Inquiry answering system, information processing device, inquiry answering method, and program

The inquiry response system addresses the inefficiency of multiple user interactions by supplementing missing information using AI, enhancing the response generation process and reducing user burden.

WO2026235072A1PCT designated stage Publication Date: 2026-11-12MITSUBISHI ELECTRIC BUILDING SOLUTIONS CORP +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/JP2026/016907
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-08
Filing Date
2026-04-28
Publication Date
2026-11-12

AI Technical Summary

Technical Problem

Existing systems that generate responses to user inquiries using AI require multiple user interactions to gather missing information, increasing the user's response burden.

Method used

An inquiry response system that includes an inquiry acquisition unit, a missing information determination unit, an additional information determination unit, and a response generation unit to supplement missing information using AI, reducing the need for user interaction.

Benefits of technology

Reduces the user's response burden by generating and supplementing missing information, thereby improving the efficiency of the inquiry response process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2026016907_12112026_PF_FP_ABST
    Figure JP2026016907_12112026_PF_FP_ABST
Patent Text Reader

Abstract

This inquiry answering system: acquires natural language data of inquiry information including an inquiry from a user; determines, on the basis of the inquiry information and in order to generate an answer, missing information that is missing; and determines supplementary information corresponding to the missing information. The system acquires user information of the user associated with the inquiry information from, for example, a database in which past answer history information is registered, and determines the supplementary information on the basis of the acquired user information. Thereafter, the system: generates supplementary inquiry information, which is obtained by adding the supplementary information to the inquiry information; inputs the supplementary inquiry information to an artificial intelligence unit; acquires an answer corresponding to the supplementary inquiry information from the artificial intelligence unit; and outputs the answer to the user.
Need to check novelty before this filing date? Find Prior Art

Description

Inquiry response system, information processing apparatus, inquiry response method, and program

[0001] The present disclosure relates to the technology of a system that generates responses to inquiries from users using functions of Artificial Intelligence (AI).

[0002] Patent Document 1 discloses an information processing technology that classifies dialogues into business dialogues and casual conversations, and generates response sentences for responding to dialogues related to business. When it is determined that information for generating a response sentence for a business dialogue is insufficient, the system of this technology generates a response sentence for inquiring about that information.

[0003] Japanese Unexamined Patent Publication No. 2024-129098

[0004] The technology of Patent Document 1 described above can inquire of the user for missing information when generating a response to an inquiry from the user using the function of artificial intelligence. However, if the system inquires of the user one by one for all information that is missing to generate the response, the number of responses required before reaching the answer desired by the user will increase.

[0005] The present disclosure has been made to solve the above-described problems, and an object thereof is to provide a technology capable of reducing the user's response burden by allowing the system to generate and supplement information that is missing from the user's inquiry in a system that generates responses to inquiries from users.

[0006] The inquiry response system of the present disclosure includes: an inquiry acquisition unit that acquires natural language data of inquiry information including a user's inquiry; a missing information determination unit that determines missing information that is insufficient to generate a response to the inquiry based on the inquiry information; an additional information determination unit that determines additional information corresponding to the missing information based on the inquiry information; an information adding unit that generates added inquiry information by adding the additional information to the inquiry information; and a response generation unit that inputs the added inquiry information to an artificial intelligence unit, acquires and outputs a response corresponding to the added inquiry information from the artificial intelligence unit.

[0007] Furthermore, the information processing device disclosed herein includes: a missing information determination unit that acquires natural language data of inquiry information including a user's inquiry and determines missing information that is insufficient to generate an answer to the inquiry based on the inquiry information; an additional information determination unit that determines additional information corresponding to the missing information based on the inquiry information; an information addition unit that generates additional inquiry information by adding the additional information to the inquiry information; and an answer generation unit that inputs the additional inquiry information to an artificial intelligence unit and obtains and outputs an answer corresponding to the additional inquiry information from the artificial intelligence unit.

[0008] Furthermore, the inquiry response method of this disclosure is an inquiry response method executed by a computer, comprising the steps of: acquiring natural language data of inquiry information including a user's inquiry; determining missing information that is insufficient to generate an answer to the inquiry based on the inquiry information; determining additional information corresponding to the missing information based on the inquiry information; generating additional inquiry information by adding the additional information to the inquiry information; and inputting the additional inquiry information into an artificial intelligence unit and obtaining and outputting an answer corresponding to the additional inquiry information from the artificial intelligence unit.

[0009] Furthermore, the program disclosed herein is a program that causes a computer to output a response to a user's inquiry, and it obtains natural language data of inquiry information including the user's inquiry, determines any missing information necessary to generate a response to the inquiry based on the inquiry information, determines any additional information corresponding to the missing information based on the inquiry information, generates additional inquiry information by adding the additional information to the inquiry information, inputs the additional inquiry information into the artificial intelligence unit, and causes the computer to obtain and output a response corresponding to the additional inquiry information from the artificial intelligence unit.

[0010] According to the technology disclosed herein, in a system that generates responses to user inquiries, the system can reduce the burden on users by generating and supplementing any missing information in the user's inquiry.

[0011] This is a diagram illustrating the outline of the inquiry response system according to the embodiment of this disclosure. This is a block diagram illustrating the internal configuration of the answer generation server according to the embodiment of this disclosure. This is a block diagram illustrating the internal configuration of the user client according to the embodiment of this disclosure. This is a block diagram showing an excerpt of some of the information processing functions of the inquiry response system. This is a diagram illustrating an example of missing information determination processing. This is a diagram illustrating an example of user information acquisition processing. This is a diagram illustrating an example of additional information determination processing. This is a diagram illustrating an example of information addition processing. This is a diagram illustrating an example of output results from answer output processing. This is a flowchart illustrating the processing flow executed in the inquiry response system according to Embodiment 1. This is a diagram illustrating a modified version of the hardware resources of the answer generation server. This is a diagram illustrating a modified version of the hardware resources of the user client. This is a block diagram showing an excerpt of some of the information processing functions of the inquiry response system according to Embodiment 2. This is a diagram illustrating an example of topic determination processing. This is a diagram illustrating an example of related information acquisition processing. This is a diagram illustrating an example of information acquisition means determination processing. This is a flowchart of the processing executed in the inquiry response system according to Embodiment 2. This is a block diagram illustrating an excerpt of some of the information processing functions of the inquiry response system according to Embodiment 3. This is a diagram illustrating an example of similar answer search processing. This is a diagram illustrating an example of necessary information extraction processing. This is a flowchart of the processes executed in the inquiry response system according to Embodiment 3.

[0012] The embodiments will be described below with reference to the drawings. Elements common to each figure are denoted by the same reference numerals, and redundant explanations are omitted.

[0013] 1. Embodiment 1. 1-1. Schematic Diagram 1 of the Inquiry Response System of Embodiment 1 is a diagram illustrating the overview of the inquiry response system according to the embodiment of this disclosure. The inquiry response system 10 is a system that generates answers to inquiries entered by users using the functions of artificial intelligence. The user is, for example, a maintenance worker who performs maintenance work on equipment such as elevators in a building, and the inquiries include, for example, content related to the handling of equipment such as maintenance or inspection of equipment in a building.

[0014] The inquiry response system 10 comprises an answer generation server 100, one or more user clients 300, and an LLM server 400. Each of the answer generation server 100, one or more user clients 300, and LLM server 400 is connected to a communication network 200, such as the Internet, a wired LAN, or a wireless LAN.

[0015] The response generation server 100 is an information processing device that generates responses to user inquiries. Figure 2 is a block diagram illustrating the internal configuration of the response generation server according to an embodiment of the present disclosure. The response generation server 100 includes a main memory 110, a communication I / F 120, an auxiliary storage device 130, and a processor 140. The response generation server 100 is configured, for example, by a PC (Personal Computer).

[0016] The communication interface 120 is an interface for connecting to the communication network 200. The communication interface 120 communicates with the user client 300 and the LLM server 400 via the communication network 200.

[0017] The auxiliary storage device 130 stores an equipment information database 131, a building information database 132, a document database 133, a past response history database 134, and a user-specific response history database 135. The equipment information database 131 stores equipment information related to the equipment installed in the building. The equipment information includes, for example, information such as the building name where it is installed, the installation location, alerts, power values, etc., associated with the equipment name.

[0018] The building information database 132 stores building information related to the building. This building information includes, for example, information such as the construction date, installed equipment, number of floors, and purpose, all linked to the building name. Note that equipment information and building information are part of the equipment information, and the equipment information database 131 and the building information database 132 are part of the database in which the equipment information is stored.

[0019] The document database 133 contains document information related to inquiries about building equipment and facilities. The past response history database 134 stores past response history information. This is history information related to past inquiries and answers associated with a user. Details of the response history information will be described later. The user-specific response history database 135 contains user-specific response history information generated from the past response history information stored in the past response history database 134.

[0020] The auxiliary storage device 130 stores the information processing program and various data necessary to operate the information processing program. The processor 140 includes a CPU (Central Processing Unit). The processor 140 reads and executes the information processing program, thereby realizing the various functions of the answer generation server 100. Details of the various functions realized will be described later. The information processing program may also be stored in the main storage device 110. Furthermore, the information processing program may be recorded on a computer-readable recording medium, or stored in a server device capable of communicating with the answer generation server 100, such as a cloud server.

[0021] The user client 300 is a device that accepts free text input in natural language containing user inquiries and outputs responses to those inquiries. Figure 3 is a block diagram illustrating the internal configuration of the user client according to the embodiment of this disclosure.

[0022] The user client 300 includes a communication interface 310, a main memory 320, an auxiliary memory 330, a processor 360, an input interface 350, and an output interface 340. The user client 300 is connected to an output device 370 and an input device 380. The user client 300, the output device 370, and the input device 380 are configured as terminal devices operated by the user, such as a PC (Personal Computer), smartphone, or tablet.

[0023] The communication interface 310 is an interface for connecting to the communication network 200. The communication interface 310 communicates with the response generation server 100 via the communication network 200.

[0024] The auxiliary storage device 330 stores the information processing program and various data necessary to operate the information processing program. The processor 360 includes a CPU (Central Processing Unit). The processor 360 reads and executes the information processing program, thereby realizing various functions of the user client 300. Details of the various functions realized will be described later. The information processing program may also be stored in the main memory device 320. Furthermore, the information processing program may be recorded on a computer-readable recording medium, or stored in a server device capable of communicating with the user client 300, such as a cloud server.

[0025] The input interface 350 is an interface that receives user input from the input device 380. Examples of input devices 380 include a mouse, keyboard, microphone, stylus, etc.

[0026] The output I / F 340 is an interface that outputs various information to the output device 370. Examples of the output device 370 include a display device such as a liquid crystal display, a speaker, etc. The output device 370 may also be configured as a touch panel that has the function of receiving operation input from an input device 380 such as a stylus pen.

[0027] The LLM server 400 is a device equipped with the functions of an artificial intelligence unit that realizes artificial intelligence with intelligent functions such as reasoning and judgment, and its operating environment. The LLM server 400 includes an LLM 401 and an LLM control unit 402. The LLM 401 stores, for example, a language learning model that takes natural language as input and obtains output results. The language learning model includes model information, which will be described later. The language learning model may also include model parameters, which are information that defines the behavior of the model, such as constraints, weighting variables, and evaluation functions.

[0028] The model may be, for example, a model called LLM (Large Language Model), BERT (Bidirectional Encoder Representations from Transformers), or GPT (Generative Pre-trained Transformer) (GPT is a registered trademark). In addition, the language learning model may be a combination of information from the above-mentioned models and information from other models, such as those shown below. Other models may include, for example, models called NN (Neural Network), CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), VAE (Variational Autoencoder), GAN (Generative Adversarial Networks), diffusion model, transformer, VLM (Visual Language Model), and CLIP (Contrastive Language Image Pre-training). The models mentioned above are not mutually exclusive; for example, LLM, VLM, BERT, and GPT are included in Transformer. Also, for example, Transformer is included in NN. Furthermore, learning algorithms and models may be combinations of multiple types. Models also include what are called multimodal models, which are trained using a combination of multiple different types of data.

[0029] When the LLM control unit 402 acquires various types of information, as described later, it outputs a generation result corresponding to the acquired information based on the language learning model stored in the LLM 401.

[0030] Furthermore, other information used by the LLM 401 and LLM control unit 402 may be pre-prepared or, if necessary, acquired via the network.

[0031] 1-2. Functional Configuration of the Inquiry Response System Next, the functional configuration of the inquiry response system 10 will be explained in more detail. Figure 4 is a block diagram showing an excerpt of some of the information processing functions of the inquiry response system. The user client 300 of the inquiry response system 10 includes an inquiry acquisition unit 361, an answer output unit 362, and an evaluation information acquisition unit 363, as functions realized by the processor 360 reading and executing an information processing program. The answer generation server 100 of the inquiry response system 10 includes a missing information determination unit 142, a user information acquisition unit 145, an additional information determination unit 146, an information additional unit 147, an answer generation unit 150, a history registration unit 153, a specific history extraction unit 152, and an answer history generation unit 151, as functions realized by the processor 140 reading and executing an information processing program.

[0032] The inquiry acquisition unit 361 is a functional block that executes the process of acquiring natural language data of inquiry information, including the user's inquiry. This process is hereinafter referred to as the "inquiry acquisition process." When a user inputs free-form inquiry information in natural language from the input device 380, the inquiry acquisition unit 361 executes the inquiry acquisition process to acquire the natural language data of the input inquiry information. The natural language data may be text data input from the input device 380, such as a keyboard, or voice data input from the input device 380, such as a microphone. In the following explanation, we will describe the functions of the inquiry answering system 10 using as an example an inquiry related to maintenance work on equipment in a building, where user "Mr. A," a maintenance worker, inputs inquiry information such as "The alert on the equipment installed in building A won't go away. Please tell me why." The acquired inquiry information is sent to the answer generation server 100 via the communication network 200.

[0033] The missing information determination unit 142 is a functional block that performs a process to determine the missing information necessary to generate an answer to a user's inquiry, based on the inquiry information. This process is hereinafter referred to as the "missing information determination process". Figure 5 is a diagram illustrating an example of the missing information determination process. In the example shown in this figure, the information necessary for the inquiry to generate an answer, and specific examples thereof, are set in advance. In the missing information determination process, the missing information determination unit 142 determines that information whose terms in the set specific examples are not included in the inquiry information is missing information necessary to generate an answer. Figure 5 illustrates a case where information regarding the equipment name, part name, answer format, building address, and contact information is determined to be missing information that is not included in the inquiry information.

[0034] Another example of the missing information determination process is that the missing information determination unit 142 may be configured to determine missing information that is not included in the query information by utilizing the functions of artificial intelligence such as the LLM server 400. For example, the LLM 402 of the LLM server 400 contains information of a model that has been learned by linking the query information with the missing information. Alternatively, the LLM 402 contains information of a model that has been learned by linking the query information with the information necessary for the query and specific examples thereof. The missing information determination unit 142 inputs the acquired query information to the LLM server 400 and obtains the missing information corresponding to the query information from the LLM server 400.

[0035] The user information acquisition unit 145 is a functional block that performs a process of searching for information related to the inquiry information from the user's past response history information, based on the response history information for each user stored in the response history database 135 for each user, and acquiring it as user information. This process is hereinafter referred to as the "user information acquisition process". Figure 6 is a diagram illustrating an example of the user information acquisition process. In the example shown in this figure, in the user information acquisition process, the user information acquisition unit 145 searches for response history information similar to the inquiry information from the response history information for user "Mr. A" stored in the response history database 135 for each user, and extracts it as user information. Here, an example is shown in which response history information for Mr. A related to "Building A" included in the inquiry information is extracted as user information. Note that the user information may be summarized using the functions of artificial intelligence such as the LLM server 400. Figure 6 illustrates user information summarized using the functions of artificial intelligence. The acquired user information is sent to the additional information determination unit 146.

[0036] The additional information determination unit 146 is a functional block that performs a process to determine the content of additional information to supplement the missing information, based on the inquiry information, missing information, and user information. This process is hereinafter referred to as the "additional information determination process". Figure 7 is a diagram illustrating an example of the additional information determination process. The missing information has, for example, a predetermined importance level for generating an answer. The LLM 402 of the LLM server 400 contains information from a model that has been learned by linking the inquiry information, missing information, user information, and additional information. In the additional information determination process, the additional information determination unit 146 inputs the inquiry information, the missing information with high importance, and the user information to the LLM server 400, and obtains additional information from the LLM server 400 that corresponds to the inquiry information, the missing information with high importance, and the user information.

[0037] Alternatively, in the additional information determination process, the additional information determination unit 146 inputs the inquiry information, missing information, and user information to the LLM server 400, thereby obtaining additional information corresponding to the inquiry information, missing information, and user information from the LLM server 400.

[0038] Alternatively, the LLM 402 of the LLM server 400 contains information of a model that has been learned by linking query information, high-priority missing information, user information, and supplementary information. In this case, during the supplementary information determination process, the supplementary information determination unit 146 inputs the query information and missing information to the LLM server 400 to determine the high-priority missing information corresponding to the query information and missing information, and obtains supplementary information corresponding to the query information, the determined high-priority missing information, and user information from the LLM server 400. The obtained supplementary information is sent to the information supplementary unit 147.

[0039] The information appending unit 147 is a functional block that executes a process to generate information by appending additional information to the query information to supplement any missing information. This process is hereinafter referred to as the "information appending process," and the information generated by the information appending process is hereinafter referred to as the "append query information." Figure 8 is a diagram illustrating an example of the information appending process. In the information appending process, the information appending unit 147 generates append query information by appending additional information to the query information. At this time, the information appending unit 147 may also unify the terminology for synonyms in the generated append query information.

[0040] As another example of the information appending process, the information appending unit 147 may use the functions of artificial intelligence such as the LLM server 400 to generate query information and appending query information corresponding to the appended information.

[0041] The answer generation unit 150 is a functional block that uses the artificial intelligence functions of the LLM server 400 to perform the process of generating answers corresponding to additional inquiry information. This process is hereinafter referred to as the "answer generation process". The LLM 402 of the LLM server 400 contains information of a model that has been learned by associating additional inquiry information with answers to inquiries. In the answer generation process, the answer generation unit 150 inputs the additional inquiry information to the LLM server 400 and obtains the answer corresponding to the additional inquiry information from the LLM server 400. The obtained answer is transmitted to the user client 300 via the communication network 200. The obtained answer is also sent to the history registration unit 153.

[0042] The response output unit 362 is a functional block that executes the process of outputting a response to a user's inquiry to the user. This process is hereinafter referred to as the "response output process". Figure 9 is a diagram illustrating an example of the output result from the response output process. In the example shown in this figure, during the response output process, the response output unit 362 displays the inquiry information, additional information, and the response on a display device, such as the output device 370.

[0043] Note that the output mode in the response output processing is not limited to the display mode, and the inquiry information, additional write information, and response may be output by voice from a speaker serving as the output device 370. Further, in the response output processing, the output of the additional write information is not essential, and for example, the inquiry information and the response may be displayed on a display device serving as the output device 370.

[0044] The evaluation information acquisition unit 363 is a functional block that executes processing for acquiring evaluation information related to a user's evaluation of a response. This processing is hereinafter referred to as "evaluation information acquisition processing". In the evaluation information acquisition processing, the evaluation information acquisition unit 363 acquires the evaluation information input by the user from the input device 380. Examples of the evaluation information include input information of an action such as "like" input by the user, input information in a free text format regarding whether the response is the desired content, and input information of reflection points or request points related to the response such as what kind of response was desired. The acquired evaluation information is transmitted to the response generation server 100 via the communication network 200.

[0045] The history registration unit 153 is a functional block that executes processing for registering response history information related to a user's inquiry in the response history database 135. This processing is hereinafter referred to as "history registration processing". The response history information includes, as information associated with the user who made the inquiry, for example, information such as inquiry information, inquiry date and time, user information, additional write information, evaluation information, response, and related materials.

[0046] The specific history extraction unit 152 is a functional block that performs a process to extract only specific answer history information from the answer history information registered in the answer history database 135 during the history registration process and register it in the answer history database 135. This process is hereinafter referred to as the "specific history extraction process". The specific answer history information here is, for example, answer history information corresponding to an action that has received a higher evaluation than the standard evaluation in the evaluation information. Alternatively, if the evaluation information is input information in the form of free text, quantitative evaluation information may be obtained from the input information using the functions of artificial intelligence such as the LLM server 400, and specific answer history information may be determined based on the obtained evaluation information. Note that the configuration of the specific history extraction unit 152 is not a mandatory configuration in the inquiry response system 10.

[0047] The response history generation unit 151 is a functional block that generates response history information for each user from the response history information registered in the response history database 135 and registers it in the response history database 135 for each user. This process is hereinafter referred to as the "response history generation process". In the response history generation process, the response history generation unit 151 may register response history information for each department to which the user belongs, rather than response history information for each user.

[0048] 1-4. Specific Processing Performed in the Inquiry Response System The specific processing performed in the inquiry response system 10 will be described below with reference to the flowchart. Figure 10 is a flowchart showing the flow of processing performed in the inquiry response system according to Embodiment 1. The processing in steps S100 to S104 of the flowchart shown in Figure 10 is performed by the processor 360 of the user client 300 executing an information processing program stored in the auxiliary storage device 330. Furthermore, the processing in steps S110 to S122 of the flowchart shown in Figure 10 is performed by the processor 140 of the answer generation server 100 executing an information processing program stored in the auxiliary storage device 130. Note that the flowchart shown in Figure 10 also represents a part of the inquiry response method in which the inquiry response system 10 outputs an answer to the user's inquiry to the user.

[0049] When a natural language inquiry is input by a user from the input device 380, the user client 300 executes an inquiry acquisition process in step S100 to acquire inquiry information. The acquired inquiry information is sent to the response generation server 100.

[0050] The response generation server 100 that has received the inquiry information executes a missing information determination process in step S110 to acquire missing information that is missing from the inquiry information. Note that in the missing information determination process, the artificial intelligence function of the LLM server 400 may be used to generate missing information corresponding to the inquiry information. When the process of step S110 is completed, the process proceeds to step S112.

[0051] In step S112, the response generation server 100 executes a user information acquisition process to acquire user information. When the process of step S112 is completed, the process proceeds to step S114.

[0052] In step S114, the response generation server 100 executes an additional information determination process to generate additional information. Note that in the additional information determination process, the artificial intelligence function of the LLM server 400 may be used to generate additional information corresponding to the inquiry information, the missing information, and the user information. When the process of step S114 is completed, the process proceeds to step S116.

[0053] In step S116, the response generation server 100 executes an information addition process to acquire added inquiry information. Note that in the information addition process, the artificial intelligence function of the LLM server 400 may be used to generate added inquiry information corresponding to the additional information and the inquiry information. When the process of step S116 is completed, the process proceeds to step S118.

[0054] In step S118, the response generation server 100 executes a response generation process to acquire a response to the user's inquiry. Here, the response generation server 100 uses the artificial intelligence function of the LLM server 400 to acquire a response corresponding to the added inquiry information. The acquired response is sent to the user client 300.

[0055] Upon receiving the response, the user client 300 executes the response output process in step S102. This outputs the inquiry information, additional information, and the response to the output device 370. Once the process in step S102 is complete, the process proceeds to step S104.

[0056] In step S104, the user client 300 executes the evaluation information acquisition process to acquire the evaluation information entered by the user from the input device 380. The acquired evaluation information is sent to the response generation server 100.

[0057] The response generation server 100, which has acquired the evaluation information, executes a history registration process in step S120 to register the response history information in the response history database 135. Once the process in step S120 is completed, the process proceeds to step S122. In step S122, the response generation server 100 executes a response history generation process to register the response history information in the corresponding response history database 135 for each user.

[0058] As described above, according to the inquiry response system 10 of this embodiment, if there is missing information in the inquiry information, supplementary information is generated to supplement the missing information. As a result, supplementary inquiry information including the supplementary information can be input to the LLM server 400 to obtain a response, thereby reducing the burden on the user to respond by not having to ask the user for missing information.

[0059] 1-5. Modified Examples The inquiry response system 10 of Embodiment 1 may adopt the following modified forms. These modifications can also be applied to the inquiry response system of other embodiments described later.

[0060] 1-5-1. Hardware resources of the response generation server 100 and user client 300

[0061] Figure 11 shows a modified example of the hardware resources of the answer generation server. In the example shown in Figure 11, the answer generation server 100 includes, for example, a processor 140, an auxiliary storage device 130, and a processing circuit 171 including dedicated hardware 170. Figure 11 shows an example in which some of the functions of the answer generation server 100 are realized by the dedicated hardware 170. All of the functions of the answer generation server 100 may also be realized by the dedicated hardware 170. As the dedicated hardware 170, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a combination thereof can be used.

[0062] Figure 12 shows a modified example of the hardware resources of a user client. In the example shown in Figure 12, the user client 300 includes, for example, a processor 360, an auxiliary storage device 330, and a processing circuit 391 including dedicated hardware 390. The configuration of the dedicated hardware 390 and the processing circuit 391 can be modified in the same way as the configuration of the dedicated hardware 170 and the processing circuit 171 in the modified configuration of the answer generation server 100 shown in Figure 11.

[0063] The auxiliary storage device 130 may be operated independently of the answer generation server 100, with the cloud or similar handling its role. Similarly, the auxiliary storage device 330 may be operated independently of the user client 300, with the cloud or similar handling its role.

[0064] Furthermore, the inquiry response system 10 may have all or part of the functions of the response generation server 100 or the user client 300 installed on a remote server.

[0065] 1-5-2. LLM Server 400 Figure 1 shows the case where the LLM server 400 is located outside the answer generation server 100 or the user client 300. However, part or all of the LLM server 400 may be located inside the answer generation server 100 or the user client 300. Also, the LLM 402 may consist of multiple databases connected via the communication network 200.

[0066] 2. Embodiment 2. In Embodiment 2, the differences from the example disclosed in Embodiment 1 will be explained in particular detail. For features not described in Embodiment 2, any of the features from the example disclosed in Embodiment 1 may be adopted.

[0067] 2-1. Functional Configuration of the Inquiry Response System in Embodiment 2 The functional configuration of the inquiry response system 10 according to Embodiment 2 will be described. Figure 13 is a block diagram showing an excerpt of some of the information processing functions of the inquiry response system according to Embodiment 2. The response generation server 100 of the inquiry response system 10 includes, in addition to the functions shown in Figure 4, a topic determination unit 141, a related information acquisition unit 144, and an information acquisition means determination unit 143, as functions realized by the processor 140 reading and executing an information processing program.

[0068] The topic determination unit 141 is a functional block that executes a process to determine the topic of the query included in the query information. This process is hereinafter referred to as the "topic determination process". Figure 14 is a diagram illustrating an example of the topic determination process. In the example shown in this figure, terms are predefined for each query topic. In the topic determination process, the topic determination unit 141 determines the topic of the query based on whether the defined topic terms are included in the query information. If the query information includes terms from multiple topics, the topic corresponding to the term that appears most frequently is determined to be the topic of the query. Figure 14 illustrates the case where "device alert" included in the query information is determined to be the topic.

[0069] As another example of topic determination processing, the topic determination unit 141 may classify the query information using a clustering algorithm such as the k-means method and estimate the topic from characteristic words. Alternatively, the topic determination unit 141 may be configured to obtain the topic corresponding to the query information using the functions of an artificial intelligence such as an LLM server 400. In this case, for example, the LLM 402 of the LLM server 400 contains information of a model that has been trained to associate query information with topics. The topic determination unit 141 inputs the acquired query information into the LLM server 400 and obtains the topic corresponding to the query information from the LLM server 400.

[0070] The acquired topics are sent to the missing information determination unit 142 and used, for example, when determining the importance of the missing information. In this case, the importance of the missing information is predetermined for each topic, and the importance corresponding to the determined missing information is set in the missing information determination process according to the determined topic. Note that in the inquiry response system 10 according to Embodiment 2, the configuration of the topic determination unit 141 is not essential.

[0071] The related information acquisition unit 144 is a functional block that executes a process to acquire related information related to the inquiry information. This process is hereinafter referred to as the "related information acquisition process." The related information here includes, for example, building information and equipment information related to the inquiry information.

[0072] Figure 15 is a diagram illustrating an example of the related information acquisition process. In the related information acquisition process, the related information acquisition unit 144 searches for information related to the query information from the building information stored in the building information database 132 and extracts it as related information. Figure 15 illustrates a case where building information related to "Building A" included in the query information is extracted. In addition, in the related information acquisition process, the related information acquisition unit 144 searches for information related to the query information and the extracted building information from the building information stored in the equipment information database 131 and extracts it as related information. Figure 15 illustrates a case where equipment information related to "Building A" included in the query information and "Equipment a" included in the extracted building information is extracted.

[0073] Furthermore, the databases referenced in the related information acquisition process are not limited to the building information database 132 and the equipment information database 131. That is, in the related information acquisition process, the related information acquisition unit 144 may search for personal information corresponding to the user of the inquiry information from a personal information database that stores information such as the user's history and job title, and extract it as related information. Also, in the related information acquisition process, the related information acquisition unit 144 may acquire further information from other databases based on information stored in one database. Furthermore, the acquired related information may be summarized using the functions of artificial intelligence such as the LLM server 400. The acquired related information is sent to the additional information determination unit 146.

[0074] The information acquisition means determination unit 143 is a functional block that executes a process to determine the information acquisition means to be used in the additional information determination process when there is missing information in the inquiry information. This process is hereinafter referred to as the "information acquisition means determination process".

[0075] Figure 16 is a diagram illustrating an example of the information acquisition means determination process. In the example shown in (A) of this figure, the information necessary to generate a response to an inquiry and the information of the means capable of acquiring each piece of information are pre-set. In the information acquisition means determination process, the information acquisition means determination unit 143 determines that the means associated with the missing information not included in the inquiry information are means capable of acquiring the information to be used in the additional information determination process.

[0076] Alternatively, as another example of the information acquisition means determination process, the information acquisition means determination unit 143 may be configured to determine means capable of acquiring the information to be used in the append information determination process by utilizing the functions of artificial intelligence such as the LLM server 400. For example, in the example shown in (B) of this figure, the LLM 402 of the LLM server 400 contains information of a model that has been learned by linking, for example, necessary query information with input / output examples from the processing of each means. The information acquisition means determination unit 143 inputs the acquired missing information to the LLM server 400, thereby acquiring means from the LLM server 400 that correspond to the input / output examples related to the missing information as means capable of acquiring the information to be used in the append information determination process.

[0077] Alternatively, as another example of information acquisition means determination processing using the functions of artificial intelligence such as the LLM server 400, for example, in the example shown in (C) in this figure, the LLM 402 of the LLM server 400 contains information of a model that has been learned by linking each means with the necessary query information that can be obtained by the processing in each means. The information acquisition means determination unit 143 inputs the acquired missing information to the LLM server 400, thereby obtaining from the LLM server 400 the means that can obtain the necessary query information corresponding to the missing information, and the means that can obtain the information to be used in the appending information determination processing.

[0078] Furthermore, if the additional information determination process determines that there is no means to obtain the information, the output device 370 of the user client 300 may output an inquiry about the missing information to prompt the user to input the missing information.

[0079] The information of the means acquired by the information acquisition means determination process is sent to the related information acquisition unit 144 and the user information acquisition unit 145, respectively. The related information acquisition unit 144 executes the related information acquisition process if it is determined by the information acquisition means determination process that the means is capable of acquiring information to be used in the append information determination process. The user information acquisition unit 145 also executes user information acquisition if it is determined by the information acquisition means determination process that the means is capable of acquiring information to be used in the append information determination process.

[0080] 2-2. Specific Processing Diagram 17 of the Inquiry Response System According to Embodiment 2 is a flowchart of the processing performed in the inquiry response system according to Embodiment 2. The processing in steps S200 to S204 of the flowchart shown in Figure 17 is performed by the processor 360 of the user client 300 executing an information processing program stored in the auxiliary storage device 330. Furthermore, the processing in steps S210 to S222 of the flowchart shown in Figure 17 is performed by the processor 140 of the answer generation server 100 executing an information processing program stored in the auxiliary storage device 130. Note that the flowchart shown in Figure 17 also represents a part of the inquiry response method in the inquiry response system 10 that outputs an answer to the user's inquiry to the user.

[0081] When a user inputs a query in natural language from the input device 380, the user client 300 executes a query acquisition process in step S200 to obtain query information. The acquired query information is sent to the answer generation server 100.

[0082] Upon receiving the inquiry information, the response generation server 100 performs a topic determination process in step S210 to obtain the topic of the inquiry information. Note that the topic determination process may utilize the artificial intelligence function of the LLM server 400 to generate a topic corresponding to the inquiry information. Once the process in step S210 is complete, the process proceeds to step S212.

[0083] In step S212, the response generation server 100 performs a missing information determination process to obtain missing information from the query information. In the missing information determination process, the response generation server 100 sets the importance level of each piece of missing information based on the topics obtained in step S210. In the missing information determination process, the artificial intelligence function of the LLM server 400 may be used to generate missing information corresponding to the query information. Once the processing in step S212 is completed, the process proceeds to step S214.

[0084] In step S214, the response generation server 100 executes an information acquisition means determination process to determine the means for acquiring the information to be used in the additional information determination process. Once the process in step S214 is completed, the process proceeds to step S216.

[0085] In step S216, the response generation server 100 executes a process to acquire information to be used in the additional information determination process using the means determined in the process of step S214. For example, if the user information acquisition unit 145 is determined in the information acquisition means determination process, the user information acquisition process is executed to acquire user information as information to be used in the additional information determination process. Also, for example, if the related information acquisition unit 144 is determined in the information acquisition means determination process, the related information acquisition process is executed to acquire related information as information to be used in the additional information determination process. Once the process of step S216 is completed, the process proceeds to step S218.

[0086] In step S218, the response generation server 100 executes an additional information determination process to generate additional information. In the additional information determination process, the artificial intelligence function of the LLM server 400 may be used to generate additional information corresponding to the inquiry information, missing information, user information, or related information. Once the process in step S218 is completed, the process proceeds to step S220.

[0087] In steps S220, S222, S202, S204, S224, and S226, the same processes as steps S114, S116, S118, S102, S104, S120, and S122 of the flowchart shown in Figure 10 are executed, respectively.

[0088] As described above, according to the inquiry response system 10 of this embodiment, when there is missing information in the inquiry information, the means for acquiring the information necessary to generate supplementary information to supplement the missing information is determined. This makes it possible to improve the accuracy of the supplementary information acquired in the supplementary information determination process.

[0089] 2-3. Modified Embodiment The inquiry response system 10 of the second embodiment may adopt the following modified forms.

[0090] The related information acquisition process in the related information acquisition unit 144 may be configured to be performed independently of the determination result of the information acquisition means determination process.

[0091] 3. Embodiment 3. In Embodiment 3, the differences from the example disclosed in Embodiment 1 will be explained in particular detail. For features not described in Embodiment 3, any of the features from the example disclosed in Embodiment 1 may be adopted.

[0092] 3-1. Functional Configuration of the Inquiry Response System in Embodiment 3 The functional configuration of the inquiry response system 10 according to Embodiment 3 will be described. Figure 18 is a block diagram showing an excerpt of some of the information processing functions of the inquiry response system according to Embodiment 3. The response generation server 100 of the inquiry response system 10 is realized by the processor 140 reading and executing an information processing program and, in addition to the functions shown in Figure 4, further includes a data search unit 160, a necessary information extraction unit 162, and a user information acquisition unit 145.

[0093] The data retrieval unit 160 is a functional block that executes the process of searching for data related to the inquiry information. This process is hereinafter referred to as the "data retrieval process," and the information obtained by the data retrieval process is hereinafter referred to as the "data retrieval information." In the data retrieval process, the data retrieval unit 160 calculates the similarity between the inquiry information and the data in the data database 133, and extracts data with a calculated similarity higher than the judgment value as data retrieval information. For example, cosine similarity can be used as the similarity measure.

[0094] Furthermore, in the document search process, the document search unit 160 may calculate the similarity between the additional inquiry information and the documents in the document database 133. Also, in the document search process, the document search unit 160 may use the artificial intelligence function of the LLM server 400 to pre-narrow down the search target documents from the documents in the document database 133 based on information such as topics. Furthermore, if the document database 133 is composed of multiple classified databases, the document search unit 160 may use the artificial intelligence function of the LLM server 400 to narrow down the database to be searched. The acquired document information is sent to the necessary information extraction unit 162.

[0095] The Similar Answer Search Unit 161 is a functional block that performs the process of searching for answers and related materials associated with inquiries similar to the inquiry information from the answer history information. This process is hereinafter referred to as the "Similar Answer Search Process," and the information obtained by the Similar Answer Search Process is hereinafter referred to as "Similar Answer Information." Figure 19 is a diagram illustrating an example of the Similar Answer Search Process. In the example of the Similar Answer Search Process shown in this figure, the Similar Answer Search Unit 161 calculates the similarity between the acquired inquiry information and the inquiries in the past answer history information stored in the answer history database 134, and extracts answers and related materials associated with inquiries whose calculated similarity is higher than the judgment value as similar answer information. For example, cosine similarity can be used as the similarity. In addition, in the Similar Answer Search Process, the Similar Answer Search Unit 161 may also calculate the similarity between the additional inquiry information and the inquiries in the past answer history information stored in the answer history database 134. The acquired similar answer information is sent to the Necessary Information Extraction Unit 162.

[0096] The necessary information extraction unit 162 is a functional block that performs the process of extracting necessary information required to generate an answer from the data search information and similar answer information. This process is hereinafter referred to as the "necessary information extraction process". Figure 20 is a diagram illustrating an example of the necessary information extraction process. In the example of the necessary information extraction process shown in this figure, the necessary information extraction unit 162 uses the artificial intelligence function of the LLM server 400 to acquire the necessary information corresponding to the additional inquiry information, data search information, and similar answer information. For example, LLM 402 of the LLM server 400 contains information of a model that has been learned by linking the additional inquiry information, data search information, and similar answer information with the necessary information. The necessary information extraction unit 162 inputs the acquired additional inquiry information, data search information, and similar answer information into the LLM server 400 to acquire the necessary information corresponding to the additional inquiry information, data search information, and similar answer information from the LLM server 400. The necessary information may also be summarized using the artificial intelligence function of the LLM server 400 or similar. Figure 20 illustrates the necessary information summarized using the artificial intelligence function. The acquired necessary information is sent to the response generation unit 150 and used for the response generation process.

[0097] 3-2. Specific Processing Diagram 21 of the Inquiry Response System According to Embodiment 3 is a flowchart of the processing performed in the inquiry response system according to Embodiment 3. The processing in steps S300 to S304 of the flowchart shown in Figure 21 is performed by the processor 360 of the user client 300 executing an information processing program stored in the auxiliary storage device 330. The processing in steps S310 to S328 of the flowchart shown in Figure 21 is performed by the processor 140 of the answer generation server 100 executing an information processing program stored in the auxiliary storage device 130. Note that the flowchart shown in Figure 21 also represents a part of the inquiry response method in the inquiry response system 10 that outputs an answer to the user's inquiry to the user.

[0098] When a user inputs a query in natural language from the input device 380, the user client 300 executes a query acquisition process in step S300 to obtain query information. The acquired query information is sent to the answer generation server 100.

[0099] Upon receiving the inquiry information, the response generation server 100 performs the same processing as steps S210 to S216 in the flowchart shown in Figure 10 during steps S310 to S316. Once the processing in step S316 is completed, the process proceeds to steps S318 and S320.

[0100] In step S318, the answer generation server 100 performs a data search process to obtain data search information. In step S320, the answer generation server 100 performs a similar answer search process to obtain similar answer information. Once steps S318 and S320 are completed, the process proceeds to step S322.

[0101] In step S322, the response generation server 100 performs a necessary information extraction process to obtain the necessary information required to generate a response. In the necessary information extraction process, the artificial intelligence function of the LLM server 400 is used to generate the necessary information corresponding to the additional inquiry information, document search information, and similar response information. Once the process in step S322 is completed, the process proceeds to step S324.

[0102] In step S324, the answer generation server 100 executes the answer generation process to obtain an answer to the user's inquiry. The LLM 402 of the LLM server 400 contains information of a model that has been learned by associating additional inquiry information and necessary information with answers. Here, the answer generation server 100 uses the artificial intelligence function of the LLM server 400 to obtain an answer corresponding to the additional inquiry information and necessary information. The obtained answer is sent to the user client 300. In the processes of steps S302 and S304, the user client 300 executes the same processes as steps S102 and S104 of the flowchart shown in Figure 10.

[0103] As described above, according to the inquiry response system 10 of this embodiment, the necessary information required to generate an answer to an inquiry is generated using information obtained from the data database 133 and the past answer history database 135. As a result, the generated necessary information is used in the answer generation process, which improves the accuracy of the generated answer.

[0104] 10 Answer System, 100 Answer Generation Server, 110 Main Memory, 120 Communication I / F, 130 Auxiliary Storage, 131 Equipment Information Database, 132 Building Information Database, 133 Document Database, 134 Past Answer History Database, 135 User-Specific Answer History Database, 140 Processor, 141 Topic Determination Unit, 142 Missing Information Determination Unit, 143 Information Acquisition Means Determination Unit, 144 Related Information Acquisition Unit, 145 User Information Acquisition Unit, 146 Additional Information Determination Unit, 147 Information Addition Unit, 150 Answer Generation Unit, 151 Answer History Generation Unit, 152 Specific History Extraction Unit, 153 History Registration Unit, 160 Document Search Unit, 161 Similar Answer Search Unit, 162 Necessary Information Extraction Unit, 170 Dedicated Hardware, 171 200 Processing circuit, 300 Communication network, 310 User client, 320 Communication I / F, 330 Main memory, 340 Auxiliary memory, 350 Output I / F, 360 Input I / F, 360 Processor, 361 Query acquisition unit, 362 Response output unit, 363 Evaluation information acquisition unit, 370 Output device, 380 Input device, 390 Dedicated hardware, 391 Processing circuit, 400 LLM server, 402 LLM control unit

Claims

1. An inquiry response system comprising: an inquiry acquisition unit that acquires natural language data of inquiry information including a user's inquiry; a missing information determination unit that determines missing information necessary to generate an answer to the inquiry based on the inquiry information; an additional information determination unit that determines additional information corresponding to the missing information based on the inquiry information; an information addition unit that generates additional inquiry information by adding the additional information to the inquiry information; and an answer generation unit that inputs the additional inquiry information to an artificial intelligence unit and acquires and outputs an answer corresponding to the additional inquiry information from the artificial intelligence unit.

2. The inquiry response system according to claim 1, further comprising a user information acquisition unit that searches for past information of the user related to the inquiry information from a database in which response history information related to past inquiries and answers associated with the user is registered and acquires it as user information, wherein the additional information determination unit is configured to determine the additional information based on the user information.

3. The inquiry response system according to claim 2, wherein the additional information determination unit is configured to input the user information and the missing information into the artificial intelligence unit and to obtain the additional information corresponding to the user information and the missing information from the artificial intelligence unit.

4. The inquiry response system according to any one of claims 1 to 3, further comprising a related information acquisition unit that acquires related information relating to the equipment included in the inquiry from a database in which equipment information is registered, based on the inquiry information, wherein the additional information determination unit is configured to determine the additional information based on the related information.

5. The inquiry response system according to any one of claims 1 to 4, further comprising a response output unit that outputs the response output from the response generation unit to the user.

6. The inquiry response system according to claim 5, wherein the response output unit is configured to output the additional information in association with the response.

7. The inquiry response system according to claim 2, further comprising: a history registration unit that registers the response history information, including the inquiry information obtained by the inquiry acquisition unit and the response obtained by the response generation unit, in the database, linked to the user.

8. The inquiry response system according to claim 7, comprising an evaluation information acquisition unit that acquires evaluation information including the user's evaluation of the response, wherein the history registration unit is configured to register the response corresponding to the evaluation information in the response history information when the evaluation information includes an evaluation higher than the standard evaluation, and not to register the response corresponding to the evaluation information in the response history information when the evaluation information includes an evaluation lower than the standard evaluation.

9. The inquiry response system according to claim 1, comprising an information acquisition means determination unit that determines means for acquiring information to be used to determine the supplementary information based on the missing information, wherein the supplementary information determination unit is configured to determine the supplementary information based on the information acquired by the processing determined by the information acquisition means determination unit.

10. The inquiry response system according to claim 9, further comprising a user information acquisition unit that searches for past information of the user related to the inquiry information from a database in which response history information related to past inquiries and answers associated with the user is registered and acquires it as user information, wherein the additional information determination unit is configured to determine the additional information based on the user information when the information acquisition means determination unit determines that processing in the user information acquisition unit is required.

11. The inquiry response system according to claim 9 or 10, comprising a related information acquisition unit that performs processing to acquire related information relating to the equipment included in the inquiry from a database in which equipment information is registered, wherein the additional information determination unit is configured to determine the additional information based on the related information when the information acquisition means determination unit determines that processing in the related information acquisition unit is performed.

12. An inquiry response system according to any one of claims 1 to 11, comprising: a document search unit that searches a database for document search information including information on documents related to the inquiry; a similar answer search unit that searches a database for similar answer information including information on similar answers related to the inquiry; and a necessary information extraction unit that extracts necessary information for generating an answer to the inquiry from the document search information and the similar answer information, wherein the answer generation unit is configured to input the additional inquiry information and the necessary information into the artificial intelligence unit, and to obtain and output an answer corresponding to the additional inquiry information and the necessary information from the artificial intelligence unit.

13. An inquiry response system according to any one of claims 1 to 12, comprising a topic determination unit for determining the topic of the inquiry based on the inquiry information, wherein the missing information determination unit is configured to determine the missing information based on the topic determined by the topic determination unit and the inquiry information.

14. The inquiry response system according to any one of claims 1 to 13, wherein the inquiry includes an inquiry relating to the maintenance of equipment installed within the facility.

15. An information processing device comprising: an information deficiency determination unit that acquires natural language data of inquiry information including a user's inquiry and determines missing information that is insufficient to generate an answer to the inquiry based on the inquiry information; an information addition determination unit that determines additional information corresponding to the missing information based on the inquiry information; an information addition unit that generates additional inquiry information by adding the additional information to the inquiry information; and an answer generation unit that inputs the additional inquiry information to an artificial intelligence unit and obtains and outputs an answer corresponding to the additional inquiry information from the artificial intelligence unit.

16. A query response method performed by a computer, comprising: a step of obtaining natural language data of query information including a user's query; a step of determining missing information that is insufficient to generate a response to the query based on the query information; a step of determining additional information corresponding to the missing information based on the query information; a step of generating additional query information by adding the additional information to the query information; and a step of inputting the additional query information into an artificial intelligence unit and obtaining and outputting a response corresponding to the additional query information from the artificial intelligence unit.

17. A program that causes a computer to output an answer to a user's inquiry, the program to: acquire natural language data of inquiry information including the user's inquiry; determine missing information necessary to generate an answer to the inquiry based on the inquiry information; determine additional information corresponding to the missing information based on the inquiry information; generate additional inquiry information by adding the additional information to the inquiry information; input the additional inquiry information into an artificial intelligence unit; and cause the computer to acquire and output an answer corresponding to the additional inquiry information from the artificial intelligence unit.