Server device, control method for server device, and program

Abstract

This system provides a server device that appropriately responds to inquiries from users performing specified tasks. [Solution] The server device comprises an acquisition means and an answer control means. The acquisition means acquires inquiry items from the user. The answer control means inputs a prompt generated using the acquired inquiry items into a first learning model, which is generated using at least a manual for a predetermined system and a collection of know-how for a predetermined business as learning data. By inputting the prompt into the first learning model, the answer control means acquires an answer to the acquired inquiry items and provides the acquired answer to the user.

Description

Technical Field

[0001] The present invention relates to a server device, a control method for the server device, and a program.

Background Art

[0002] There are response generation devices that utilize large language models and the like.

[0003] For example, Patent Document 1 describes that a response text is output such that the expression of the response text is easy for the user to understand and the content conforms to the purpose and context of the dialogue. The response generation device of Patent Document 1 includes a response generation unit, a response correction unit, a response expression appropriateness evaluation unit, a response content validity evaluation unit, and a selection unit. The response generation unit generates a response text for the input text. The response correction unit corrects the response text generated by the response generation unit. The response expression appropriateness evaluation unit determines whether the expression of the response text is appropriate for the user among the response texts in the group of response texts corrected by the response correction unit. The response content validity evaluation unit determines whether the content of the response text is appropriate for the user's past interaction content among the response texts in the group of response texts corrected by the response correction unit. The selection unit determines the response text to be output from the group of corrected response texts based on the evaluation by the response expression appropriateness evaluation unit and the evaluation by the response content validity evaluation unit.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Large-scale language models are sometimes used to build question-answering systems for systems and equipment operation, with the aim of supporting employees' work. The content of the responses output by the large-scale language model depends on the training data used to generate it. For example, a large-scale language model generated based on a system or equipment operation manual can answer questions about the operation of that system or equipment, but it may not be able to provide appropriate answers to questions about other matters. However, supporting employees' work requires that appropriate answers be provided for matters other than system and equipment operation.

[0006] Furthermore, Patent Document 1 merely discloses a technology for outputting response text that is in line with the purpose and context of the dialogue. Therefore, applying the technology disclosed in Patent Document 1 will not satisfy the above requirements.

[0007] The primary objective of this invention is to provide a server device, a control method for the server device, and a program that contribute to appropriately responding to inquiries from users performing predetermined tasks. [Means for solving the problem]

[0008] According to a first aspect of the present invention, a server device is provided that includes an acquisition means for acquiring inquiries from a user, and an answer control means for acquiring answers to the acquired inquiries and providing the acquired answers to the user by inputting prompts generated using the acquired inquiries into a first learning model generated using at least a manual for a predetermined system and a collection of know-how for a predetermined operation as learning data.

[0009] A second aspect of the present invention provides a server device control method comprising: an acquisition step of acquiring inquiry items from a user; and an answer control step of inputting a prompt generated using the acquired inquiry items into a first learning model generated using at least a manual for a predetermined system and a collection of know-how for a predetermined business as learning data, thereby acquiring an answer to the acquired inquiry items and providing the acquired answer to the user.

[0010] According to a third aspect of the present invention, a program is provided for a computer mounted on a server device to perform an acquisition process for acquiring inquiries from a user, and an answer control process for acquiring answers to the acquired inquiries and providing the acquired answers to the user by inputting prompts generated using the acquired inquiries into a first learning model generated using at least a manual for a predetermined system and a collection of know-how for a predetermined business as learning data. [Effects of the Invention]

[0011] According to each aspect of the present invention, a server device, a control method for the server device, and a program are provided that contribute to appropriately responding to inquiries from users performing predetermined tasks. However, the effects of the present invention are not limited to those described above. The present invention may produce other effects in lieu of or in conjunction with the effects described above. [Brief explanation of the drawing]

[0012] [Figure 1] Figure 1 is a diagram illustrating the outline of one embodiment. [Figure 2] Figure 2 is a flowchart illustrating the operation overview of one embodiment. [Figure 3] Figure 3 shows an example of a schematic configuration of an information processing system according to the embodiment of this disclosure. [Figure 4] Figure 4 is a diagram illustrating the operation of an information processing system according to an embodiment of this disclosure. [Figure 5] Figure 5 shows an example of the display on a hotel terminal according to the embodiment of this disclosure. [Figure 6] Figure 6 shows an example of the processing configuration of a server device according to the present disclosure. [Figure 7] Figure 7 is a flowchart showing an example of the operation of the inquiry response control unit according to the present disclosure. [Figure 8] Figure 8 is a sequence diagram showing an example of the operation of an information processing system according to the embodiment of this disclosure. [Figure 9] Figure 9 is a diagram illustrating the operation of an information processing system according to a modified embodiment of the present disclosure. [Figure 10] Figure 10 shows an example of the processing configuration of a server device according to a modified embodiment of the present disclosure. [Figure 11] Figure 11 shows an example of the hardware configuration of the server device related to this disclosure. [Modes for carrying out the invention]

[0013] First, an overview of one embodiment will be described. The reference numerals in the drawings attached to this overview are provided for convenience as examples to aid understanding, and this overview is not intended to be limiting in any way. Furthermore, unless otherwise specified, the blocks shown in each drawing represent functional units, not hardware units. The connecting lines between blocks in each drawing include both bidirectional and unidirectional lines. Unidirectional arrows schematically indicate the flow of the main signal (data) and do not exclude bidirectional flow. In this specification and in the drawings, elements that can be similarly described are given the same reference numerals to avoid redundant explanation.

[0014] The server device 100 according to an embodiment includes an acquisition unit 101 and a response control unit 102 (see FIG. 1). The acquisition unit 101 acquires an inquiry item from a user (step S1 in FIG. 2). The response control unit 102 inputs a prompt generated using the acquired inquiry item into a first learning model generated using at least a manual related to a predetermined system and a collection of know-how related to a predetermined operation as learning data (step S2). By inputting the prompt into the first learning model, the response control unit 102 acquires a response to the acquired inquiry item and provides the acquired response to the user (step S3).

[0015] The server device 100 uses, as learning data, not only the manual of the system but also the know-how of the operation to generate a learning model for answering user inquiries. As a result, more useful information (responses) can be provided to users who have business troubles or the like. That is, the server device 100 can provide an appropriate response to an inquiry from a user who performs a predetermined operation.

[0016] Specific embodiments will be described in more detail below with reference to the drawings.

[0017] [First Embodiment] The first embodiment will be described in more detail with reference to the drawings.

[0018] The information processing system according to the first embodiment realizes an inquiry response system that supports employees working in a predetermined industry or business. In the first embodiment, an inquiry response system for employees working in a hotel or the like will be described.

[0019] [System Configuration] FIG. 3 is a diagram showing an example of the schematic configuration of an information processing system according to an embodiment of the present disclosure. As shown in FIG. 3, the information processing system includes a plurality of hotels A to C. The same operator operates hotels A to C. Alternatively, hotels A to C may be hotels operated by partnering operators.

[0020] The information processing system includes a server device 10 and multiple hotel terminals 20.

[0021] Server device 10 performs the main operations of the hotel. For example, server device 10 stores reservation information for guests of each hotel. Server device 10 uses the reservation information to perform check-in and check-out procedures for guests. Hotels A to C each use a common hotel system for guest reservation management and other purposes.

[0022] Furthermore, the server device 10 is equipped with functions to support hotel employees. Specifically, the server device 10 is equipped with functions to answer questions and requests from hotel employees regarding hotel operations and front desk operations.

[0023] The hotel terminal 20 is installed at the front desk or other locations in each hotel. The hotel terminal 20 may be, for example, a personal computer or a tablet device. Alternatively, the hotel terminal 20 may be a mobile device such as a smartphone. Each hotel employee may also possess a hotel terminal 20.

[0024] The hotel terminal 20 is a terminal that serves as an interface for hotel staff who require support. Hotel staff who need to deal with problems or other issues input inquiries such as questions and requests into the hotel terminal 20. The hotel staff obtain answers to their questions from the server device 10 via the hotel terminal 20.

[0025] The devices shown in Figure 3 are interconnected. Specifically, the server device 10 and the hotel terminal 20 are connected by wired or wireless communication means and are configured to communicate with each other.

[0026] The configuration of the information processing system shown in Figure 3 is illustrative and not intended to limit its configuration. For example, the system may include multiple server devices 10. Load balancing and redundancy may be achieved by the multiple server devices 10. Alternatively, the information processing system may include one hotel. Alternatively, each hotel may have multiple hotel terminals 20 installed.

[0027] [General operation] Next, we will describe the general operation of the information processing system according to the first embodiment.

[0028] <Support from hotel staff> Hotel employees perform their duties at the hotel where they work. In the course of performing their duties, hotel employees may have questions regarding the operation of hotel systems. For example, a hotel employee might want to know how to display a list of hotel reservations.

[0029] For example, a hotel system may include a reservation system (managing online reservations, telephone reservations, reservations from travel agencies, etc.), a lodging management system (check-in / check-out, room allocation, guest information management, etc.), a payment system (credit card payments, invoice issuance, billing, etc.), a customer relationship management system (managing customer information, operating loyalty programs, etc.), a restaurant reservation and management system (managing restaurant reservations, table allocation, order management, etc.), an event and conference management system (reserving banquet halls and conference rooms, facility management, etc.), a human resources management system (managing employee shifts, payroll, etc.), an inventory management system (managing inventory of ingredients, amenities, consumables, etc.), a concierge service system (providing tourist information, reservation assistance services, etc.), and a laundry management system (managing hotel linens and guest clothing cleaning).

[0030] Alternatively, hotel staff may be asked by guests to resolve problems related to their rooms. Hotel staff may also want to know how to resolve such requests (problems with rooms, etc.). Examples of room-related problems include equipment malfunctions (malfunctions or defects in in-room equipment such as air conditioners, televisions, and showers), noise problems (complaints from guests about noise from adjacent rooms or corridors), and loss or theft of valuables.

[0031] Regarding the operation of the hotel system and troubleshooting issues in guest rooms, hotel staff enter their inquiries into the hotel terminal 20 (see Figure 4). For example, hotel staff enter questions or other inquiries into the hotel terminal 20. For example, hotel staff may enter questions or other inquiries into the hotel terminal 20 by typing text. For example, hotel staff may enter questions or other inquiries into the hotel terminal 20 by inputting voice, images, or videos.

[0032] The hotel terminal 20 notifies the server device 10 of the acquired inquiry details. Specifically, the hotel terminal 20 sends an "inquiry response request" containing the inquiry details and the hotel ID to the server device 10.

[0033] For example, the "Inquiry Response Request" may include the employee ID of the hotel employee who made the inquiry. Alternatively, the employee ID of the hotel employee who made the inquiry may be included in the "Inquiry Response Request" by identifying the hotel employee who made the inquiry through facial recognition. In other words, only hotel employees who can be authenticated by facial recognition may be allowed to make inquiries.

[0034] The hotel ID is used to identify the hotel where the hotel terminal 20 is installed. The hotel ID can be the hotel's name or a predetermined code. Alternatively, the MAC (Media Access Control) address or IP (Internet Protocol) address of the hotel terminal 20 can be used as the hotel ID.

[0035] The hotel ID is shared between the server device 10 and the hotel terminal 20 by any method. For example, the system administrator determines the hotel ID and sets the determined hotel ID on the server device 10. The system administrator also sets the determined hotel ID on the hotel terminal 20.

[0036] Server device 10 processes inquiry response requests received from hotel terminal 20. Server device 10 processes inquiry response requests using a learning model obtained through machine learning.

[0037] The server device 10 identifies the hotel where the source of the inquiry response request (the hotel terminal 20 that sent the inquiry response request) is located, based on the hotel ID included in the inquiry response request.

[0038] The server device 10 generates a prompt that clearly indicates the inquiry is from the identified hotel and instructs the server to respond to the inquiry. The server device 10 inputs the generated prompt into a learning model to obtain the response to the inquiry from the hotel employee.

[0039] Upon obtaining a response, the server device 10 sends an affirmative response (a response to the inquiry response request) containing the obtained response to the hotel terminal 20.

[0040] Upon receiving an affirmative response, the hotel terminal 20 presents the answer contained in that affirmative response to the hotel employee. The hotel terminal 20 outputs the answer as audio or displays it on a screen. For example, the hotel terminal 20 may output or display the answer in multiple formats, such as text, images, audio, and video.

[0041] For example, as shown in Figure 5, the hotel terminal 20 displays the inquiry and the answer to that inquiry on a display or the like.

[0042] <Generating a Learning Model> Here, system administrators, for example, generate a learning model using hotel system manuals and collections of know-how accumulated at each hotel (such as know-how collections related to front desk operations and case reports) as training data.

[0043] For example, a system administrator generates a Large Language Model (LLM). The system administrator generates a learning model that has learned how to use the hotel system and how to resolve problems in guest rooms (know-how). The generated learning model is implemented on the server device 10.

[0044] Next, we will describe the details of each device included in the information processing system according to the first embodiment.

[0045] [Server equipment] Figure 6 shows an example of the processing configuration (processing module) of the server device 10 according to the embodiment disclosed herein. Referring to Figure 6, the server device 10 comprises a communication control unit 201, a learning model management unit 202, an inquiry response control unit 203, and a storage unit 204.

[0046] The communication control unit 201 is a means for controlling communication with other devices. For example, the communication control unit 201 receives data (packets) from the hotel terminal 20. The communication control unit 201 also transmits data to the hotel terminal 20. The communication control unit 201 passes the data received from other devices to other processing modules. The communication control unit 201 transmits the data obtained from other processing modules to other devices. In this way, other processing modules send and receive data with other devices via the communication control unit 201. The communication control unit 201 has the function of a receiving unit that receives data from other devices and the function of a transmitting unit that transmits data to other devices.

[0047] The learning model management unit 202 is a means for controlling and managing the learning model.

[0048] The learning model management unit 202 acquires the learning model prepared by the system administrator or other relevant parties. For example, the learning model management unit 202 acquires the learning model using a GUI. Alternatively, the learning model management unit 202 may acquire the learning model via a USB (Universal Serial Bus) memory or the like.

[0049] Next, we will provide an overview of the learning model and how to generate it.

[0050] A learning model is a model that generates responses to requests. For example, when a learning model receives a prompt (query) generated based on a request from a user, it outputs a response corresponding to that prompt.

[0051] For example, a learning model is composed of a language model. This language model may, but is not limited to, what is called an LLM (Large Language Model).

[0052] A language model is a machine learning model (also called a generative model) that takes language as input and outputs language. A language model learns the relationships between words in a text and generates related strings from a given string. By using a language model trained on sentences and texts in various contexts, it is possible to generate related strings with appropriate content related to the given string.

[0053] For example, let's consider the case where a language model is used in question answering. The language model accepts the question "What kind of country is Japan?" as input. The language model generates a string such as "Japan is an island nation in the Northern Hemisphere" as an answer to the question.

[0054] The method for training a language model is not particularly limited, but one example is one that is trained to output at least one sentence containing the input string. Specifically, the language model is a Generative Pre-trained Transformer (GPT) that outputs a sentence containing the input string by predicting the string that is most likely to follow the input string.

[0055] Other examples of language models include T5 (Text-to-Text Transfer Transformer), BERT (Bidirectional Encoder Representations from Transformers), RoBERTa (Robustly optimized BERT approach), and ELECTRA (Efficiently Learning an Encoder that Classifies Token Replacements Accurately).

[0056] The content generated by a language model is not limited to strings. For example, a language model may generate image data, video data, audio data, or other data formats corresponding to the input string.

[0057] The learning model (language model) is generated based on the training data. Specifically, a large-scale language model is generated using hotel system manuals, collections of know-how accumulated by each hotel, etc. In other words, the learning model implemented in the server device 10 is generated using information on how to use the hotel system and the unique know-how of each hotel (for example, solutions to problems in each guest room) as training data.

[0058] Alternatively, a learning model may be generated by utilizing an existing learning model (language model). For example, transfer learning (fine-tuning) may be performed, in which the weights of a previously generated learning model are trained with new training data. Specifically, an existing language model may be used, and additional training may be performed with unique training data (datasets) to give the learning model distinctive features. For example, a learning model that answers inquiries from hotel employees may be generated by preparing hotel system manuals and hotel-specific know-how collections as training data, and then adding this training data to a basic learning model.

[0059] Furthermore, the learning model may be updated in conjunction with the update of the learning data. For example, the update of the learning data (learning model) may be performed after a predetermined period has elapsed since the update of the learning data (learning model). Know-how and other information accumulated during that predetermined period may be reflected in the learning model. Furthermore, such updates of the learning data and learning model may be performed automatically.

[0060] The inquiry response control unit 203 is a means for controlling inquiries received from hotel employees. The inquiry response control unit 203 processes inquiry response requests received from the hotel terminal 20.

[0061] The inquiry response control unit 203 has functions as an acquisition means and functions as an answer control means. As an acquisition means, the inquiry response control unit 203 acquires inquiry items from the user. As an answer control means, the inquiry response control unit 203 inputs a prompt generated using the acquired inquiry items into a first learning model, which is generated using at least a manual for a predetermined system and a collection of know-how for a predetermined operation as learning data. By inputting the prompt into the learning model, the inquiry response control unit 203 acquires an answer to the acquired inquiry items and provides the acquired answer to the user (hotel employee).

[0062] The first learning model is a learning model generated using a hotel system manual and a collection of hotel operation know-how as training data. The inquiry response control unit 203 uses, for example, a large-scale language model as a learning model to obtain answers to inquiries received from users.

[0063] The operation of the inquiry response control unit 203 will be explained with reference to Figure 7.

[0064] Upon receiving an inquiry response request, the inquiry response control unit 203 identifies the hotel that made the inquiry (the hotel where the hotel terminal 20 is installed) based on the hotel ID included in the request (step S101).

[0065] For example, the inquiry response control unit 203 refers to table information that stores hotel IDs and hotel names in association with each other to identify the hotel that made the inquiry.

[0066] The inquiry response control unit 203 generates a prompt that includes information about the identified hotel and instructs the creation of an answer to the inquiry items included in the inquiry response request (generate prompt; step S102).

[0067] For example, the inquiry response control unit 203 generates a prompt such as, "We have received an inquiry from Hotel A. Please respond to the inquiry item, 'Please tell me how to display a list of booked guests.'" This prompt is generated by setting the hotel name and inquiry item in a pre-prepared template. Alternatively, the inquiry response control unit 203 may generate prompts based on inquiries from hotel employees. Specifically, the inquiry response control unit 203 may generate prompts that reflect the inquiries of hotel employees after engaging in multiple exchanges (such as chat) with them.

[0068] Alternatively, the inquiry response control unit 203 generates a prompt such as, "We have received an inquiry from Hotel A. Please respond to the inquiry item: 'The water pipe in guest room 202 is clogged. Please tell us how to resolve this.'"

[0069] The inquiry response control unit 203 inputs the generated prompt into a learning model (large-scale language model) to obtain an answer to the inquiry from the hotel employee (step S103).

[0070] Furthermore, the learning model will output the same answer to inquiries regarding the use of the hotel system, regardless of the hotel making the inquiry. In contrast, for inquiries concerning matters specific to the hotel making the inquiry, the learning model will output an answer tailored to that hotel.

[0071] More specifically, for inquiries regarding the use of the hotel system, the system outputs answers based on the hotel system manual used as training data. In contrast, for hotel-specific problems, the system outputs answers based on the know-how collection of each hotel.

[0072] The inquiry response control unit 203 transmits the acquired response to the hotel terminal 20. Specifically, the inquiry response control unit 203 transmits a response to the inquiry response request to the hotel terminal 20 (step S104).

[0073] More specifically, the inquiry response control unit 203 sends an affirmative response to the hotel terminal 20 that includes the answer obtained from the learning model. If an appropriate answer cannot be obtained from the learning model (for example, if the learning model provides an answer such as "cannot answer"), the inquiry response control unit 203 sends a negative response to the hotel terminal 20 indicating that it cannot answer the inquiry.

[0074] For example, if the learning model cannot provide an appropriate answer, the inquiry response control unit 203 may provide answers to similar questions or related information. The inquiry response control unit 203 may also ask the hotel employee who made the inquiry additional questions. Furthermore, the inquiry response control unit 203 may transfer the inquiry to a human operator. Finally, the inquiry response control unit 203 may introduce employees or departments that may be able to provide an answer.

[0075] The memory unit 204 is a means for storing information necessary for the operation of the server device 10.

[0076] [Hotel terminal] Examples of the hotel terminal 20 include mobile devices such as smartphones, mobile phones, game consoles, and tablets, as well as computers (personal computers, laptops). The hotel terminal 20 can be any device or equipment as long as it can receive user input and communicate with the server device 10, etc. Furthermore, the configuration of the hotel terminal 20 is obvious to those skilled in the art, so a detailed explanation is omitted. The hotel terminal 20 has a dedicated application installed for sending inquiries to the server device 10 and displaying answers.

[0077] [System operation] Next, the operation of the information processing system according to the first embodiment will be described.

[0078] Refer to Figure 8 to explain how the information processing system works when responses from a large-scale language model are provided to hotel employees.

[0079] The hotel terminal 20 sends an inquiry response request, including the inquiry details and the hotel ID, to the server device 10 (step S01).

[0080] The server device 10 generates a response to the query using a pre-prepared large-scale language model (step S02).

[0081] The server device 10 sends an affirmative response containing the generated answer to the hotel terminal 20 (step S03).

[0082] The hotel terminal 20 presents the response included in the received affirmative response to the hotel employee (step S04).

[0083] Next, a modified example of the first embodiment will be described.

[0084] <Example 1> In the above embodiment, a case was described in which a learning model (large-scale language model) is generated based on the hotel system manual and a collection of know-how for each hotel. However, the learning model may be generated using other information as training data instead of, or in addition to, the hotel system manual and know-how collection.

[0085] For example, a customer service manual common to all hotels may be used as training data to generate a learning model. Alternatively, an emergency response manual may be used as training data to generate a learning model.

[0086] Thus, a learning model can be generated using manuals and documents covering all aspects of hotel operations as training data. As a result, the learning model (large-scale language model) can provide more useful information for the work of hotel employees.

[0087] <Modification 2> The server device 10 may have multiple learning models (large-scale language models) implemented to answer inquiries from hotel employees. For example, the server device 10 may have a learning model (system) for answering inquiries about hotel systems common to all hotels, and a learning model for answering inquiries specific to each hotel.

[0088] A learning model for answering inquiries specific to hotels may be a single learning model generated using a collection of know-how from each hotel as training data.

[0089] Alternatively, a learning model may be generated for each hotel. For example, a learning model (Hotel A) generated using the know-how of Hotel A as training data, a learning model (Hotel B) generated using the know-how of Hotel B as training data, etc., may be implemented on the server device 10.

[0090] When the server device 10 receives an inquiry response request, it may select a learning model to obtain an answer based on the inquiry items and hotel ID included in the request. In the example above, if the inquiry concerns the hotel system, the inquiry response control unit 203 selects a learning model (system) generated based on the hotel system manual. Alternatively, if the inquiry concerns the operations (know-how) of hotel A, the inquiry response control unit 203 selects a learning model (hotel A).

[0091] The inquiry response control unit 203 inputs a prompt to the selected learning model (large-scale language model) and obtains an answer to the inquiry from the hotel employee.

[0092] Thus, the inquiry response control unit 203 may obtain answers to inquiries received from hotel employees by inputting prompts generated using the acquired inquiry items into the second or third learning model instead of the first learning model. The second learning model is a learning model generated using a manual for a predetermined system as learning data. The third learning model is a learning model generated using a collection of know-how related to a predetermined business as learning data. Furthermore, the inquiry response control unit 203 may select a learning model from the second and third learning models to input prompts generated using the acquired inquiry items, based on the inquiry items received from hotel employees.

[0093] Furthermore, by generating a learning model based on a collection of know-how specific to each hotel, it becomes easy to update the learning model generated based on that know-how collection. In addition, since a learning model is prepared for each hotel, it becomes easy to customize the learning model for that particular hotel.

[0094] <Variation 3> In the above embodiment, the server device 10 was described in a case where it provides information in response to an inquiry from a user (hotel employee). However, the server device 10 may also provide information to hotel employees in response to an event (incident) that occurs at the hotel.

[0095] In this case, each hotel is equipped with devices and sensors to detect events occurring within the hotel. For example, as shown in Figure 9, a fire alarm 31, security cameras 32, and microphones 33 are installed inside the hotel. The fire alarm 31, security cameras 32, and microphones 33 are configured to communicate with the server device 10.

[0096] Furthermore, the server device 10 includes an information provision control unit 205 (see Figure 10). The information provision control unit 205 is a means for controlling the provision of information to users (hotel employees, etc.).

[0097] When the fire alarm 31 detects the occurrence of a fire, it transmits this fact, along with the hotel ID, to the server device 10.

[0098] The information provision control unit 205 identifies the hotel where the fire occurred based on the hotel ID received from the fire alarm 31. The information provision control unit 205 generates information regarding the actions that hotel employees (hotel employees of the hotel where the fire occurred) should take if a fire occurs at the identified hotel.

[0099] In this process, the information provision control unit 205 generates information to be provided to hotel employees using a learning model (large-scale language model). For example, the information provision control unit 205 generates a prompt such as, "A fire has broken out at Hotel A. Please tell us what actions hotel employees should take."

[0100] The information provision control unit 205 inputs the generated prompt into a large-scale language model and obtains information to be provided to hotel employees. For example, a learning model generated using an emergency response manual as training data can output an appropriate response to the above prompt. For example, the information provision control unit 205 obtains information such as, "First, please contact the fire department. Next, please evacuate the guests safely," from the large-scale language model.

[0101] The information provision control unit 205 transmits the generated information to the hotel terminals 20. For example, the information provision control unit 205 transmits the generated information to all or some of the hotel terminals 20 installed in the hotel where the fire occurred.

[0102] For example, the information provision control unit 205 refers to table information that stores the IP addresses of the hotel terminals 20 installed in each hotel, and identifies the destination to which the generated information is to be sent. The information provision control unit 205 then sends the information to the identified destination (hotel terminal 20).

[0103] Alternatively, the server device 10 may detect an event occurring at the hotel (for example, an emergency) based on image data obtained from the security camera 32 and audio data obtained from the microphone 33.

[0104] In this case, the security camera 32 transmits image data (video data, still image data) and the hotel ID to the server device 10 in real time. The microphone 33 also transmits audio data and the hotel ID to the server device 10 in real time.

[0105] The information provision control unit 205 detects events occurring at the hotel based on acquired image data and audio data. For example, the information provision control unit 205 uses image data to detect fire outbreaks, intruder intrusions, etc. Alternatively, the information provision control unit 205 detects fire outbreaks based on the warning sounds from the fire alarm 31 picked up by the microphone 33.

[0106] The information provision control unit 205 may also use a learning model obtained through machine learning to detect events (emergencies) occurring within the hotel.

[0107] The learning model is obtained through machine learning using a large amount of training data in which image data (image data of the hotel interior) is labeled (emergency situation; e.g., intruder intrusion, fire outbreak). Any algorithm such as support vector machines, boosting, or neural networks can be used to generate the learning model. Since the above-mentioned algorithms such as support vector machines can be publicly known, their explanation will be omitted. A learning model using audio data can be generated in a similar manner.

[0108] The information provision control unit 205 acquires information from a learning model (large-scale language model) regarding the actions that hotel employees should take in response to events (emergencies) detected based on image data and audio data. The information provision control unit 205 then transmits the acquired information (provided information) to the hotel terminal 20.

[0109] In this manner, when a predetermined event occurs in a predetermined area where predetermined operations are performed (for example, a hotel), the information provision control unit 205 inputs a prompt generated in response to the event into the fourth learning model. The fourth learning model is a learning model generated using at least an emergency response manual as learning data. By inputting the prompt into the fourth learning model, the information provision control unit 205 acquires information to be provided to the user (hotel employee) and provides the acquired information to the user.

[0110] Furthermore, the information provision control unit 205 detects the occurrence of a predetermined event using at least one of the following: image data obtained from the camera device, audio data obtained from a microphone installed in a predetermined area, and information obtained from a predetermined device installed in a predetermined area. The information provision control unit 205 transmits the information acquired from the learning model (large-scale language model) to the hotel terminal 20 installed in the predetermined area.

[0111] As described above, the server device 10 may detect events occurring at the hotel and provide hotel employees with information on the actions they should take in response to the detected events. In other words, the server device 10 may implement a push-type information provision system triggered by events occurring at the hotel. Through the operation of such a server device 10, hotel employees can act calmly even in the event of an emergency.

[0112] <Modification 4> When the hotel terminal 20 receives an inquiry from a hotel employee, it may also obtain the category of the inquiry. For example, the hotel terminal 20 may obtain categories such as "inquiry about the hotel system," "inquiry about room amenities," and "inquiry from a guest." The hotel terminal 20 may also send an inquiry response request to the server device 10, which includes the obtained category along with the inquiry.

[0113] The server device 10 may use the acquired category to generate a prompt. For example, if the category "Hotel System Inquiry" is acquired, the inquiry response control unit 203 may generate a prompt such as, "Hotel A has made an inquiry regarding the hotel system. Please answer the inquiry item 'Please tell me how to display a list of reservations.'" By clarifying the instructions for the large-scale language model, the inquiry response control unit 203 can obtain a more accurate answer.

[0114] Alternatively, the server device 10 may aggregate the number of inquiries and the content of inquiries for each category. For example, the server device 10 may store inquiries related to the hotel system, inquiries from guests (content of guest complaints), etc., in a predetermined database. The information stored in the database may be used to improve hotel operations.

[0115] <Modification 5> The above embodiment describes a case where support for hotel employees is provided using a learning model (large-scale language model). The server device 10 may provide support for hotel employees using means other than the learning model, either in place of or in addition to the learning model.

[0116] For example, the server device 10 may use information obtained from the internet (external servers on the network) to support hotel staff. For example, the server device 10 may obtain weather forecasts, evacuation information in the event of a disaster, etc., from external servers. For example, in order to answer a question such as "What is the weather like tomorrow?", the server device 10 may obtain weather forecasts from external servers. Alternatively, the server device 10 may obtain traffic information, nearby event information, etc., from external servers.

[0117] As described above, the server device 10 according to the first embodiment uses a learning model generated using not only the hotel system manual but also a collection of hotel operation know-how as training data to answer user inquiries. As a result, more useful information (answers) is provided to hotel employees who are experiencing operational problems. Furthermore, since the server device 10 uses a learning model generated using know-how specific to each hotel as training data to answer inquiries from hotel employees, the know-how of veteran employees can be passed on to less experienced employees.

[0118] Next, we will describe the hardware of each device that makes up the information processing system. Figure 11 shows an example of the hardware configuration of server device 10.

[0119] The server device 10 can be configured using an information processing device (a so-called computer), and has the configuration illustrated in Figure 11. For example, the server device 10 includes a processor 311, memory 312, input / output interface 313, and communication interface 314, etc. The components of the processor 311, etc., are connected by an internal bus or the like and are configured to communicate with each other.

[0120] However, the configuration shown in Figure 11 is not intended to limit the hardware configuration of the server device 10. The server device 10 may include hardware not shown, and it may not have to have an input / output interface 313 if necessary. Also, the number of processors 311 etc. included in the server device 10 is not intended to be limited to the example in Figure 11; for example, multiple processors 311 may be included in the server device 10.

[0121] The processor 311 is a programmable device such as a CPU (Central Processing Unit), MPU (Micro Processing Unit), or DSP (Digital Signal Processor). Alternatively, the processor 311 may be a device such as an FPGA (Field Programmable Gate Array) or ASIC (Application Specific Integrated Circuit). The processor 311 executes various programs, including an operating system (OS).

[0122] Memory 312 includes RAM (Random Access Memory), ROM (Read Only Memory), HDD (Hard Disk Drive), SSD (Solid State Drive), etc. Memory 312 stores the OS program, application programs, and various data.

[0123] The input / output interface 313 is an interface for a display device or input device (not shown). The display device is, for example, a liquid crystal display. The input device is, for example, a device that accepts user input such as a keyboard, mouse, or touch panel.

[0124] The communication interface 314 is a circuit, module, etc., that communicates with other devices. For example, the communication interface 314 includes a NIC (Network Interface Card), etc.

[0125] The functions of the server device 10 are realized by various processing modules. These processing modules are realized, for example, by the processor 311 executing a program stored in memory 312. The program can also be recorded on a computer-readable storage medium. The storage medium can be a non-transitory material such as semiconductor memory, hard disk, magnetic recording medium, or optical recording medium. In other words, the present invention can also be embodied as a computer program product. Furthermore, the program can be downloaded via a network or updated using the storage medium on which the program is stored. Moreover, the processing module may be realized by a semiconductor chip.

[0126] Furthermore, the hotel terminal 20 and the like can also be configured using an information processing device, similar to the server device 10. Since their basic hardware configuration is no different from that of the server device 10, a detailed explanation will be omitted.

[0127] The server device 10 is equipped with a computer, and its functions can be realized by having the computer execute a program. Furthermore, the server device 10 executes a control method for the server device 10 using this program.

[0128] [Differentiation] The configuration and operation of the information processing system described in the above embodiment are illustrative examples and are not intended to limit the system configuration.

[0129] The input / output interface of the hotel terminal 20 may have a multimodal input function that accepts input in multiple modalities such as text, voice, images, and video.

[0130] Text input is performed via the keyboard or touchscreen of the hotel terminal 20. Voice input is performed using the microphone built into the hotel terminal 20 or an external microphone, and is converted to text by a deep learning-based speech recognition model. This speech recognition model is fine-tuned to handle specialized terminology and proper nouns specific to hotel operations.

[0131] Image input is performed using the camera on the hotel terminal 20 or an externally connected camera device. The input images are analyzed by an image recognition model using a convolutional neural network (CNN). For example, when making an inquiry about equipment problems in a guest room, inputting an image of the faulty part allows for a more accurate understanding of the situation.

[0132] The same camera device is used for video input, performing image recognition frame by frame, as well as analyzing the data as a time series. This can be used, for example, to detect suspicious behavioral patterns of people inside a hotel.

[0133] Input data from each modality is normalized and processed, including denoising, by a dedicated preprocessing module. It is then integrated by a cross-modal learning model and converted into a single feature vector. This feature vector is used as input to the first learning model.

[0134] For example, if a hotel employee makes a voice inquiry saying, "The air conditioner in my room is broken," and simultaneously sends a photo of the broken air conditioner, the voice is converted to text, and the image is categorized into the object (air conditioner) and its condition (type of malfunction). This information is then integrated and processed into a more detailed and accurate inquiry.

[0135] If an error occurs in input recognition, the system will ask the hotel employee who made the inquiry for clarification. For example, if the reliability of the speech recognition is low, the recognition result will be displayed as text, providing an opportunity for correction. Also, if the object cannot be identified by image recognition, the system will request additional information or photos from a different angle from the hotel employee who made the inquiry.

[0136] This multimodal input feature allows hotel staff to make inquiries and reports more naturally and efficiently, and enables the system to provide more accurate and contextually relevant responses. Furthermore, this feature facilitates the communication of visual issues and complex situations that were difficult to convey with traditional text-based inquiry systems, contributing to improved efficiency and quality in hotel operations.

[0137] The above embodiment described an information processing system that supports employees working at a hotel. However, the information processing system disclosed herein may also support employees working in other industries or sectors. For example, the information processing system may support employees working at retail stores, restaurants, educational institutions, etc.

[0138] The server device 10 may provide hotel employees with answers obtained from the learning model, clearly indicating whether the answers are based on the hotel system manual or on a collection of know-how. In this case, the inquiry response control unit 203 of the server device 10 should input a prompt to the learning model instructing it to output an answer that clearly indicates the learning data that forms the basis of the answer. That is, the server device 10 may distinguish between hotel employee actions defined in the manual and hotel employee actions that are supplemented and customized with know-how and present them to the hotel employees.

[0139] The server device 10 may present multiple answers to a single inquiry to the hotel employee. In this case, the server device 10 may aggregate the answers adopted by the hotel employee (actions suggested by the server device 10) for each attribute of the hotel employee. For example, the server device 10 may identify the hotel employee with the inquiry using facial recognition and store the identified hotel employee's information, such as age, gender, and years of service, in association with the adopted answer. The server device 10 may use the stored information to calculate the adoption rate for each hotel employee and answer.

[0140] In the above embodiment, a server-client system was described in which the server device 10 is the server and the hotel terminal 20 is the client. However, the functions of the server device 10 may be realized by the hotel terminal 20 on which a predetermined application is installed. That is, the acquisition of inquiries from hotel employees, the generation of prompts, the acquisition and provision of answers may all be completed at the hotel terminal 20.

[0141] Alternatively, some functions of the server device 10 may be implemented in another device. More specifically, it is sufficient if the "inquiry response control unit (inquiry response control means)" etc. described above are implemented in any device included in the system.

[0142] The responses output by the learning model may include drawings and audio. The server device 10 may provide responses including drawings (images) to hotel employees via the hotel terminal 20.

[0143] The form of data transmission and reception between each device (server device 10, hotel terminal 20) is not particularly limited, but the data transmitted and received between these devices may be encrypted. Inquiries and answers are transmitted and received between these devices, and it is desirable that encrypted data be transmitted and received in order to properly protect this information. For example, communication between the server device 10 and each hotel terminal 20 may be encrypted end-to-end using the industry-standard TLS (Transport Layer Security) protocol. Inquiry history and answer data stored in the database within the server device 10 may be encrypted at the database level, and only administrators with access rights may be able to decrypt it. The encryption key may be managed by a hardware security module (HSM) and updated periodically. All encryption and decryption operations are recorded in an audit log and used to detect unauthorized access.

[0144] In the flowcharts (sequence diagrams) used in the above description, multiple processes (processes) are shown in order, but the execution order of the processes performed in the embodiment is not limited to the order in which they are shown. In the embodiment, the order of the illustrated processes can be changed to the extent that it does not impair the content, for example, by executing each process in parallel.

[0145] The embodiments described above are explained in detail to facilitate understanding of the disclosure, and it is not intended that all the configurations described above are necessary. Furthermore, when multiple embodiments are described, each embodiment may be used individually or in combination. For example, it is possible to replace parts of the configuration of one embodiment with those of another embodiment, or to add configurations from other embodiments to the configuration of one embodiment. In addition, it is possible to add, delete, or replace parts of the configuration of one embodiment with those of another.

[0146] As described above, the industrial applicability of the present invention is clear, and it is particularly suitable for information processing systems related to support for hotel employees and the like.

[0147] Some or all of the above embodiments may also be described as follows, but are not limited to the following:

[0148] [Note 1] Methods for obtaining inquiries from users, A response control means that inputs a prompt generated using the acquired inquiry into a first learning model generated using at least a manual for a predetermined system and a collection of know-how for a predetermined task as learning data, thereby obtaining an answer to the acquired inquiry and providing the acquired answer to the user. A server device equipped with the following features.

[0149] [Note 2] The server device described in Appendix 1, wherein the response control means obtains an answer to the acquired inquiry by inputting a prompt generated using the acquired inquiry to a second learning model generated using a manual for a predetermined system as learning data, or a third learning model generated using a collection of know-how for a predetermined business as learning data, instead of the first learning model.

[0150] [Note 3] The server device as described in Appendix 2, wherein the response control means selects a learning model from the second learning model and the third learning model that inputs a prompt generated using the acquired inquiry, based on the acquired inquiry.

[0151] [Note 4] The server device described in Appendix 1 further comprises an information provision control means that, when a predetermined event occurs in a predetermined area where the predetermined operations are performed, inputs a prompt generated in response to the event into a fourth learning model to acquire information to be provided to the user and provides the acquired information to the user.

[0152] [Note 5] The server device described in Appendix 4, wherein the information provision control means detects the occurrence of a predetermined event using at least one of the following: image data obtained from a camera installed in the predetermined area, audio data obtained from a microphone installed in the predetermined area, and information obtained from a predetermined device installed in the predetermined area.

[0153] [Note 6] The information provision control means is a server device as described in Appendix 5, which transmits information acquired from the learning model to a terminal installed in the predetermined area.

[0154] [Note 7] The first learning model is a server device as described in any one of the appendices 1 to 6, which is a learning model generated using a manual on the hotel system and a collection of know-how on hotel operations as learning data.

[0155] [Note 8] The response control means is a server device according to any one of the appendices 1 to 6, which uses a large-scale language model as the learning model to obtain a response to the acquired inquiry.

[0156] [Note 9] The process of obtaining inquiries from users, and The response control step involves inputting a prompt generated using the acquired inquiry into a first learning model generated using at least a manual for a predetermined system and a collection of know-how for a predetermined task as learning data, thereby obtaining an answer to the acquired inquiry and providing the acquired answer to the user. A control method for a server device, comprising the following features.

[0157] [Note 10] The server device control method described in Appendix 9, wherein the response control step involves inputting a prompt generated using the acquired inquiry items to a second learning model generated using a manual for a predetermined system as learning data, or to a third learning model generated using a collection of know-how for a predetermined business as learning data, instead of the first learning model, thereby obtaining an answer to the acquired inquiry items.

[0158] [Note 11] The response control step is a control method for a server device as described in Appendix 10, in which, based on the acquired query items, a learning model is selected from the second learning model and the third learning model to input a prompt generated using the acquired query items.

[0159] [Note 12] A control method for a server device as described in Appendix 9, further comprising an information provision control step, in which, when a predetermined event occurs in a predetermined area where the predetermined operations are performed, a prompt generated in accordance with the event occurs is input to a fourth learning model to obtain information to be provided to the user, and the obtained information is provided to the user.

[0160] [Note 13] The control method for a server device as described in Appendix 12, wherein the information provision control step detects the occurrence of a predetermined event using at least one of the following: image data obtained from a camera installed in the predetermined area, audio data obtained from a microphone installed in the predetermined area, and information obtained from a predetermined device installed in the predetermined area.

[0161] [Note 14] The information provision control step is a control method for a server device as described in Appendix 13, which transmits information acquired from the learning model to a terminal installed in the predetermined area.

[0162] [Note 15] The control method for a server device as described in any one of the appendices 9 to 14, wherein the first learning model is a learning model generated using a manual for the hotel system and a collection of know-how for hotel operations as learning data.

[0163] [Note 16] The response control step is a server device control method according to any one of the appendices 9 to 14, wherein the response control step uses a large-scale language model as the learning model to obtain a response to the acquired inquiry.

[0164] [Note 17] The computer installed in the server device The process of obtaining inquiry details from users, The response control process involves inputting a prompt generated using the acquired inquiry into a first learning model generated using at least a manual for a predetermined system and a collection of know-how for a predetermined task as learning data, thereby obtaining an answer to the acquired inquiry and providing the acquired answer to the user. A program to execute.

[0165] [Note 18] The program described in Appendix 17, wherein the response control process obtains an answer to the acquired inquiry by inputting a prompt generated using the acquired inquiry to a second learning model generated using a manual for a predetermined system as learning data, or a third learning model generated using a collection of know-how for a predetermined business as learning data, instead of the first learning model.

[0166] [Note 19] The response control process is the program described in Appendix 18, which, based on the acquired query items, selects a learning model from the second and third learning models that inputs a prompt generated using the acquired query items.

[0167] [Note 20] The program described in Appendix 17 further executes an information provision control process, which, when a predetermined event occurs in a predetermined area where the predetermined operations are performed, inputs a prompt generated in response to the event into a fourth learning model to obtain information to be provided to the user and provides the obtained information to the user.

[0168] [Note 21] The information provision control process is a program as described in Appendix 20, which detects the occurrence of a predetermined event using at least one of the following: image data obtained from a camera installed in the predetermined area, audio data obtained from a microphone installed in the predetermined area, and information obtained from a predetermined device installed in the predetermined area.

[0169] [Note 22] The information provision control process is the program described in Appendix 21, which transmits the information obtained from the learning model to a terminal installed in the predetermined area.

[0170] [Note 23] The first learning model is a program described in any one of the appendices 17 to 22, which is a learning model generated using a manual on the hotel system and a collection of know-how on hotel operations as learning data.

[0171] [Note 24] The response control process is a program described in any one of the appendices 17 to 22, which uses a large-scale language model as the learning model to obtain a response to the acquired inquiry.

[0172] Furthermore, some or all of the configurations described in Appendices 2 to 8, which are subordinate to Appendice 1 above, may also be subordinate to Appendices 9 and 17 in the same way as those described in Appendices 2 to 8. Moreover, not limited to Appendices 1, 9 and 17, some or all of the configurations described as appendices may also be subordinate to various hardware, software, various recording means for recording software, or systems, without departing from the embodiments described above.

[0173] Furthermore, each disclosure of the above-mentioned prior art documents cited herein is incorporated herein by reference. Although embodiments of the present invention have been described above, the present invention is not limited to these embodiments. It will be understood by those skilled in the art that these embodiments are merely illustrative and that various modifications are possible without departing from the scope and spirit of the present invention. That is, the present invention naturally includes the entire disclosure, including the claims, and various modifications and alterations that can be made by those skilled in the art in accordance with the technical idea. [Explanation of Symbols]

[0174] 10 Server devices 20 Hotel Terminals 31 Fire alarm 32 security cameras 33 Mike 100 Server Devices 101 Acquisition method 102 Response control means 201 Communication Control Unit 202 Learning Model Management Department 203 Inquiry Response Control Unit 204 Storage section 205 Information Provision Control Unit 311 Processors 312 memory 313 Input / Output Interfaces 314 Communication Interface

Claims

1. Methods for obtaining inquiries from users, A response control means that inputs a prompt generated using the acquired inquiry into a first learning model generated using at least a manual for a predetermined system and a collection of know-how for a predetermined task as learning data, thereby obtaining an answer to the acquired inquiry and providing the acquired answer to the user. A server device equipped with the following features.

2. The server device according to claim 1, wherein the response control means obtains an answer to the acquired inquiry by inputting a prompt generated using the acquired inquiry to a second learning model generated using a manual for a predetermined system as learning data, or a third learning model generated using a collection of know-how for a predetermined business as learning data, instead of the first learning model.

3. The server device according to claim 2, wherein the response control means selects a learning model from the second learning model and the third learning model that inputs a prompt generated using the acquired inquiry items, based on the acquired inquiry items.

4. The server device according to claim 1, further comprising an information provision control means that, when a predetermined event occurs in a predetermined area where the predetermined operations are performed, inputs a prompt generated in response to the event into a fourth learning model to acquire information to be provided to the user and provides the acquired information to the user.

5. The server device according to claim 4, wherein the information provision control means detects the occurrence of a predetermined event using at least one of the following: image data obtained from a camera installed in the predetermined area, audio data obtained from a microphone installed in the predetermined area, and information obtained from a predetermined device installed in the predetermined area.

6. The server device according to claim 5, wherein the information provision control means transmits information acquired from the learning model to a terminal installed in the predetermined area.

7. The server device according to any one of claims 1 to 6, wherein the first learning model is a learning model generated using a manual for the hotel system and a collection of know-how for hotel operations as learning data.

8. The server device according to any one of claims 1 to 6, wherein the response control means uses a large-scale language model as the learning model to obtain a response to the acquired inquiry.

9. The process of obtaining inquiries from users, and The response control step involves inputting a prompt generated using the acquired inquiry into a first learning model generated using at least a manual for a predetermined system and a collection of know-how for a predetermined task as learning data, thereby obtaining an answer to the acquired inquiry and providing the acquired answer to the user. A control method for a server device, comprising the following features.

10. The computer installed in the server device The process of obtaining inquiry details from users, The response control process involves inputting a prompt generated using the acquired inquiry into a first learning model generated using at least a manual for a predetermined system and a collection of know-how for a predetermined task as learning data, thereby obtaining an answer to the acquired inquiry and providing the acquired answer to the user. A program to execute.

Images