically generates text or answers based on given prompts or input data.

[1873] A "prompt" is an instruction or input sentence used by a generative AI model to generate a response, and it serves as a guideline for providing appropriate output based on its content.

[1874] This invention is a system that supports customer service in physical stores. Its purpose is to enable store staff to respond quickly and accurately to customer complaints and inquiries. A specific example of the system is described below.

[1875] Hardware and software to be used

[1876] The system uses the following hardware and software.

[1877] Hardware: Smartphones, smart glasses, head-mounted displays

[1878] Software: Python, TensorFlow, Transformers package, SpeechRecognition package

[1879] System Configuration

[1880] 1. Means of receiving voice complaints from customers:

[1881] Staff members wear smartphones, smart glasses, or head-mounted displays to receive customer complaints and inquiries.

[1882] 2. Means for recording received audio:

[1883] The received audio data is recorded on the staff member's terminal and sent to the server in real time.

[1884] 3. Methods for transcribing recorded audio data:

[1885] The server converts the received audio data into text data using the SpeechRecognition package. This allows the content of complaints and inquiries to be obtained as text.

[1886] 4. Means for analyzing transcribed text to understand the content of the complaint and the customer's feelings:

[1887] The server analyzes the converted text data using the Transformers package and a natural language processing model to understand the content of the complaint, while simultaneously recognizing the customer's emotions using an emotion engine.

[1888] 5. Means for generating an appropriate apology response based on the nature of the complaint and the customer's feelings:

[1889] Based on the analyzed complaint details and customer emotions, the server uses a generative AI model to generate an appropriate apology response. For example, if the complaint is "the product was broken" and the customer's emotion is recognized as "anger," an apology response such as "We sincerely apologize. We will arrange for a replacement immediately" will be generated.

[1890] 6. Means of providing the generated apology response:

[1891] The generated apology response is sent from the server to the staff member's terminal and displayed on their smartphone, smart glasses, or head-mounted display. The staff member uses this response as a reference when dealing with the customer.

[1892] Example of a prompt

[1893] Examples of prompt statements that can actually be used include the following:

[1894] "Analyze the customer's emotions from the following text: 'The product was broken. Please do something about it.'"

[1895] Specific example

[1896] Examples of customer service:

[1897] When a staff member receives a customer complaint using smart glasses, the audio is recorded by the smart glasses and sent to a server in real time. The server converts the received audio into text and analyzes the emotions. Then, an appropriate apology response is displayed on the staff member's glasses. The staff member uses the displayed apology response to address the customer and resolve the issue quickly.

[1898] Thus, the present invention aims to improve customer satisfaction in physical stores by enabling appropriate and prompt responses that take customer emotions into consideration.

[1899] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1900] Step 1:

[1901] Receive voice complaints from customers.

[1902] Input: A customer makes a verbal complaint to a store staff member.

[1903] Specific operation: The store staff member's smartphone, smart glasses, or head-mounted display receives audio in real time.

[1904] Output: Received audio data.

[1905] Step 2:

[1906] Record the received audio.

[1907] Input: Audio data received in Step 1.

[1908] Specific operation: The device (smartphone, smart glasses, head-mounted display) records audio and sends the recorded data to the server.

[1909] Output: Recorded audio data.

[1910] Step 3:

[1911] Transcribe the recorded audio data into text.

[1912] Input: Audio data recorded in Step 2.

[1913] Specific operation: The server uses a speech recognition API to convert the audio data into text data.

[1914] Output: Transcripted text data.

[1915] Step 4:

[1916] The transcribed text is analyzed to understand the content of the complaint and the customer's feelings.

[1917] Input: Text data obtained in Step 3.

[1918] Specific operation: The server uses a natural language processing (NLP) model to analyze the content of the complaint from the text data. At the same time, it uses an emotion engine to recognize the customer's emotions.

[1919] Output: Analyzed complaint details and perceived customer sentiment.

[1920] Step 5:

[1921] Based on the nature of the complaint and the customer's feelings, generate an appropriate apology response.

[1922] Input: Complaint details and customer sentiment analyzed in Step 4.

[1923] Specific operation: The server uses an AI model to generate an appropriate apology response based on the entered complaint content and emotions. For example, if the complaint is "the product was broken" and the emotion "anger" is recognized, an apology response such as "We sincerely apologize. We will arrange for a replacement immediately." will be generated.

[1924] Output: Generated apology response.

[1925] Step 6:

[1926] Provide a generated apology response.

[1927] Input: The apology response generated in Step 5.

[1928] Specific operation: The server sends the generated apology response to the staff member's device (smartphone, smart glasses, or head-mounted display). The staff member can then use the apology response as a reference when dealing with the customer.

[1929] Output: The apology response displayed on the terminal.

[1930] This allows the system to provide quick and appropriate responses in physical stores, taking customer emotions into consideration, thereby improving customer satisfaction.

[1931] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1932] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1933] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[1934] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1935] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[1936] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[1937] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[1938] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[1939] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[1940] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[1941] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[1942] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[1943] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[1944] 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.

[1945] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[1946] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[1947] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[1948] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[1949] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[1950] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[1951] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted as being incorporated by reference.

[1952] The following is further disclosed regarding the embodiments described above.

[1953] (Claim 1)

[1954] A means of receiving customer complaints via voice,

[1955] A means of recording the received audio,

[1956] Methods for transcribing recorded audio data,

[1957] A means of analyzing the transcribed text to understand the content of the complaint,

[1958] A means of generating an appropriate apology response based on the content of the complaint,

[1959] A means of providing a generated apology response,

[1960] A system that includes this.

[1961] (Claim 2)

[1962] A means of receiving complaints and inquiries via email,

[1963] A means of analyzing the content of received emails to understand complaints and inquiries,

[1964] A means of generating appropriate response content based on complaints and inquiries,

[1965] A means of providing the generated reply content,

[1966] The system according to claim 1, further comprising:

[1967] (Claim 3)

[1968] A means for analyzing transcribed text and email content using a natural language processing model,

[1969] A means for automatically generating reply content and apology responses based on the analysis results,

[1970] The system according to claim 1, further comprising:

[1971] "Example 1"

[1972] (Claim 1)

[1973] A means of receiving customer complaints via voice,

[1974] A means of recording the received audio,

[1975] Methods for transcribing recorded audio data,

[1976] A means of analyzing the transcribed text to understand the content of the complaint,

[1977] A means of generating an appropriate apology response based on the content of the complaint,

[1978] A means of providing a generated apology response,

[1979] A means of transmitting recorded audio data to a server in real time,

[1980] A means of converting speech to text data using a speech recognition API,

[1981] A means of analyzing text data using a natural language processing model,

[1982] A method for automatically generating apology responses using a generative AI model,

[1983] A system that includes this.

[1984] (Claim 2)

[1985] A means of receiving complaints and inquiries via email,

[1986] A means of analyzing the content of received emails to understand complaints and inquiries,

[1987] A means of generating appropriate response content based on complaints and inquiries,

[1988] A means of providing the generated reply content,

[1989] A method for analyzing email content using a natural language processing model,

[1990] A method for automatically generating reply content using a generative AI model,

[1991] The system according to claim 1, including the following:

[1992] (Claim 3)

[1993] A means for analyzing transcribed text and email content using a natural language processing model,

[1994] A means for automatically generating reply content and apology responses based on the analysis results,

[1995] Means for performing analysis and generation using speech recognition APIs, natural language processing models, and generative AI models,

[1996] The system according to claim 1, including the following:

[1997] "Application Example 1"

[1998] (Claim 1)

[1999] A means of receiving customer complaints via voice,

[2000] A means of recording the received audio,

[2001] Methods for transcribing recorded audio data,

[2002] A means of analyzing the transcribed text to understand the content of the complaint,

[2003] A means of generating an appropriate apology response based on the content of the complaint,

[2004] A means of providing a generated apology response,

[2005] A system including speech recognition means, natural language processing means, and response generation means for automatically analyzing complaints and inquiries from viewers and automatically generating appropriate responses.

[2006] (Claim 2)

[2007] A means of receiving complaints and inquiries via email,

[2008] A means of analyzing the content of received emails to understand complaints and inquiries,

[2009] A means of generating appropriate response content based on complaints and inquiries,

[2010] A means of providing the generated reply content,

[2011] The system according to claim 1, further comprising means for analyzing audio complaints and inquiries from viewers in real time and displaying automatically generated responses.

[2012] (Claim 3)

[2013] A means for analyzing transcribed text and email content using a natural language processing model,

[2014] A means for automatically generating reply content and apology responses based on the analysis results,

[2015] The system according to claim 1, further comprising means for analyzing the emotions of viewers based on the content of their complaints and inquiries and generating an automated response accordingly.

[2016] "Example 2 of combining an emotion engine"

[2017] (Claim 1)

[2018] A means of receiving customer complaints via voice,

[2019] A means of recording the received audio,

[2020] Methods for transcribing recorded audio data,

[2021] A means of analyzing the transcribed text to understand the content of the complaint,

[2022] Using natural language processing models and sentiment analysis engines, we analyze transcribed text to understand customer emotions.

[2023] A means of generating an appropriate apology response based on the content of the complaint and the customer's feelings,

[2024] A means of providing a generated apology response,

[2025] A system that includes this.

[2026] (Claim 2)

[2027] A means of receiving complaints and inquiries via email,

[2028] A means of analyzing the content of received emails to understand complaints, inquiries, and customer sentiment,

[2029] A means of generating appropriate response content based on the content of complaints and inquiries and the customer's feelings,

[2030] A means of providing the generated reply content,

[2031] The system according to claim 1, further comprising:

[2032] (Claim 3)

[2033] A means for automatically generating response content and apology responses based on the content of the complaint and the customer's emotions using a generative AI model,

[2034] A means of instructing a generative AI model to give an appropriate response using a prompt statement,

[2035] The system according to claim 1, further comprising:

[2036] "Application example 2 when combining with an emotional engine"

[2037] (Claim 1)

[2038] A means of receiving customer complaints via voice,

[2039] A means of recording the received audio,

[2040] Methods for transcribing recorded audio data,

[2041] A means of analyzing transcribed text to understand the content of the complaint and the customer's feelings,

[2042] A means of generating an appropriate apology response based on the content of the complaint and the customer's feelings,

[2043] A means of providing a generated apology response,

[2044] A system that includes this.

[2045] (Claim 2)

[2046] A means of receiving complaints and inquiries via email,

[2047] A means of analyzing the content of received emails to understand complaints, inquiries, and customer sentiment,

[2048] A means of generating appropriate response content based on the content of complaints and inquiries and the customer's feelings,

[2049] A means of providing the generated reply content,

[2050] The system according to claim 1, including the following:

[2051] (Claim 3)

[2052] A means for analyzing transcribed text and email content using a natural language processing model,

[2053] A means for automatically generating reply content and apology responses based on the analysis results,

[2054] A means for designing and using a generative AI model and prompt sentences that analyze customer emotions and automatically generate appropriate responses,

[2055] The system according to claim 1, including the following: [Explanation of Symbols]

[2056] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. A means of receiving customer complaints via voice, A means of recording the received audio, Methods for transcribing recorded audio data, A means of analyzing the transcribed text to understand the content of the complaint, A means of generating an appropriate apology response based on the content of the complaint, A means of providing a generated apology response, A system that includes this.

2. A means of receiving complaints and inquiries via email, A means of analyzing the content of received emails to understand complaints and inquiries, A means of generating appropriate response content based on complaints and inquiries, A means of providing the generated reply content, The system according to claim 1, further comprising:

3. A means for analyzing transcribed text and email content using a natural language processing model, A means for automatically generating reply content and apology responses based on the analysis results, The system according to claim 1, further comprising:

Images