Method, program and information processing system for providing visitor with information on event

Abstract

To provide a method, a program and an information processing system capable of providing a visitor of an event of an exhibition or the like with accurate information meeting desires of the visitor.SOLUTION: A method for providing a visitor with information on an event by an information processing device 1 includes the steps for: acquiring questions of the visitor about the event; acquiring one or more pieces of event information having relevance to question sentences from the plurality of pieces of event information on the event; generating a first prompt to give an instruction to a large-scale language model so as to generate answer sentences to the question sentences on the basis of the acquired one or more pieces of event information; and performing processing for providing the visitor with answer sentences generated by the large-scale language model according to the first prompt.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to a method, a program, and an information processing system for providing information about an event to visitors.

Background Art

[0002] There is known a system that enables visitors to easily find an exhibition booth where information desired by the visitors can be obtained. The following patent document describes an exhibition guidance system that conducts a questionnaire on fields of interest in advance for those who plan to visit an exhibition, and based on the results of this questionnaire, guides the exhibition positions of fields of interest at the exhibition venue. According to this system, visitors can immediately know where in the exhibition venue the information they are looking for is located as soon as they visit the exhibition.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the above-described conventional system, since the guidance of the exhibition booth is performed based on the results of the questionnaire on fields of interest prepared in advance, fields of interest of visitors that are not included in the questionnaire are not considered at all. Therefore, there is a problem that accurate guidance cannot be provided for the various demands of visitors.

[0005] The present invention has been made in view of such circumstances, and an object thereof is to provide a method, a program, and an information processing system that can provide accurate information that meets the demands of visitors to an event such as an exhibition.

Means for Solving the Problems

[0006] The method according to the first aspect of the present invention is a method for an information processing system to provide information regarding an event to visitors, the method comprising: a step of acquiring a question text of a visitor regarding the event; a step of acquiring, from a plurality of event information regarding the event, one or more pieces of event information having relevance to the question text; a step of generating a first prompt for instructing a large language model to generate an answer text for the question text based on the acquired one or more pieces of event information; and a step of performing a process of providing the answer text generated by the large language model in response to the first prompt to the visitors.

[0007] The program according to the second aspect of the present invention is a program including instructions for causing an information processing system to perform a process of providing information regarding an event to visitors, and the process that the information processing system performs according to the instructions includes: a step of acquiring a question text of a visitor regarding the event; a step of acquiring, from a plurality of event information regarding the event, one or more pieces of event information having relevance to the question text; a step of generating a first prompt for instructing a large language model to generate an answer text for the question text based on the acquired one or more pieces of event information; and a step of performing a process of providing the answer text generated by the large language model in response to the first prompt to the visitors.

[0008] The information processing system according to the third aspect of the present invention is an information processing system that performs a process of providing information regarding an event to visitors, the information processing system comprising: a processing unit; and a storage unit that stores instructions to be executed in the processing unit, and the process that the processing unit performs according to the instructions includes: a step of acquiring a question text of a visitor regarding the event; a step of acquiring, from a plurality of event information regarding the event, one or more pieces of event information having relevance to the question text; a step of generating a first prompt for instructing a large language model to generate an answer text for the question text based on the acquired one or more pieces of event information; and a step of performing a process of providing the answer text generated by the large language model in response to the first prompt to the visitors.

[0009] An information processing system according to a fourth aspect of the present invention is an information processing system that performs processing for providing information related to an event to visitors, the system including means for acquiring a question sentence of a visitor related to the event, means for acquiring, from a plurality of event information related to the event, one or more pieces of event information having relevance to the question sentence, means for generating a first prompt for instructing a large language model to generate an answer sentence for the question sentence based on the acquired one or more pieces of event information, and means for performing processing for providing the answer sentence generated by the large language model according to the first prompt to the visitors.

Advantages of the Invention

[0010] According to the present invention, it is possible to provide a method, a program, and an information processing system capable of providing accurate information that meets the needs of visitors to an event such as an exhibition.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

MODE FOR CARRYING OUT THE INVENTION

[0012] FIG. 1 is a diagram showing an example of the configuration of the system according to the present embodiment. The system shown in the example of FIG. 1 includes an information processing apparatus 1, a visitor terminal device 4, a participant terminal device 5, an operator terminal device 6, and a large language model 7 that can communicate with each other via a communication network 9 such as the Internet. The information processing apparatus 1 is an example of the information processing system of the present invention.

[0013]

[0014] [Information Processing Apparatus 1] The information processing apparatus 1 performs a process of providing information related to an event in response to a request from the visitor terminal device 4, a process of recording the behavior of visitors at an event in progress, a process of obtaining the matching degree between participants and visitors in response to a request from the participant terminal device 5, and the like.

[0015] For example, the information processing apparatus 1 is configured to include one or a plurality of computers. The information processing apparatus 1 shown in the example of FIG. 1 has a communication unit 11, a storage unit 12, and a processing unit 13.

[0016] The communication unit 11 communicates with other devices (visitor terminal device 4, participant terminal device 5, operator terminal device 6, large language model 7, etc.) via the communication network 9. The communication unit 11 includes a device (such as a network interface card) that communicates in accordance with a predetermined communication standard such as Ethernet (registered trademark) or wireless LAN.

[0017] The storage unit 12 stores one or more programs 121 including instructions executed by the processing unit 13, data temporarily stored during the process of the processing by the processing unit 13, data used in the processing of the processing unit 13, data obtained as a result of the processing of the processing unit 13, and the like. The storage unit 12 may include, for example, a main storage device (RAM, ROM, etc.) and an auxiliary storage device (flash memory, SSD, hard disk, memory card, optical disk, etc.). The storage unit 12 may be composed of one memory or a plurality of memories. When the storage unit 12 is composed of a plurality of memories, each memory is connected to the processing unit 13 via a computer bus or other arbitrary data transmission means.

[0018] The processing unit 13 comprehensively controls the overall operation of the information processing apparatus 1 and executes predetermined information processing. The processing unit 13 includes, for example, one or more processors (CPU (central processing unit), MPU (micro-processing unit), DSP (digital signal processor), etc.) that execute processing in accordance with the instructions of one or more programs 121 stored in the storage unit 12. The processing unit 13 operates as a computer when one or more processors execute the instructions of one or more programs 121 stored in the storage unit 12. The information processing apparatus 1 may have a plurality of such computers, and at least a part of the processing according to the present embodiment may be executed by a plurality of computers in cooperation.

[0019] The processing unit 13 may include one or more dedicated hardware components (such as an ASIC (application specific integrated circuit), FPGA (field-programmable gate array), etc.) configured to implement specific functions. In this case, the processing unit 13 may execute all the processes described in this embodiment on a computer, or at least some of the processes may be executed on dedicated hardware.

[0020] The program 121 may be recorded, for example, on a computer-readable recording medium (such as an optical disk, memory card, USB memory, or other non-transitory tangible medium). The processing unit 13 may read at least a part of one or more programs 121 recorded on such a recording medium by a recording medium reading device (such as an optical disk device) or an interface device (such as a USB interface) not shown in the figure, and write it to the storage unit 12. Alternatively, the processing unit 13 may download at least a part of one or more programs 121 from another device connected to the communication network 9 via the communication unit 11 and write it to the storage unit 12. The one or more programs 121 may include instructions for causing the processing unit 13 to execute at least a part of the processes according to the present embodiment described later.

[0021] The storage device 2 stores various information used in the processing of the information processing device 1. The information processing device 1 and the storage device 2 can communicate via an arbitrary communication path (LAN, dedicated line network, Internet, etc.). For example, the storage device 2 may be included in a file server, a database server, a cloud server, etc. that accepts access from a plurality of devices, or may be a dedicated storage device that can be accessed only by the information processing device 1. In the example of FIG. 1, the storage device 2 stores a visitor database 21, a participant database 22, a target management database 23, a panel management database 24, a content management database 25, an interaction history database 26, an action record database 27, and a vector database 28. The storage device 2 also stores a plurality of event information 29 that is information related to each event, and a plurality of contents 30 prepared for visitors by participants and the like. In the following description, the database may be abbreviated as "DB".

[0022] The visitor DB 21 includes a plurality of visitor information corresponding to a plurality of visitors. One piece of visitor information includes information about one visitor. One piece of visitor information includes, for example, at least a part of the following information. · Information for identifying an individual visitor (visitor ID) · Date and time when the visitor information was registered · Attributes of the visitor (age, gender, occupation, position, etc.)

[0023] The participant DB 22 includes a plurality of participant information corresponding to a plurality of participants who participate in the event. One piece of participant information includes information about one participant (such as an exhibitor at an exhibition). One piece of participant information includes, for example, at least a part of the following information. · Information for identifying an individual participant (participant ID) · Date and time when the participant information was registered · Name of the participant (company name, etc.) · Contact information of the participant (address, phone number, email address, etc.)

[0024] The target management DB 23 includes a plurality of target information regarding the targets (products, services, etc. provided by participants) of the actions of the visitors at the event. One piece of target information includes information regarding one target (product, service, etc.) related to one participant. One piece of target information includes, for example, at least a part of the following information. · Information for identifying an individual target (target ID) · Name of the target (brand name, model name, etc.) · Category to which the target belongs (type of product or service, etc.) · Participant ID indicating the participant related to the target

[0025] The visitors to the event can obtain information regarding a plurality of targets (products, services, etc.) registered in the target management DB 23 at the event.

[0026] The panel management DB 24 includes a plurality of panel information regarding the panels arranged at the event venue to provide various information and materials, etc. to the visitors. One piece of panel information includes information regarding one panel. One piece of panel information includes, for example, at least a part of the following information. · Information for identifying an individual panel (panel ID) · One or more target IDs indicating the targets related to the panel · One or more content IDs indicating the content related to the panel · Installation location of the panel (management number of the exhibition booth, etc.)

[0027] For the panel indicated by the panel ID, an optical code such as a QR code (registered trademark) or a barcode is printed, and the panel ID is included in this optical code. The optical code of the panel is read by the visitor terminal device 4 of the visitor who visited the event venue and is used for registering the action record information described later.

[0028] The content management DB 25 includes a plurality of content information regarding the content provided to the visitors. One piece of content information includes information regarding one content. One piece of content information includes, for example, at least a part of the following information. · Information for identifying individual contents (content ID) · Information regarding the content of the content (name of the content, type, data size, etc.) · One or more target IDs indicating the targets (products, services, etc.) related to the content · One or more participant IDs indicating the participants (exhibitors, etc.) related to the content · Storage location of the content

[0029] The "storage location of the content" is information regarding the location where the content (content 30 in FIG. 2) is stored in the storage device 2, and includes, for example, the directory and file name where the content 30 is stored. The storage location of the content 30 is not limited to the storage device 2, and may also be other devices (data servers, etc.) connected to the communication network 9.

[0030] The dialogue history DB 26 includes a plurality of dialogue history information including the history of questions regarding the events performed by the visitors and the answers from the information processing device 1 thereto. One piece of dialogue history information includes the history of questions and answers by one visitor. One piece of dialogue history information includes, for example, at least a part of the following information. · Visitor ID indicating the visitor related to the dialogue history · One or more question texts by the visitor · Answer texts for each of the one or more question texts · Event information 29 extracted for each of the one or more question texts Instead of the event information 29, an event information ID may be included.

[0031] The action record DB 27 includes a plurality of action record information corresponding to a plurality of visitors. One piece of action record information includes a record regarding a predetermined action performed by one visitor at the event. One piece of action record information includes, for example, at least a part of the following information. · Visitor ID indicating the one visitor who performed a predetermined action at the event · Date and time when the visitor performed a predetermined action at the event · One or more target IDs indicating the target (product, service, etc.) of the predetermined action · Participant ID indicating participants related to the target of a predetermined action · Type of action in an event · Content of the action in the event

[0032] The "type of action" includes, for example, the optical code of the admission ticket being read by the participant terminal device 5 operated by an explainer or the like when visiting an exhibition booth or the like at the event venue, receiving content related to products or the like (viewing, downloading, etc.), inputting evaluations or comments on the target of products or the like, answering questionnaires, and the like. The "content of the action" indicates the content of a predetermined action indicated by the "type of action", and includes, for example, information on the content (content ID, etc.) in the case of downloading content.

[0033] The vector DB 28 includes a plurality of vector information corresponding to a plurality of event information 29. One vector information includes information on a vector expressing the content (meaning, etc.) of one event information 29. One vector information includes, for example, at least a part of the following information. · Information (event information ID) for identifying individual event information 29 · A vector obtained as a distributed representation of event information 29 · One or more target IDs indicating targets (products, services, etc.) related to event information 29 · One or more participant IDs indicating participants (exhibitors, etc.) related to event information 29

[0034] For example, the vector of the event information 29 is obtained in advance using a known learning model such as Word2vec or BERT (bidirectional encoder representations from transformers) and registered in the vector DB 28.

[0035] Event information 29 includes various information related to the event. For example, event information 29 includes information about products, services, etc. provided to visitors (introduction texts of products and services, explanatory texts of technologies related to products and services, explanatory texts about industry and market trends, etc.), information about participants in the event (introduction texts of participating companies providing products and services, locations of exhibition booths, seminar schedules, etc.). An event information ID is assigned to each individual event information 29.

[0036] Content 30 is content for visitors related to the objects of the visitors' actions (exhibited products, services, etc.), and includes, for example, document files, image files, video files, URLs for content, etc. A content ID is assigned to each individual content 30.

[0037] [Visitor terminal device 4] The visitor terminal device 4 is a device operated by a visitor to the event, and is, for example, a device equipped with an information communication function such as a smartphone, a tablet, or a personal computer. The system shown in FIG. 1 has a plurality of visitor terminal devices 4 corresponding to a plurality of visitors. The visitor terminal device 4 includes a processing unit, a storage unit, and a communication unit similar to the communication unit 11, the storage unit 12, and the processing unit 13 of the information processing device 1, and also includes an input unit (an input device such as a touch panel, a mouse, or a keyboard) for inputting a user's instruction to the processing unit, a display unit (a display device such as a liquid crystal display) for displaying information, etc. Further, the visitor terminal device 4 includes a camera capable of reading optical codes such as QR codes (registered trademarks) and barcodes.

[0038] [Participant terminal device 5] The participant terminal device 5 is a device operated by participants (exhibitors, etc.) of an event, and is a device equipped with an information communication function such as a smartphone, a tablet, or a personal computer. The system shown in FIG. 1 has a plurality of participant terminal devices 5 corresponding to a plurality of participants. The participant terminal device 5 includes a processing unit, a storage unit, and a communication unit similar to the communication unit 11, the storage unit 12, and the processing unit 13 of the information processing device 1, and also includes an input unit (input devices such as a touch panel, a mouse, and a keyboard) for inputting a user's instruction to the processing unit, a display unit (display device such as a liquid crystal display) for displaying information, etc. Further, the participant terminal device 5 includes a camera capable of reading optical codes such as QR codes (registered trademarks) and barcodes.

[0039] [Operator terminal device 6] The operator terminal device 6 is a device operated by an operator (event organizer, operation staff, etc.) related to the holding and operation of an event, and is a device equipped with an information communication function such as a smartphone, a tablet, or a personal computer. The operator terminal device 6 includes a processing unit, a storage unit, and a communication unit similar to the communication unit 11, the storage unit 12, and the processing unit 13 of the information processing device 1, and also includes an input unit (input devices such as a touch panel, a mouse, and a keyboard) for inputting a user's instruction to the processing unit, a display unit (display device such as a liquid crystal display) for displaying information, etc. Further, the operator terminal device 6 includes a camera capable of reading optical codes such as QR codes (registered trademarks) and barcodes.

[0040] [Large language model 7] The large language model 7 is a system configured to mimic human language understanding ability by training a large amount of text data on an artificial neural network having a large amount of parameters (for example, billions to trillions). The large language model 7 can perform a wide range of natural language processing tasks such as grasping the pattern and context of a sentence, answering questions, generating sentences, and translation according to a given prompt. As the large language model 7, for example, the GPT series (Chat GPT, etc.) operated by OpenAI in the United States can be used.

[0041] Here, the operation of the system shown in FIG. 1 having the above-described configuration will be described. FIG. 3 is a flowchart for explaining an example of the processing of the system shown in FIG. 1, and shows an example of the processing for providing a response sentence to a visitor's question.

[0042] The visitor terminal device 4 displays the screen of the web application provided by the information processing device 1 or the screen of the dedicated application, and inputs a question sentence of the visitor regarding the event (ST100). When the information processing device 1 acquires a question sentence from a visitor from the visitor terminal device 4 (ST105), this question sentence is added to the dialogue history information (dialogue history DB26) corresponding to the one visitor (ST110).

[0043] The information processing device 1 performs a process (ST115 to ST130) of acquiring one or more pieces of event information 29 having relevance to the question sentence from the visitors from among the plurality of pieces of event information 29 stored in the storage device 2.

[0044] Specifically, the information processing device 1 acquires a vector (hereinafter sometimes referred to as a "question sentence vector") representing the content of the question sentence from the visitors (ST115). For example, the information processing device 1 converts the question sentence from the visitors into a question sentence vector using the same learning model (such as Word2vec, BERT, etc.) as that used when acquiring the vectors registered in the vector DB28.

[0045] Note that the process of converting text such as a question sentence into a vector may be executed in the information processing device 1, or the process may be executed by an external server that provides the function of the vectorization learning model, and the information processing device 1 may acquire the vector of the processing result.

[0046] The information processing apparatus 1 searches the vector DB 28 for vectors similar to the question text vector acquired in step ST115, and identifies one or more event information IDs associated with the similar vectors in the vector DB 28. For example, the information processing apparatus 1 calculates the similarity between vectors based on the cosine similarity or other evaluation metrics, extracts the top one or more vectors with the highest similarity, and identifies one or more event information IDs associated with the extracted vectors. In this case, the information processing apparatus 1 may extract a predetermined number of vectors with the highest similarity, or may extract a number of vectors not exceeding the maximum number of the top vectors whose similarity is higher than a predetermined threshold value (if there are no vectors whose similarity is higher than the threshold value, a predetermined number of vectors with the highest similarity). The information processing apparatus 1 acquires (ST130) one or more event information 29 corresponding to the one or more event information IDs identified in the vector DB 28 from the storage device 2 as being relevant to the question text.

[0047] When one or more event information 29 relevant to the question text by one visitor are acquired, the information processing apparatus 1 adds this event information 29 (or event information ID) to the dialogue history information (dialogue history DB 26) corresponding to the one visitor (ST140).

[0048] The information processing apparatus 1 generates a prompt (hereinafter sometimes referred to as the "first prompt") for instructing the large language model 7 to generate a response text for the question text based on the one or more event information 29 acquired.

[0049] FIG. 4 is a diagram showing an example of the first prompt for instructing the generation of a response text for the question text. The first prompt shown in FIG. 4 includes an instruction sentence for giving an overall instruction to the large language model 7, the current question text acquired in step ST105, and the event information 29 (reference information) acquired in step ST130 for the current question text.

[0050] Also, if a visitor has asked questions before, the first prompt generated for the questions from this visitor includes the previous question-and-answer exchanges as the dialogue history. When a set of text consisting of a question sentence, event information 29 obtained for this question sentence, and an answer sentence of the large language model 7 for the question sentence is called a dialogue sentence, the first prompt shown in FIG. 4 includes the same number of dialogue sentences as the number of past questions by the visitor as the dialogue history. The information processing apparatus 1 acquires the dialogue history of one visitor from the dialogue history information (dialogue history DB 26) corresponding to the one visitor. That is, when the information processing apparatus 1 acquires one question sentence from one visitor, based on the dialogue history information corresponding to the one visitor and one or more pieces of event information 29 obtained for the one question sentence, it generates a first prompt that gives an instruction to generate an answer sentence for the one question sentence.

[0051] The information processing apparatus 1 provides the generated first prompt to the large language model 7 (ST165), and the large language model 7 generates an answer sentence according to the first prompt (ST170).

[0052] When an answer sentence for a question sentence of one visitor is generated by the large language model 7, the information processing apparatus 1 adds the generated answer sentence to the dialogue history information (dialogue history DB 26) corresponding to the one visitor (ST180).

[0053] The information processing apparatus 1 provides the answer sentence generated for the question sentence of one visitor to the visitor terminal device 4 of the one visitor (ST185). The visitor terminal device 4 displays the answer sentence provided from the information processing apparatus 1 on the screen of the web application of the information processing apparatus 1 or the screen of a dedicated application (ST190).

[0054] Thus, according to this embodiment, a first prompt is generated to instruct the large language model 7 to generate an answer sentence for the question sentence based on one or more pieces of event information 29 related to the question sentence of the visitor regarding the event. The answer sentence generated in response to this first prompt is provided to the visitor terminal device 4 of the visitor. Thereby, even when various questions are received from the visitor, an accurate answer sentence for the question sentence can be generated based on the highly accurate event information 29 as information regarding the event and provided to the visitor. Therefore, accurate information that meets the requirements of each individual can be provided to the visitors who visit the event.

[0055] Also, according to this embodiment, when one question sentence is acquired from one visitor, a first prompt is generated to instruct the generation of an answer sentence for the one question sentence based on the dialogue history information (dialogue history DB 26) corresponding to the one visitor and one or more pieces of event information 29 acquired for the one question sentence. Thereby, since the answer sentence of the large language model 7 is generated based on the previous question-and-answer exchanges, information that better meets the requirements of the visitor can be provided.

[0056] Next, a modified example of the above-described process for acquiring an answer sentence corresponding to the question sentence will be described.

[0057] FIG. 5 is a flowchart for explaining a modified example of the process of generating an answer to a visitor's question. The flowchart shown in FIG. 5 is obtained by replacing step ST145 in the flowchart shown in FIG. 3 with step ST145A, and the other steps are the same as those in the flowchart shown in FIG. 3.

[0058] In the modified example shown in FIG. 5, when the information processing apparatus 1 generates a first prompt so as to generate a response sentence for a first question sentence based on one piece of dialogue history information (ST145A), it determines whether a plurality of pieces of event information 29 including one or more pieces of event information 29 included in the one piece of dialogue history information and pieces of event information 29 acquired as having relevance to the one question sentence include a plurality of identical pieces of event information 29 (that is, whether the same piece of event information 29 overlaps). When a plurality of identical pieces of event information 29 are included in the plurality of pieces of event information 29, the information processing apparatus 1 generates a first prompt so that a part of the plurality of identical pieces of event information 29 is omitted. For example, the information processing apparatus 1 generates a first prompt so that at least a part of the remaining event information 29 (event information 29 corresponding to a question sentence after the oldest question sentence) excluding the event information 29 corresponding to the oldest question sentence among the plurality of identical pieces of event information 29 is omitted so that the process of the dialogue can be easily grasped correctly in the large language model 7.

[0059] FIG. 6 is a diagram showing an example of a first prompt, and shows an example in the case where the same piece of event information 29 has relevance to a plurality of question sentences. In the example of FIG. 6, the same piece of event information 29 is acquired as having relevance to the first question sentence and the seventh question sentence, respectively. In this case, the information processing apparatus 1 does not add the event information 29 having relevance to the seventh question sentence as "reference information 7" to the dialogue history information. Instead, the information processing apparatus 1 describes in the first prompt that the event information 29 (reference information 7) having relevance to the seventh question sentence is the same as the event information 29 (reference information 1) having relevance to the first question sentence.

[0060] According to this modified example, by preventing the event information 29 with the same content from being duplicated in the first prompt, the number of characters of the first prompt can be suppressed, so it is easier to avoid the situation where the number of characters (the number of tokens described later) of the first prompt reaches the limit of the processing of the large language model 7.

[0061] FIG. 7 is a flowchart for explaining another modified example of the process of generating an answer to a visitor's question. The flowchart shown in FIG. 7 is obtained by adding steps ST150 to ST160 to the flowchart shown in FIG. 6, and the other steps are the same as those in the flowchart shown in FIG. 6.

[0062] In the modified example shown in FIG. 7, the information processing apparatus 1 obtains an estimate of the number of tokens, which is the text processing unit in the large language model 7, for the first prompt generated in step ST145A (ST150). The estimate of the number of tokens can be approximately calculated using a library function or the like provided for the large language model 7. When the estimate of the number of tokens exceeds a predetermined threshold (Yes in ST155), the information processing apparatus 1 reduces the number of characters of the first prompt provided to the large language model 7 so that the estimate of the number of tokens is below the threshold (ST160). When it is determined in step ST155 that the estimate of the number of tokens of the first prompt is below the predetermined threshold (No in ST155), the information processing apparatus 1 proceeds to step ST165 and provides the information processing apparatus 1 to the large language model 7.

[0063] When reducing the number of characters of the first prompt in step ST160, for example, the information processing apparatus 1 may omit or delete past dialogue sentences that are considered to be of relatively low importance in the dialogue history.

[0064] Specifically, the information processing apparatus 1 may perform at least a part of the following (1) to (4). (1) Delete at least a part of the oldest one or more question sentences and the oldest one or more answer sentences in the first prompt. (2) Replace at least a part of the oldest one or more question sentences and the oldest one or more answer sentences in the first prompt with a summarized sentence. (3) Delete the remaining event information 29 excluding one or more event information 29 obtained according to the newest one or more question sentences included in the first prompt. Replace the remaining event information 29, excluding one or more pieces of event information 29 obtained according to the one or more latest question sentences included in the first prompt, with a summarized sentence.

[0065] When replacing the sentences (question sentences, answer sentences, event information 29) targeted in the above (1) to (4) with summarized sentences, the information processing apparatus 1 may have the large language model 7 summarize this targeted sentence.

[0066] Also, when the information processing apparatus 1 reduces the number of characters of the first prompt in step ST160, it may replace the entire first prompt generated in step ST145 (or ST145A) with a summarized sentence. Also in this case, the information processing apparatus 1 may have the large language model 7 summarize the first prompt.

[0067] Next, a process of recording the actions of visitors in the system shown in FIG. 1 and obtaining the matching degree between the visitors and the participants based on the action record will be described.

[0068] FIG. 8 is a diagram for explaining an outline of a process related to the collection of the action record of a visitor. The event operator prepares an admission ticket C1 to be distributed to each visitor. An optical code (QR code (registered trademark), barcode, etc.) including identification information (visitor ID) assigned in advance to each visitor is printed on the admission ticket C1. The name of the visitor, the name of the event, the venue name, the opening date, etc. may be printed on the admission ticket C1.

[0069] When the event operator receives a visitor, for example, at the entrance of the event venue, the operator terminal device 6 reads the optical code printed on the admission ticket C1 handed to the visitor, and transmits the visitor ID included in the read optical code from the operator terminal device 6 to the information processing apparatus 1. The information processing apparatus 1 generates action record information including the visitor ID read from the admission ticket C1 and registers it in the action record DB27. Thereby, it is recorded in the action record DB27 that the visitor has entered the event venue.

[0070] On the other hand, participants in the event (such as exhibitors) upload, using the participant terminal device 5, to the information processing device 1 the content (such as electronic documents in PDF format) to be provided to visitors at the event instead of printed materials. In this case, the information processing device 1 stores the content 30 uploaded from the participant terminal device 5 in the storage device 2. Further, the information processing device 1 generates content information indicating the association between the content 30 and the target (products, services, etc.) and the association between the content 30 and the participant in response to an instruction from the participant terminal device 5, and registers it in the content management DB 25.

[0071] The participant installs the panel C2 posted together with the target such as products at the event venue in their own exhibition booth or the like. The panel C2 has an optical code printed thereon that is required when a visitor who visits the participant's exhibition booth or the like acquires content (such as electronic documents) related to the products or the like displayed there. The participant uses the participant terminal device 5 to transmit to the information processing device 1 the identification information of the panel C2 (panel ID), the information of the target associated with the panel C2 (target ID, etc.), the information of the content associated with the panel C2 (content ID, etc.), the information of the installation location of the panel C2, and the like. In this case, the information processing device 1 generates panel information indicating the association between the panel ID received from the participant terminal device 5 and the target (products, services, etc.), the association between the panel ID and the content, and the association between the panel ID and the installation location of the panel C2, and registers it in the panel management DB 24.

[0072] Also, the participant reads, using the participant terminal device 5, the optical code printed on the admission ticket C1 of the visitor who was received at the exhibition booth or the like, and transmits the visitor ID included in the read optical code from the participant terminal device 5 to the information processing device 1. The information processing device 1 generates action record information including the visitor ID read from the admission ticket C1 and registers it in the action record DB 27. Thereby, it is recorded in the action record DB 27 that the visitor visited the participant's exhibition booth or the like.

[0073] In this case, the participant terminal device 5 transmits an instruction to permit the download, etc. of the specified content to the information processing device 1 for the visitors, and the information processing device 1 may generate action record information to permit the download, etc. of the content specified by this instruction for the visitors and register it in the action record DB 27. Thereby, it becomes possible for the participant to provide appropriate content to the visitors while hearing the requests of the visitors.

[0074] Also in this case, the information processing device 1 may acquire the matching degree between the visitor and the participant, which will be described later, and provide it to the participant terminal device 5. Thereby, the participant can proceed with the conversation with the visitor after grasping the matching degree provided from the information processing device 1.

[0075] During the event, the participant can acquire and view information regarding the visiting status of visitors to the exhibition booth, etc. (the number of visitors who visited, the number of content downloads, etc.) from the information processing device 1 by the participant terminal device 5 at any time. Thereby, the participant can grasp the visiting status of visitors in real time at their own exhibition booth, the display location of products, etc. Also, after the event, the participant can acquire, by the participant terminal device 5, the personal information of the visitors who visited their own exhibition booth, etc., and the personal information of the visitors who downloaded the content related to their own products, etc. from the server of the operator, etc.

[0076] Attendees carry an admission ticket C1 with an optical code containing the attendee ID at the event venue. When an attendee wants to ask a question about the event to the information processing device 1 or perform an operation to obtain content related to products, etc., the attendee reads the optical code printed on their own admission ticket C1 with the attendee terminal device 4, and based on the address information (such as URI) contained in the read optical code, accesses the web application server by the information processing device 1. The information processing device 1 authenticates the attendee operating the attendee terminal device 4 based on the optical code information (such as attendee ID) of the admission ticket C1 obtained from the attendee terminal device 4, and causes a dedicated screen for the attendee (a screen of the web application) to be displayed on the attendee terminal device 4.

[0077] When an attendee finds an object of interest (products, services, etc.) at the event venue, the attendee reads the optical code of the panel C2 installed near the display location from the web application screen of the attendee terminal device 4, and transmits the panel ID contained in the read optical code to the information processing device 1. When the information processing device 1 receives the panel ID of the panel C2 read on the web application screen of the attendee terminal device 4, it generates action record information indicating that the optical code of the panel C2 has been read by the attendee, and registers it in the action record DB27. The information processing device 1 permits the download, viewing, email transfer, etc. of the content associated with the panel ID for the attendee in whom this action record information is recorded. For example, the information processing device 1 displays a list of content permitted for download, etc. by the action record information on the web application screen of the attendee terminal device 4.

[0078] The visitor views a list of the content for which downloading, etc. has been permitted on the screen of the web application of the visitor terminal device 4, and selects the desired content from the list. The visitor terminal device 4 requests the information processing device 1 for the content related to the target selected by the visitor. In response to this request, the information processing device 1 reads out the content requested by the visitor from the storage device 2 and performs a process of providing it to the visitor terminal device 4 that is the request source. That is, the information processing device 1 performs a process of displaying the content on the visitor terminal device 4 that is the request source, or a process of causing the visitor terminal device 4 that is the request source to download the content. Further, when the transfer destination email address is specified by the visitor terminal device 4, the information processing device 1 performs a process of transferring the content read out from the storage device 2 (or information such as a URL for downloading the same) to the specified email address.

[0079] FIG. 9 is a flowchart for explaining an example of a process of collecting the behavior record of a visitor.

[0080] The operator terminal device 6 reads the optical code printed on the admission ticket C1 handed to the visitor at the entrance of the event venue or the like, and transmits the visitor ID included in the read optical code to the information processing device 1 (ST200). The information processing device 1 generates behavior record information including the visitor ID read from the admission ticket C1, and registers it in the behavior record DB27 (ST205). By this behavior record information, it is recorded that the visitor has entered the event venue.

[0081] The participant terminal device 5 reads the optical code of the admission ticket C1 of the visitor who has visited the exhibition booth or the like of the participant in the event venue, and transmits the visitor ID included in the read optical code to the information processing device 1 (ST210). The information processing device 1 generates behavior record information including the visitor ID read from the admission ticket C1, and registers it in the behavior record DB27 (ST215). By this behavior record information, it is recorded that the visitor has visited a specific participant's exhibition booth or the like.

[0082] When the participant terminal device 5 of one participant reads the visitor ID of one visitor from the admission ticket C1, the information processing device 1 acquires the matching degree between the one participant and the one visitor (ST220) and provides it to the participant terminal device 5 (ST225). The participant terminal device 5 displays the matching degree provided from the information processing device 1 (ST230). Thereby, when a participant interacts with a visitor at an exhibition booth or the like, the participant can proceed with the conversation while grasping the matching degree with that visitor. The process of acquiring the matching degree in step ST220 will be described later with reference to FIGS. 10 to 12.

[0083] The visitor terminal device 4 reads the optical code printed on the admission ticket C1 of the visitor and accesses the web application server by the information processing device 1 based on the address information included in the read optical code (ST240). The information processing device 1 authenticates the visitor operating the visitor terminal device 4 based on the optical code information (visitor ID, etc.) of the admission ticket C1 acquired from the visitor terminal device 4 and causes the visitor terminal device 4 to display a dedicated screen (web application screen) for the visitor (ST245).

[0084] The visitor terminal device 4 reads the optical code of the panel C2 posted together with the target (product, service, etc.) at the exhibition booth or the like visited by the visitor and transmits the panel ID included in the optical code to the information processing device 1 (ST250). When the information processing device 1 receives the panel ID of the panel C2 read on the web application screen of the visitor terminal device 4, the information processing device 1 generates action record information indicating that the optical code of the panel C2 has been read by the visitor and registers it in the action record DB27 (ST255). The information processing device 1 permits the visitor to download the content corresponding to this content ID based on the content ID associated with the panel ID in the action record information.

[0085] The information processing device 1 causes a list of content for which downloading etc. is permitted based on the action record information to be displayed on the screen of the web application of the visitor terminal device 4. The visitor selects specific content from among the available content displayed on the screen of the web application, and inputs an instruction to the visitor terminal device 4 to request that the content be provided by a specified method (such as browsing, downloading, email transfer, etc.). When this instruction from the visitor is input, the visitor terminal device 4 requests the information processing device 1 to provide the specific content by the specified method (ST260). In response to the request from the visitor terminal device 4, the information processing device 1 executes content providing processing to provide the selected specific content to the visitor by the specified method. For example, the information processing device 1 performs processing to display the content on the visitor terminal device 4, processing to download the content to the visitor terminal device 4, processing to transfer the content to an arbitrary address by email, etc., in response to the request from the visitor terminal device 4 operated by the visitor. When the information processing device 1 performs processing to provide content to the visitor in response to the request from the visitor terminal device 4, it generates action record information indicating that the visitor has received the content provision, and registers it in the action record DB27 (ST265).

[0086] Also, the visitor can input an evaluation of an arbitrary target (product, service, etc.) on the screen of the web application by the information processing device 1. When an evaluation of the target is input, for example, by pressing the "like" button, the visitor terminal device 4 transmits the evaluation to the information processing device 1 (ST270). When the information processing device 1 receives an evaluation of a single target input by a single visitor, it generates action record information indicating that the evaluation of the target has been input by the visitor, and registers it in the action record DB27 (ST275).

[0087] Furthermore, the visitors can answer the questionnaires created by participants, event organizers, etc. on the screen of the web application by the information processing apparatus 1. When, for example, an answer to a questionnaire regarding a specific target (product, service, etc.) is input to the visitor terminal device 4, the answer to the questionnaire is transmitted to the information processing apparatus 1 (ST280). When the information processing apparatus 1 receives an answer to a questionnaire input by one visitor, the information processing apparatus 1 generates action record information indicating that the answer to the questionnaire has been input by the visitor, and registers the action record information in the action record DB27 (ST285).

[0088] FIG. 10 is a flowchart for explaining an example of a process (ST220 in FIG. 9) of obtaining the matching degree between a visitor and a participant.

[0089] When the optical code of the admission ticket C1 is read in the participant terminal device 5 of one participant and the information processing apparatus 1 receives the visitor ID of one visitor from the participant terminal device 5, the information processing apparatus 1 performs a process (ST220) of obtaining the matching degree regarding the degree to which the one participant and the one visitor match. That is, when the information processing apparatus 1 receives a request (matching degree request) for obtaining the matching degree with one visitor from the participant terminal device 5 of one participant, the information processing apparatus 1 performs a process of obtaining the matching degree between the one participant and the one visitor.

[0090] In the process of obtaining the matching degree, the information processing apparatus 1 performs a process of obtaining a dialogue history evaluation, which is an evaluation of the matching degree based on the dialogue history of the visitor (ST300 to ST325). In this case, the information processing apparatus 1 obtains event information related to one participant designated as the partner for the matching degree from the storage device 2 (ST300). For example, the information processing apparatus 1 specifies one or more pieces of event information 29 associated with one participant ID in the vector information of the vector DB28, and obtains the specified one or more pieces of event information 29 from the storage device 2.

[0091] Further, the information processing apparatus 1 acquires the dialogue history information of one visitor designated as a matching partner from the dialogue history DB 26, and acquires the history of question sentences included in this dialogue history information (ST305).

[0092] Based on the event information 29 acquired in step ST300 and the history of question sentences acquired in step ST305, the information processing apparatus 1 generates a prompt (hereinafter sometimes referred to as the "second prompt") for instructing the large language model 7 to evaluate the matching degree between one participant and one visitor designated as a matching partner. For example, the information processing apparatus 1 acquires the name (company name, etc.) of one participant who is a matching partner from the participant information (participant DB 22). Then, the information processing apparatus 1 evaluates the degree of matching between this participant (company name, etc.) and the visitor who wrote the question sentence based on the event information 29 (ST300) and the history of question sentences (ST305), and generates a second prompt for instructing to represent the evaluation result as a numerical value within a predetermined range (for example, a numerical value from 0 to 1).

[0093] The information processing apparatus 1 provides the generated second prompt to the large language model 7 (ST315), and the large language model 7 generates an evaluation result of the matching degree corresponding to the second prompt (ST320). The information processing apparatus 1 acquires the evaluation result of the matching degree generated according to the second prompt from the large language model 7 as the "dialogue history evaluation" (ST325).

[0094] Further, in the process of acquiring the matching degree, the information processing apparatus 1 performs a process of acquiring a visitor attribute evaluation, which is an evaluation of the matching degree based on the attributes of the visitor (ST300, ST330 to ST350). When the information processing apparatus 1 performs the process of acquiring the visitor attribute evaluation, it uses the one or more pieces of event information 29 already acquired in ST300. Also, the information processing apparatus 1 acquires information (age, gender, occupation, position, etc.) regarding the attributes of this visitor from the visitor information (visitor DB 21) of one visitor designated as a matching target (ST330).

[0095] The information processing apparatus 1 generates a prompt (hereinafter sometimes referred to as the "third prompt") for instructing the large language model 7 to evaluate the matching degree between a certain participant designated as a matching partner and a certain visitor based on the event information 29 acquired in step ST300 and the information regarding the attributes of the visitors acquired in step ST330 (ST335). For example, the information processing apparatus 1 evaluates the degree of matching between a participant (such as a company name) and a visitor who wrote a question sentence based on the event information 29 (ST300) and the attributes of the visitor (ST330), and generates a third prompt for instructing to represent the evaluation result by a numerical value within a predetermined range (for example, a numerical value from 0 to 1).

[0096] The information processing apparatus 1 provides the generated third prompt to the large language model 7 (ST340), and the large language model 7 generates an evaluation result of the matching degree according to the third prompt (ST345). The information processing apparatus 1 acquires, from the large language model 7, the evaluation result of the matching degree generated according to the third prompt as the "visitor attribute evaluation" (ST350).

[0097] Furthermore, in the process of acquiring the matching degree, the information processing apparatus 1 calculates an index based on the action record information (action record DB27) (ST355, ST360).

[0098] That is, when a certain participant and a certain visitor are designated as matching partners, for each of one or more objects (products, services, etc.) related to the certain participant, the information processing apparatus 1 calculates a parameter (action parameter) regarding a predetermined action performed by the certain visitor based on the action record information. Also, the information processing apparatus 1 calculates, for each object, an interest index indicating the degree of interest of the visitor in the object based on a group of action parameters calculated for each object related to the certain participant. Then, the information processing apparatus 1 calculates an index (first matching index) related to the matching degree between the certain participant and the certain visitor based on the interest indexes calculated for each object related to the certain participant (ST355).

[0099] In addition, when one participant and one visitor are specified as the matching partners, the information processing apparatus 1 identifies, based on a plurality of pieces of target information registered in the target management DB 23, one or more targets related to the one participant that have the same category (such as the type of product or service) as the one or more targets related to the one participant, and one or more targets related to another participant different from the one participant. Then, for each of the one or more identified targets, the information processing apparatus 1 calculates action parameters for a predetermined action performed by the one visitor based on the action record information. Further, the information processing apparatus 1 calculates, for each target related to the one participant, an interest index indicating the degree of interest of the visitor in the target based on a group of action parameters calculated for each target related to the one participant. Then, the information processing apparatus 1 calculates an index (second matching index) related to the matching degree between the another participant and the one visitor based on the interest indexes calculated for each target related to the another participant (ST360).

[0100] FIG. 11 is a flowchart for explaining an example of a process (ST355: FIG. 10) of calculating a first matching index.

[0101] First, the information processing apparatus 1 identifies, in the target management DB 23, one or more targets (products, services, etc.) related to one participant specified as the matching partner (ST400). That is, the information processing apparatus 1 identifies, in the target management DB 23, target information including the participant ID of the one participant.

[0102] The information processing apparatus 1 selects one of the one or more targets identified in step ST400 (ST405), and acquires, from the action record DB 27, action record information regarding the selected one target and including the action record of the one visitor specified as the matching partner (ST410).

[0103] The information processing apparatus 1 calculates one or more action parameters (action parameter group) based on the action record information (record of the action performed by the one visitor regarding the one target) acquired from the action record DB 27 in step ST410 (ST415).

[0104] The action parameter is a parameter related to a predetermined action taken by a visitor, and includes the number of times a visitor has taken a predetermined action with respect to one object, the time spent by a visitor on a predetermined action with respect to one object, and the like. Specifically, the action parameter includes at least a part of the parameters as described below. · Presence or absence of reading an optical code of a panel related to one object · Number of times of viewing content related to one object · Number of times of downloading content related to one object · Number of times of email forwarding of content related to one object · Evaluation of one object (such as whether the "like" button on the website is pressed) · Presence or absence of answering a questionnaire related to one object · Presence or absence of entering a comment related to one object · Time spent staying at the display location of one object (Example) Elapsed time from the time when the optical code of a panel related to one object is read to the time when action record information related to another object is recorded

[0105] Based on one or more action parameters (action parameter group) calculated in step ST415, the information processing apparatus 1 calculates an "interest index", which is an index related to the degree of interest (degree of interest) of a visitor in one object (ST420). For example, the information processing apparatus 1 calculates, as the interest index, the sum of a plurality of products obtained by multiplying each of the plurality of action parameters calculated in step ST415 by a weight coefficient.

[0106] When the (n + 1) parameters (action parameters) calculated in step ST415 are "P0", "P1",..., "Pk",..., "Pn" (where k represents an integer satisfying 0 ≦ k ≦ n), the weight coefficient multiplied by the parameter Pk is "Wk", and the interest index is "I", the interest index I is expressed by the following formula (1).

[0107]

Equation

[0108] If there is an unselected target among the targets specified in step ST400 (Yes in ST425), the information processing apparatus 1 selects one of them (ST430) and repeats the processes of steps ST410 to ST420 described above. When the processes of steps ST410 to ST420 have been performed for all the targets specified in step ST400 (No in ST425), the information processing apparatus 1 proceeds to step ST435.

[0109] When the information processing apparatus 1 proceeds to step ST435, it calculates a first matching index, which is an index related to the matching degree between one participant and one visitor, based on the interest indices calculated for each target specified in step ST400. The first matching index may be, for example, the sum of the interest indices calculated for each target, or the sum of the products of the weight coefficients set for each target and the interest indices for all the targets.

[0110] FIG. 12 is a flowchart for explaining an example of the process (ST360: FIG. 10) of calculating the second matching index.

[0111] First, when one participant and one visitor are specified as the matching partners, the information processing apparatus 1 identifies one or more targets that have the same category as one or more targets related to the one participant and one or more targets related to another participant different from the one participant based on a plurality of target information registered in the target management DB 23 (ST500). That is, the information processing apparatus 1 acquires the target information of one or more targets related to the one participant from the target management DB 23 and identifies the categories (types of products or services, etc.) of the targets included in the acquired target information. Then, the information processing apparatus 1 identifies, in the target management DB 23, the target information that includes any of the identified categories and includes a participant ID different from the one participant.

[0112] After identifying one or more targets in step ST500, the information processing apparatus 1 calculates one or more interest indicators corresponding to the identified one or more targets by performing the processes of steps ST505 to ST530 similar to steps ST405 to ST430 (FIG. 11).

[0113] When the information processing apparatus 1 calculates one or more interest indicators corresponding to one or more targets in step ST500, based on these calculated interest indicators, it calculates a second matching indicator which is an indicator related to the degree of matching between another participant related to a target in the same category as a certain participant and a certain visitor. The second matching indicator may be, for example, the sum of the interest indicators calculated for each target, or the sum of the products of the weight coefficients set for each target and the interest indicators added up for all targets.

[0114] The second matching indicator is, for example, an indicator related to the degree of matching between another participant (a competitor) handling products in the same category as a certain participant designated as a matching partner and a visitor.

[0115] Return to FIG. 10. Based on the dialogue history evaluation acquired in step ST325, the visitor attribute evaluation acquired in step ST350, the first matching indicator acquired in step ST355, and the second matching indicator acquired in step ST360, the information processing apparatus 1 acquires the degree of matching between a certain participant and a certain visitor. For example, the information processing apparatus 1 may calculate, as the degree of matching, a value obtained by multiplying the values of these evaluations and indicators by respective predetermined weight coefficients and then summing them up.

[0116] Thus, according to this embodiment, based on the history of the questions asked by the visitors and the event information related to the participants, the large language model 7 gives an evaluation result of the matching degree (dialogue history evaluation), and the matching degree is obtained based on this dialogue history evaluation. Therefore, since it becomes possible to directly obtain the matching degree from the desires and needs of the visitors included in the questions, it becomes easier to obtain an appropriate matching degree that reflects the actual desires of the visitors.

[0117] Also, according to this embodiment, based on the attributes of the visitors and the event information related to the participants, the large language model 7 gives an evaluation result of the matching degree (visitor attribute evaluation), and the matching degree is obtained based on this visitor attribute evaluation. Therefore, since it is possible to obtain the matching degree taking into account the attributes of the visitors, it becomes easier to obtain the matching degree that reflects the potential desires of the visitors.

[0118] Also, according to this embodiment, based on the records of the actions of the visitors targeted at the objects (products, services, etc.) related to the participants, an index (interest index) regarding the degree of interest in the objects related to the participants is calculated for each object, and based on the interest index of each object, an index regarding the matching degree (first matching index) is calculated, and the final matching degree is obtained based on the first matching index. Therefore, it becomes easier to obtain a highly accurate matching degree based on the actions of the visitors in the event.

[0119] Also according to the present embodiment, objects (products, services, etc.) that are in the same category as the object related to one participant designated as the matching partner and that are related to one or more participants other than the said one participant are identified. Then, based on the records of the actions of visitors targeting the said one or more objects, an index (interest index) regarding the degree of interest in the objects related to the said other participant is calculated for each object, and based on the interest index of each object, an index (second matching index) regarding the matching degree with the said other participant is calculated, and the final matching degree is obtained based on the second matching index. Therefore, although no direct actions are observed regarding the object related to the said one participant, if there are visitors who are acting regarding objects that are in the same category as the object related to the said one participant and that are related to other participants, it is possible to correctly evaluate that a certain degree of matching is recognized between such visitors and the said one participant.

[0120] Note that the present invention is not limited only to the above-described embodiment and includes various variations.

[0121] In the process of the flowchart shown in FIG. 10, the matching degree is calculated based on two evaluation results (dialogue history evaluation, visitor attribute evaluation) by the large language model 7 and two matching indexes (first matching index, second matching index) based on the action records. However, in other embodiments of the present invention, the matching degree may be calculated based on one or more of these evaluations and the deviation of the matching indexes.

[0122] In the above-described embodiment, the function of the large language model is provided by a computer system (large language model 7) separate from the information processing apparatus 1, but at least a part of the function of the large language model may be realized in the information processing apparatus 1.

[0123] Part of the processing of the information processing apparatus 1 in the above-described embodiment may be executed by another apparatus (such as the visitor terminal apparatus 4). For example, when part of the above-described processing of the information processing apparatus 1 is executed in the visitor terminal apparatus 4, an information processing system including the computer of the information processing apparatus 1 and the computer of the visitor terminal apparatus 4 is configured, and it can be said that the processing according to this embodiment is executed in this information processing system.

[0124] The configuration of the databases (21 to 28) in the above-described embodiment is an example, and this embodiment is not limited to this example. That is, part of the above-described databases may be replaced with one or more other databases. Further, the information (such as dialogue history information and action history information) handled in the processing of this embodiment does not necessarily have to be included in the records of one database, and may be distributed, for example, in the records of two or more databases.

Explanation of Reference Numerals

[0125] 1... Information processing apparatus, 11... Communication unit, 12... Storage unit, 121... Program, 13... Processing unit, 2... Storage device, 21... Visitor DB, 22... Participant DB, 23... Target management DB, 24... Panel management DB, 25... Content management DB, 26... Dialogue history DB, 27... Action record DB, 28... Vector DB, 29... Event information, 30... Content, 4... Visitor terminal apparatus, 5... Participant terminal apparatus, 6... Operator terminal apparatus, 7... Large language model, 9... Communication network

Claims

1. A method for an information processing system to provide information related to an event to visitors, comprising: obtaining a question text of the visitor regarding the event; obtaining, from a plurality of event information regarding the event, one or more pieces of the event information having relevance to the question text; generating a first prompt for instructing a large language model to generate an answer text for the question text based on the obtained one or more pieces of the event information; performing a process of providing the answer text generated by the large language model in response to the first prompt to the visitor; The method comprising the above steps.

2. Each of the plurality of pieces of the event information is associated with one vector representing the content of the information, the vector representing the content of the question text is referred to as a question text vector, the step of obtaining the event information includes obtaining, from the plurality of pieces of the event information, one or more pieces of the event information associated with a vector similar to the question text vector, The method according to claim 1.

3. The information processing system is accessible to a storage device, the storage device stores a plurality of dialogue history information corresponding to a plurality of the visitors, one of the dialogue history information includes one or more of the question texts by one visitor, one or more of the event information obtained for the one or more question texts, and one or more of the answer texts generated by the large language model for the one or more question texts, when generating the first prompt for one of the question texts by one visitor, the step of generating the first prompt includes generating the first prompt for instructing the large language model to generate an answer text for the one question text based on the dialogue history information corresponding to the one visitor and the one or more pieces of the event information obtained for the one question text, The method according to claim 1.

4. when the question text by one visitor is obtained, adding the obtained question text to the dialogue history information corresponding to the one visitor; when the event information for the question text by one visitor is obtained, adding the obtained event information to the dialogue history information corresponding to the one visitor; When the answer sentence to the question sentence by one of the visitors is generated by the large language model, a step of adding the generated answer sentence to the dialogue history information corresponding to the one visitor is included. The method according to claim 3.

5. When the step of generating the first prompt generates the first prompt so as to generate an answer sentence to one question sentence based on one piece of the dialogue history information, if a plurality of the same event information is included in the plurality of event information including the one or more pieces of event information included in the one piece of dialogue history information and the event information obtained as having relevance to the one question sentence, the step of generating the first prompt includes generating the first prompt so that a part of the plurality of the same event information is omitted. The method according to claim 4.

6. The step of generating the first prompt includes obtaining an estimate of the number of tokens, which is a text processing unit in the large language model, for the generated first prompt, and when the estimated number of tokens exceeds a predetermined threshold, reducing the number of characters of the first prompt provided to the large language model so that the estimated number of tokens is below the threshold. The method according to claim 4.

7. In the step of generating the first prompt, reducing the number of characters of the first prompt provided to the large language model includes deleting at least a part of the oldest one or more question sentences and the oldest one or more answer sentences in the first prompt; replacing at least a part of the oldest one or more question sentences and the oldest one or more answer sentences in the first prompt with a summarized sentence; deleting the remaining event information except for one or more pieces of event information obtained according to the newest one or more question sentences included in the first prompt; replacing the remaining event information except for one or more pieces of event information obtained according to the newest one or more question sentences included in the first prompt with a summarized sentence; replacing the whole of the first prompt with a summarized sentence including at least one of. The method according to claim 6.

8. At least a part of the plurality of pieces of event information includes information related to one or more participants participating in the event. When receiving a matching degree requirement for obtaining the degree of matching between one of the participants and one of the visitors, based on one or more pieces of the event information related to the one participant and at least the history of the question texts included in the conversation history information of the one visitor, a second prompt is generated to instruct the large language model to evaluate the matching degree between the one participant and the one visitor. Based on the conversation history evaluation which is the evaluation result of the matching degree generated by the large language model according to the second prompt, a step of obtaining the matching degree. A step of performing a process of providing the obtained matching degree to the requester of the matching degree requirement. The method according to claim 3.

9. The storage device stores a plurality of visitor information corresponding to a plurality of the visitors. One piece of the visitor information includes information regarding the attributes of one of the visitors. When receiving the matching degree requirement regarding the matching degree between one of the participants and one of the visitors, based on one or more pieces of the event information related to the one participant and the information regarding the attributes of the one visitor included in the visitor information of the one visitor, a step of generating a third prompt to instruct the large language model to evaluate the matching degree between the one participant and the one visitor. The step of obtaining the matching degree includes obtaining the matching degree based on the visitor attribute evaluation which is the evaluation result of the matching degree generated by the large language model according to the third prompt and the conversation history evaluation. The method according to claim 8.

10. The visitor can obtain information regarding a plurality of objects in the event. A plurality of participants participating in the event are each related to at least one of the objects. The storage device stores a plurality of action record information. One piece of the action record information is information including a record regarding a predetermined action aimed at one of the objects performed by one of the visitors in one or more of the events. A parameter regarding the predetermined action taken by one visitor with respect to one of the said objects, which is a parameter indicating the number of times of the predetermined action taken by one visitor with respect to one of the said objects, and / or a parameter indicating the time spent on the said predetermined action is called an action parameter. The degree of interest of one visitor in one of the said objects is called the degree of interest. An index regarding the degree of interest is called an interest index. The interest index is calculated based on a group of action parameters which are one or more of the said action parameters. When receiving the matching degree requirement regarding the matching degree between one participant and one visitor, for each of one or more of the said objects related to the one participant, calculate the group of action parameters based on one or more of the said action record information including the record of the predetermined action taken by the one visitor, calculate one or more of the interest indexes corresponding to the one or more of the said objects based on the group of action parameters calculated for each of the one or more of the said objects, and calculate an index related to the matching degree between the one participant and the one visitor as the first matching index based on the one or more of the calculated interest indexes. The step of obtaining the matching degree includes obtaining the matching degree based on the first matching index and the dialogue history evaluation. The method according to claim 8.

11. Each of the said objects is classified into one of a plurality of predetermined categories. The storage device includes a plurality of object information corresponding to a plurality of the said objects. Each of the said object information includes the category into which the object is classified. When receiving the matching degree requirement for the matching degree between one of the participants and one of the visitors, the step of calculating the index is based on a plurality of pieces of the target information to identify one or more targets related to the one participant that are the same as the category, and one or more targets related to another participant different from the one participant. For each of the identified one or more targets, the action parameter group is calculated based on one or more pieces of the action record information including the record of the predetermined action performed by the one visitor, and based on the action parameter group calculated for each of the one or more targets, one or more interest indexes corresponding to the one or more targets are calculated. Based on the calculated one or more interest indexes, an index related to the matching degree between the another participant and the one visitor is calculated as a second matching index. The step of obtaining the matching degree includes obtaining the matching degree based on the first matching index, the second matching index, and the dialogue history evaluation. The method according to claim 10.

12. The method includes a step of obtaining action record information including a record of the predetermined action performed by the visitor in the event being held and storing the action record information in the storage device. The method according to claim 10.

13. A program including an instruction for causing an information processing system to perform a process of providing information related to an event to a visitor, The process performed by the information processing system according to the instruction includes each step of the method described in any one of claims 1 to 12. Program.

14. An information processing system that performs a process of providing information related to an event to a visitor, A processing unit; It has a storage unit that stores instructions executed in the processing unit, The process performed by the processing unit according to the instruction includes each step of the method described in any one of claims 1 to 12. Information processing system.

15. An information processing system that performs a process of providing information related to an event to a visitor, It is provided with means for performing each step of the method described in any one of claims 1 to 12. Information processing system.

Images