System

Abstract

A system is provided.SOLUTION: A system comprising: means for receiving a query; means for analyzing the received query to identify query content; means for accessing an in-house system to obtain relevant information based on the identified query content; means for generating an answer to the query using a generative AI model based on the obtained information; means for evaluating the reliability of the generated answer; means for determining whether the answer is appropriate based on the evaluation results and escalating to a human representative if not appropriate; means for sending a final answer to the user; and means for storing the query and answer in a log.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] In the current situation where a lot of time and effort is spent on inquiries from corporate customers, there is a problem that it becomes difficult for sales representatives to concentrate on their core sales activities. Furthermore, delays in customer responses can decrease customer satisfaction and have a negative impact on a company's digitalization efforts. Furthermore, if inquiries cannot be responded to optimally, there is an increased risk of incorrect information being provided. These issues need to be resolved. [Means for solving the problem]

[0005] The present invention solves these problems by including a means for receiving an inquiry, a means for analyzing the received inquiry to identify the inquiry content, and a means for accessing an internal system to obtain related information based on the identified inquiry content. It also includes a means for generating an answer to the inquiry using a generative AI model based on the acquired information, and a means for evaluating the reliability of the generated answer. It also includes a means for determining whether the answer is appropriate based on the evaluation result, and escalating the issue to a human representative if it is not appropriate, and further includes a means for sending the final answer to the user and a means for saving the inquiry and the answer in a log.

[0006] The "means for receiving an inquiry" refers to a system or software module for receiving inquiry information such as a text message or voice input from a user.

[0007] The "means for analyzing the inquiry and identifying the content of the inquiry" refers to a program or algorithm that uses natural language processing technology or the like to analyze the inquiry information received from the user and understand its intent and content.

[0008] "Means of accessing internal systems to obtain relevant information based on the identified inquiry" refers to APIs or data access mechanisms for obtaining the necessary information from internal databases or other internal systems depending on the inquiry.

[0009] "Means for generating responses to inquiries using generative AI models" refers to computational models or generative algorithms that use artificial intelligence to create accurate responses based on acquired information.

[0010] The "means for evaluating the reliability of the generated answer" refers to an evaluation criterion or scoring system for measuring the reliability of the generated answer and determining whether it is appropriate.

[0011] "Means for escalation to a human agent if not appropriate" refers to a notification or forwarding mechanism to transfer the inquiry to a human agent if the generated answer is not reliable.

[0012] The "means for sending a final answer to the user" refers to a communication means or message transmission mechanism for returning a reliable answer to the user.

[0013] A "means for logging queries and responses" is a data storage system or logging mechanism that records the queries and responses exchanged and stores them for future reference. [Brief explanation of the drawings]

[0014] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0015] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0016] First, the terms used in the following description will be explained.

[0017] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0018] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0019] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0020] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0021] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0022] [First embodiment]

[0023] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0024] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0025] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0026] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0027] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0028] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0029] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0030] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0031] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0032] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0033] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0034] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0035] The present invention is a system for effectively responding to inquiries from corporate customers, and includes a series of processes for receiving inquiries, analyzing them, and generating appropriate responses. An embodiment of this system will be described in detail below.

[0036] First, the user sends a query to the corporate chatbot. For example, the user might enter a query such as, "Please tell me the renewal period for this phone number." The device receives this query in real time and sends it to the server.

[0037] The server then analyzes the received inquiry using natural language processing technology to accurately understand the intent of the inquiry. For example, it can extract the keyword "renewal period" from the inquiry and identify it as a question about the renewal period for a phone number.

[0038] After identifying the query, the server accesses the internal system via API to retrieve relevant information. Specifically, the server queries the internal database to retrieve necessary data, such as "phone number renewal period." This data is then formatted by the server and passed to the generative AI model.

[0039] Based on the acquired information, the server uses a generative AI model to generate an appropriate response to the user's inquiry. The generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0040] The server evaluates the reliability of the generated answer. It calculates a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, it returns the answer to the user as is. On the other hand, if the confidence is low, the server escalates the answer to a human agent.

[0041] The server sends the final, verified answer to the user. If a reliable answer is generated automatically or if a suitable answer is prepared after human verification, the content is sent to the user's device. For example, "Your phone number renewal period ends on December 31, 2023."

[0042] Finally, the server stores all interactions in a log, including the inquiry content, the generated response, escalation history, etc. The stored log is used to improve the system and improve the accuracy of inquiry responses in the future.

[0043] This system reduces the time sales representatives spend on tedious inquiries, allowing them to focus on important sales activities. Providing fast and accurate information also contributes to improving customer satisfaction. Thus, the present invention is extremely useful as a means of streamlining inquiries from corporate customers and providing more advanced customer service.

[0044] The processing flow will be explained below.

[0045] Step 1:

[0046] A user sends a query to the enterprise chatbot, for example, "What is the renewal period for this phone number?"

[0047] Step 2:

[0048] The device receives this inquiry in real time and sends the inquiry content to the server. Specifically, it creates an HTTP request and sends it to the server.

[0049] Step 3:

[0050] The server analyzes the received inquiry, which involves using natural language processing (NLP) techniques to understand the intent of the text. It extracts the keyword "renewal period" from the inquiry and identifies it as a request related to the renewal period of a phone number.

[0051] Step 4:

[0052] The server then accesses the internal system via API to retrieve relevant information based on the identified inquiry. Specifically, the server sends a request to the appropriate API endpoint of the internal database to retrieve information about the "phone number renewal period."

[0053] Step 5:

[0054] The server formats the information and passes it to the generative AI model. For example, it converts the information into text data in the format "This phone number's renewal period ends December 31, 2023."

[0055] Step 6:

[0056] The server uses a generative AI model to generate an appropriate response to the inquiry. The generative AI model creates a response in natural language based on the acquired information. It generates a response in the format, such as, "Your phone number renewal period ends on December 31, 2023."

[0057] Step 7:

[0058] The server evaluates the reliability of the generated answer by calculating the confidence score of the generative AI model and checking whether the score exceeds a certain threshold.

[0059] Step 8:

[0060] The server determines whether the answer is appropriate based on the evaluation results. If the answer is highly reliable, it is sent to the user. If it is inappropriate, it is escalated to a sales representative.

[0061] Step 9:

[0062] The server then sends the final verified answer to the user, either authentic or verified by a human agent, via HTTP response or real-time messaging protocol to the user's device.

[0063] Step 10:

[0064] The server stores all interactions in a log, including inquiries, information obtained, responses generated, escalation history, etc. The log is used for future analysis and system improvement.

[0065] Example 1

[0066] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0067] Conventional inquiry response systems have difficulty providing quick and accurate responses to user inquiries, and often require human intervention. As a result, inquiries take time to respond, leading to issues such as a decline in customer satisfaction. Furthermore, there is a lack of efficient management of data generated during the inquiry response process and its use for future improvements.

[0068] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0069] In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an internal system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative artificial intelligence model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and, if not, for escalating to a human representative, means for sending the final answer to the user, and means for storing the inquiry and the answer in a record. This makes it possible to respond to user inquiries quickly and accurately, improving customer satisfaction and contributing to future system improvements.

[0070] The "means for receiving an inquiry" refers to a technique or device for inputting inquiry information such as text or voice sent from a user into a server via the Internet.

[0071] The "means of analyzing the received inquiry and identifying the content of the inquiry" refers to technology or algorithms that use natural language processing technology to break down the text data of the received inquiry and extract and understand the subject and keywords of the inquiry.

[0072] "Means of accessing internal systems to obtain relevant information based on the identified inquiry" refers to the technology or procedures by which a server sends requests to an internal company database or API to search for and obtain information related to the inquiry.

[0073] "Means for generating answers to inquiries using a generative AI model" refers to technologies or systems that use a generative AI model (e.g., GPT-3) based on acquired information to create answers in natural language format to user inquiries.

[0074] "Means for assessing the reliability of generated answers" refers to technologies or algorithms that evaluate the reliability and accuracy of answers output by generative AI models using criteria such as confidence scores.

[0075] "Means of determining whether the answer is appropriate based on the evaluation results, and escalating to a human operator if it is not appropriate" refers to the technology or process of determining whether the answer meets certain criteria based on the evaluated reliability score, and if it does not, handing over the processing to a human operator.

[0076] The "means for sending the final answer to the user" refers to a technique or means for returning an answer that has passed the reliability evaluation to the user in real time.

[0077] "Means for recording inquiries and responses" means techniques or methods for recording all inquiry and response interactions in log files or databases and storing them for future analysis and improvement.

[0078] The present invention provides a system for effectively responding to inquiries from corporate customers, which includes a series of processes for receiving and analyzing inquiries, and generating and returning appropriate answers.

[0079] First, the user sends a query to the corporate chatbot. For example, they might ask, "What is the renewal period for this phone number?" The device receives this query in real time and sends it to a server via the Internet.

[0080] The server analyzes the received query using natural language processing techniques (e.g., spaCy or NLTK). The server tokenizes the query text, extracts the keyword "renewal period," and understands that the query is about the renewal period for a phone number.

[0081] The server then accesses the internal system via an API to retrieve relevant information. Specifically, the server queries the internal database (e.g., Microsoft SQL Server or MySQL) to retrieve the necessary data. The retrieved information is then formatted and passed to the generative AI model.

[0082] The server uses a generative AI model (for example, OpenAI's GPT-3) to generate an appropriate response to the user's inquiry. Based on the acquired information, the generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0083] The server evaluates the reliability of the generated answer, calculates a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, returns the answer to the user as is. On the other hand, if the confidence is low, the answer is escalated to a human representative.

[0084] Finally, the server sends the final answer that passes the evaluation to the user. The terminal receives the answer and displays it on the chatbot interface. For example, "Your phone number renewal period ends on December 31, 2023."

[0085] The server also stores all inquiries and responses in a log, which includes the inquiry content, the generated responses, escalation history, etc. The stored logs are used to improve the system and the accuracy of inquiries in the future.

[0086] As a specific example, if a user sends a query such as "Please tell me the renewal period for this phone number," the device sends the query to the server, which uses natural language processing technology to extract the keyword "renewal period." The server accesses the internal system to obtain the corresponding information, and based on that, uses a generative AI model to generate the answer "The renewal period for your phone number is December 31, 2023." The answer that passes the evaluation is sent to the user, and the content of the query and answer is saved in a log.

[0087] Example prompt sentence:

[0088] User: How long do I need to renew this phone number?

[0089] Terminal: (sends user query to server)

[0090] Server: (analyzes the received query)

[0091] Server: (retrieving corresponding information from internal database)

[0092] Server: (Generate an answer using a generative AI model based on the acquired information)

[0093] Server: Your phone number is due for renewal on December 31, 2023.

[0094] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0095] Step 1:

[0096] The user sends a query to the chatbot. The user accesses the chatbot's interface, enters a question in the text field, and presses the send button. For example, the user might enter, "Please tell me the renewal period for this phone number." This sentence is the user's input.

[0097] Step 2:

[0098] The terminal sends a query to the server. The terminal receives input from the user and sends it as an HTTP POST request to the server's query acceptance endpoint. The input here is the user's query statement, and the output is the HTTP request sent to the server.

[0099] Step 3:

[0100] The server analyzes the query. The server analyzes the received text data and uses natural language processing techniques to understand the intent of the query. Specifically, the server tokenizes the text and extracts keywords and phrases. For example, it extracts the keyword "renewal period" and identifies it as being about the renewal period for a phone number. The input to this step is the query received by the server, and the output is the extracted keywords and their intent.

[0101] Step 4:

[0102] Based on the identified query, the server retrieves information from the internal system via API. The server sends a query to the internal database based on the keyword "phone number renewal period." For example, the server uses an SQL query to search for "phone number renewal period." The input for this step is the identified keyword, and the output is the information retrieved from the database.

[0103] Step 5:

[0104] The server generates an answer using a generative AI model based on the information acquired. The server inputs the acquired data into a generative AI model (e.g., GPT-3) and generates an answer in natural language format. Specifically, the server inputs the prompt "What is the renewal period for your phone number?" and obtains the generated answer "Your phone number renewal period is December 31, 2023." The inputs for this step are the acquired information and the prompt, and the output is the generated answer.

[0105] Step 6:

[0106] The server evaluates the reliability of the generated answer. The server calculates a confidence score for the answer and determines whether it exceeds a certain threshold. The reliability evaluation uses the scoring function of the generative AI model. For example, if the confidence score is evaluated as 0.85, which is above the set threshold of 0.80, the answer is sent as is. The input of this step is the generated answer, and the output is the result of the reliability evaluation.

[0107] Step 7:

[0108] The server sends a final answer to the user based on the confidence rating. If a reliable answer is generated, the server sends it back to the device as an HTTP response. The device displays the received answer in the chatbot interface. For example: "Your phone number renewal period is December 31, 2023." The input of this step is the evaluated answer, and the output is the answer displayed in the user's interface.

[0109] Step 8:

[0110] The server stores queries and responses in a log. The server records all interactions in a log file or database. The recorded data includes the query content, generated responses, trust assessment results, escalation history, etc. The input of this step is all processing results, and the output is the stored log data.

[0111] (Application example 1)

[0112] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0113] With conventional systems, it is difficult to respond appropriately and quickly to inquiries from corporate customers. Furthermore, due to the wide range of inquiries, sales representatives spend a lot of time responding to inquiries, preventing them from devoting sufficient time to their core sales activities. Furthermore, due to insufficient reliability evaluation of the generated answers and insufficient escalation procedures, customer satisfaction may decline.

[0114] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0115] In this invention, the server includes: means for receiving inquiries; means for analyzing the received inquiries to identify the inquiry content; means for accessing an internal system to obtain related information based on the identified inquiry content; means for generating an answer to the inquiry using a generative AI model based on the obtained information; means for evaluating the reliability of the generated answer; means for determining whether the answer is appropriate based on the evaluation result and, if not, for escalating to a human representative; means for sending the final answer to the user; means for saving the inquiry and the answer in a log; means implemented as a smartphone application that receives inquiries related to a specific service, analyzes them, and generates an answer; means for inputting and displaying the inquiry content through a user interface; and means for generating an answer by providing a prompt sentence to the generative AI model based on the inquiry content. This enables prompt and accurate answers to inquiries from corporate customers, allowing sales representatives to focus on their core business activities. Furthermore, the reliability evaluation of the generated answers and the escalation procedures are strengthened, thereby improving customer satisfaction.

[0116] The "means for receiving an inquiry" refers to a device or program for receiving an inquiry from a user in real time and transferring it to a server.

[0117] "Means for analyzing received inquiries and identifying the content of the inquiries" refers to devices or programs that use natural language processing technology to analyze the content of inquiries from users and accurately understand their intent.

[0118] "Means for accessing an internal system to obtain related information based on the identified inquiry content" refers to a device or program that works in conjunction with an API or database to enable the server to obtain information related to the identified inquiry content from the internal system.

[0119] "Means for generating a response to an inquiry using a generative AI model based on acquired information" refers to a device or program that uses acquired information to generate a response to an inquiry using a generative AI model.

[0120] "Means for evaluating the reliability of the generated answers" refers to devices or programs that calculate confidence scores and other factors to evaluate the accuracy and validity of answers generated by generative AI models.

[0121] "Means for determining whether the answer is appropriate based on the evaluation results, and escalating to a human staff member if it is not appropriate" refers to a device or program that determines the appropriateness of the answer generated based on the evaluation results, and if it is not appropriate, hands it over to a human staff member.

[0122] "Means for sending a final answer to the user" refers to a device or program that sends the content of a reliable answer to the user when a reliable answer is generated or when an appropriate answer is prepared after human verification.

[0123] "Means for saving inquiries and responses in a log" refers to a device or program that saves all interactions, inquiry content, generated responses, and escalation history in a log, and uses this information to improve the system and increase the accuracy of inquiry responses.

[0124] "Means implemented as a smartphone application that receives inquiries related to a specific service, analyzes them, and generates a response" refers to a device or program that receives inquiries related to a specific service through a smartphone application, analyzes the content of the inquiries, and generates a response.

[0125] The "means for inputting and displaying inquiry details through a user interface" refers to an interface through which a user inputs inquiry details and displays responses to those inquiries.

[0126] "Means for generating an answer by providing a prompt sentence to a generative AI model based on the content of the inquiry" refers to a device or program that generates a prompt sentence based on the content of the user's inquiry, passes it to a generative AI model, and generates an answer.

[0127] The following is a detailed description of an embodiment of the present invention: The system of the present invention includes a smartphone application and a server for quickly and accurately responding to inquiries from corporate customers.

[0128] The overall system overview is as follows:

[0129] 1. When a user enters an inquiry using a smartphone application, the inquiry is received by the device in real time.

[0130] 2. The received query is forwarded to the server, which analyzes the query using natural language processing technology to properly identify the query content and intent.

[0131] 3. Based on the identified query, the server accesses internal databases and systems to retrieve the relevant information needed.

[0132] 4. Based on the acquired information, the server uses a generative AI model to generate an appropriate answer. At the time of generation, the prompt sentence is provided to the generative AI model.

[0133] 5. The reliability of the generated answer is assessed by a confidence score: if the confidence is above a certain threshold, it is automatically sent to the user; if it is lower, it is escalated to a human agent.

[0134] 6. The final answer is sent to the user, and the system keeps a log of all interactions.

[0135] Hardware and software used

[0136] Hardware: Cloud servers (e.g., Amazon Web Services, Google Cloud Platform), smartphones (iOS or Android)

[0137] Software: Flask (a lightweight Python web framework), OpenAI's GPT-3 API, natural language processing libraries (e.g., spaCy, NLTK)

[0138] Example of processing flow

[0139] As a specific example, let us consider an inquiry about the release date of a new movie.

[0140] Example inquiry: "When is the new movie coming out?"

[0141] The server receives this query and uses natural language processing technology to identify the "release date of new movies."

[0142] Based on the identified information, the company accesses internal databases to retrieve relevant information.

[0143] The acquired information is provided to the generative AI model as a prompt sentence.

[0144] Prompt Sentence Examples

[0145] Examples of prompts are:

[0146] Text format

[0147] User Asks: When is the new movie coming out?

[0148] answer:

[0149] By passing this prompt to the OpenAI GPT-3 API, a specific answer such as "The release date of the new movie will be November 1, 2023" will be generated.

[0150] The above process makes it possible to respond to inquiries from corporate customers quickly and accurately. This system reduces the burden on sales representatives in responding to inquiries, allowing them to focus more on full-scale sales activities. In addition, by strengthening the reliability evaluation of responses and escalation procedures, customer satisfaction can be improved.

[0151] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0152] Step 1:

[0153] The device receives inquiries from users in real time. At this time, the user inputs the inquiry content through the smartphone application interface. The input inquiry is saved in text format on the device and temporarily stored within the device.

[0154] Input: User's inquiry (text format)

[0155] Output: Received inquiry (text format)

[0156] Step 2:

[0157] The terminal sends the received query to the server using a communication protocol such as an HTTP POST request. The server receives this request and prepares to analyze the query.

[0158] Input: Received inquiry (text format)

[0159] Output: Query content forwarded to the server (text format)

[0160] Step 3:

[0161] The server analyzes the received query and identifies the query content. Here, natural language processing technology (e.g., spaCy or NLTK) is used to extract the query intent and keywords. For example, the keyword "release date of new movies" is identified.

[0162] Input: Inquiry content forwarded to the server (text format)

[0163] Output: Identified inquiry content and keywords (text format)

[0164] Step 4:

[0165] The server accesses the company's internal systems based on the identified inquiry content and retrieves relevant information, such as data such as "release dates for new movies" using the company's internal databases and APIs.

[0166] Input: Identified inquiry content and keywords (text format)

[0167] Output: Relevant information retrieved (in text, possibly structured format)

[0168] Step 5:

[0169] Based on the acquired information, the server provides a prompt sentence to a generative AI model (e.g., OpenAI GPT-3) to generate an answer to the query. At this time, the prompt sentence is input to the generative AI model as a sentence with a specific structure.

[0170] Input: Retrieved relevant information (text format)

[0171] Output: Answers generated by the generative AI model (in text format)

[0172] Step 6:

[0173] The server evaluates the reliability of the generated answer using an algorithm that calculates a confidence score, determines whether the confidence score exceeds a certain threshold, and escalates it if necessary.

[0174] Input: Generated answer (text format)

[0175] Output: Reliability assessment results and confidence scores (numerical data)

[0176] Step 7:

[0177] If the final answer passes the trust assessment, the server sends it to the user, which returns it to the device as an HTTP response. If the trust is low, the request is escalated to a human agent.

[0178] Input: Reliability assessment result and answer (text format)

[0179] Output: Final answer (text format)

[0180] Step 8:

[0181] The server stores all interactions in a log, including the inquiry content, generated answers, reliability assessment results, escalation history, etc. The log is used for future system improvements and analysis.

[0182] Input: All correspondence (text format)

[0183] Output: Saved logs (database entries)

[0184] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0185] The present invention is a system for effectively responding to inquiries from corporate customers, which includes a series of processes for receiving inquiries, analyzing them, and generating appropriate answers, as well as an emotion engine for recognizing user emotions. An embodiment of this system will be described in detail below.

[0186] First, the user sends a query to the corporate chatbot. For example, they enter a query such as, "Please tell me the renewal period for this phone number." The device receives this query in real time and sends the query to the server.

[0187] The server then analyzes the received inquiry using natural language processing (NLP) techniques to accurately understand the intent of the inquiry. For example, it can extract the keyword "renewal period" from the inquiry and identify it as a query about the renewal period for a phone number.

[0188] After identifying the query, the server accesses internal systems via APIs to retrieve relevant information. Specifically, the server queries an internal database to retrieve information about, for example, phone number renewal periods. This data is then formatted by the server and passed to the generative AI model.

[0189] Based on the acquired information, the server uses a generative AI model to generate an appropriate response to the user's inquiry. The generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0190] The server then evaluates the reliability of the generated answer, calculating a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, the server returns the answer to the user as is. On the other hand, if the confidence score is low, the server escalates the answer to a human agent.

[0191] Furthermore, the present invention includes an emotion engine for recognizing the user's emotions. The server analyzes the language and emotional expressions contained in the user's inquiry, and if it determines that the user is expressing anger or frustration, it immediately escalates the inquiry to a human agent. The emotion engine adjusts the tone of the response depending on the user's emotions, for example, generating a more polite and reassuring response if the user is expressing frustration.

[0192] The server sends the final, verified answer to the user. The answer, either trusted or verified by a human agent, is sent to the user's device via an HTTP response or real-time messaging protocol. Example: "Your phone number is due for renewal on December 31, 2023."

[0193] Finally, the server stores all interactions in a log, including the inquiry content, information obtained, responses generated, escalation history, and the results of the analysis of recognized emotions. The stored logs are used to improve the system and enhance the accuracy of inquiry responses in the future.

[0194] As described above, the present invention reduces the time that sales representatives spend on tedious inquiries, creating an environment in which they can concentrate on important sales activities. Furthermore, the introduction of an emotion engine enables responses that take into consideration the user's emotions, contributing to improved customer satisfaction. Thus, the present invention is extremely useful as a means for streamlining inquiries from corporate customers and providing more advanced customer service.

[0195] The processing flow will be explained below.

[0196] Step 1:

[0197] A user sends a query to the enterprise chatbot, for example, "What is the renewal period for this phone number?"

[0198] Step 2:

[0199] The device receives this inquiry in real time and sends the inquiry content to the server. Specifically, it creates an HTTP request and sends it to the server.

[0200] Step 3:

[0201] The server analyzes the received inquiry, which involves using natural language processing (NLP) techniques to understand the intent of the text. It extracts the keyword "renewal period" from the inquiry and identifies it as a request related to the renewal period of a phone number.

[0202] Step 4:

[0203] The server then accesses the internal system via API to retrieve relevant information based on the identified inquiry. Specifically, the server sends a request to the appropriate API endpoint of the internal database to retrieve information about the "phone number renewal period."

[0204] Step 5:

[0205] The server formats the information and passes it to the generative AI model. For example, it converts the information into text data in the format "This phone number's renewal period ends December 31, 2023."

[0206] Step 6:

[0207] The server uses a generative AI model to generate an appropriate response to the inquiry. The generative AI model creates a response in natural language based on the acquired information. It generates a response in the format, such as, "Your phone number renewal period ends on December 31, 2023."

[0208] Step 7:

[0209] The server evaluates the reliability of the generated answer by calculating the confidence score of the generative AI model and checking whether the score exceeds a certain threshold.

[0210] Step 8:

[0211] The server determines whether the answer is appropriate based on the evaluation results. If the answer is highly reliable, it is sent to the user. If it is inappropriate, it is escalated to a sales representative.

[0212] Step 9:

[0213] The server uses an emotion engine to recognize the emotions contained in the user's query, analyzing the language and emotional expressions to determine whether the user is expressing anger or frustration.

[0214] Step 10:

[0215] The server adjusts the tone of the response depending on the user's emotions, for example, generating a more polite and reassuring response if the user is expressing anger.

[0216] Step 11:

[0217] The server sets escalation conditions based on the recognized emotions, and if the user expresses anger or frustration, the call is automatically escalated to a human agent.

[0218] Step 12:

[0219] The server then sends the final, verified answer to the user. The answer, either trusted or verified by a human agent, is sent to the user's device via an HTTP response or real-time messaging protocol. Example: "Your phone number renewal period ends December 31, 2023."

[0220] Step 13:

[0221] The server stores all interactions in a log, including the inquiry, the information obtained, the response generated, the escalation history, the analysis results of the recognized sentiment, etc. The stored log is used for future analysis and system improvement.

[0222] Example 2

[0223] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0224] Conventional corporate chatbot systems have several limitations in analyzing user inquiries and generating answers. For example, there are insufficient means to ensure the reliability of answers and a lack of consideration for user feelings, which can lead to a decline in the quality of inquiry responses. In addition, the system's automated responses are often unreliable, resulting in excessive reliance on human agents. These issues have led to issues such as delayed response times and reduced user satisfaction.

[0225] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an internal system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative AI model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and escalating the inquiry to a human agent if it is not appropriate, means for performing emotion analysis and escalating the inquiry to a human agent based on the user's emotions, means for sending a final answer to the user, and means for saving the inquiry and the answer in a log. This improves the accuracy of inquiry analysis and answer generation, enabling quick and reliable responses that also take user emotions into consideration.

[0226] The "means for receiving an inquiry" is a mechanism for transferring the contents of an inquiry entered by a user to a server.

[0227] The "means for analyzing the received inquiry and identifying the content of the inquiry" is a mechanism for analyzing the received inquiry using natural language processing technology, and extracting and identifying the intent and content of the inquiry.

[0228] "Means for accessing internal systems to obtain relevant information based on the identified inquiry" refers to a mechanism for accessing internal databases and other internal resources to obtain the necessary information based on the identified inquiry.

[0229] "Means for generating an answer to a query using a generative AI model based on acquired information" refers to a mechanism for generating an appropriate answer to a user's query using a generative AI model with acquired information as input.

[0230] The "means for assessing the reliability of the generated answer" is a mechanism for calculating a confidence score for the answer generated by the generative AI model and assessing its reliability.

[0231] "Means for determining whether the response is appropriate based on the evaluation results, and escalating to a human representative if it is not appropriate" is a mechanism for determining the appropriateness of the response based on the results of a reliability evaluation, and transferring the inquiry to a human representative if necessary.

[0232] The "means for performing emotion analysis and escalating inquiries to a human agent based on the user's emotions" is a mechanism for analyzing the emotions contained in a user's inquiry and transferring the inquiry to a human agent based on the results.

[0233] The "means for transmitting the final answer to the user" is a mechanism for transmitting the final confirmed answer to the user's terminal.

[0234] The "means for saving inquiries and responses in a log" is a mechanism for saving the inquiry content, acquired information, generated responses, and the results of escalation and sentiment analysis in a log.

[0235] MODE FOR CARRYING OUT THE INVENTION

[0236] The present invention is a system for responding effectively and quickly to inquiries from corporate customers. This system not only receives inquiries, analyzes them, and generates appropriate responses, but also includes an emotion engine that recognizes the emotions of users.

[0237] System hardware and software configuration

[0238] 1. User device: The device (PC, smartphone, tablet, etc.) through which the user sends an inquiry via the chatbot.

[0239] 2. Server: The central computing resource that performs various processes, such as query analysis, information retrieval, answer generation, credibility assessment, and sentiment analysis.

[0240] Natural Language Processing (NLP): Uses Google Cloud Natural Language API.

[0241] Internal system access: Access to our internal database (MySQL).

[0242] Generative AI model: OpenAI's GPT-4 is used.

[0243] Sentiment analysis: Using Azure Text Analytics.

[0244] System Operation

[0245] The user enters the question "Please tell me the renewal period for this phone number" into the chatbot interface. The device receives this query in real time and sends it to the server.

[0246] The server analyzes the received query using the Google Cloud Natural Language API. Based on the analysis results, it identifies the keyword "renewal period" and determines that the query is about the phone number renewal period.

[0247] The server then queries the company's internal database using MySQL to retrieve relevant data. Based on this information, the server uses OpenAI's GPT-4 model to generate an appropriate answer, such as "Your phone number is due for renewal on December 31, 2023."

[0248] To assess the reliability of the generated answer, the server calculates a confidence score and automatically sends the answer to the user if it exceeds a set threshold (CERTAINTY_THRESHOLD), or escalates the answer to a human agent if the confidence is low.

[0249] Additionally, the server uses Azure Text Analytics to analyze the user's emotions and, if it determines that the user is expressing anger or frustration, it will immediately escalate the call to a human agent.

[0250] The final answer is sent from the server to the device via an HTTP response or WebSocket protocol. For example, the answer may say, "Your phone number renewal period ends on December 31, 2023."

[0251] All interactions are logged by the server, including the inquiry, the information obtained, the response generated, the escalation history, and the results of sentiment analysis, which can be used to improve the system and its accuracy.

[0252] Examples of concrete examples and prompts

[0253] For example, if a user types "What is the expiration date of my contract?", the server uses NLP technology to identify the keyword "contract expiration date" and retrieves relevant information from the company's internal database. The generative AI model creates an answer such as "Your contract expires on March 31, 2024," which is sent to the user after passing a reliability assessment. If the user expresses strong dissatisfaction, the emotion engine will be activated and a human agent will manually process the request.

[0254] An example prompt is:

[0255] "Please tell me the renewal period for this phone number."

[0256] Please tell me the expiration date of the contract.

[0257] As described above, the present invention is extremely useful as a means for improving the efficiency of responses to inquiries from corporate customers and providing more advanced customer service.

[0258] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0259] Step 1:

[0260] A user inputs an inquiry into a corporate chatbot and clicks the send button. For example, they might input "Please tell me the renewal period for this phone number." The specific action is when the user inputs text into the chat window and clicks the send button. The input volume is the content of the user's inquiry, and the output volume is when this inquiry is transferred to the server.

[0261] Step 2:

[0262] The terminal receives the query. The terminal receives the query from the user in real time, converts the content into JSON format, and sends it to the server. The input is the text received from the user, and the output is JSON format data. Specifically, the terminal analyzes and structures the content of the query, and sends it to the server as an API request.

[0263] Step 3:

[0264] The server sends the received JSON data to the Google Cloud Natural Language API to identify the intent of the query. The input is JSON data, and the output is the analysis results, such as keywords and intent. Specifically, the server sends an API request to Google Cloud and receives the analysis results.

[0265] Step 4:

[0266] The server queries the company's internal database based on the analysis results to retrieve relevant information. The input is the analyzed keywords and intent, and the output is the relevant information as a result of the database query. Specifically, the server executes an SQL query to extract the required data.

[0267] Step 5:

[0268] Based on the information acquired by the server, it sends prompts to OpenAI's GPT-4 model to generate an appropriate answer to the query. The input is the information acquired from the database and the query content, and the output is the generated answer text. Specifically, the server provides the prompt to GPT-4 and executes an API request to generate the answer.

[0269] Step 6:

[0270] The server evaluates the reliability of the generated answer. The input is the generated answer text, and the output is the confidence score. Specifically, the server calculates the confidence of the answer and compares it with the configured threshold (CERTAINTY_THRESHOLD).

[0271] Step 7:

[0272] The server determines the appropriateness of the answer based on the evaluation results. The inputs are the confidence score and a threshold, and the output is the destination of the answer (automatic sending or escalation). Specifically, if the confidence exceeds the threshold, the answer is sent automatically, and if not, the answer is escalated to a human representative.

[0273] Step 8:

[0274] The server uses Azure Text Analytics to analyze user sentiment. The input is the user's query text, and the output is the sentiment analysis result. The specific operation is to send the query content to the API and receive the sentiment analysis result.

[0275] Step 9:

[0276] Based on the emotion analysis results, the server determines whether to escalate the inquiry to a person in charge. The input is the emotion analysis result, and the output is an instruction to escalate. Specifically, if anger or dissatisfaction is strong, an instruction to escalate is sent to a person in charge.

[0277] Step 10:

[0278] The server sends the final answer to the user. The input is the final confirmed answer text, and the output is the data sent to the user terminal. The specific operation is to send the answer using an HTTP response or the WebSocket protocol.

[0279] Step 11:

[0280] The server stores all interactions in a log. The inputs include the inquiry content, acquired information, generated answers, evaluation results, and sentiment analysis results, while the output is the stored log data. Specific operations include storing this data in a log file or database.

[0281] (Application example 2)

[0282] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0283] Conventional inquiry response systems have the problem of being unable to properly assess a user's emotions and respond accordingly when analyzing the content of inquiries and generating responses. This can result in an inability to respond appropriately to inquiries from users who are particularly emotional, which can lead to a decline in customer satisfaction. Furthermore, in some cases, escalation may not be performed appropriately if the response to an inquiry is unreliable, which also leads to a decline in customer satisfaction.

[0284] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an in-house system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative AI model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and escalating the answer to a human representative if it is not appropriate, means for analyzing emotions and adjusting the tone of the answer based on the analysis, means for sending the final answer to the user, and means for saving the inquiry and the answer in a log. This makes it possible to generate an appropriate answer that takes user emotions into consideration and provide a highly reliable answer.

[0285] The "means for receiving inquiries" refers to a device or software mechanism for electronically receiving inquiries from users.

[0286] "Means for analyzing received inquiries and identifying the content of the inquiries" refers to a mechanism for analyzing the content of received inquiries using natural language processing technology and identifying their intent and purpose.

[0287] "Means for accessing internal systems to obtain relevant information based on the identified inquiry content" refers to a mechanism for accessing internal systems or databases to obtain data related to the inquiry content.

[0288] "Means for generating responses to inquiries using a generative AI model based on acquired information" refers to a mechanism that uses a generative AI model (such as GPT-3) to generate appropriate responses in natural language format based on acquired data.

[0289] The "means for assessing the reliability of the generated answer" is a mechanism for calculating a confidence score to assess the accuracy and appropriateness of the generated answer.

[0290] "Means of determining whether the answer is appropriate based on the evaluation results, and escalating it to a human representative if it is not appropriate" is a mechanism that transfers the evaluated answer to a human representative if it does not meet certain reliability standards.

[0291] "Means for analyzing emotions and adjusting the tone of the response based on that" is a mechanism that analyzes emotions from the user's statements and text and generates a response in a tone that corresponds to those emotions.

[0292] The "means for transmitting the final answer to the user" is a mechanism for transmitting the generated final answer to the user using a communication means.

[0293] "Means for logging inquiries and responses" refers to a mechanism for storing all inquiries and responses as records.

[0294] The present invention is a system for effectively responding to user inquiries, which analyzes the inquiry, generates an answer, adjusts the tone of the response based on sentiment analysis, etc. An embodiment of this system will be described in detail below.

[0295] First, a user sends a query through a smartphone application. For example, a query such as "Please tell me the release date of the latest drama." The device receives this query in real time and transmits the query content to the server.

[0296] The server then analyzes the received query using natural language processing (NLP) techniques to determine the intent of the query, for example by extracting the keyword "publication date" and determining that it is related to a specific piece of content.

[0297] After identifying the query, the server accesses the company's internal systems via API to retrieve relevant information. Specifically, the server queries the company's internal database to retrieve information about the release dates of "new dramas," for example. This data is then formatted by the server and passed to the generative AI model.

[0298] Based on the acquired information, the server uses a generative AI model (e.g., GPT-3) to generate an appropriate answer to the query. The generated answer is provided in the form of natural language, such as "Thank you for your question. The release date of 'New Drama' is October 21, 2023."

[0299] The server then evaluates the reliability of the generated answer using a confidence score, and if the confidence exceeds a certain threshold, it returns the answer to the user directly, otherwise it escalates the answer to a human agent.

[0300] Furthermore, the server uses an emotion engine to analyze the user's emotions. It analyzes the language and emotional expressions contained in the inquiry, and if the user expresses dissatisfaction or anger, it immediately escalates the inquiry or adjusts the tone of the response. For example, if a user says, "I want an answer right away!", the server will respond with a polite tone, such as, "I'm sorry. I understand your dissatisfaction. I'll look into it right away."

[0301] Finally, the verified answer is sent to the user. Answers with high confidence or verified by a human agent are sent to the user's device via HTTP responses or real-time messaging protocols.

[0302] Finally, the server stores all interactions in a log, including the inquiry content, information obtained, responses generated, escalation history, sentiment analysis results, etc. These logs are used to improve the system and improve the accuracy of inquiry responses in the future.

[0303] Example prompt:

[0304] User Question: "What is the release date of the latest drama?"

[0305] The system responds: "The content in question is 'New Drama' and its release date is October 21, 2023."

[0306] User Question: "I want an answer right now!"

[0307] The system responds: "We're sorry, we understand your frustration. We'll look into it right away."

[0308] In this way, the embodiment of the present invention makes it possible to generate an appropriate answer that takes into account the user's feelings and to provide a highly reliable answer.

[0309] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0310] Step 1:

[0311] The user sends a query. For example, the user inputs a query such as "Please tell me the release date of the latest drama" through a smartphone application. The device receives this query in real time and transmits the query to the server.

[0312] Step 2:

[0313] The server analyzes the query. The server uses natural language processing (NLP) technology to analyze the received query and identify the intent of the query. Specifically, it extracts the keyword "publication date" and determines that the query is about the publication date of the content. The input for this analysis is the user's query, and the output is the analysis result (the intent of the query).

[0314] Step 3:

[0315] The server accesses the internal system to obtain information. Based on the specified query, the server accesses the internal database via API to obtain related information (e.g., content release date). The input of this step is the analysis result, and the output is the obtained information (e.g., the release date of the drama).

[0316] Step 4:

[0317] The server generates an answer using a generative AI model. Based on the acquired information, the server uses a generative AI model (e.g., GPT-3) to generate an answer in natural language format. For example, it creates an answer such as, "Thank you for your question. The release date of 'New Drama' is October 21, 2023." The input of this step is the acquired information, and the output is the generated answer.

[0318] Step 5:

[0319] The server evaluates the reliability of the generated answer. The server calculates a confidence score for the generated answer and determines whether this score exceeds a certain threshold. If the score is low, it escalates to a human agent. The input to this step is the generated answer and its confidence score, and the output is a branching process depending on whether escalation is necessary.

[0320] Step 6:

[0321] The server analyzes the emotion and adjusts the tone of the response. The server uses an emotion engine to analyze the emotion from the user's query. For example, if the user says, "I want an answer right away!", it determines that the user is expressing anger or dissatisfaction. As a result, the server creates a response in a tone such as, "I'm sorry. I understand your dissatisfaction. We will look into it right away." The input for this step is the user's query, and the output is a response that takes the emotion into consideration.

[0322] Step 7:

[0323] The server sends a final answer to the user. The server sends the answer that is determined to be reliable or the answer that has been confirmed after escalation to the user's device. This final answer is delivered via an HTTP response or real-time messaging protocol. The input of this step is the confirmed answer text, and the output is the answer sent to the user's device.

[0324] Step 8:

[0325] The server stores the queries and responses in a log. The server records and stores all interactions as log data. This data includes the query content, information obtained, responses generated, escalation history, and sentiment analysis results. The log data is used for future system improvements. The input of this step is a record of all interactions, and the output is the stored log data.

[0326] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0327] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0328] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0329] [Second embodiment]

[0330] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0331] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0332] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0333] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0334] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0335] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0336] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0337] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0338] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0339] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0340] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0341] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0342] The present invention is a system for effectively responding to inquiries from corporate customers, and includes a series of processes for receiving inquiries, analyzing them, and generating appropriate responses. An embodiment of this system will be described in detail below.

[0343] First, the user sends a query to the corporate chatbot. For example, the user might enter a query such as, "Please tell me the renewal period for this phone number." The device receives this query in real time and sends it to the server.

[0344] The server then analyzes the received inquiry using natural language processing technology to accurately understand the intent of the inquiry. For example, it can extract the keyword "renewal period" from the inquiry and identify it as a question about the renewal period for a phone number.

[0345] After identifying the query, the server accesses the internal system via API to retrieve relevant information. Specifically, the server queries the internal database to retrieve necessary data, such as "phone number renewal period." This data is then formatted by the server and passed to the generative AI model.

[0346] Based on the acquired information, the server uses a generative AI model to generate an appropriate response to the user's inquiry. The generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0347] The server evaluates the reliability of the generated answer. It calculates a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, it returns the answer to the user as is. On the other hand, if the confidence is low, the server escalates the answer to a human agent.

[0348] The server sends the final, verified answer to the user. If a reliable answer is generated automatically or if a suitable answer is prepared after human verification, the content is sent to the user's device. For example, "Your phone number renewal period ends on December 31, 2023."

[0349] Finally, the server stores all interactions in a log, including the inquiry content, the generated response, escalation history, etc. The stored log is used to improve the system and improve the accuracy of inquiry responses in the future.

[0350] This system reduces the time sales representatives spend on tedious inquiries, allowing them to focus on important sales activities. Providing fast and accurate information also contributes to improving customer satisfaction. Thus, the present invention is extremely useful as a means of streamlining inquiries from corporate customers and providing more advanced customer service.

[0351] The processing flow will be explained below.

[0352] Step 1:

[0353] A user sends a query to the enterprise chatbot, for example, "What is the renewal period for this phone number?"

[0354] Step 2:

[0355] The device receives this inquiry in real time and sends the inquiry content to the server. Specifically, it creates an HTTP request and sends it to the server.

[0356] Step 3:

[0357] The server analyzes the received inquiry, which involves using natural language processing (NLP) techniques to understand the intent of the text. It extracts the keyword "renewal period" from the inquiry and identifies it as a request related to the renewal period of a phone number.

[0358] Step 4:

[0359] The server then accesses the internal system via API to retrieve relevant information based on the identified inquiry. Specifically, the server sends a request to the appropriate API endpoint of the internal database to retrieve information about the "phone number renewal period."

[0360] Step 5:

[0361] The server formats the information and passes it to the generative AI model. For example, it converts the information into text data in the format "This phone number's renewal period ends December 31, 2023."

[0362] Step 6:

[0363] The server uses a generative AI model to generate an appropriate response to the inquiry. The generative AI model creates a response in natural language based on the acquired information. It generates a response in the format, such as, "Your phone number renewal period ends on December 31, 2023."

[0364] Step 7:

[0365] The server evaluates the reliability of the generated answer by calculating the confidence score of the generative AI model and checking whether the score exceeds a certain threshold.

[0366] Step 8:

[0367] The server determines whether the answer is appropriate based on the evaluation results. If the answer is highly reliable, it is sent to the user. If it is inappropriate, it is escalated to a sales representative.

[0368] Step 9:

[0369] The server then sends the final verified answer to the user, either authentic or verified by a human agent, via HTTP response or real-time messaging protocol to the user's device.

[0370] Step 10:

[0371] The server stores all interactions in a log, including inquiries, information obtained, responses generated, escalation history, etc. The log is used for future analysis and system improvement.

[0372] Example 1

[0373] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0374] Conventional inquiry response systems have difficulty providing quick and accurate responses to user inquiries, and often require human intervention. As a result, inquiries take time to respond, leading to issues such as a decline in customer satisfaction. Furthermore, there is a lack of efficient management of data generated during the inquiry response process and its use for future improvements.

[0375] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0376] In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an internal system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative artificial intelligence model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and, if not, for escalating to a human representative, means for sending the final answer to the user, and means for storing the inquiry and the answer in a record. This makes it possible to respond to user inquiries quickly and accurately, improving customer satisfaction and contributing to future system improvements.

[0377] The "means for receiving an inquiry" refers to a technique or device for inputting inquiry information such as text or voice sent from a user into a server via the Internet.

[0378] The "means of analyzing the received inquiry and identifying the content of the inquiry" refers to technology or algorithms that use natural language processing technology to break down the text data of the received inquiry and extract and understand the subject and keywords of the inquiry.

[0379] "Means of accessing internal systems to obtain relevant information based on the identified inquiry" refers to the technology or procedures by which a server sends requests to an internal company database or API to search for and obtain information related to the inquiry.

[0380] "Means for generating answers to inquiries using a generative AI model" refers to technologies or systems that use a generative AI model (e.g., GPT-3) based on acquired information to create answers in natural language format to user inquiries.

[0381] "Means for assessing the reliability of generated answers" refers to technologies or algorithms that evaluate the reliability and accuracy of answers output by generative AI models using criteria such as confidence scores.

[0382] "Means of determining whether the answer is appropriate based on the evaluation results, and escalating to a human operator if it is not appropriate" refers to the technology or process of determining whether the answer meets certain criteria based on the evaluated reliability score, and if it does not, handing over the processing to a human operator.

[0383] The "means for sending the final answer to the user" refers to a technique or means for returning an answer that has passed the reliability evaluation to the user in real time.

[0384] "Means for recording inquiries and responses" means techniques or methods for recording all inquiry and response interactions in log files or databases and storing them for future analysis and improvement.

[0385] The present invention provides a system for effectively responding to inquiries from corporate customers, which includes a series of processes for receiving and analyzing inquiries, and generating and returning appropriate answers.

[0386] First, the user sends a query to the corporate chatbot. For example, they might ask, "What is the renewal period for this phone number?" The device receives this query in real time and sends it to a server via the Internet.

[0387] The server analyzes the received query using natural language processing techniques (e.g., spaCy or NLTK). The server tokenizes the query text, extracts the keyword "renewal period," and understands that the query is about the renewal period for a phone number.

[0388] The server then accesses the internal system via an API to retrieve relevant information. Specifically, the server queries the internal database (e.g., Microsoft SQL Server or MySQL) to retrieve the necessary data. The retrieved information is then formatted and passed to the generative AI model.

[0389] The server uses a generative AI model (for example, OpenAI's GPT-3) to generate an appropriate response to the user's inquiry. Based on the acquired information, the generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0390] The server evaluates the reliability of the generated answer, calculates a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, returns the answer to the user as is. On the other hand, if the confidence is low, the answer is escalated to a human representative.

[0391] Finally, the server sends the final answer that passes the evaluation to the user. The terminal receives the answer and displays it on the chatbot interface. For example, "Your phone number renewal period ends on December 31, 2023."

[0392] The server also stores all inquiries and responses in a log, which includes the inquiry content, the generated responses, escalation history, etc. The stored logs are used to improve the system and the accuracy of inquiries in the future.

[0393] As a specific example, if a user sends a query such as "Please tell me the renewal period for this phone number," the device sends the query to the server, which uses natural language processing technology to extract the keyword "renewal period." The server accesses the internal system to obtain the corresponding information, and based on that, uses a generative AI model to generate the answer "The renewal period for your phone number is December 31, 2023." The answer that passes the evaluation is sent to the user, and the content of the query and answer is saved in a log.

[0394] Example prompt sentence:

[0395] User: How long do I need to renew this phone number?

[0396] Terminal: (sends user query to server)

[0397] Server: (analyzes the received query)

[0398] Server: (retrieving corresponding information from internal database)

[0399] Server: (Generate an answer using a generative AI model based on the acquired information)

[0400] Server: Your phone number is due for renewal on December 31, 2023.

[0401] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0402] Step 1:

[0403] The user sends a query to the chatbot. The user accesses the chatbot's interface, enters a question in the text field, and presses the send button. For example, the user might enter, "Please tell me the renewal period for this phone number." This sentence is the user's input.

[0404] Step 2:

[0405] The terminal sends a query to the server. The terminal receives input from the user and sends it as an HTTP POST request to the server's query acceptance endpoint. The input here is the user's query statement, and the output is the HTTP request sent to the server.

[0406] Step 3:

[0407] The server analyzes the query. The server analyzes the received text data and uses natural language processing techniques to understand the intent of the query. Specifically, the server tokenizes the text and extracts keywords and phrases. For example, it extracts the keyword "renewal period" and identifies it as being about the renewal period for a phone number. The input to this step is the query received by the server, and the output is the extracted keywords and their intent.

[0408] Step 4:

[0409] Based on the identified query, the server retrieves information from the internal system via API. The server sends a query to the internal database based on the keyword "phone number renewal period." For example, the server uses an SQL query to search for "phone number renewal period." The input for this step is the identified keyword, and the output is the information retrieved from the database.

[0410] Step 5:

[0411] The server generates an answer using a generative AI model based on the information acquired. The server inputs the acquired data into a generative AI model (e.g., GPT-3) and generates an answer in natural language format. Specifically, the server inputs the prompt "What is the renewal period for your phone number?" and obtains the generated answer "Your phone number renewal period is December 31, 2023." The inputs for this step are the acquired information and the prompt, and the output is the generated answer.

[0412] Step 6:

[0413] The server evaluates the reliability of the generated answer. The server calculates a confidence score for the answer and determines whether it exceeds a certain threshold. The reliability evaluation uses the scoring function of the generative AI model. For example, if the confidence score is evaluated as 0.85, which is above the set threshold of 0.80, the answer is sent as is. The input of this step is the generated answer, and the output is the result of the reliability evaluation.

[0414] Step 7:

[0415] The server sends a final answer to the user based on the confidence rating. If a reliable answer is generated, the server sends it back to the device as an HTTP response. The device displays the received answer in the chatbot interface. For example: "Your phone number renewal period is December 31, 2023." The input of this step is the evaluated answer, and the output is the answer displayed in the user's interface.

[0416] Step 8:

[0417] The server stores queries and responses in a log. The server records all interactions in a log file or database. The recorded data includes the query content, generated responses, trust assessment results, escalation history, etc. The input of this step is all processing results, and the output is the stored log data.

[0418] (Application example 1)

[0419] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0420] With conventional systems, it is difficult to respond appropriately and quickly to inquiries from corporate customers. Furthermore, due to the wide range of inquiries, sales representatives spend a lot of time responding to inquiries, preventing them from devoting sufficient time to their core sales activities. Furthermore, due to insufficient reliability evaluation of the generated answers and insufficient escalation procedures, customer satisfaction may decline.

[0421] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0422] In this invention, the server includes: means for receiving inquiries; means for analyzing the received inquiries to identify the inquiry content; means for accessing an internal system to obtain related information based on the identified inquiry content; means for generating an answer to the inquiry using a generative AI model based on the obtained information; means for evaluating the reliability of the generated answer; means for determining whether the answer is appropriate based on the evaluation result and, if not, for escalating to a human representative; means for sending the final answer to the user; means for saving the inquiry and the answer in a log; means implemented as a smartphone application that receives inquiries related to a specific service, analyzes them, and generates an answer; means for inputting and displaying the inquiry content through a user interface; and means for generating an answer by providing a prompt sentence to the generative AI model based on the inquiry content. This enables prompt and accurate answers to inquiries from corporate customers, allowing sales representatives to focus on their core business activities. Furthermore, the reliability evaluation of the generated answers and the escalation procedures are strengthened, thereby improving customer satisfaction.

[0423] The "means for receiving an inquiry" refers to a device or program for receiving an inquiry from a user in real time and transferring it to a server.

[0424] "Means for analyzing received inquiries and identifying the content of the inquiries" refers to devices or programs that use natural language processing technology to analyze the content of inquiries from users and accurately understand their intent.

[0425] "Means for accessing an internal system to obtain related information based on the identified inquiry content" refers to a device or program that works in conjunction with an API or database to enable the server to obtain information related to the identified inquiry content from the internal system.

[0426] "Means for generating a response to an inquiry using a generative AI model based on acquired information" refers to a device or program that uses acquired information to generate a response to an inquiry using a generative AI model.

[0427] "Means for evaluating the reliability of the generated answers" refers to devices or programs that calculate confidence scores and other factors to evaluate the accuracy and validity of answers generated by generative AI models.

[0428] "Means for determining whether the answer is appropriate based on the evaluation results, and escalating to a human staff member if it is not appropriate" refers to a device or program that determines the appropriateness of the answer generated based on the evaluation results, and if it is not appropriate, hands it over to a human staff member.

[0429] "Means for sending a final answer to the user" refers to a device or program that sends the content of a reliable answer to the user when a reliable answer is generated or when an appropriate answer is prepared after human verification.

[0430] "Means for saving inquiries and responses in a log" refers to a device or program that saves all interactions, inquiry content, generated responses, and escalation history in a log, and uses this information to improve the system and increase the accuracy of inquiry responses.

[0431] "Means implemented as a smartphone application that receives inquiries related to a specific service, analyzes them, and generates a response" refers to a device or program that receives inquiries related to a specific service through a smartphone application, analyzes the content of the inquiries, and generates a response.

[0432] The "means for inputting and displaying inquiry details through a user interface" refers to an interface through which a user inputs inquiry details and displays responses to those inquiries.

[0433] "Means for generating an answer by providing a prompt sentence to a generative AI model based on the content of the inquiry" refers to a device or program that generates a prompt sentence based on the content of the user's inquiry, passes it to a generative AI model, and generates an answer.

[0434] The following is a detailed description of an embodiment of the present invention: The system of the present invention includes a smartphone application and a server for quickly and accurately responding to inquiries from corporate customers.

[0435] The overall system overview is as follows:

[0436] 1. When a user enters an inquiry using a smartphone application, the inquiry is received by the device in real time.

[0437] 2. The received query is forwarded to the server, which analyzes the query using natural language processing technology to properly identify the query content and intent.

[0438] 3. Based on the identified query, the server accesses internal databases and systems to retrieve the relevant information needed.

[0439] 4. Based on the acquired information, the server uses a generative AI model to generate an appropriate answer. At the time of generation, the prompt sentence is provided to the generative AI model.

[0440] 5. The reliability of the generated answer is assessed by a confidence score: if the confidence is above a certain threshold, it is automatically sent to the user; if it is lower, it is escalated to a human agent.

[0441] 6. The final answer is sent to the user, and the system keeps a log of all interactions.

[0442] Hardware and software used

[0443] Hardware: Cloud servers (e.g., Amazon Web Services, Google Cloud Platform), smartphones (iOS or Android)

[0444] Software: Flask (a lightweight Python web framework), OpenAI's GPT-3 API, natural language processing libraries (e.g., spaCy, NLTK)

[0445] Example of processing flow

[0446] As a specific example, let us consider an inquiry about the release date of a new movie.

[0447] Example inquiry: "When is the new movie coming out?"

[0448] The server receives this query and uses natural language processing technology to identify the "release date of new movies."

[0449] Based on the identified information, the company accesses internal databases to retrieve relevant information.

[0450] The acquired information is provided to the generative AI model as a prompt sentence.

[0451] Prompt Sentence Examples

[0452] Examples of prompts are:

[0453] Text format

[0454] User Asks: When is the new movie coming out?

[0455] answer:

[0456] By passing this prompt to the OpenAI GPT-3 API, a specific answer such as "The release date of the new movie will be November 1, 2023" will be generated.

[0457] The above process makes it possible to respond to inquiries from corporate customers quickly and accurately. This system reduces the burden on sales representatives in responding to inquiries, allowing them to focus more on full-scale sales activities. In addition, by strengthening the reliability evaluation of responses and escalation procedures, customer satisfaction can be improved.

[0458] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0459] Step 1:

[0460] The device receives inquiries from users in real time. At this time, the user inputs the inquiry content through the smartphone application interface. The input inquiry is saved in text format on the device and temporarily stored within the device.

[0461] Input: User's inquiry (text format)

[0462] Output: Received inquiry (text format)

[0463] Step 2:

[0464] The terminal sends the received query to the server using a communication protocol such as an HTTP POST request. The server receives this request and prepares to analyze the query.

[0465] Input: Received inquiry (text format)

[0466] Output: Query content forwarded to the server (text format)

[0467] Step 3:

[0468] The server analyzes the received query and identifies the query content. Here, natural language processing technology (e.g., spaCy or NLTK) is used to extract the query intent and keywords. For example, the keyword "release date of new movies" is identified.

[0469] Input: Inquiry content forwarded to the server (text format)

[0470] Output: Identified inquiry content and keywords (text format)

[0471] Step 4:

[0472] The server accesses the company's internal systems based on the identified inquiry content and retrieves relevant information, such as data such as "release dates for new movies" using the company's internal databases and APIs.

[0473] Input: Identified inquiry content and keywords (text format)

[0474] Output: Relevant information retrieved (in text, possibly structured format)

[0475] Step 5:

[0476] Based on the acquired information, the server provides a prompt sentence to a generative AI model (e.g., OpenAI GPT-3) to generate an answer to the query. At this time, the prompt sentence is input to the generative AI model as a sentence with a specific structure.

[0477] Input: Retrieved relevant information (text format)

[0478] Output: Answers generated by the generative AI model (in text format)

[0479] Step 6:

[0480] The server evaluates the reliability of the generated answer using an algorithm that calculates a confidence score, determines whether the confidence score exceeds a certain threshold, and escalates it if necessary.

[0481] Input: Generated answer (text format)

[0482] Output: Reliability assessment results and confidence scores (numerical data)

[0483] Step 7:

[0484] If the final answer passes the trust assessment, the server sends it to the user, which returns it to the device as an HTTP response. If the trust is low, the request is escalated to a human agent.

[0485] Input: Reliability assessment result and answer (text format)

[0486] Output: Final answer (text format)

[0487] Step 8:

[0488] The server stores all interactions in a log, including the inquiry content, generated answers, reliability assessment results, escalation history, etc. The log is used for future system improvements and analysis.

[0489] Input: All correspondence (text format)

[0490] Output: Saved logs (database entries)

[0491] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0492] The present invention is a system for effectively responding to inquiries from corporate customers, which includes a series of processes for receiving inquiries, analyzing them, and generating appropriate answers, as well as an emotion engine for recognizing user emotions. An embodiment of this system will be described in detail below.

[0493] First, the user sends a query to the corporate chatbot. For example, they enter a query such as, "Please tell me the renewal period for this phone number." The device receives this query in real time and sends the query to the server.

[0494] The server then analyzes the received inquiry using natural language processing (NLP) techniques to accurately understand the intent of the inquiry. For example, it can extract the keyword "renewal period" from the inquiry and identify it as a query about the renewal period for a phone number.

[0495] After identifying the query, the server accesses internal systems via APIs to retrieve relevant information. Specifically, the server queries an internal database to retrieve information about, for example, phone number renewal periods. This data is then formatted by the server and passed to the generative AI model.

[0496] Based on the acquired information, the server uses a generative AI model to generate an appropriate response to the user's inquiry. The generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0497] The server then evaluates the reliability of the generated answer, calculating a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, the server returns the answer to the user as is. On the other hand, if the confidence score is low, the server escalates the answer to a human agent.

[0498] Furthermore, the present invention includes an emotion engine for recognizing the user's emotions. The server analyzes the language and emotional expressions contained in the user's inquiry, and if it determines that the user is expressing anger or frustration, it immediately escalates the inquiry to a human agent. The emotion engine adjusts the tone of the response depending on the user's emotions, for example, generating a more polite and reassuring response if the user is expressing frustration.

[0499] The server sends the final, verified answer to the user. The answer, either trusted or verified by a human agent, is sent to the user's device via an HTTP response or real-time messaging protocol. Example: "Your phone number is due for renewal on December 31, 2023."

[0500] Finally, the server stores all interactions in a log, including the inquiry content, information obtained, responses generated, escalation history, and the results of the analysis of recognized emotions. The stored logs are used to improve the system and enhance the accuracy of inquiry responses in the future.

[0501] As described above, the present invention reduces the time that sales representatives spend on tedious inquiries, creating an environment in which they can concentrate on important sales activities. Furthermore, the introduction of an emotion engine enables responses that take into consideration the user's emotions, contributing to improved customer satisfaction. Thus, the present invention is extremely useful as a means for streamlining inquiries from corporate customers and providing more advanced customer service.

[0502] The processing flow will be explained below.

[0503] Step 1:

[0504] A user sends a query to the enterprise chatbot, for example, "What is the renewal period for this phone number?"

[0505] Step 2:

[0506] The device receives this inquiry in real time and sends the inquiry content to the server. Specifically, it creates an HTTP request and sends it to the server.

[0507] Step 3:

[0508] The server analyzes the received inquiry, which involves using natural language processing (NLP) techniques to understand the intent of the text. It extracts the keyword "renewal period" from the inquiry and identifies it as a request related to the renewal period of a phone number.

[0509] Step 4:

[0510] The server then accesses the internal system via API to retrieve relevant information based on the identified inquiry. Specifically, the server sends a request to the appropriate API endpoint of the internal database to retrieve information about the "phone number renewal period."

[0511] Step 5:

[0512] The server formats the information and passes it to the generative AI model. For example, it converts the information into text data in the format "This phone number's renewal period ends December 31, 2023."

[0513] Step 6:

[0514] The server uses a generative AI model to generate an appropriate response to the inquiry. The generative AI model creates a response in natural language based on the acquired information. It generates a response in the format, such as, "Your phone number renewal period ends on December 31, 2023."

[0515] Step 7:

[0516] The server evaluates the reliability of the generated answer by calculating the confidence score of the generative AI model and checking whether the score exceeds a certain threshold.

[0517] Step 8:

[0518] The server determines whether the answer is appropriate based on the evaluation results. If the answer is highly reliable, it is sent to the user. If it is inappropriate, it is escalated to a sales representative.

[0519] Step 9:

[0520] The server uses an emotion engine to recognize the emotions contained in the user's query, analyzing the language and emotional expressions to determine whether the user is expressing anger or frustration.

[0521] Step 10:

[0522] The server adjusts the tone of the response depending on the user's emotions, for example, generating a more polite and reassuring response if the user is expressing anger.

[0523] Step 11:

[0524] The server sets escalation conditions based on the recognized emotions, and if the user expresses anger or frustration, the call is automatically escalated to a human agent.

[0525] Step 12:

[0526] The server then sends the final, verified answer to the user. The answer, either trusted or verified by a human agent, is sent to the user's device via an HTTP response or real-time messaging protocol. Example: "Your phone number renewal period ends December 31, 2023."

[0527] Step 13:

[0528] The server stores all interactions in a log, including the inquiry, the information obtained, the response generated, the escalation history, the analysis results of the recognized sentiment, etc. The stored log is used for future analysis and system improvement.

[0529] Example 2

[0530] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0531] Conventional corporate chatbot systems have several limitations in analyzing user inquiries and generating answers. For example, there are insufficient means to ensure the reliability of answers and a lack of consideration for user feelings, which can lead to a decline in the quality of inquiry responses. In addition, the system's automated responses are often unreliable, resulting in excessive reliance on human agents. These issues have led to issues such as delayed response times and reduced user satisfaction.

[0532] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an internal system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative AI model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and escalating the inquiry to a human agent if it is not appropriate, means for performing emotion analysis and escalating the inquiry to a human agent based on the user's emotions, means for sending a final answer to the user, and means for saving the inquiry and the answer in a log. This improves the accuracy of inquiry analysis and answer generation, enabling quick and reliable responses that also take user emotions into consideration.

[0533] The "means for receiving an inquiry" is a mechanism for transferring the contents of an inquiry entered by a user to a server.

[0534] The "means for analyzing the received inquiry and identifying the content of the inquiry" is a mechanism for analyzing the received inquiry using natural language processing technology, and extracting and identifying the intent and content of the inquiry.

[0535] "Means for accessing internal systems to obtain relevant information based on the identified inquiry" refers to a mechanism for accessing internal databases and other internal resources to obtain the necessary information based on the identified inquiry.

[0536] "Means for generating an answer to a query using a generative AI model based on acquired information" refers to a mechanism for generating an appropriate answer to a user's query using a generative AI model with acquired information as input.

[0537] The "means for assessing the reliability of the generated answer" is a mechanism for calculating a confidence score for the answer generated by the generative AI model and assessing its reliability.

[0538] "Means for determining whether the response is appropriate based on the evaluation results, and escalating to a human representative if it is not appropriate" is a mechanism for determining the appropriateness of the response based on the results of a reliability evaluation, and transferring the inquiry to a human representative if necessary.

[0539] The "means for performing emotion analysis and escalating inquiries to a human agent based on the user's emotions" is a mechanism for analyzing the emotions contained in a user's inquiry and transferring the inquiry to a human agent based on the results.

[0540] The "means for transmitting the final answer to the user" is a mechanism for transmitting the final confirmed answer to the user's terminal.

[0541] The "means for saving inquiries and responses in a log" is a mechanism for saving the inquiry content, acquired information, generated responses, and the results of escalation and sentiment analysis in a log.

[0542] MODE FOR CARRYING OUT THE INVENTION

[0543] The present invention is a system for responding effectively and quickly to inquiries from corporate customers. This system not only receives inquiries, analyzes them, and generates appropriate responses, but also includes an emotion engine that recognizes the emotions of users.

[0544] System hardware and software configuration

[0545] 1. User device: The device (PC, smartphone, tablet, etc.) through which the user sends an inquiry via the chatbot.

[0546] 2. Server: The central computing resource that performs various processes, such as query analysis, information retrieval, answer generation, credibility assessment, and sentiment analysis.

[0547] Natural Language Processing (NLP): Uses Google Cloud Natural Language API.

[0548] Internal system access: Access to our internal database (MySQL).

[0549] Generative AI model: OpenAI's GPT-4 is used.

[0550] Sentiment analysis: Using Azure Text Analytics.

[0551] System Operation

[0552] The user enters the question "Please tell me the renewal period for this phone number" into the chatbot interface. The device receives this query in real time and sends it to the server.

[0553] The server analyzes the received query using the Google Cloud Natural Language API. Based on the analysis results, it identifies the keyword "renewal period" and determines that the query is about the phone number renewal period.

[0554] The server then queries the company's internal database using MySQL to retrieve relevant data. Based on this information, the server uses OpenAI's GPT-4 model to generate an appropriate answer, such as "Your phone number is due for renewal on December 31, 2023."

[0555] To assess the reliability of the generated answer, the server calculates a confidence score and automatically sends the answer to the user if it exceeds a set threshold (CERTAINTY_THRESHOLD), or escalates the answer to a human agent if the confidence is low.

[0556] Additionally, the server uses Azure Text Analytics to analyze the user's emotions and, if it determines that the user is expressing anger or frustration, it will immediately escalate the call to a human agent.

[0557] The final answer is sent from the server to the device via an HTTP response or WebSocket protocol. For example, the answer may say, "Your phone number renewal period ends on December 31, 2023."

[0558] All interactions are logged by the server, including the inquiry, the information obtained, the response generated, the escalation history, and the results of sentiment analysis, which can be used to improve the system and its accuracy.

[0559] Examples of concrete examples and prompts

[0560] For example, if a user types "What is the expiration date of my contract?", the server uses NLP technology to identify the keyword "contract expiration date" and retrieves relevant information from the company's internal database. The generative AI model creates an answer such as "Your contract expires on March 31, 2024," which is sent to the user after passing a reliability assessment. If the user expresses strong dissatisfaction, the emotion engine will be activated and a human agent will manually process the request.

[0561] An example prompt is:

[0562] "Please tell me the renewal period for this phone number."

[0563] Please tell me the expiration date of the contract.

[0564] As described above, the present invention is extremely useful as a means for improving the efficiency of responses to inquiries from corporate customers and providing more advanced customer service.

[0565] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0566] Step 1:

[0567] A user inputs an inquiry into a corporate chatbot and clicks the send button. For example, they might input "Please tell me the renewal period for this phone number." The specific action is when the user inputs text into the chat window and clicks the send button. The input volume is the content of the user's inquiry, and the output volume is when this inquiry is transferred to the server.

[0568] Step 2:

[0569] The terminal receives the query. The terminal receives the query from the user in real time, converts the content into JSON format, and sends it to the server. The input is the text received from the user, and the output is JSON format data. Specifically, the terminal analyzes and structures the content of the query, and sends it to the server as an API request.

[0570] Step 3:

[0571] The server sends the received JSON data to the Google Cloud Natural Language API to identify the intent of the query. The input is JSON data, and the output is the analysis results, such as keywords and intent. Specifically, the server sends an API request to Google Cloud and receives the analysis results.

[0572] Step 4:

[0573] The server queries the company's internal database based on the analysis results to retrieve relevant information. The input is the analyzed keywords and intent, and the output is the relevant information as a result of the database query. Specifically, the server executes an SQL query to extract the required data.

[0574] Step 5:

[0575] Based on the information acquired by the server, it sends prompts to OpenAI's GPT-4 model to generate an appropriate answer to the query. The input is the information acquired from the database and the query content, and the output is the generated answer text. Specifically, the server provides the prompt to GPT-4 and executes an API request to generate the answer.

[0576] Step 6:

[0577] The server evaluates the reliability of the generated answer. The input is the generated answer text, and the output is the confidence score. Specifically, the server calculates the confidence of the answer and compares it with the configured threshold (CERTAINTY_THRESHOLD).

[0578] Step 7:

[0579] The server determines the appropriateness of the answer based on the evaluation results. The inputs are the confidence score and a threshold, and the output is the destination of the answer (automatic sending or escalation). Specifically, if the confidence exceeds the threshold, the answer is sent automatically, and if not, the answer is escalated to a human representative.

[0580] Step 8:

[0581] The server uses Azure Text Analytics to analyze user sentiment. The input is the user's query text, and the output is the sentiment analysis result. The specific operation is to send the query content to the API and receive the sentiment analysis result.

[0582] Step 9:

[0583] Based on the emotion analysis results, the server determines whether to escalate the inquiry to a person in charge. The input is the emotion analysis result, and the output is an instruction to escalate. Specifically, if anger or dissatisfaction is strong, an instruction to escalate is sent to a person in charge.

[0584] Step 10:

[0585] The server sends the final answer to the user. The input is the final confirmed answer text, and the output is the data sent to the user terminal. The specific operation is to send the answer using an HTTP response or the WebSocket protocol.

[0586] Step 11:

[0587] The server stores all interactions in a log. The inputs include the inquiry content, acquired information, generated answers, evaluation results, and sentiment analysis results, while the output is the stored log data. Specific operations include storing this data in a log file or database.

[0588] (Application example 2)

[0589] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0590] Conventional inquiry response systems have the problem of being unable to properly assess a user's emotions and respond accordingly when analyzing the content of inquiries and generating responses. This can result in an inability to respond appropriately to inquiries from users who are particularly emotional, which can lead to a decline in customer satisfaction. Furthermore, in some cases, escalation may not be performed appropriately if the response to an inquiry is unreliable, which also leads to a decline in customer satisfaction.

[0591] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an in-house system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative AI model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and escalating the answer to a human representative if it is not appropriate, means for analyzing emotions and adjusting the tone of the answer based on the analysis, means for sending the final answer to the user, and means for saving the inquiry and the answer in a log. This makes it possible to generate an appropriate answer that takes user emotions into consideration and provide a highly reliable answer.

[0592] The "means for receiving inquiries" refers to a device or software mechanism for electronically receiving inquiries from users.

[0593] "Means for analyzing received inquiries and identifying the content of the inquiries" refers to a mechanism for analyzing the content of received inquiries using natural language processing technology and identifying their intent and purpose.

[0594] "Means for accessing internal systems to obtain relevant information based on the identified inquiry content" refers to a mechanism for accessing internal systems or databases to obtain data related to the inquiry content.

[0595] "Means for generating responses to inquiries using a generative AI model based on acquired information" refers to a mechanism that uses a generative AI model (such as GPT-3) to generate appropriate responses in natural language format based on acquired data.

[0596] The "means for assessing the reliability of the generated answer" is a mechanism for calculating a confidence score to assess the accuracy and appropriateness of the generated answer.

[0597] "Means of determining whether the answer is appropriate based on the evaluation results, and escalating it to a human representative if it is not appropriate" is a mechanism that transfers the evaluated answer to a human representative if it does not meet certain reliability standards.

[0598] "Means for analyzing emotions and adjusting the tone of the response based on that" is a mechanism that analyzes emotions from the user's statements and text and generates a response in a tone that corresponds to those emotions.

[0599] The "means for transmitting the final answer to the user" is a mechanism for transmitting the generated final answer to the user using a communication means.

[0600] "Means for logging inquiries and responses" refers to a mechanism for storing all inquiries and responses as records.

[0601] The present invention is a system for effectively responding to user inquiries, which analyzes the inquiry, generates an answer, adjusts the tone of the response based on sentiment analysis, etc. An embodiment of this system will be described in detail below.

[0602] First, a user sends a query through a smartphone application. For example, a query such as "Please tell me the release date of the latest drama." The device receives this query in real time and transmits the query content to the server.

[0603] The server then analyzes the received query using natural language processing (NLP) techniques to determine the intent of the query, for example by extracting the keyword "publication date" and determining that it is related to a specific piece of content.

[0604] After identifying the query, the server accesses the company's internal systems via API to retrieve relevant information. Specifically, the server queries the company's internal database to retrieve information about the release dates of "new dramas," for example. This data is then formatted by the server and passed to the generative AI model.

[0605] Based on the acquired information, the server uses a generative AI model (e.g., GPT-3) to generate an appropriate answer to the query. The generated answer is provided in the form of natural language, such as "Thank you for your question. The release date of 'New Drama' is October 21, 2023."

[0606] The server then evaluates the reliability of the generated answer using a confidence score, and if the confidence exceeds a certain threshold, it returns the answer to the user directly, otherwise it escalates the answer to a human agent.

[0607] Furthermore, the server uses an emotion engine to analyze the user's emotions. It analyzes the language and emotional expressions contained in the inquiry, and if the user expresses dissatisfaction or anger, it immediately escalates the inquiry or adjusts the tone of the response. For example, if a user says, "I want an answer right away!", the server will respond with a polite tone, such as, "I'm sorry. I understand your dissatisfaction. I'll look into it right away."

[0608] Finally, the verified answer is sent to the user. Answers with high confidence or verified by a human agent are sent to the user's device via HTTP responses or real-time messaging protocols.

[0609] Finally, the server stores all interactions in a log, including the inquiry content, information obtained, responses generated, escalation history, sentiment analysis results, etc. These logs are used to improve the system and improve the accuracy of inquiry responses in the future.

[0610] Example prompt:

[0611] User Question: "What is the release date of the latest drama?"

[0612] The system responds: "The content in question is 'New Drama' and its release date is October 21, 2023."

[0613] User Question: "I want an answer right now!"

[0614] The system responds: "We're sorry, we understand your frustration. We'll look into it right away."

[0615] In this way, the embodiment of the present invention makes it possible to generate an appropriate answer that takes into account the user's feelings and to provide a highly reliable answer.

[0616] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0617] Step 1:

[0618] The user sends a query. For example, the user inputs a query such as "Please tell me the release date of the latest drama" through a smartphone application. The device receives this query in real time and transmits the query to the server.

[0619] Step 2:

[0620] The server analyzes the query. The server uses natural language processing (NLP) technology to analyze the received query and identify the intent of the query. Specifically, it extracts the keyword "publication date" and determines that the query is about the publication date of the content. The input for this analysis is the user's query, and the output is the analysis result (the intent of the query).

[0621] Step 3:

[0622] The server accesses the internal system to obtain information. Based on the specified query, the server accesses the internal database via API to obtain related information (e.g., content release date). The input of this step is the analysis result, and the output is the obtained information (e.g., the release date of the drama).

[0623] Step 4:

[0624] The server generates an answer using a generative AI model. Based on the acquired information, the server uses a generative AI model (e.g., GPT-3) to generate an answer in natural language format. For example, it creates an answer such as, "Thank you for your question. The release date of 'New Drama' is October 21, 2023." The input of this step is the acquired information, and the output is the generated answer.

[0625] Step 5:

[0626] The server evaluates the reliability of the generated answer. The server calculates a confidence score for the generated answer and determines whether this score exceeds a certain threshold. If the score is low, it escalates to a human agent. The input to this step is the generated answer and its confidence score, and the output is a branching process depending on whether escalation is necessary.

[0627] Step 6:

[0628] The server analyzes the emotion and adjusts the tone of the response. The server uses an emotion engine to analyze the emotion from the user's query. For example, if the user says, "I want an answer right away!", it determines that the user is expressing anger or dissatisfaction. As a result, the server creates a response in a tone such as, "I'm sorry. I understand your dissatisfaction. We will look into it right away." The input for this step is the user's query, and the output is a response that takes the emotion into consideration.

[0629] Step 7:

[0630] The server sends a final answer to the user. The server sends the answer that is determined to be reliable or the answer that has been confirmed after escalation to the user's device. This final answer is delivered via an HTTP response or real-time messaging protocol. The input of this step is the confirmed answer text, and the output is the answer sent to the user's device.

[0631] Step 8:

[0632] The server stores the queries and responses in a log. The server records and stores all interactions as log data. This data includes the query content, information obtained, responses generated, escalation history, and sentiment analysis results. The log data is used for future system improvements. The input of this step is a record of all interactions, and the output is the stored log data.

[0633] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0634] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0635] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0636] [Third embodiment]

[0637] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0638] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0639] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0640] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0641] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0642] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0643] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0644] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0645] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0646] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0647] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0648] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0649] The present invention is a system for effectively responding to inquiries from corporate customers, and includes a series of processes for receiving inquiries, analyzing them, and generating appropriate responses. An embodiment of this system will be described in detail below.

[0650] First, the user sends a query to the corporate chatbot. For example, the user might enter a query such as, "Please tell me the renewal period for this phone number." The device receives this query in real time and sends it to the server.

[0651] The server then analyzes the received inquiry using natural language processing technology to accurately understand the intent of the inquiry. For example, it can extract the keyword "renewal period" from the inquiry and identify it as a question about the renewal period for a phone number.

[0652] After identifying the query, the server accesses the internal system via API to retrieve relevant information. Specifically, the server queries the internal database to retrieve necessary data, such as "phone number renewal period." This data is then formatted by the server and passed to the generative AI model.

[0653] Based on the acquired information, the server uses a generative AI model to generate an appropriate response to the user's inquiry. The generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0654] The server evaluates the reliability of the generated answer. It calculates a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, it returns the answer to the user as is. On the other hand, if the confidence is low, the server escalates the answer to a human agent.

[0655] The server sends the final, verified answer to the user. If a reliable answer is generated automatically or if a suitable answer is prepared after human verification, the content is sent to the user's device. For example, "Your phone number renewal period ends on December 31, 2023."

[0656] Finally, the server stores all interactions in a log, including the inquiry content, the generated response, escalation history, etc. The stored log is used to improve the system and improve the accuracy of inquiry responses in the future.

[0657] This system reduces the time sales representatives spend on tedious inquiries, allowing them to focus on important sales activities. Providing fast and accurate information also contributes to improving customer satisfaction. Thus, the present invention is extremely useful as a means of streamlining inquiries from corporate customers and providing more advanced customer service.

[0658] The processing flow will be explained below.

[0659] Step 1:

[0660] A user sends a query to the enterprise chatbot, for example, "What is the renewal period for this phone number?"

[0661] Step 2:

[0662] The device receives this inquiry in real time and sends the inquiry content to the server. Specifically, it creates an HTTP request and sends it to the server.

[0663] Step 3:

[0664] The server analyzes the received inquiry, which involves using natural language processing (NLP) techniques to understand the intent of the text. It extracts the keyword "renewal period" from the inquiry and identifies it as a request related to the renewal period of a phone number.

[0665] Step 4:

[0666] The server then accesses the internal system via API to retrieve relevant information based on the identified inquiry. Specifically, the server sends a request to the appropriate API endpoint of the internal database to retrieve information about the "phone number renewal period."

[0667] Step 5:

[0668] The server formats the information and passes it to the generative AI model. For example, it converts the information into text data in the format "This phone number's renewal period ends December 31, 2023."

[0669] Step 6:

[0670] The server uses a generative AI model to generate an appropriate response to the inquiry. The generative AI model creates a response in natural language based on the acquired information. It generates a response in the format, such as, "Your phone number renewal period ends on December 31, 2023."

[0671] Step 7:

[0672] The server evaluates the reliability of the generated answer by calculating the confidence score of the generative AI model and checking whether the score exceeds a certain threshold.

[0673] Step 8:

[0674] The server determines whether the answer is appropriate based on the evaluation results. If the answer is highly reliable, it is sent to the user. If it is inappropriate, it is escalated to a sales representative.

[0675] Step 9:

[0676] The server then sends the final verified answer to the user, either authentic or verified by a human agent, via HTTP response or real-time messaging protocol to the user's device.

[0677] Step 10:

[0678] The server stores all interactions in a log, including inquiries, information obtained, responses generated, escalation history, etc. The log is used for future analysis and system improvement.

[0679] Example 1

[0680] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0681] Conventional inquiry response systems have difficulty providing quick and accurate responses to user inquiries, and often require human intervention. As a result, inquiries take time to respond, leading to issues such as a decline in customer satisfaction. Furthermore, there is a lack of efficient management of data generated during the inquiry response process and its use for future improvements.

[0682] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0683] In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an internal system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative artificial intelligence model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and, if not, for escalating to a human representative, means for sending the final answer to the user, and means for storing the inquiry and the answer in a record. This makes it possible to respond to user inquiries quickly and accurately, improving customer satisfaction and contributing to future system improvements.

[0684] The "means for receiving an inquiry" refers to a technique or device for inputting inquiry information such as text or voice sent from a user into a server via the Internet.

[0685] The "means of analyzing the received inquiry and identifying the content of the inquiry" refers to technology or algorithms that use natural language processing technology to break down the text data of the received inquiry and extract and understand the subject and keywords of the inquiry.

[0686] "Means of accessing internal systems to obtain relevant information based on the identified inquiry" refers to the technology or procedures by which a server sends requests to an internal company database or API to search for and obtain information related to the inquiry.

[0687] "Means for generating answers to inquiries using a generative AI model" refers to technologies or systems that use a generative AI model (e.g., GPT-3) based on acquired information to create answers in natural language format to user inquiries.

[0688] "Means for assessing the reliability of generated answers" refers to technologies or algorithms that evaluate the reliability and accuracy of answers output by generative AI models using criteria such as confidence scores.

[0689] "Means of determining whether the answer is appropriate based on the evaluation results, and escalating to a human operator if it is not appropriate" refers to the technology or process of determining whether the answer meets certain criteria based on the evaluated reliability score, and if it does not, handing over the processing to a human operator.

[0690] The "means for sending the final answer to the user" refers to a technique or means for returning an answer that has passed the reliability evaluation to the user in real time.

[0691] "Means for recording inquiries and responses" means techniques or methods for recording all inquiry and response interactions in log files or databases and storing them for future analysis and improvement.

[0692] The present invention provides a system for effectively responding to inquiries from corporate customers, which includes a series of processes for receiving and analyzing inquiries, and generating and returning appropriate answers.

[0693] First, the user sends a query to the corporate chatbot. For example, they might ask, "What is the renewal period for this phone number?" The device receives this query in real time and sends it to a server via the Internet.

[0694] The server analyzes the received query using natural language processing techniques (e.g., spaCy or NLTK). The server tokenizes the query text, extracts the keyword "renewal period," and understands that the query is about the renewal period for a phone number.

[0695] The server then accesses the internal system via an API to retrieve relevant information. Specifically, the server queries the internal database (e.g., Microsoft SQL Server or MySQL) to retrieve the necessary data. The retrieved information is then formatted and passed to the generative AI model.

[0696] The server uses a generative AI model (for example, OpenAI's GPT-3) to generate an appropriate response to the user's inquiry. Based on the acquired information, the generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0697] The server evaluates the reliability of the generated answer, calculates a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, returns the answer to the user as is. On the other hand, if the confidence is low, the answer is escalated to a human representative.

[0698] Finally, the server sends the final answer that passes the evaluation to the user. The terminal receives the answer and displays it on the chatbot interface. For example, "Your phone number renewal period ends on December 31, 2023."

[0699] The server also stores all inquiries and responses in a log, which includes the inquiry content, the generated responses, escalation history, etc. The stored logs are used to improve the system and the accuracy of inquiries in the future.

[0700] As a specific example, if a user sends a query such as "Please tell me the renewal period for this phone number," the device sends the query to the server, which uses natural language processing technology to extract the keyword "renewal period." The server accesses the internal system to obtain the corresponding information, and based on that, uses a generative AI model to generate the answer "The renewal period for your phone number is December 31, 2023." The answer that passes the evaluation is sent to the user, and the content of the query and answer is saved in a log.

[0701] Example prompt sentence:

[0702] User: How long do I need to renew this phone number?

[0703] Terminal: (sends user query to server)

[0704] Server: (analyzes the received query)

[0705] Server: (retrieving corresponding information from internal database)

[0706] Server: (Generate an answer using a generative AI model based on the acquired information)

[0707] Server: Your phone number is due for renewal on December 31, 2023.

[0708] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0709] Step 1:

[0710] The user sends a query to the chatbot. The user accesses the chatbot's interface, enters a question in the text field, and presses the send button. For example, the user might enter, "Please tell me the renewal period for this phone number." This sentence is the user's input.

[0711] Step 2:

[0712] The terminal sends a query to the server. The terminal receives input from the user and sends it as an HTTP POST request to the server's query acceptance endpoint. The input here is the user's query statement, and the output is the HTTP request sent to the server.

[0713] Step 3:

[0714] The server analyzes the query. The server analyzes the received text data and uses natural language processing techniques to understand the intent of the query. Specifically, the server tokenizes the text and extracts keywords and phrases. For example, it extracts the keyword "renewal period" and identifies it as being about the renewal period for a phone number. The input to this step is the query received by the server, and the output is the extracted keywords and their intent.

[0715] Step 4:

[0716] Based on the identified query, the server retrieves information from the internal system via API. The server sends a query to the internal database based on the keyword "phone number renewal period." For example, the server uses an SQL query to search for "phone number renewal period." The input for this step is the identified keyword, and the output is the information retrieved from the database.

[0717] Step 5:

[0718] The server generates an answer using a generative AI model based on the information acquired. The server inputs the acquired data into a generative AI model (e.g., GPT-3) and generates an answer in natural language format. Specifically, the server inputs the prompt "What is the renewal period for your phone number?" and obtains the generated answer "Your phone number renewal period is December 31, 2023." The inputs for this step are the acquired information and the prompt, and the output is the generated answer.

[0719] Step 6:

[0720] The server evaluates the reliability of the generated answer. The server calculates a confidence score for the answer and determines whether it exceeds a certain threshold. The reliability evaluation uses the scoring function of the generative AI model. For example, if the confidence score is evaluated as 0.85, which is above the set threshold of 0.80, the answer is sent as is. The input of this step is the generated answer, and the output is the result of the reliability evaluation.

[0721] Step 7:

[0722] The server sends a final answer to the user based on the confidence rating. If a reliable answer is generated, the server sends it back to the device as an HTTP response. The device displays the received answer in the chatbot interface. For example: "Your phone number renewal period is December 31, 2023." The input of this step is the evaluated answer, and the output is the answer displayed in the user's interface.

[0723] Step 8:

[0724] The server stores queries and responses in a log. The server records all interactions in a log file or database. The recorded data includes the query content, generated responses, trust assessment results, escalation history, etc. The input of this step is all processing results, and the output is the stored log data.

[0725] (Application example 1)

[0726] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0727] With conventional systems, it is difficult to respond appropriately and quickly to inquiries from corporate customers. Furthermore, due to the wide range of inquiries, sales representatives spend a lot of time responding to inquiries, preventing them from devoting sufficient time to their core sales activities. Furthermore, due to insufficient reliability evaluation of the generated answers and insufficient escalation procedures, customer satisfaction may decline.

[0728] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0729] In this invention, the server includes: means for receiving inquiries; means for analyzing the received inquiries to identify the inquiry content; means for accessing an internal system to obtain related information based on the identified inquiry content; means for generating an answer to the inquiry using a generative AI model based on the obtained information; means for evaluating the reliability of the generated answer; means for determining whether the answer is appropriate based on the evaluation result and, if not, for escalating to a human representative; means for sending the final answer to the user; means for saving the inquiry and the answer in a log; means implemented as a smartphone application that receives inquiries related to a specific service, analyzes them, and generates an answer; means for inputting and displaying the inquiry content through a user interface; and means for generating an answer by providing a prompt sentence to the generative AI model based on the inquiry content. This enables prompt and accurate answers to inquiries from corporate customers, allowing sales representatives to focus on their core business activities. Furthermore, the reliability evaluation of the generated answers and the escalation procedures are strengthened, thereby improving customer satisfaction.

[0730] The "means for receiving an inquiry" refers to a device or program for receiving an inquiry from a user in real time and transferring it to a server.

[0731] "Means for analyzing received inquiries and identifying the content of the inquiries" refers to devices or programs that use natural language processing technology to analyze the content of inquiries from users and accurately understand their intent.

[0732] "Means for accessing an internal system to obtain related information based on the identified inquiry content" refers to a device or program that works in conjunction with an API or database to enable the server to obtain information related to the identified inquiry content from the internal system.

[0733] "Means for generating a response to an inquiry using a generative AI model based on acquired information" refers to a device or program that uses acquired information to generate a response to an inquiry using a generative AI model.

[0734] "Means for evaluating the reliability of the generated answers" refers to devices or programs that calculate confidence scores and other factors to evaluate the accuracy and validity of answers generated by generative AI models.

[0735] "Means for determining whether the answer is appropriate based on the evaluation results, and escalating to a human staff member if it is not appropriate" refers to a device or program that determines the appropriateness of the answer generated based on the evaluation results, and if it is not appropriate, hands it over to a human staff member.

[0736] "Means for sending a final answer to the user" refers to a device or program that sends the content of a reliable answer to the user when a reliable answer is generated or when an appropriate answer is prepared after human verification.

[0737] "Means for saving inquiries and responses in a log" refers to a device or program that saves all interactions, inquiry content, generated responses, and escalation history in a log, and uses this information to improve the system and increase the accuracy of inquiry responses.

[0738] "Means implemented as a smartphone application that receives inquiries related to a specific service, analyzes them, and generates a response" refers to a device or program that receives inquiries related to a specific service through a smartphone application, analyzes the content of the inquiries, and generates a response.

[0739] The "means for inputting and displaying inquiry details through a user interface" refers to an interface through which a user inputs inquiry details and displays responses to those inquiries.

[0740] "Means for generating an answer by providing a prompt sentence to a generative AI model based on the content of the inquiry" refers to a device or program that generates a prompt sentence based on the content of the user's inquiry, passes it to a generative AI model, and generates an answer.

[0741] The following is a detailed description of an embodiment of the present invention: The system of the present invention includes a smartphone application and a server for quickly and accurately responding to inquiries from corporate customers.

[0742] The overall system overview is as follows:

[0743] 1. When a user enters an inquiry using a smartphone application, the inquiry is received by the device in real time.

[0744] 2. The received query is forwarded to the server, which analyzes the query using natural language processing technology to properly identify the query content and intent.

[0745] 3. Based on the identified query, the server accesses internal databases and systems to retrieve the relevant information needed.

[0746] 4. Based on the acquired information, the server uses a generative AI model to generate an appropriate answer. At the time of generation, the prompt sentence is provided to the generative AI model.

[0747] 5. The reliability of the generated answer is assessed by a confidence score: if the confidence is above a certain threshold, it is automatically sent to the user; if it is lower, it is escalated to a human agent.

[0748] 6. The final answer is sent to the user, and the system keeps a log of all interactions.

[0749] Hardware and software used

[0750] Hardware: Cloud servers (e.g., Amazon Web Services, Google Cloud Platform), smartphones (iOS or Android)

[0751] Software: Flask (a lightweight Python web framework), OpenAI's GPT-3 API, natural language processing libraries (e.g., spaCy, NLTK)

[0752] Example of processing flow

[0753] As a specific example, let us consider an inquiry about the release date of a new movie.

[0754] Example inquiry: "When is the new movie coming out?"

[0755] The server receives this query and uses natural language processing technology to identify the "release date of new movies."

[0756] Based on the identified information, the company accesses internal databases to retrieve relevant information.

[0757] The acquired information is provided to the generative AI model as a prompt sentence.

[0758] Prompt Sentence Examples

[0759] Examples of prompts are:

[0760] Text format

[0761] User Asks: When is the new movie coming out?

[0762] answer:

[0763] By passing this prompt to the OpenAI GPT-3 API, a specific answer such as "The release date of the new movie will be November 1, 2023" will be generated.

[0764] The above process makes it possible to respond to inquiries from corporate customers quickly and accurately. This system reduces the burden on sales representatives in responding to inquiries, allowing them to focus more on full-scale sales activities. In addition, by strengthening the reliability evaluation of responses and escalation procedures, customer satisfaction can be improved.

[0765] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0766] Step 1:

[0767] The device receives inquiries from users in real time. At this time, the user inputs the inquiry content through the smartphone application interface. The input inquiry is saved in text format on the device and temporarily stored within the device.

[0768] Input: User's inquiry (text format)

[0769] Output: Received inquiry (text format)

[0770] Step 2:

[0771] The terminal sends the received query to the server using a communication protocol such as an HTTP POST request. The server receives this request and prepares to analyze the query.

[0772] Input: Received inquiry (text format)

[0773] Output: Query content forwarded to the server (text format)

[0774] Step 3:

[0775] The server analyzes the received query and identifies the query content. Here, natural language processing technology (e.g., spaCy or NLTK) is used to extract the query intent and keywords. For example, the keyword "release date of new movies" is identified.

[0776] Input: Inquiry content forwarded to the server (text format)

[0777] Output: Identified inquiry content and keywords (text format)

[0778] Step 4:

[0779] The server accesses the company's internal systems based on the identified inquiry content and retrieves relevant information, such as data such as "release dates for new movies" using the company's internal databases and APIs.

[0780] Input: Identified inquiry content and keywords (text format)

[0781] Output: Relevant information retrieved (in text, possibly structured format)

[0782] Step 5:

[0783] Based on the acquired information, the server provides a prompt sentence to a generative AI model (e.g., OpenAI GPT-3) to generate an answer to the query. At this time, the prompt sentence is input to the generative AI model as a sentence with a specific structure.

[0784] Input: Retrieved relevant information (text format)

[0785] Output: Answers generated by the generative AI model (in text format)

[0786] Step 6:

[0787] The server evaluates the reliability of the generated answer using an algorithm that calculates a confidence score, determines whether the confidence score exceeds a certain threshold, and escalates it if necessary.

[0788] Input: Generated answer (text format)

[0789] Output: Reliability assessment results and confidence scores (numerical data)

[0790] Step 7:

[0791] If the final answer passes the trust assessment, the server sends it to the user, which returns it to the device as an HTTP response. If the trust is low, the request is escalated to a human agent.

[0792] Input: Reliability assessment result and answer (text format)

[0793] Output: Final answer (text format)

[0794] Step 8:

[0795] The server stores all interactions in a log, including the inquiry content, generated answers, reliability assessment results, escalation history, etc. The log is used for future system improvements and analysis.

[0796] Input: All correspondence (text format)

[0797] Output: Saved logs (database entries)

[0798] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0799] The present invention is a system for effectively responding to inquiries from corporate customers, which includes a series of processes for receiving inquiries, analyzing them, and generating appropriate answers, as well as an emotion engine for recognizing user emotions. An embodiment of this system will be described in detail below.

[0800] First, the user sends a query to the corporate chatbot. For example, they enter a query such as, "Please tell me the renewal period for this phone number." The device receives this query in real time and sends the query to the server.

[0801] The server then analyzes the received inquiry using natural language processing (NLP) techniques to accurately understand the intent of the inquiry. For example, it can extract the keyword "renewal period" from the inquiry and identify it as a query about the renewal period for a phone number.

[0802] After identifying the query, the server accesses internal systems via APIs to retrieve relevant information. Specifically, the server queries an internal database to retrieve information about, for example, phone number renewal periods. This data is then formatted by the server and passed to the generative AI model.

[0803] Based on the acquired information, the server uses a generative AI model to generate an appropriate response to the user's inquiry. The generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0804] The server then evaluates the reliability of the generated answer, calculating a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, the server returns the answer to the user as is. On the other hand, if the confidence score is low, the server escalates the answer to a human agent.

[0805] Furthermore, the present invention includes an emotion engine for recognizing the user's emotions. The server analyzes the language and emotional expressions contained in the user's inquiry, and if it determines that the user is expressing anger or frustration, it immediately escalates the inquiry to a human agent. The emotion engine adjusts the tone of the response depending on the user's emotions, for example, generating a more polite and reassuring response if the user is expressing frustration.

[0806] The server sends the final, verified answer to the user. The answer, either trusted or verified by a human agent, is sent to the user's device via an HTTP response or real-time messaging protocol. Example: "Your phone number is due for renewal on December 31, 2023."

[0807] Finally, the server stores all interactions in a log, including the inquiry content, information obtained, responses generated, escalation history, and the results of the analysis of recognized emotions. The stored logs are used to improve the system and enhance the accuracy of inquiry responses in the future.

[0808] As described above, the present invention reduces the time that sales representatives spend on tedious inquiries, creating an environment in which they can concentrate on important sales activities. Furthermore, the introduction of an emotion engine enables responses that take into consideration the user's emotions, contributing to improved customer satisfaction. Thus, the present invention is extremely useful as a means for streamlining inquiries from corporate customers and providing more advanced customer service.

[0809] The processing flow will be explained below.

[0810] Step 1:

[0811] A user sends a query to the enterprise chatbot, for example, "What is the renewal period for this phone number?"

[0812] Step 2:

[0813] The device receives this inquiry in real time and sends the inquiry content to the server. Specifically, it creates an HTTP request and sends it to the server.

[0814] Step 3:

[0815] The server analyzes the received inquiry, which involves using natural language processing (NLP) techniques to understand the intent of the text. It extracts the keyword "renewal period" from the inquiry and identifies it as a request related to the renewal period of a phone number.

[0816] Step 4:

[0817] The server then accesses the internal system via API to retrieve relevant information based on the identified inquiry. Specifically, the server sends a request to the appropriate API endpoint of the internal database to retrieve information about the "phone number renewal period."

[0818] Step 5:

[0819] The server formats the information and passes it to the generative AI model. For example, it converts the information into text data in the format "This phone number's renewal period ends December 31, 2023."

[0820] Step 6:

[0821] The server uses a generative AI model to generate an appropriate response to the inquiry. The generative AI model creates a response in natural language based on the acquired information. It generates a response in the format, such as, "Your phone number renewal period ends on December 31, 2023."

[0822] Step 7:

[0823] The server evaluates the reliability of the generated answer by calculating the confidence score of the generative AI model and checking whether the score exceeds a certain threshold.

[0824] Step 8:

[0825] The server determines whether the answer is appropriate based on the evaluation results. If the answer is highly reliable, it is sent to the user. If it is inappropriate, it is escalated to a sales representative.

[0826] Step 9:

[0827] The server uses an emotion engine to recognize the emotions contained in the user's query, analyzing the language and emotional expressions to determine whether the user is expressing anger or frustration.

[0828] Step 10:

[0829] The server adjusts the tone of the response depending on the user's emotions, for example, generating a more polite and reassuring response if the user is expressing anger.

[0830] Step 11:

[0831] The server sets escalation conditions based on the recognized emotions, and if the user expresses anger or frustration, the call is automatically escalated to a human agent.

[0832] Step 12:

[0833] The server then sends the final, verified answer to the user. The answer, either trusted or verified by a human agent, is sent to the user's device via an HTTP response or real-time messaging protocol. Example: "Your phone number renewal period ends December 31, 2023."

[0834] Step 13:

[0835] The server stores all interactions in a log, including the inquiry, the information obtained, the response generated, the escalation history, the analysis results of the recognized sentiment, etc. The stored log is used for future analysis and system improvement.

[0836] Example 2

[0837] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0838] Conventional corporate chatbot systems have several limitations in analyzing user inquiries and generating answers. For example, there are insufficient means to ensure the reliability of answers and a lack of consideration for user feelings, which can lead to a decline in the quality of inquiry responses. In addition, the system's automated responses are often unreliable, resulting in excessive reliance on human agents. These issues have led to issues such as delayed response times and reduced user satisfaction.

[0839] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an internal system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative AI model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and escalating the inquiry to a human agent if it is not appropriate, means for performing emotion analysis and escalating the inquiry to a human agent based on the user's emotions, means for sending a final answer to the user, and means for saving the inquiry and the answer in a log. This improves the accuracy of inquiry analysis and answer generation, enabling quick and reliable responses that also take user emotions into consideration.

[0840] The "means for receiving an inquiry" is a mechanism for transferring the contents of an inquiry entered by a user to a server.

[0841] The "means for analyzing the received inquiry and identifying the content of the inquiry" is a mechanism for analyzing the received inquiry using natural language processing technology, and extracting and identifying the intent and content of the inquiry.

[0842] "Means for accessing internal systems to obtain relevant information based on the identified inquiry" refers to a mechanism for accessing internal databases and other internal resources to obtain the necessary information based on the identified inquiry.

[0843] "Means for generating an answer to a query using a generative AI model based on acquired information" refers to a mechanism for generating an appropriate answer to a user's query using a generative AI model with acquired information as input.

[0844] The "means for assessing the reliability of the generated answer" is a mechanism for calculating a confidence score for the answer generated by the generative AI model and assessing its reliability.

[0845] "Means for determining whether the response is appropriate based on the evaluation results, and escalating to a human representative if it is not appropriate" is a mechanism for determining the appropriateness of the response based on the results of a reliability evaluation, and transferring the inquiry to a human representative if necessary.

[0846] The "means for performing emotion analysis and escalating inquiries to a human agent based on the user's emotions" is a mechanism for analyzing the emotions contained in a user's inquiry and transferring the inquiry to a human agent based on the results.

[0847] The "means for transmitting the final answer to the user" is a mechanism for transmitting the final confirmed answer to the user's terminal.

[0848] The "means for saving inquiries and responses in a log" is a mechanism for saving the inquiry content, acquired information, generated responses, and the results of escalation and sentiment analysis in a log.

[0849] MODE FOR CARRYING OUT THE INVENTION

[0850] The present invention is a system for responding effectively and quickly to inquiries from corporate customers. This system not only receives inquiries, analyzes them, and generates appropriate responses, but also includes an emotion engine that recognizes the emotions of users.

[0851] System hardware and software configuration

[0852] 1. User device: The device (PC, smartphone, tablet, etc.) through which the user sends an inquiry via the chatbot.

[0853] 2. Server: The central computing resource that performs various processes, such as query analysis, information retrieval, answer generation, credibility assessment, and sentiment analysis.

[0854] Natural Language Processing (NLP): Uses Google Cloud Natural Language API.

[0855] Internal system access: Access to our internal database (MySQL).

[0856] Generative AI model: OpenAI's GPT-4 is used.

[0857] Sentiment analysis: Using Azure Text Analytics.

[0858] System Operation

[0859] The user enters the question "Please tell me the renewal period for this phone number" into the chatbot interface. The device receives this query in real time and sends it to the server.

[0860] The server analyzes the received query using the Google Cloud Natural Language API. Based on the analysis results, it identifies the keyword "renewal period" and determines that the query is about the phone number renewal period.

[0861] The server then queries the company's internal database using MySQL to retrieve relevant data. Based on this information, the server uses OpenAI's GPT-4 model to generate an appropriate answer, such as "Your phone number is due for renewal on December 31, 2023."

[0862] To assess the reliability of the generated answer, the server calculates a confidence score and automatically sends the answer to the user if it exceeds a set threshold (CERTAINTY_THRESHOLD), or escalates the answer to a human agent if the confidence is low.

[0863] Additionally, the server uses Azure Text Analytics to analyze the user's emotions and, if it determines that the user is expressing anger or frustration, it will immediately escalate the call to a human agent.

[0864] The final answer is sent from the server to the device via an HTTP response or WebSocket protocol. For example, the answer may say, "Your phone number renewal period ends on December 31, 2023."

[0865] All interactions are logged by the server, including the inquiry, the information obtained, the response generated, the escalation history, and the results of sentiment analysis, which can be used to improve the system and its accuracy.

[0866] Examples of concrete examples and prompts

[0867] For example, if a user types "What is the expiration date of my contract?", the server uses NLP technology to identify the keyword "contract expiration date" and retrieves relevant information from the company's internal database. The generative AI model creates an answer such as "Your contract expires on March 31, 2024," which is sent to the user after passing a reliability assessment. If the user expresses strong dissatisfaction, the emotion engine will be activated and a human agent will manually process the request.

[0868] An example prompt is:

[0869] "Please tell me the renewal period for this phone number."

[0870] Please tell me the expiration date of the contract.

[0871] As described above, the present invention is extremely useful as a means for improving the efficiency of responses to inquiries from corporate customers and providing more advanced customer service.

[0872] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0873] Step 1:

[0874] A user inputs an inquiry into a corporate chatbot and clicks the send button. For example, they might input "Please tell me the renewal period for this phone number." The specific action is when the user inputs text into the chat window and clicks the send button. The input volume is the content of the user's inquiry, and the output volume is when this inquiry is transferred to the server.

[0875] Step 2:

[0876] The terminal receives the query. The terminal receives the query from the user in real time, converts the content into JSON format, and sends it to the server. The input is the text received from the user, and the output is JSON format data. Specifically, the terminal analyzes and structures the content of the query, and sends it to the server as an API request.

[0877] Step 3:

[0878] The server sends the received JSON data to the Google Cloud Natural Language API to identify the intent of the query. The input is JSON data, and the output is the analysis results, such as keywords and intent. Specifically, the server sends an API request to Google Cloud and receives the analysis results.

[0879] Step 4:

[0880] The server queries the company's internal database based on the analysis results to retrieve relevant information. The input is the analyzed keywords and intent, and the output is the relevant information as a result of the database query. Specifically, the server executes an SQL query to extract the required data.

[0881] Step 5:

[0882] Based on the information acquired by the server, it sends prompts to OpenAI's GPT-4 model to generate an appropriate answer to the query. The input is the information acquired from the database and the query content, and the output is the generated answer text. Specifically, the server provides the prompt to GPT-4 and executes an API request to generate the answer.

[0883] Step 6:

[0884] The server evaluates the reliability of the generated answer. The input is the generated answer text, and the output is the confidence score. Specifically, the server calculates the confidence of the answer and compares it with the configured threshold (CERTAINTY_THRESHOLD).

[0885] Step 7:

[0886] The server determines the appropriateness of the answer based on the evaluation results. The inputs are the confidence score and a threshold, and the output is the destination of the answer (automatic sending or escalation). Specifically, if the confidence exceeds the threshold, the answer is sent automatically, and if not, the answer is escalated to a human representative.

[0887] Step 8:

[0888] The server uses Azure Text Analytics to analyze user sentiment. The input is the user's query text, and the output is the sentiment analysis result. The specific operation is to send the query content to the API and receive the sentiment analysis result.

[0889] Step 9:

[0890] Based on the emotion analysis results, the server determines whether to escalate the inquiry to a person in charge. The input is the emotion analysis result, and the output is an instruction to escalate. Specifically, if anger or dissatisfaction is strong, an instruction to escalate is sent to a person in charge.

[0891] Step 10:

[0892] The server sends the final answer to the user. The input is the final confirmed answer text, and the output is the data sent to the user terminal. The specific operation is to send the answer using an HTTP response or the WebSocket protocol.

[0893] Step 11:

[0894] The server stores all interactions in a log. The inputs include the inquiry content, acquired information, generated answers, evaluation results, and sentiment analysis results, while the output is the stored log data. Specific operations include storing this data in a log file or database.

[0895] (Application example 2)

[0896] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0897] Conventional inquiry response systems have the problem of being unable to properly assess a user's emotions and respond accordingly when analyzing the content of inquiries and generating responses. This can result in an inability to respond appropriately to inquiries from users who are particularly emotional, which can lead to a decline in customer satisfaction. Furthermore, in some cases, escalation may not be performed appropriately if the response to an inquiry is unreliable, which also leads to a decline in customer satisfaction.

[0898] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an in-house system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative AI model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and escalating the answer to a human representative if it is not appropriate, means for analyzing emotions and adjusting the tone of the answer based on the analysis, means for sending the final answer to the user, and means for saving the inquiry and the answer in a log. This makes it possible to generate an appropriate answer that takes user emotions into consideration and provide a highly reliable answer.

[0899] The "means for receiving inquiries" refers to a device or software mechanism for electronically receiving inquiries from users.

[0900] "Means for analyzing received inquiries and identifying the content of the inquiries" refers to a mechanism for analyzing the content of received inquiries using natural language processing technology and identifying their intent and purpose.

[0901] "Means for accessing internal systems to obtain relevant information based on the identified inquiry content" refers to a mechanism for accessing internal systems or databases to obtain data related to the inquiry content.

[0902] "Means for generating responses to inquiries using a generative AI model based on acquired information" refers to a mechanism that uses a generative AI model (such as GPT-3) to generate appropriate responses in natural language format based on acquired data.

[0903] The "means for assessing the reliability of the generated answer" is a mechanism for calculating a confidence score to assess the accuracy and appropriateness of the generated answer.

[0904] "Means of determining whether the answer is appropriate based on the evaluation results, and escalating it to a human representative if it is not appropriate" is a mechanism that transfers the evaluated answer to a human representative if it does not meet certain reliability standards.

[0905] "Means for analyzing emotions and adjusting the tone of the response based on that" is a mechanism that analyzes emotions from the user's statements and text and generates a response in a tone that corresponds to those emotions.

[0906] The "means for transmitting the final answer to the user" is a mechanism for transmitting the generated final answer to the user using a communication means.

[0907] "Means for logging inquiries and responses" refers to a mechanism for storing all inquiries and responses as records.

[0908] The present invention is a system for effectively responding to user inquiries, which analyzes the inquiry, generates an answer, adjusts the tone of the response based on sentiment analysis, etc. An embodiment of this system will be described in detail below.

[0909] First, a user sends a query through a smartphone application. For example, a query such as "Please tell me the release date of the latest drama." The device receives this query in real time and transmits the query content to the server.

[0910] The server then analyzes the received query using natural language processing (NLP) techniques to determine the intent of the query, for example by extracting the keyword "publication date" and determining that it is related to a specific piece of content.

[0911] After identifying the query, the server accesses the company's internal systems via API to retrieve relevant information. Specifically, the server queries the company's internal database to retrieve information about the release dates of "new dramas," for example. This data is then formatted by the server and passed to the generative AI model.

[0912] Based on the acquired information, the server uses a generative AI model (e.g., GPT-3) to generate an appropriate answer to the query. The generated answer is provided in the form of natural language, such as "Thank you for your question. The release date of 'New Drama' is October 21, 2023."

[0913] The server then evaluates the reliability of the generated answer using a confidence score, and if the confidence exceeds a certain threshold, it returns the answer to the user directly, otherwise it escalates the answer to a human agent.

[0914] Furthermore, the server uses an emotion engine to analyze the user's emotions. It analyzes the language and emotional expressions contained in the inquiry, and if the user expresses dissatisfaction or anger, it immediately escalates the inquiry or adjusts the tone of the response. For example, if a user says, "I want an answer right away!", the server will respond with a polite tone, such as, "I'm sorry. I understand your dissatisfaction. I'll look into it right away."

[0915] Finally, the verified answer is sent to the user. Answers with high confidence or verified by a human agent are sent to the user's device via HTTP responses or real-time messaging protocols.

[0916] Finally, the server stores all interactions in a log, including the inquiry content, information obtained, responses generated, escalation history, sentiment analysis results, etc. These logs are used to improve the system and improve the accuracy of inquiry responses in the future.

[0917] Example prompt:

[0918] User Question: "What is the release date of the latest drama?"

[0919] The system responds: "The content in question is 'New Drama' and its release date is October 21, 2023."

[0920] User Question: "I want an answer right now!"

[0921] The system responds: "We're sorry, we understand your frustration. We'll look into it right away."

[0922] In this way, the embodiment of the present invention makes it possible to generate an appropriate answer that takes into account the user's feelings and to provide a highly reliable answer.

[0923] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0924] Step 1:

[0925] The user sends a query. For example, the user inputs a query such as "Please tell me the release date of the latest drama" through a smartphone application. The device receives this query in real time and transmits the query to the server.

[0926] Step 2:

[0927] The server analyzes the query. The server uses natural language processing (NLP) technology to analyze the received query and identify the intent of the query. Specifically, it extracts the keyword "publication date" and determines that the query is about the publication date of the content. The input for this analysis is the user's query, and the output is the analysis result (the intent of the query).

[0928] Step 3:

[0929] The server accesses the internal system to obtain information. Based on the specified query, the server accesses the internal database via API to obtain related information (e.g., content release date). The input of this step is the analysis result, and the output is the obtained information (e.g., the release date of the drama).

[0930] Step 4:

[0931] The server generates an answer using a generative AI model. Based on the acquired information, the server uses a generative AI model (e.g., GPT-3) to generate an answer in natural language format. For example, it creates an answer such as, "Thank you for your question. The release date of 'New Drama' is October 21, 2023." The input of this step is the acquired information, and the output is the generated answer.

[0932] Step 5:

[0933] The server evaluates the reliability of the generated answer. The server calculates a confidence score for the generated answer and determines whether this score exceeds a certain threshold. If the score is low, it escalates to a human agent. The input to this step is the generated answer and its confidence score, and the output is a branching process depending on whether escalation is necessary.

[0934] Step 6:

[0935] The server analyzes the emotion and adjusts the tone of the response. The server uses an emotion engine to analyze the emotion from the user's query. For example, if the user says, "I want an answer right away!", it determines that the user is expressing anger or dissatisfaction. As a result, the server creates a response in a tone such as, "I'm sorry. I understand your dissatisfaction. We will look into it right away." The input for this step is the user's query, and the output is a response that takes the emotion into consideration.

[0936] Step 7:

[0937] The server sends a final answer to the user. The server sends the answer that is determined to be reliable or the answer that has been confirmed after escalation to the user's device. This final answer is delivered via an HTTP response or real-time messaging protocol. The input of this step is the confirmed answer text, and the output is the answer sent to the user's device.

[0938] Step 8:

[0939] The server stores the queries and responses in a log. The server records and stores all interactions as log data. This data includes the query content, information obtained, responses generated, escalation history, and sentiment analysis results. The log data is used for future system improvements. The input of this step is a record of all interactions, and the output is the stored log data.

[0940] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0941] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0942] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[0943] [Fourth embodiment]

[0944] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[0945] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0946] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0947] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[0948] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0949] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0950] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0951] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[0952] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0953] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0954] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0955] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0956] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0957] The present invention is a system for effectively responding to inquiries from corporate customers, and includes a series of processes for receiving inquiries, analyzing them, and generating appropriate responses. An embodiment of this system will be described in detail below.

[0958] First, the user sends a query to the corporate chatbot. For example, the user might enter a query such as, "Please tell me the renewal period for this phone number." The device receives this query in real time and sends it to the server.

[0959] The server then analyzes the received inquiry using natural language processing technology to accurately understand the intent of the inquiry. For example, it can extract the keyword "renewal period" from the inquiry and identify it as a question about the renewal period for a phone number.

[0960] After identifying the query, the server accesses the internal system via API to retrieve relevant information. Specifically, the server queries the internal database to retrieve necessary data, such as "phone number renewal period." This data is then formatted by the server and passed to the generative AI model.

[0961] Based on the acquired information, the server uses a generative AI model to generate an appropriate response to the user's inquiry. The generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[0962] The server evaluates the reliability of the generated answer. It calculates a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, it returns the answer to the user as is. On the other hand, if the confidence is low, the server escalates the answer to a human agent.

[0963] The server sends the final, verified answer to the user. If a reliable answer is generated automatically or if a suitable answer is prepared after human verification, the content is sent to the user's device. For example, "Your phone number renewal period ends on December 31, 2023."

[0964] Finally, the server stores all interactions in a log, including the inquiry content, the generated response, escalation history, etc. The stored log is used to improve the system and improve the accuracy of inquiry responses in the future.

[0965] This system reduces the time sales representatives spend on tedious inquiries, allowing them to focus on important sales activities. Providing fast and accurate information also contributes to improving customer satisfaction. Thus, the present invention is extremely useful as a means of streamlining inquiries from corporate customers and providing more advanced customer service.

[0966] The processing flow will be explained below.

[0967] Step 1:

[0968] A user sends a query to the enterprise chatbot, for example, "What is the renewal period for this phone number?"

[0969] Step 2:

[0970] The device receives this inquiry in real time and sends the inquiry content to the server. Specifically, it creates an HTTP request and sends it to the server.

[0971] Step 3:

[0972] The server analyzes the received inquiry, which involves using natural language processing (NLP) techniques to understand the intent of the text. It extracts the keyword "renewal period" from the inquiry and identifies it as a request related to the renewal period of a phone number.

[0973] Step 4:

[0974] The server then accesses the internal system via API to retrieve relevant information based on the identified inquiry. Specifically, the server sends a request to the appropriate API endpoint of the internal database to retrieve information about the "phone number renewal period."

[0975] Step 5:

[0976] The server formats the information and passes it to the generative AI model. For example, it converts the information into text data in the format "This phone number's renewal period ends December 31, 2023."

[0977] Step 6:

[0978] The server uses a generative AI model to generate an appropriate response to the inquiry. The generative AI model creates a response in natural language based on the acquired information. It generates a response in the format, such as, "Your phone number renewal period ends on December 31, 2023."

[0979] Step 7:

[0980] The server evaluates the reliability of the generated answer by calculating the confidence score of the generative AI model and checking whether the score exceeds a certain threshold.

[0981] Step 8:

[0982] The server determines whether the answer is appropriate based on the evaluation results. If the answer is highly reliable, it is sent to the user. If it is inappropriate, it is escalated to a sales representative.

[0983] Step 9:

[0984] The server then sends the final verified answer to the user, either authentic or verified by a human agent, via HTTP response or real-time messaging protocol to the user's device.

[0985] Step 10:

[0986] The server stores all interactions in a log, including inquiries, information obtained, responses generated, escalation history, etc. The log is used for future analysis and system improvement.

[0987] Example 1

[0988] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0989] Conventional inquiry response systems have difficulty providing quick and accurate responses to user inquiries, and often require human intervention. As a result, inquiries take time to respond, leading to issues such as a decline in customer satisfaction. Furthermore, there is a lack of efficient management of data generated during the inquiry response process and its use for future improvements.

[0990] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0991] In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an internal system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative artificial intelligence model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and, if not, for escalating to a human representative, means for sending the final answer to the user, and means for storing the inquiry and the answer in a record. This makes it possible to respond to user inquiries quickly and accurately, improving customer satisfaction and contributing to future system improvements.

[0992] The "means for receiving an inquiry" refers to a technique or device for inputting inquiry information such as text or voice sent from a user into a server via the Internet.

[0993] The "means of analyzing the received inquiry and identifying the content of the inquiry" refers to technology or algorithms that use natural language processing technology to break down the text data of the received inquiry and extract and understand the subject and keywords of the inquiry.

[0994] "Means of accessing internal systems to obtain relevant information based on the identified inquiry" refers to the technology or procedures by which a server sends requests to an internal company database or API to search for and obtain information related to the inquiry.

[0995] "Means for generating answers to inquiries using a generative AI model" refers to technologies or systems that use a generative AI model (e.g., GPT-3) based on acquired information to create answers in natural language format to user inquiries.

[0996] "Means for assessing the reliability of generated answers" refers to technologies or algorithms that evaluate the reliability and accuracy of answers output by generative AI models using criteria such as confidence scores.

[0997] "Means of determining whether the answer is appropriate based on the evaluation results, and escalating to a human operator if it is not appropriate" refers to the technology or process of determining whether the answer meets certain criteria based on the evaluated reliability score, and if it does not, handing over the processing to a human operator.

[0998] The "means for sending the final answer to the user" refers to a technique or means for returning an answer that has passed the reliability evaluation to the user in real time.

[0999] "Means for recording inquiries and responses" means techniques or methods for recording all inquiry and response interactions in log files or databases and storing them for future analysis and improvement.

[1000] The present invention provides a system for effectively responding to inquiries from corporate customers, which includes a series of processes for receiving and analyzing inquiries, and generating and returning appropriate answers.

[1001] First, the user sends a query to the corporate chatbot. For example, they might ask, "What is the renewal period for this phone number?" The device receives this query in real time and sends it to a server via the Internet.

[1002] The server analyzes the received query using natural language processing techniques (e.g., spaCy or NLTK). The server tokenizes the query text, extracts the keyword "renewal period," and understands that the query is about the renewal period for a phone number.

[1003] The server then accesses the internal system via an API to retrieve relevant information. Specifically, the server queries the internal database (e.g., Microsoft SQL Server or MySQL) to retrieve the necessary data. The retrieved information is then formatted and passed to the generative AI model.

[1004] The server uses a generative AI model (for example, OpenAI's GPT-3) to generate an appropriate response to the user's inquiry. Based on the acquired information, the generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[1005] The server evaluates the reliability of the generated answer, calculates a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, returns the answer to the user as is. On the other hand, if the confidence is low, the answer is escalated to a human representative.

[1006] Finally, the server sends the final answer that passes the evaluation to the user. The terminal receives the answer and displays it on the chatbot interface. For example, "Your phone number renewal period ends on December 31, 2023."

[1007] The server also stores all inquiries and responses in a log, which includes the inquiry content, the generated responses, escalation history, etc. The stored logs are used to improve the system and the accuracy of inquiries in the future.

[1008] As a specific example, if a user sends a query such as "Please tell me the renewal period for this phone number," the device sends the query to the server, which uses natural language processing technology to extract the keyword "renewal period." The server accesses the internal system to obtain the corresponding information, and based on that, uses a generative AI model to generate the answer "The renewal period for your phone number is December 31, 2023." The answer that passes the evaluation is sent to the user, and the content of the query and answer is saved in a log.

[1009] Example prompt sentence:

[1010] User: How long do I need to renew this phone number?

[1011] Terminal: (sends user query to server)

[1012] Server: (analyzes the received query)

[1013] Server: (retrieving corresponding information from internal database)

[1014] Server: (Generate an answer using a generative AI model based on the acquired information)

[1015] Server: Your phone number is due for renewal on December 31, 2023.

[1016] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1017] Step 1:

[1018] The user sends a query to the chatbot. The user accesses the chatbot's interface, enters a question in the text field, and presses the send button. For example, the user might enter, "Please tell me the renewal period for this phone number." This sentence is the user's input.

[1019] Step 2:

[1020] The terminal sends a query to the server. The terminal receives input from the user and sends it as an HTTP POST request to the server's query acceptance endpoint. The input here is the user's query statement, and the output is the HTTP request sent to the server.

[1021] Step 3:

[1022] The server analyzes the query. The server analyzes the received text data and uses natural language processing techniques to understand the intent of the query. Specifically, the server tokenizes the text and extracts keywords and phrases. For example, it extracts the keyword "renewal period" and identifies it as being about the renewal period for a phone number. The input to this step is the query received by the server, and the output is the extracted keywords and their intent.

[1023] Step 4:

[1024] Based on the identified query, the server retrieves information from the internal system via API. The server sends a query to the internal database based on the keyword "phone number renewal period." For example, the server uses an SQL query to search for "phone number renewal period." The input for this step is the identified keyword, and the output is the information retrieved from the database.

[1025] Step 5:

[1026] The server generates an answer using a generative AI model based on the information acquired. The server inputs the acquired data into a generative AI model (e.g., GPT-3) and generates an answer in natural language format. Specifically, the server inputs the prompt "What is the renewal period for your phone number?" and obtains the generated answer "Your phone number renewal period is December 31, 2023." The inputs for this step are the acquired information and the prompt, and the output is the generated answer.

[1027] Step 6:

[1028] The server evaluates the reliability of the generated answer. The server calculates a confidence score for the answer and determines whether it exceeds a certain threshold. The reliability evaluation uses the scoring function of the generative AI model. For example, if the confidence score is evaluated as 0.85, which is above the set threshold of 0.80, the answer is sent as is. The input of this step is the generated answer, and the output is the result of the reliability evaluation.

[1029] Step 7:

[1030] The server sends a final answer to the user based on the confidence rating. If a reliable answer is generated, the server sends it back to the device as an HTTP response. The device displays the received answer in the chatbot interface. For example: "Your phone number renewal period is December 31, 2023." The input of this step is the evaluated answer, and the output is the answer displayed in the user's interface.

[1031] Step 8:

[1032] The server stores queries and responses in a log. The server records all interactions in a log file or database. The recorded data includes the query content, generated responses, trust assessment results, escalation history, etc. The input of this step is all processing results, and the output is the stored log data.

[1033] (Application example 1)

[1034] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1035] With conventional systems, it is difficult to respond appropriately and quickly to inquiries from corporate customers. Furthermore, due to the wide range of inquiries, sales representatives spend a lot of time responding to inquiries, preventing them from devoting sufficient time to their core sales activities. Furthermore, due to insufficient reliability evaluation of the generated answers and insufficient escalation procedures, customer satisfaction may decline.

[1036] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1037] In this invention, the server includes: means for receiving inquiries; means for analyzing the received inquiries to identify the inquiry content; means for accessing an internal system to obtain related information based on the identified inquiry content; means for generating an answer to the inquiry using a generative AI model based on the obtained information; means for evaluating the reliability of the generated answer; means for determining whether the answer is appropriate based on the evaluation result and, if not, for escalating to a human representative; means for sending the final answer to the user; means for saving the inquiry and the answer in a log; means implemented as a smartphone application that receives inquiries related to a specific service, analyzes them, and generates an answer; means for inputting and displaying the inquiry content through a user interface; and means for generating an answer by providing a prompt sentence to the generative AI model based on the inquiry content. This enables prompt and accurate answers to inquiries from corporate customers, allowing sales representatives to focus on their core business activities. Furthermore, the reliability evaluation of the generated answers and the escalation procedures are strengthened, thereby improving customer satisfaction.

[1038] The "means for receiving an inquiry" refers to a device or program for receiving an inquiry from a user in real time and transferring it to a server.

[1039] "Means for analyzing received inquiries and identifying the content of the inquiries" refers to devices or programs that use natural language processing technology to analyze the content of inquiries from users and accurately understand their intent.

[1040] "Means for accessing an internal system to obtain related information based on the identified inquiry content" refers to a device or program that works in conjunction with an API or database to enable the server to obtain information related to the identified inquiry content from the internal system.

[1041] "Means for generating a response to an inquiry using a generative AI model based on acquired information" refers to a device or program that uses acquired information to generate a response to an inquiry using a generative AI model.

[1042] "Means for evaluating the reliability of the generated answers" refers to devices or programs that calculate confidence scores and other factors to evaluate the accuracy and validity of answers generated by generative AI models.

[1043] "Means for determining whether the answer is appropriate based on the evaluation results, and escalating to a human staff member if it is not appropriate" refers to a device or program that determines the appropriateness of the answer generated based on the evaluation results, and if it is not appropriate, hands it over to a human staff member.

[1044] "Means for sending a final answer to the user" refers to a device or program that sends the content of a reliable answer to the user when a reliable answer is generated or when an appropriate answer is prepared after human verification.

[1045] "Means for saving inquiries and responses in a log" refers to a device or program that saves all interactions, inquiry content, generated responses, and escalation history in a log, and uses this information to improve the system and increase the accuracy of inquiry responses.

[1046] "Means implemented as a smartphone application that receives inquiries related to a specific service, analyzes them, and generates a response" refers to a device or program that receives inquiries related to a specific service through a smartphone application, analyzes the content of the inquiries, and generates a response.

[1047] The "means for inputting and displaying inquiry details through a user interface" refers to an interface through which a user inputs inquiry details and displays responses to those inquiries.

[1048] "Means for generating an answer by providing a prompt sentence to a generative AI model based on the content of the inquiry" refers to a device or program that generates a prompt sentence based on the content of the user's inquiry, passes it to a generative AI model, and generates an answer.

[1049] The following is a detailed description of an embodiment of the present invention: The system of the present invention includes a smartphone application and a server for quickly and accurately responding to inquiries from corporate customers.

[1050] The overall system overview is as follows:

[1051] 1. When a user enters an inquiry using a smartphone application, the inquiry is received by the device in real time.

[1052] 2. The received query is forwarded to the server, which analyzes the query using natural language processing technology to properly identify the query content and intent.

[1053] 3. Based on the identified query, the server accesses internal databases and systems to retrieve the relevant information needed.

[1054] 4. Based on the acquired information, the server uses a generative AI model to generate an appropriate answer. At the time of generation, the prompt sentence is provided to the generative AI model.

[1055] 5. The reliability of the generated answer is assessed by a confidence score: if the confidence is above a certain threshold, it is automatically sent to the user; if it is lower, it is escalated to a human agent.

[1056] 6. The final answer is sent to the user, and the system keeps a log of all interactions.

[1057] Hardware and software used

[1058] Hardware: Cloud servers (e.g., Amazon Web Services, Google Cloud Platform), smartphones (iOS or Android)

[1059] Software: Flask (a lightweight Python web framework), OpenAI's GPT-3 API, natural language processing libraries (e.g., spaCy, NLTK)

[1060] Example of processing flow

[1061] As a specific example, let us consider an inquiry about the release date of a new movie.

[1062] Example inquiry: "When is the new movie coming out?"

[1063] The server receives this query and uses natural language processing technology to identify the "release date of new movies."

[1064] Based on the identified information, the company accesses internal databases to retrieve relevant information.

[1065] The acquired information is provided to the generative AI model as a prompt sentence.

[1066] Prompt Sentence Examples

[1067] Examples of prompts are:

[1068] Text format

[1069] User Asks: When is the new movie coming out?

[1070] answer:

[1071] By passing this prompt to the OpenAI GPT-3 API, a specific answer such as "The release date of the new movie will be November 1, 2023" will be generated.

[1072] The above process makes it possible to respond to inquiries from corporate customers quickly and accurately. This system reduces the burden on sales representatives in responding to inquiries, allowing them to focus more on full-scale sales activities. In addition, by strengthening the reliability evaluation of responses and escalation procedures, customer satisfaction can be improved.

[1073] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1074] Step 1:

[1075] The device receives inquiries from users in real time. At this time, the user inputs the inquiry content through the smartphone application interface. The input inquiry is saved in text format on the device and temporarily stored within the device.

[1076] Input: User's inquiry (text format)

[1077] Output: Received inquiry (text format)

[1078] Step 2:

[1079] The terminal sends the received query to the server using a communication protocol such as an HTTP POST request. The server receives this request and prepares to analyze the query.

[1080] Input: Received inquiry (text format)

[1081] Output: Query content forwarded to the server (text format)

[1082] Step 3:

[1083] The server analyzes the received query and identifies the query content. Here, natural language processing technology (e.g., spaCy or NLTK) is used to extract the query intent and keywords. For example, the keyword "release date of new movies" is identified.

[1084] Input: Inquiry content forwarded to the server (text format)

[1085] Output: Identified inquiry content and keywords (text format)

[1086] Step 4:

[1087] The server accesses the company's internal systems based on the identified inquiry content and retrieves relevant information, such as data such as "release dates for new movies" using the company's internal databases and APIs.

[1088] Input: Identified inquiry content and keywords (text format)

[1089] Output: Relevant information retrieved (in text, possibly structured format)

[1090] Step 5:

[1091] Based on the acquired information, the server provides a prompt sentence to a generative AI model (e.g., OpenAI GPT-3) to generate an answer to the query. At this time, the prompt sentence is input to the generative AI model as a sentence with a specific structure.

[1092] Input: Retrieved relevant information (text format)

[1093] Output: Answers generated by the generative AI model (in text format)

[1094] Step 6:

[1095] The server evaluates the reliability of the generated answer using an algorithm that calculates a confidence score, determines whether the confidence score exceeds a certain threshold, and escalates it if necessary.

[1096] Input: Generated answer (text format)

[1097] Output: Reliability assessment results and confidence scores (numerical data)

[1098] Step 7:

[1099] If the final answer passes the trust assessment, the server sends it to the user, which returns it to the device as an HTTP response. If the trust is low, the request is escalated to a human agent.

[1100] Input: Reliability assessment result and answer (text format)

[1101] Output: Final answer (text format)

[1102] Step 8:

[1103] The server stores all interactions in a log, including the inquiry content, generated answers, reliability assessment results, escalation history, etc. The log is used for future system improvements and analysis.

[1104] Input: All correspondence (text format)

[1105] Output: Saved logs (database entries)

[1106] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1107] The present invention is a system for effectively responding to inquiries from corporate customers, which includes a series of processes for receiving inquiries, analyzing them, and generating appropriate answers, as well as an emotion engine for recognizing user emotions. An embodiment of this system will be described in detail below.

[1108] First, the user sends a query to the corporate chatbot. For example, they enter a query such as, "Please tell me the renewal period for this phone number." The device receives this query in real time and sends the query to the server.

[1109] The server then analyzes the received inquiry using natural language processing (NLP) techniques to accurately understand the intent of the inquiry. For example, it can extract the keyword "renewal period" from the inquiry and identify it as a query about the renewal period for a phone number.

[1110] After identifying the query, the server accesses internal systems via APIs to retrieve relevant information. Specifically, the server queries an internal database to retrieve information about, for example, phone number renewal periods. This data is then formatted by the server and passed to the generative AI model.

[1111] Based on the acquired information, the server uses a generative AI model to generate an appropriate response to the user's inquiry. The generative AI model creates a response in the form of natural language, such as "Your phone number renewal period ends on December 31, 2023."

[1112] The server then evaluates the reliability of the generated answer, calculating a confidence score for the answer output by the generative AI model, and if the score exceeds a certain threshold, the server returns the answer to the user as is. On the other hand, if the confidence score is low, the server escalates the answer to a human agent.

[1113] Furthermore, the present invention includes an emotion engine for recognizing the user's emotions. The server analyzes the language and emotional expressions contained in the user's inquiry, and if it determines that the user is expressing anger or frustration, it immediately escalates the inquiry to a human agent. The emotion engine adjusts the tone of the response depending on the user's emotions, for example, generating a more polite and reassuring response if the user is expressing frustration.

[1114] The server sends the final, verified answer to the user. The answer, either trusted or verified by a human agent, is sent to the user's device via an HTTP response or real-time messaging protocol. Example: "Your phone number is due for renewal on December 31, 2023."

[1115] Finally, the server stores all interactions in a log, including the inquiry content, information obtained, responses generated, escalation history, and the results of the analysis of recognized emotions. The stored logs are used to improve the system and enhance the accuracy of inquiry responses in the future.

[1116] As described above, the present invention reduces the time that sales representatives spend on tedious inquiries, creating an environment in which they can concentrate on important sales activities. Furthermore, the introduction of an emotion engine enables responses that take into consideration the user's emotions, contributing to improved customer satisfaction. Thus, the present invention is extremely useful as a means for streamlining inquiries from corporate customers and providing more advanced customer service.

[1117] The processing flow will be explained below.

[1118] Step 1:

[1119] A user sends a query to the enterprise chatbot, for example, "What is the renewal period for this phone number?"

[1120] Step 2:

[1121] The device receives this inquiry in real time and sends the inquiry content to the server. Specifically, it creates an HTTP request and sends it to the server.

[1122] Step 3:

[1123] The server analyzes the received inquiry, which involves using natural language processing (NLP) techniques to understand the intent of the text. It extracts the keyword "renewal period" from the inquiry and identifies it as a request related to the renewal period of a phone number.

[1124] Step 4:

[1125] The server then accesses the internal system via API to retrieve relevant information based on the identified inquiry. Specifically, the server sends a request to the appropriate API endpoint of the internal database to retrieve information about the "phone number renewal period."

[1126] Step 5:

[1127] The server formats the information and passes it to the generative AI model. For example, it converts the information into text data in the format "This phone number's renewal period ends December 31, 2023."

[1128] Step 6:

[1129] The server uses a generative AI model to generate an appropriate response to the inquiry. The generative AI model creates a response in natural language based on the acquired information. It generates a response in the format, such as, "Your phone number renewal period ends on December 31, 2023."

[1130] Step 7:

[1131] The server evaluates the reliability of the generated answer by calculating the confidence score of the generative AI model and checking whether the score exceeds a certain threshold.

[1132] Step 8:

[1133] The server determines whether the answer is appropriate based on the evaluation results. If the answer is highly reliable, it is sent to the user. If it is inappropriate, it is escalated to a sales representative.

[1134] Step 9:

[1135] The server uses an emotion engine to recognize the emotions contained in the user's query, analyzing the language and emotional expressions to determine whether the user is expressing anger or frustration.

[1136] Step 10:

[1137] The server adjusts the tone of the response depending on the user's emotions, for example, generating a more polite and reassuring response if the user is expressing anger.

[1138] Step 11:

[1139] The server sets escalation conditions based on the recognized emotions, and if the user expresses anger or frustration, the call is automatically escalated to a human agent.

[1140] Step 12:

[1141] The server then sends the final, verified answer to the user. The answer, either trusted or verified by a human agent, is sent to the user's device via an HTTP response or real-time messaging protocol. Example: "Your phone number renewal period ends December 31, 2023."

[1142] Step 13:

[1143] The server stores all interactions in a log, including the inquiry, the information obtained, the response generated, the escalation history, the analysis results of the recognized sentiment, etc. The stored log is used for future analysis and system improvement.

[1144] Example 2

[1145] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1146] Conventional corporate chatbot systems have several limitations in analyzing user inquiries and generating answers. For example, there are insufficient means to ensure the reliability of answers and a lack of consideration for user feelings, which can lead to a decline in the quality of inquiry responses. In addition, the system's automated responses are often unreliable, resulting in excessive reliance on human agents. These issues have led to issues such as delayed response times and reduced user satisfaction.

[1147] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an internal system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative AI model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and escalating the inquiry to a human agent if it is not appropriate, means for performing emotion analysis and escalating the inquiry to a human agent based on the user's emotions, means for sending a final answer to the user, and means for saving the inquiry and the answer in a log. This improves the accuracy of inquiry analysis and answer generation, enabling quick and reliable responses that also take user emotions into consideration.

[1148] The "means for receiving an inquiry" is a mechanism for transferring the contents of an inquiry entered by a user to a server.

[1149] The "means for analyzing the received inquiry and identifying the content of the inquiry" is a mechanism for analyzing the received inquiry using natural language processing technology, and extracting and identifying the intent and content of the inquiry.

[1150] "Means for accessing internal systems to obtain relevant information based on the identified inquiry" refers to a mechanism for accessing internal databases and other internal resources to obtain the necessary information based on the identified inquiry.

[1151] "Means for generating an answer to a query using a generative AI model based on acquired information" refers to a mechanism for generating an appropriate answer to a user's query using a generative AI model with acquired information as input.

[1152] The "means for assessing the reliability of the generated answer" is a mechanism for calculating a confidence score for the answer generated by the generative AI model and assessing its reliability.

[1153] "Means for determining whether the response is appropriate based on the evaluation results, and escalating to a human representative if it is not appropriate" is a mechanism for determining the appropriateness of the response based on the results of a reliability evaluation, and transferring the inquiry to a human representative if necessary.

[1154] The "means for performing emotion analysis and escalating inquiries to a human agent based on the user's emotions" is a mechanism for analyzing the emotions contained in a user's inquiry and transferring the inquiry to a human agent based on the results.

[1155] The "means for transmitting the final answer to the user" is a mechanism for transmitting the final confirmed answer to the user's terminal.

[1156] The "means for saving inquiries and responses in a log" is a mechanism for saving the inquiry content, acquired information, generated responses, and the results of escalation and sentiment analysis in a log.

[1157] MODE FOR CARRYING OUT THE INVENTION

[1158] The present invention is a system for responding effectively and quickly to inquiries from corporate customers. This system not only receives inquiries, analyzes them, and generates appropriate responses, but also includes an emotion engine that recognizes the emotions of users.

[1159] System hardware and software configuration

[1160] 1. User device: The device (PC, smartphone, tablet, etc.) through which the user sends an inquiry via the chatbot.

[1161] 2. Server: The central computing resource that performs various processes, such as query analysis, information retrieval, answer generation, credibility assessment, and sentiment analysis.

[1162] Natural Language Processing (NLP): Uses Google Cloud Natural Language API.

[1163] Internal system access: Access to our internal database (MySQL).

[1164] Generative AI model: OpenAI's GPT-4 is used.

[1165] Sentiment analysis: Using Azure Text Analytics.

[1166] System Operation

[1167] The user enters the question "Please tell me the renewal period for this phone number" into the chatbot interface. The device receives this query in real time and sends it to the server.

[1168] The server analyzes the received query using the Google Cloud Natural Language API. Based on the analysis results, it identifies the keyword "renewal period" and determines that the query is about the phone number renewal period.

[1169] The server then queries the company's internal database using MySQL to retrieve relevant data. Based on this information, the server uses OpenAI's GPT-4 model to generate an appropriate answer, such as "Your phone number is due for renewal on December 31, 2023."

[1170] To assess the reliability of the generated answer, the server calculates a confidence score and automatically sends the answer to the user if it exceeds a set threshold (CERTAINTY_THRESHOLD), or escalates the answer to a human agent if the confidence is low.

[1171] Additionally, the server uses Azure Text Analytics to analyze the user's emotions and, if it determines that the user is expressing anger or frustration, it will immediately escalate the call to a human agent.

[1172] The final answer is sent from the server to the device via an HTTP response or WebSocket protocol. For example, the answer may say, "Your phone number renewal period ends on December 31, 2023."

[1173] All interactions are logged by the server, including the inquiry, the information obtained, the response generated, the escalation history, and the results of sentiment analysis, which can be used to improve the system and its accuracy.

[1174] Examples of concrete examples and prompts

[1175] For example, if a user types "What is the expiration date of my contract?", the server uses NLP technology to identify the keyword "contract expiration date" and retrieves relevant information from the company's internal database. The generative AI model creates an answer such as "Your contract expires on March 31, 2024," which is sent to the user after passing a reliability assessment. If the user expresses strong dissatisfaction, the emotion engine will be activated and a human agent will manually process the request.

[1176] An example prompt is:

[1177] "Please tell me the renewal period for this phone number."

[1178] Please tell me the expiration date of the contract.

[1179] As described above, the present invention is extremely useful as a means for improving the efficiency of responses to inquiries from corporate customers and providing more advanced customer service.

[1180] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1181] Step 1:

[1182] A user inputs an inquiry into a corporate chatbot and clicks the send button. For example, they might input "Please tell me the renewal period for this phone number." The specific action is when the user inputs text into the chat window and clicks the send button. The input volume is the content of the user's inquiry, and the output volume is when this inquiry is transferred to the server.

[1183] Step 2:

[1184] The terminal receives the query. The terminal receives the query from the user in real time, converts the content into JSON format, and sends it to the server. The input is the text received from the user, and the output is JSON format data. Specifically, the terminal analyzes and structures the content of the query, and sends it to the server as an API request.

[1185] Step 3:

[1186] The server sends the received JSON data to the Google Cloud Natural Language API to identify the intent of the query. The input is JSON data, and the output is the analysis results, such as keywords and intent. Specifically, the server sends an API request to Google Cloud and receives the analysis results.

[1187] Step 4:

[1188] The server queries the company's internal database based on the analysis results to retrieve relevant information. The input is the analyzed keywords and intent, and the output is the relevant information as a result of the database query. Specifically, the server executes an SQL query to extract the required data.

[1189] Step 5:

[1190] Based on the information acquired by the server, it sends prompts to OpenAI's GPT-4 model to generate an appropriate answer to the query. The input is the information acquired from the database and the query content, and the output is the generated answer text. Specifically, the server provides the prompt to GPT-4 and executes an API request to generate the answer.

[1191] Step 6:

[1192] The server evaluates the reliability of the generated answer. The input is the generated answer text, and the output is the confidence score. Specifically, the server calculates the confidence of the answer and compares it with the configured threshold (CERTAINTY_THRESHOLD).

[1193] Step 7:

[1194] The server determines the appropriateness of the answer based on the evaluation results. The inputs are the confidence score and a threshold, and the output is the destination of the answer (automatic sending or escalation). Specifically, if the confidence exceeds the threshold, the answer is sent automatically, and if not, the answer is escalated to a human representative.

[1195] Step 8:

[1196] The server uses Azure Text Analytics to analyze user sentiment. The input is the user's query text, and the output is the sentiment analysis result. The specific operation is to send the query content to the API and receive the sentiment analysis result.

[1197] Step 9:

[1198] Based on the emotion analysis results, the server determines whether to escalate the inquiry to a person in charge. The input is the emotion analysis result, and the output is an instruction to escalate. Specifically, if anger or dissatisfaction is strong, an instruction to escalate is sent to a person in charge.

[1199] Step 10:

[1200] The server sends the final answer to the user. The input is the final confirmed answer text, and the output is the data sent to the user terminal. The specific operation is to send the answer using an HTTP response or the WebSocket protocol.

[1201] Step 11:

[1202] The server stores all interactions in a log. The inputs include the inquiry content, acquired information, generated answers, evaluation results, and sentiment analysis results, while the output is the stored log data. Specific operations include storing this data in a log file or database.

[1203] (Application example 2)

[1204] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1205] Conventional inquiry response systems have the problem of being unable to properly assess a user's emotions and respond accordingly when analyzing the content of inquiries and generating responses. This can result in an inability to respond appropriately to inquiries from users who are particularly emotional, which can lead to a decline in customer satisfaction. Furthermore, in some cases, escalation may not be performed appropriately if the response to an inquiry is unreliable, which also leads to a decline in customer satisfaction.

[1206] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving an inquiry, means for analyzing the received inquiry to identify the inquiry content, means for accessing an in-house system to acquire related information based on the identified inquiry content, means for generating an answer to the inquiry using a generative AI model based on the acquired information, means for evaluating the reliability of the generated answer, means for determining whether the answer is appropriate based on the evaluation result and escalating the answer to a human representative if it is not appropriate, means for analyzing emotions and adjusting the tone of the answer based on the analysis, means for sending the final answer to the user, and means for saving the inquiry and the answer in a log. This makes it possible to generate an appropriate answer that takes user emotions into consideration and provide a highly reliable answer.

[1207] The "means for receiving inquiries" refers to a device or software mechanism for electronically receiving inquiries from users.

[1208] "Means for analyzing received inquiries and identifying the content of the inquiries" refers to a mechanism for analyzing the content of received inquiries using natural language processing technology and identifying their intent and purpose.

[1209] "Means for accessing internal systems to obtain relevant information based on the identified inquiry content" refers to a mechanism for accessing internal systems or databases to obtain data related to the inquiry content.

[1210] "Means for generating responses to inquiries using a generative AI model based on acquired information" refers to a mechanism that uses a generative AI model (such as GPT-3) to generate appropriate responses in natural language format based on acquired data.

[1211] The "means for assessing the reliability of the generated answer" is a mechanism for calculating a confidence score to assess the accuracy and appropriateness of the generated answer.

[1212] "Means of determining whether the answer is appropriate based on the evaluation results, and escalating it to a human representative if it is not appropriate" is a mechanism that transfers the evaluated answer to a human representative if it does not meet certain reliability standards.

[1213] "Means for analyzing emotions and adjusting the tone of the response based on that" is a mechanism that analyzes emotions from the user's statements and text and generates a response in a tone that corresponds to those emotions.

[1214] The "means for transmitting the final answer to the user" is a mechanism for transmitting the generated final answer to the user using a communication means.

[1215] "Means for logging inquiries and responses" refers to a mechanism for storing all inquiries and responses as records.

[1216] The present invention is a system for effectively responding to user inquiries, which analyzes the inquiry, generates an answer, adjusts the tone of the response based on sentiment analysis, etc. An embodiment of this system will be described in detail below.

[1217] First, a user sends a query through a smartphone application. For example, a query such as "Please tell me the release date of the latest drama." The device receives this query in real time and transmits the query content to the server.

[1218] The server then analyzes the received query using natural language processing (NLP) techniques to determine the intent of the query, for example by extracting the keyword "publication date" and determining that it is related to a specific piece of content.

[1219] After identifying the query, the server accesses the company's internal systems via API to retrieve relevant information. Specifically, the server queries the company's internal database to retrieve information about the release dates of "new dramas," for example. This data is then formatted by the server and passed to the generative AI model.

[1220] Based on the acquired information, the server uses a generative AI model (e.g., GPT-3) to generate an appropriate answer to the query. The generated answer is provided in the form of natural language, such as "Thank you for your question. The release date of 'New Drama' is October 21, 2023."

[1221] The server then evaluates the reliability of the generated answer using a confidence score, and if the confidence exceeds a certain threshold, it returns the answer to the user directly, otherwise it escalates the answer to a human agent.

[1222] Furthermore, the server uses an emotion engine to analyze the user's emotions. It analyzes the language and emotional expressions contained in the inquiry, and if the user expresses dissatisfaction or anger, it immediately escalates the inquiry or adjusts the tone of the response. For example, if a user says, "I want an answer right away!", the server will respond with a polite tone, such as, "I'm sorry. I understand your dissatisfaction. I'll look into it right away."

[1223] Finally, the verified answer is sent to the user. Answers with high confidence or verified by a human agent are sent to the user's device via HTTP responses or real-time messaging protocols.

[1224] Finally, the server stores all interactions in a log, including the inquiry content, information obtained, responses generated, escalation history, sentiment analysis results, etc. These logs are used to improve the system and improve the accuracy of inquiry responses in the future.

[1225] Example prompt:

[1226] User Question: "What is the release date of the latest drama?"

[1227] The system responds: "The content in question is 'New Drama' and its release date is October 21, 2023."

[1228] User Question: "I want an answer right now!"

[1229] The system responds: "We're sorry, we understand your frustration. We'll look into it right away."

[1230] In this way, the embodiment of the present invention makes it possible to generate an appropriate answer that takes into account the user's feelings and to provide a highly reliable answer.

[1231] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1232] Step 1:

[1233] The user sends a query. For example, the user inputs a query such as "Please tell me the release date of the latest drama" through a smartphone application. The device receives this query in real time and transmits the query to the server.

[1234] Step 2:

[1235] The server analyzes the query. The server uses natural language processing (NLP) technology to analyze the received query and identify the intent of the query. Specifically, it extracts the keyword "publication date" and determines that the query is about the publication date of the content. The input for this analysis is the user's query, and the output is the analysis result (the intent of the query).

[1236] Step 3:

[1237] The server accesses the internal system to obtain information. Based on the specified query, the server accesses the internal database via API to obtain related information (e.g., content release date). The input of this step is the analysis result, and the output is the obtained information (e.g., the release date of the drama).

[1238] Step 4:

[1239] The server generates an answer using a generative AI model. Based on the acquired information, the server uses a generative AI model (e.g., GPT-3) to generate an answer in natural language format. For example, it creates an answer such as, "Thank you for your question. The release date of 'New Drama' is October 21, 2023." The input of this step is the acquired information, and the output is the generated answer.

[1240] Step 5:

[1241] The server evaluates the reliability of the generated answer. The server calculates a confidence score for the generated answer and determines whether this score exceeds a certain threshold. If the score is low, it escalates to a human agent. The input to this step is the generated answer and its confidence score, and the output is a branching process depending on whether escalation is necessary.

[1242] Step 6:

[1243] The server analyzes the emotion and adjusts the tone of the response. The server uses an emotion engine to analyze the emotion from the user's query. For example, if the user says, "I want an answer right away!", it determines that the user is expressing anger or dissatisfaction. As a result, the server creates a response in a tone such as, "I'm sorry. I understand your dissatisfaction. We will look into it right away." The input for this step is the user's query, and the output is a response that takes the emotion into consideration.

[1244] Step 7:

[1245] The server sends a final answer to the user. The server sends the answer that is determined to be reliable or the answer that has been confirmed after escalation to the user's device. This final answer is delivered via an HTTP response or real-time messaging protocol. The input of this step is the confirmed answer text, and the output is the answer sent to the user's device.

[1246] Step 8:

[1247] The server stores the queries and responses in a log. The server records and stores all interactions as log data. This data includes the query content, information obtained, responses generated, escalation history, and sentiment analysis results. The log data is used for future system improvements. The input of this step is a record of all interactions, and the output is the stored log data.

[1248] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1249] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1250] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1251] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1252] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1253] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1254] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1255] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1256] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1257] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1258] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1259] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1260] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1261] 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.

[1262] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1263] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1264] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1265] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1266] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1267] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1268] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1269] The following is further disclosed regarding the above embodiment.

[1270] (Claim 1)

[1271] means for receiving inquiries;

[1272] A means for analyzing the received inquiry to identify the content of the inquiry;

[1273] A means of accessing internal systems to obtain relevant information based on the identified inquiry; and

[1274] A means for generating a response to the inquiry using a generative AI model based on the acquired information;

[1275] a means of assessing the reliability of the generated answers;

[1276] Based on the evaluation results, it determines whether the answer is appropriate, and if it is not appropriate, it escalates it to a human agent.

[1277] means for transmitting the final answer to the user;

[1278] a means of logging inquiries and responses;

[1279] A system including:

[1280] (Claim 2)

[1281] 2. The system of claim 1, wherein the means for assessing the reliability of the generated answer calculates a confidence score and automatically submits the answer only if it exceeds a certain confidence threshold.

[1282] (Claim 3)

[1283] 2. The system according to claim 1, wherein the query analysis means uses natural language processing technology.

[1284] "Example 1"

[1285] (Claim 1)

[1286] means for receiving inquiries;

[1287] A means for analyzing the received inquiry to identify the content of the inquiry;

[1288] a means of accessing internal systems to obtain relevant information based on the identified inquiry;

[1289] means for generating a response to the inquiry using a generative artificial intelligence model based on the acquired information;

[1290] a means of assessing the reliability of the generated answers;

[1291] Based on the evaluation results, it determines whether the answer is appropriate, and if it is not appropriate, it escalates it to a human agent.

[1292] means for transmitting the final answer to the user;

[1293] a means of keeping records of inquiries and responses;

[1294] A system including:

[1295] (Claim 2)

[1296] 2. The system of claim 1, wherein the means for assessing the reliability of the generated answer calculates a confidence score and automatically transmits the answer only if it exceeds a certain confidence threshold.

[1297] (Claim 3)

[1298] 2. The system according to claim 1, wherein the query analysis means uses natural language processing technology.

[1299] "Application Example 1"

[1300] (Claim 1)

[1301] means for receiving inquiries;

[1302] A means for analyzing the received inquiry to identify the content of the inquiry;

[1303] A means of accessing internal systems to obtain relevant information based on the identified inquiry; and

[1304] A means for generating a response to the inquiry using a generative AI model based on the acquired information;

[1305] a means of assessing the reliability of the generated answers;

[1306] Based on the evaluation results, it determines whether the answer is appropriate, and if it is not appropriate, it escalates it to a human agent.

[1307] means for transmitting the final answer to the user;

[1308] a means of logging inquiries and responses;

[1309] means implemented as a smartphone application for receiving, analyzing, and generating responses to inquiries related to a particular service;

[1310] means for inputting and displaying inquiries through a user interface;

[1311] A means for generating an answer by providing a prompt sentence to a generative AI model based on the query content;

[1312] A system including:

[1313] (Claim 2)

[1314] 2. The system of claim 1, wherein the means for assessing the reliability of the generated answer calculates a confidence score and automatically submits the answer only if it exceeds a certain confidence threshold.

[1315] (Claim 3)

[1316] 2. The system according to claim 1, wherein the query analysis means uses natural language processing technology.

[1317] "Example 2: Combining Emotion Engines"

[1318] (Claim 1)

[1319] means for receiving inquiries;

[1320] A means for analyzing the received inquiry to identify the content of the inquiry;

[1321] a means of accessing internal systems to obtain relevant information based on the identified inquiry;

[1322] A means for generating a response to the inquiry using a generative AI model based on the acquired information;

[1323] a means of assessing the reliability of the generated answers;

[1324] Based on the evaluation results, it determines whether the answer is appropriate, and if it is not appropriate, it escalates it to a human agent.

[1325] A means for performing sentiment analysis and escalating inquiries to personnel based on user sentiment;

[1326] means for transmitting the final answer to the user;

[1327] a means of logging inquiries and responses;

[1328] A system including:

[1329] (Claim 2)

[1330] 2. The system of claim 1, wherein the means for assessing the reliability of the generated answer calculates a confidence score and automatically submits the answer only if it exceeds a certain confidence threshold.

[1331] (Claim 3)

[1332] 2. The system according to claim 1, wherein the query analysis means uses natural language processing technology.

[1333] "Application example 2 when combining emotion engines"

[1334] (Claim 1)

[1335] means for receiving inquiries;

[1336] A means for analyzing the received inquiry to identify the content of the inquiry;

[1337] A means of accessing internal systems to obtain relevant information based on the identified inquiry; and

[1338] A means for generating a response to the inquiry using a generative AI model based on the acquired information;

[1339] a means of assessing the reliability of the generated answers;

[1340] Based on the evaluation results, it determines whether the answer is appropriate, and if it is not appropriate, it escalates it to a human agent.

[1341] A means of analyzing sentiment and adjusting the tone of responses accordingly;

[1342] means for transmitting the final answer to the user;

[1343] a means of logging inquiries and responses;

[1344] A system including:

[1345] (Claim 2)

[1346] 2. The system of claim 1, wherein the means for assessing the reliability of the generated answer calculates a confidence score and automatically submits the answer only if it exceeds a certain confidence threshold.

[1347] (Claim 3)

[1348] 2. The system according to claim 1, wherein the query analysis means uses natural language processing technology. [Explanation of symbols]

[1349] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for receiving inquiries; A means for analyzing the received inquiry to identify the content of the inquiry; A means of accessing internal systems to obtain relevant information based on the identified inquiry; and A means for generating a response to the inquiry using a generative AI model based on the acquired information; a means of assessing the reliability of the generated answers; Based on the evaluation results, it determines whether the answer is appropriate, and if it is not appropriate, it escalates it to a human agent. means for transmitting the final answer to the user; a means of logging inquiries and responses; A system including:

2. 2. The system of claim 1, wherein the means for assessing the reliability of the generated answer calculates a confidence score and automatically transmits the answer only if it exceeds a certain confidence threshold.

3. 2. The system according to claim 1, wherein the query analysis means uses natural language processing technology.

Images