Information processing device, information processing system, information processing method, and program

Abstract

To provide an information processing device capable of improving the accuracy of processing from source text, an information processing system, an information processing method, and a program.SOLUTION: An information processing device comprises: a source text acquisition unit which acquires the source text input by an administrator; a guideline information acquisition unit which acquires guideline information related to the processing of the source text acquired by the source text acquisition unit from a storage unit; a processed text acquisition unit which acquires processed text obtained by processing the source text acquired by the source text acquisition unit by transmitting the source text acquired by the source text acquisition unit and the guideline information acquired by the guideline information acquisition unit to a processing model; and an evaluation unit which evaluates the processed text acquired by the processed text acquisition unit by transmitting the guidance information acquired by the guideline information acquisition unit and the processed text acquired by the processed text acquisition unit to an evaluation model.SELECTED DRAWING: Figure 9

Description

Technical Field

[0001] The present invention relates to an information processing apparatus, an information processing system, an information processing method, and a program.

Background Art

[0002] Conventionally, for example, as shown in Patent Document 1, there is known an information processing system that obtains a summary sentence as a processed sentence by inputting a product description sentence as the original text into a learning model on a product sales page on an e-commerce site. In such an e-commerce site, a product description sentence is input by an administrator of the product to be traded.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in such an information processing system, depending on the original text input by the administrator, it may not be processed as desired in accordance with the operation of the information processing system. For this reason, it is desired to improve the accuracy of processing from the original text.

Means for Solving the Problems

[0005] The information processing apparatus for solving the above problems includes an original text acquisition unit that acquires the original text input by an administrator, a guideline information acquisition unit that acquires guideline information regarding the processing of the original text acquired by the original text acquisition unit from a storage unit, a processed text acquisition unit that acquires a processed text obtained by processing the original text acquired by the original text acquisition unit by transmitting the original text acquired by the original text acquisition unit and the guideline information acquired by the guideline information acquisition unit to a processing model, and an evaluation unit that evaluates the processed text acquired by the processed text acquisition unit by transmitting the original text acquired by the original text acquisition unit, the guideline information acquired by the guideline information acquisition unit, and the processed text acquired by the processed text acquisition unit to an evaluation model.

[0006] The information management system for solving the above problems includes an original text acquisition unit that acquires the original text input by an administrator, a guideline information acquisition unit that acquires guideline information regarding the processing of the original text acquired by the original text acquisition unit from a storage unit, a processed text acquisition unit that acquires a processed text obtained by processing the original text acquired by the original text acquisition unit by transmitting the original text acquired by the original text acquisition unit and the guideline information acquired by the guideline information acquisition unit to a processing model, and an evaluation unit that evaluates the processed text acquired by the processed text acquisition unit by transmitting the original text acquired by the original text acquisition unit, the guideline information acquired by the guideline information acquisition unit, and the processed text acquired by the processed text acquisition unit to an evaluation model.

[0007] The information management method for solving the above problems is executed by one or more computers to acquire the original text input by an administrator, acquire the guideline information regarding the processing of the original text from a storage unit, acquire a processed text obtained by processing the original text by transmitting the original text and the guideline information to a processing model, and evaluate the processed text by transmitting the original text, the guideline information, and the processed text to an evaluation model.

[0008] A program for solving the above problems causes one or more computers to acquire the original text input by an administrator, acquire pointer information regarding the processing of the original text from a storage unit, send the original text and the pointer information to a processing model to acquire a processed text obtained by processing the original text, and send the original text, the pointer information, and the processed text to an evaluation model to evaluate the processed text.

Effect of the Invention

[0009] According to the present invention, the accuracy of processing from the original text can be improved.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Mode for Carrying Out the Invention

[0011] [First Embodiment] An embodiment of an information processing apparatus, an information processing system, an information processing method, and a program will be described.

[0012] <Configuration of Information Processing System 10> As shown in FIG. 1, the information processing system 10 is a system that processes information. The information processing system 10 is a system that provides information to users who use the information processing system 10. The information processing system 10 is a system that controls services provided to users.

[0013] The information processing system 10 includes an information processing apparatus 11. The information processing apparatus 11 is an apparatus that processes information. The information processing apparatus 11 controls information provided to users. The information processing apparatus 11 controls services provided to users. The information processing apparatus 11 is managed by an administrator of the information processing apparatus 11. The information processing apparatus 11 is managed by an operator of the information processing system 10.

[0014] The information processing system 10 may include a travel management server 12. The information processing system 10 may include a plurality of travel management servers 12. The travel management server 12 is a server that provides travel services to users. The travel management server 12 is a server that manages information related to travel. The travel management server 12 manages information input by an administrator of a transaction target. The administrator of the transaction target is a person who provides travel services. To give a specific example, the transaction target includes hotels, restaurants, and travel documents, and transactions include reservations, purchases, sales, rentals, and uses.

[0015] The plurality of travel management servers 12 may include a first travel management server 12A. The first travel management server 12A is a server that manages information related to travel in at least the first country. The first travel management server 12A is a server that manages information related to travel in which the first language is used. For example, the first country may be Japan and the first language may be Japanese.

[0016] The plurality of travel management servers 12 may include a second travel management server 12B. The second travel management server 12B may be a server that manages information related to travel in a second country. The second travel management server 12B may be a server that manages information related to travel in a first country. The second travel management server 12B is a server that manages information related to travel in which a second language is used. For example, the second country may be the United States of America and the second language may be English.

[0017] The information processing system 10 may include a large language model (LLM) server 17. Hereinafter, the large language model server 17 is referred to as the LLM server 17. The information processing system 10 may include a plurality of LLM servers 17. The plurality of LLM servers 17 may include a first LLM server 17A and a second LLM server 17B.

[0018] The LLM server 17 is a server equipped with a large language model. A large language model is a language model constructed by a large amount of data and deep learning technology. The LLM server 17 has a function of controlling various texts using generative artificial intelligence. The LLM server 17 is equipped with various learning models of generative artificial intelligence.

[0019] Taking a specific example, the LLM server 17 has a function of summarizing the input text. The LLM server 17 is equipped with a summarization model using generative artificial intelligence. The summarization model is a learning model related to text summarization. The summarization model corresponds to an example of a processing model. The processing model is a learning model related to text processing.

[0020] The LLM server 17 has a function of translating the input text into a predetermined language. The LLM server 17 has a translation model using generative artificial intelligence. The translation model is a learning model related to text translation.

[0021] The LLM server 17 has a function of evaluating the input text by comparing the input text. The LLM server 17 has an evaluation model using generative artificial intelligence. The evaluation model is a learning model that evaluates the input text by comparing the input text.

[0022] The information processing system 10 may include one or more terminal devices 18. As a specific example, the information processing system 10 may include a user terminal device. The user terminal device is a terminal device used by a user. The user terminal device may be a portable terminal device that can be carried by the user. The user is also a person who has a trading plan with the trading partner.

[0023] The information processing device 11, one or more travel management servers 12, one or more LLM servers 17, and one or more terminal devices 18 are communicably connected via a network 19. At least any combination of the information processing device 11, one or more travel management servers 12, one or more LLM servers 17, and one or more terminal devices 18 may be directly communicably connected without passing through the network 19. Hereinafter, the description of passing through the network 19 for communication between each device will be omitted.

[0024] <Configuration of the terminal device 18> As shown in FIG. 2, the terminal device 18 is composed of one or more computers. The terminal device 18 includes a terminal control unit 20, a terminal storage unit 21, a terminal communication unit 22, a terminal input unit 23, and a terminal display unit 24.

[0025] The terminal control unit 20 is configured to control the terminal device 18. The terminal control unit 20 includes an arithmetic unit and a main storage medium. The arithmetic unit loads an OS (Operating System) and a program from the terminal storage unit 21 to the main storage medium, and executes the instructions retrieved from the main storage medium. The arithmetic unit is one or more circuitry. The circuitry may be a CPU (Central Processing Unit), a GPU (Graphic Processing Unit), or an NPU (Neural network Processing Unit). The terminal storage unit 21 is a secondary storage medium and may store a program. The program may be a dedicated application for using the information processing system 10.

[0026] The terminal communication unit 22 is implemented as hardware, software, or a combination thereof. The terminal communication unit 22 transmits and receives data to and from the information processing device 11. The terminal input unit 23 is configured to input information in response to a user operation. The terminal input unit 23 may be a touch panel integrated with the terminal display unit 24. The terminal input unit 23 may be a pointing device of an operation button. The terminal display unit 24 displays information in response to an output instruction from the terminal control unit 20.

[0027] <Configuration of the information processing device 11> As shown in FIG. 3, the information processing device 11 is composed of one or more computers. The information processing device 11 includes a control unit 30, a storage unit 31, and a communication unit 32. The control unit 30 is configured to control the information processing device 11. The control unit 30 includes an arithmetic unit and a main storage medium. The control unit 30 has the same configuration as the terminal control unit 20, and the description of the arithmetic unit and the main storage medium is omitted. The storage unit 31 is a secondary storage medium and stores information for managing the information processing device 11.

[0028] The communication unit 32 is implemented as hardware, software, or a combination thereof. The communication unit 32 transmits and receives data to and from one or more travel management servers 12 and one or more LLM servers 17. The information processing apparatus 11 may include an operation unit for a provider who provides information to perform an input operation, and may include a display unit for displaying information.

[0029] The travel management server 12 and the LLM server 17 are configured in the same manner as the information processing apparatus 11. That is, the travel management server 12 and the LLM server 17 have a configuration similar to the control unit 30, the storage unit 31, and the communication unit 32 in the information processing apparatus 11. The description of the configuration of the travel management server 12 and the LLM server 17 is omitted.

[0030] <Processing Circuit of Information Processing Apparatus 11> In the information processing apparatus 11, the control unit 30 constitutes a processing circuit by executing a program. The control unit 30 includes, as a processing circuit, a description text acquisition unit 30A, a pointer information acquisition unit 30B, an information transmission unit 30C, and an information reception unit 30D. The control unit 30 includes, as a processing circuit, a summary text acquisition unit 30E, a translation text acquisition unit 30F, an evaluation unit 30G, an information providing unit 30H, and a pointer correction unit 30I.

[0031] The description text acquisition unit 30A acquires the description text of the transaction target. Hereinafter, the description text of the transaction target is simply referred to as the description text. The description text is input by the administrator of the transaction target. The description text is an example of the original text before processing. That is, the description text acquisition unit 30A acquires the original text. The description text acquisition unit 30A corresponds to an example of the original text acquisition unit.

[0032] The description text acquisition unit 30A may acquire the description text of the travel service input by the administrator of the transaction target from the first travel management server 12A. That is, the description text acquisition unit 30A may acquire the description text of the travel service regarding the travel in the first country and in the first language.

[0033] The pointer information acquisition unit 30B acquires pointer information from the storage unit 31. The pointer information is information indicating pointers regarding the summary of the description text. That is, the pointer information is information indicating pointers regarding the processing of the original text. The pointer information does not include the information itself that requests the summary of the description text, but includes information indicating pointers regarding the content of the summary of the description text. That is, the pointer information does not include the information itself that requests the processing of the original text, but includes information indicating pointers regarding the content of the processing of the original text. The pointer information includes information regarding the selection of the description text. The pointer information includes information regarding the order of the description text. The pointer information includes information regarding the sentence expression of the description text.

[0034] The pointer information is information set by the operator of the information processing system 10, different from the administrator of the transaction target who input the description text. The pointer information is information set by the administrator of the information processing device 11.

[0035] The information transmission unit 30C is configured to transmit information. The information transmission unit 30C is configured to transmit information to the LLM server 17. To give a specific example, the information transmission unit 30C transmits the description text and the pointer information to the LLM server 17. Thereby, the information transmission unit 30C transmits the description text and the pointer information to the summary model of the generative artificial intelligence. The information transmission unit 30C may be configured to transmit information via an API (application programming interface).

[0036] The information reception unit 30D is configured to receive information. When the information transmission unit 30C transmits information to the LLM server 17, the information reception unit 30D is configured to receive information from the LLM server 17. To give a specific example, when the information transmission unit 30C transmits the description text and the pointer information to the LLM server 17, the information reception unit 30D receives the summary text obtained by summarizing the description text from the LLM server 17. The information reception unit 30D may be configured to receive information via an API (application programming interface).

[0037] The abstract text acquisition unit 30E acquires the abstract text of the transaction target that summarizes the description text. Hereinafter, the abstract text of the transaction target is simply referred to as the abstract text. The abstract text is an example of the processed text after processing. That is, the abstract text acquisition unit 30E acquires the processed text obtained by processing the original text. The abstract text acquisition unit 30E corresponds to an example of the processed text acquisition unit.

[0038] In particular, the abstract text acquisition unit 30E acquires the abstract text that summarizes the description text by sending the description text and the pointer information to the abstract model of the generative artificial intelligence. Also, the abstract text acquisition unit 30E may acquire the abstract text step by step. Taking a specific example, when the abstract text acquisition unit 30E acquires the first abstract text that summarizes the description text, the second abstract text that summarizes the description text may be acquired by sending the first abstract text and the pointer information to the abstract model of the generative artificial intelligence.

[0039] The translated text acquisition unit 30F acquires the translated text of the transaction target that translates the abstract text. That is, the translated text acquisition unit 30F acquires the translated text obtained by translating the processed text. In particular, the translated text acquisition unit 30F acquires the translated text of the transaction target that translates the abstract text by sending the abstract text to the translation model of the generative artificial intelligence.

[0040] Taking a specific example, the translated text acquisition unit 30F sends the abstract text of the travel service in the first language regarding the travel in the first country to the translation model of the generative artificial intelligence, and acquires the translated text of the travel service in the second language regarding the travel in the first country.

[0041] The evaluation unit 30G evaluates the abstract text acquired based on the pointer information. The evaluation unit 30G evaluates the abstract text acquired based on the pointer information by sending the description text, the pointer information, and the abstract text to the evaluation model of the generative artificial intelligence. That is, the evaluation unit 30G evaluates the processed text acquired based on the pointer information by sending the original text, the pointer information, and the processed text to the evaluation model of the generative artificial intelligence.

[0042] The evaluation unit 30G may perform an evaluation of the first summary text obtained based on the guideline information by transmitting the description text, the guideline information, and the first summary text to the evaluation model of the generative artificial intelligence. The evaluation unit 30G may perform an evaluation of the second summary text obtained based on the guideline information by transmitting the description text, the guideline information, and the second summary text to the evaluation model of the generative artificial intelligence. The evaluation unit 30G may perform an evaluation of the second summary text obtained based on the guideline information by transmitting the first summary text, the guideline information, and the second summary text to the evaluation model of the generative artificial intelligence. The evaluation unit 30G may perform an evaluation of the second summary text obtained based on the guideline information by transmitting the description text, the first summary text, the guideline information, and the second summary text to the evaluation model of the generative artificial intelligence.

[0043] The information providing unit 30H provides the summary text so that it can be viewed by the intended user of the summary text. That is, the information providing unit 30H provides the processed text so that it can be viewed. The information providing unit 30H provides the summary text so that it can be viewed by the intended user of the transaction target for each specified item. The specified item may be an item corresponding to the guideline item described later included in the guideline information.

[0044] In particular, the information providing unit 30H may transmit the summary text to the second travel management server 12B so that it can be viewed by the intended user of the transaction target. That is, the information providing unit 30H may provide the translated text in the second language of the transaction target obtained by translating the summary text in the first language so that it can be viewed. In this way, when the description text acquisition unit 30A acquires the description text in the first language from the first travel management server 12A, the information providing unit 30H can transmit the translated text in the second language of the transaction target to the second travel management server 12B. That is, the information providing unit 30H may provide the translated text of the summary text translated into a different language so that it can be viewed by the intended user of the transaction target.

[0045] The guideline modification unit 30I modifies the guideline information. In particular, the guideline modification unit 30I modifies the guideline information regarding the summary of the description text based on the result evaluated by the evaluation unit 30G. That is, the guideline modification unit 30I modifies the guideline information regarding the processing of the original text based on the result evaluated by the evaluation unit 30G.

[0046] <Information processing of information processing system 10> Here, with reference to FIG. 4, the information processing of the information processing system 10 will be described. As shown in FIG. 4, in the information processing system 10, the first travel management server 12A manages explanatory texts regarding the first country. In particular, the first travel management server 12A manages in a browsable manner the explanatory texts of the travel services of the first country in the first language. The first travel management server 12A provides the explanatory texts of the travel services of the first country in the first language to the transaction prospective party in response to a request from the terminal device 18 held by the transaction prospective party.

[0047] The second travel management server 12B manages explanatory texts regarding the second country. In particular, the second travel management server 12B manages in a browsable manner the explanatory texts of the travel services of the second country in the second language. The second travel management server 12B provides the explanatory texts of the travel services of the second country in the second language to the transaction prospective party in response to a request from the terminal device 18 held by the transaction prospective party.

[0048] The information processing device 11 manages pointer information. The information processing device 11 acquires explanatory texts from the first travel management server 12A through communication with the first travel management server 12A. The information processing device 11 transmits a summarization request for requesting summarization, the explanatory texts, and the pointer information to the LLM server 17.

[0049] The LLM server 17 manages a summarization model. When the LLM server 17 receives a summarization request from the information processing device 11, it inputs the summarization request, the explanatory texts, and the pointer information into the summarization model of the generative artificial intelligence. As a result, the explanatory texts and the pointer information are input into the summarization model of the generative artificial intelligence. The LLM server 17 transmits the summarization result output from the summarization model to the information processing device 11. The summarization result includes a summary text in the first language.

[0050] The information processing device 11 receives the summary result from the LLM server 17. The information processing device 11 receives the summary sentence in the first language included in the summary result from the LLM server 17. Thereby, the information processing device 11 obtains the summary sentence by transmitting the explanatory text and the pointer information to the summary model of the generative artificial intelligence.

[0051] The information processing device 11 may transmit the summary request, the pointer information, and the first summary sentence obtained from the LLM server 17 to the LLM server 17. In this case, the information processing device 11 obtains the second summary sentence by transmitting the first summary sentence and the pointer information to the summary model of the generative artificial intelligence in the same manner as when the first summary sentence is obtained.

[0052] The information processing device 11 transmits a translation request that requests translation, the summary sentence obtained from the LLM server 17, and information indicating the language after translation to the LLM server 17. In this case, the information processing device 11 transmits the translation request, the summary sentence in the first language obtained from the LLM server 17, and information indicating the second language as the language after translation to the LLM server 17.

[0053] The LLM server 17 manages the translation model. When the LLM server 17 receives a translation request from the information processing device 11, the LLM server 17 inputs the translation request, the summary sentence, and the language after translation into the translation model of the generative artificial intelligence. Thereby, the summary sentence is input into the translation model of the generative artificial intelligence. The LLM server 17 transmits the translation result output from the translation model to the information processing device 11. The translation result includes the translated sentence of the trading target in the second language.

[0054] The information processing device 11 receives the translation result including the translated sentence from the LLM server 17. The information processing device 11 receives the translation result including the translated sentence of the trading target in the second language from the LLM server 17. Thereby, the information processing device 11 obtains the translated sentence by transmitting the summary sentence to the translation model of the generative artificial intelligence. The information processing device 11 transmits the obtained translated sentence of the trading target to the second travel management server 12B.

[0055] The second travel management server 12B receives a translated text of the transaction target from the information processing device 11. The second travel management server 12B manages in a browsable manner a translated text of the travel service of the first country in the second language, in addition to the explanatory text of the travel service of the second country in the second language. The translated text of the travel service of the first country in the second language is a text in which the summary text of the travel service of the first country is translated into the second language. That is, the second travel management server 12B manages in a browsable manner the summary text of the travel service of the first country translated into the second language. The second travel management server 12B provides the summary text of the travel service of the first country in the second language to the transaction planner in response to a request from the terminal device 18 held by the transaction planner.

[0056] The information processing device 11 transmits an evaluation request for requesting an evaluation, an explanatory text, pointer information, and a summary text to the LLM server 17. The information processing device 11 may transmit the evaluation request, the explanatory text, the pointer information, and the first summary text to the LLM server 17. The information processing device 11 may transmit the evaluation request, the explanatory text, the pointer information, and the second summary text to the LLM server 17. The information processing device 11 may transmit the evaluation request, the first summary text, the pointer information, and the second summary text to the LLM server 17. The information processing device 11 may transmit the evaluation request, the explanatory text, the pointer information, the first summary text, and the second summary text to the LLM server 17.

[0057] The LLM server 17 manages an evaluation model. When the LLM server 17 receives an evaluation request from the information processing device 11, it inputs the evaluation request, the explanatory text, the pointer information, and the summary text into the evaluation model of the generative artificial intelligence. As a result, the explanatory text, the pointer information, and the summary text are input into the evaluation model of the generative artificial intelligence. The LLM server 17 transmits the evaluation result output from the evaluation model to the information processing device 11.

[0058] The information processing device 11 receives the evaluation result from the LLM server 17. As a result, the information processing device 11 obtains an evaluation result for the summary text summarized from the explanatory text based on the pointer information by transmitting the explanatory text, the pointer information, and the summary text to the evaluation model of the generative artificial intelligence. The information processing device 11 may correct the pointer information based on the evaluation result.

[0059] <Pointer information> Next, an example of the pointer information will be described in detail with reference to FIG. 5. As shown in FIG. 5, the pointer information is stored in the storage unit 31. The pointer information includes a plurality of pieces of information indicating pointers regarding the summary. In the pointer information, a pointer type and a pointer item are associated with each other. The pointer type is information indicating the type of the pointer regarding the summary. The pointer item is information indicating the item of the pointer regarding the summary.

[0060] The pointer information includes first pointer information 31A and second pointer information 31B. The first pointer information 31A includes pointer information regarding the selection of the explanatory text. The first pointer information 31A includes pointer information regarding the order of the explanatory text. The first pointer information 31A includes pointer information regarding the prescribed items for summarizing the summary text. The second pointer information 31B includes pointer information regarding the sentence expression of the explanatory text.

[0061] Taking a specific example, in the first pointer information 31A, the pointer types include maintenance and deletion. Maintenance is a pointer type indicating leaving the text for the pointer item. The pointer items associated with maintenance may include meal content, privileges, and services. Deletion is a pointer type indicating not leaving the text for the pointer item. The pointer items associated with deletion may include overall facilities, customer comments, and the entire facility.

[0062] In the first pointer information 31A, the pointer type includes order. In the order, the order of the pointer items is determined. The order may be in the order of meal content, privileges, and services. In the order, it is determined to summarize the pointer items as prescribed items.

[0063] In the second pointer information 31B, the pointer type includes sentence expression. The sentence expression is a pointer type indicating modifying the sentence expression. The pointer items associated with the sentence expression may include subject complement, metaphorical expression, and region-specific expression. The region-specific expression is an expression considering the cultural background of the region.

[0064] <Specific Example of Description, Abstract, and Translation> Next, with reference to FIG. 6, a specific example of a description, an abstract, and a translation will be described. As shown in FIG. 6, as a specific example of the description, the first paragraph 40 can be cited. The first paragraph 40 is a paragraph input by the administrator of the transaction target that provides travel services. The first paragraph 40 includes paragraphs 41 to 48.

[0065] The first paragraph 40 includes paragraphs 41 to 43 about benefits. The first paragraph 40 includes paragraph 44 about the overall facilities of the hotel. The first paragraph 40 includes paragraphs 45 to 47 about the meal content. The first paragraph 40 includes paragraph 48 about customer feedback.

[0066] As a specific example of the abstract, the second paragraph 50 can be cited. The second paragraph 50 is a paragraph summarized from the first paragraph 40 based on the guideline information. The second paragraph 50 includes paragraphs 51, 52, 55 to 57.

[0067] The second paragraph 50 includes paragraphs 55 to 57 about the meal content. The second paragraph 50 includes paragraphs 51, 52 about benefits. Thus, the description whose guideline type is maintained remains in the abstract based on the guideline information.

[0068] The second paragraph 50 does not include paragraph 44 about the overall facilities of the hotel. The second paragraph 50 does not include paragraph 48 about customer feedback. Thus, the description whose guideline type is deleted does not remain in the abstract based on the guideline information.

[0069] The second paragraph 50 is a paragraph that summarizes paragraphs 55 to 57 about the meal content as a prescribed item. The second paragraph 50 is a paragraph that summarizes paragraphs 51, 52 about benefits as a prescribed item. Thus, the abstract is adjusted for each prescribed item based on the guideline information.

[0070] The second text 50 is a text whose order of the texts 55 - 57 about the meal content and the texts 51, 52 about the privileges has been adjusted. Specifically, the texts 55 - 57 about the meal content have been adjusted to an order prior to the texts 51, 52 about the privileges. Thus, the summary text is adjusted according to the order defined as the pointer type based on the pointer information.

[0071] The second text 50 is a text in which the subject has been complemented from the first text 40. The second text 50 is a text in which the abbreviations have been complemented from the first text 40. Specifically, text 52 is a text adjusted from the expression "with limited-time privilege for consecutive stays" in text 42 as an explanatory text by complementing the abbreviation and using the expression "with limited-time privilege for consecutive stayers" with the complemented subject "consecutive stayers".

[0072] The second text 50 is a text in which the metaphorical expression has been adjusted from the first text 40. Specifically, text 51 is a text adjusted from the expression "with a privilege friendly to women" in text 41 as an explanatory text to the expression "with a women-only privilege".

[0073] The second text 50 is a text in which the region-specific expression has been adjusted from the first text 40. Specifically, text 56 is a text adjusted from the expression "Japanese beef" in text 46 as an explanatory text to the expression "Wagyu". Thus, when there is a sense of discomfort in a direct translation, it can be translated into Roman letters by adding the Roman letter notation instead of a direct translation.

[0074] Text 57 is a text in which the subject has been complemented such as "we". Text 57 is a text adjusted from the expression "elementary school students and below" in text 47 as an explanatory text to the expression "12 years old and below". Thus, since the age differs for each country regarding schools and grades, it can be appropriately understood by foreigners by converting to an age common to all countries.

[0075] The second text 50 is the text 43 as the explanatory text with the expression "applicable to the prefectural residents" deleted. Although the text 43 is about the benefits, it has been deleted because it is a region-specific expression.

[0076] As a specific example of the translated text, the third text 60 can be cited. The third text 60 is the text obtained by translating the second text 50 in Japanese into English. The third text 60 includes texts 61, 62, 65 to 67.

[0077] In the third text 60, similar to the second text 50, texts 65 to 67 about the meal content are summarized as the specified items. In the third text 60, similar to the second text 50, texts 61, 62 about the benefits are summarized as the specified items.

[0078] <Evaluation result information 31C> Next, with reference to FIG. 7, an example of the evaluation result information 31C showing the evaluation result will be described in detail.

[0079] As shown in FIG. 7, the evaluation result information 31C is stored in the storage unit 31. The evaluation result information 31C is managed in the information processing device 11. The evaluation result information 31C is information including the evaluation result obtained from the LLM server 17. The evaluation result information 31C may be information generated each time a summary text is obtained from the explanatory text.

[0080] The evaluation result information 31C includes the results based on the explanatory text, the pointer information, and the summary text. The evaluation result information 31C may include the results based on the explanatory text, the pointer information, and the second summary text. The evaluation result information 31C may include both the results based on the explanatory text, the pointer information, and the first summary text, and the results based on the first summary text, the pointer information, and the second summary text.

[0081] In the evaluation result information 31C, the pointer type, the pointer item, the evaluation coefficient, the evaluation result and the evaluation value from the first LLM server 17A, and the evaluation result and the evaluation value from the second LLM server 17B are associated with each other.

[0082] The evaluation coefficient is a coefficient for weighting the evaluation results. The evaluation coefficient can be set by the operator of the information processing system 10. For the evaluation results that the operator wants to emphasize, a coefficient larger than that for the evaluation results that the operator wants to de-emphasize is set.

[0083] The evaluation results are the results output from the LLM server 17. The evaluation results include appropriate evaluation results and inappropriate evaluation results. The appropriate evaluation results are information indicating the results evaluated as being appropriately summarized. The inappropriate evaluation results are information indicating the results evaluated as being inappropriately summarized.

[0084] The evaluation results include invalid information. The invalid information is information indicating that it is not an evaluation target. The invalid information is information indicating that the words corresponding to the pointer type and pointer item are not included in the description. Specifically, in the pointer item associated with the maintenance as the pointer type, when the words corresponding to the pointer item are not included in the description, invalid information is associated as the evaluation result. In the pointer item associated with the deletion as the pointer type, when the words corresponding to the pointer item are not included in the description, invalid information is associated as the evaluation result.

[0085] The evaluation value is information indicating the multiplication result of the evaluation coefficient and the evaluation result. For the evaluation results and evaluation values from the first LLM server 17A and the evaluation results and evaluation values from the second LLM server 17B, the comprehensive evaluation result and the comprehensive evaluation value are associated.

[0086] As a specific example, in the evaluation results, "1" is associated as the appropriate evaluation result, "-1" is associated as the inappropriate evaluation result, and "0" is associated as the invalid information. The evaluation value is associated with the multiplication result of the evaluation result and the evaluation coefficient. As the comprehensive evaluation result, the average of all the evaluation results that are the evaluation targets is associated, but the sum of all the evaluation results that are the evaluation targets may also be associated. As the comprehensive evaluation value, the average of all the evaluation values that are the evaluation targets is associated, but the sum of all the evaluation values that are the evaluation targets may also be associated.

[0087] In this way, based on the evaluation value and the comprehensive evaluation value, the evaluation result for the summary text summarized from the explanatory text based on the guideline information can be recognized. Based on the evaluation value and the comprehensive evaluation value, the evaluation result for the second summary text summarized from the first summary text based on the guideline information can be recognized.

[0088] <Summary control process> Next, the summary control process will be described with reference to FIG. 8. The summary control process is a process executed by the control unit 30 when the summary condition is satisfied. The summary condition may be satisfied by the instruction of the operator of the information processing system 10.

[0089] As shown in FIG. 8, in step S10, the control unit 30 executes an explanatory text acquisition process. In the explanatory text acquisition process, the control unit 30 acquires an explanatory text from the first travel management server 12A through communication with the first travel management server 12A. In this case, the control unit 30 acquires an explanatory text in the first language regarding the travel service in the first country from the first travel management server 12A. The explanatory text is input by the administrator who provides the travel service in the first travel management server 12A.

[0090] In this way, the control unit 30 acquires the explanatory text input by the administrator of the transaction target. That is, the control unit 30 acquires the original text input by the administrator who is the information provider. In step S11, the control unit 30 executes a first guideline information acquisition process. In the first guideline information acquisition process, the control unit 30 acquires the first guideline information 31A by reading the first guideline information 31A from the storage unit 31.

[0091] In this way, the control unit 30 acquires the guideline information regarding the summary of the explanatory text from the storage unit 31. That is, the control unit 30 acquires the guideline information regarding the processing of the original text from the storage unit 31. In step S12, the control unit 30 executes the first summary request transmission process. In the first summary request transmission process, the control unit 30 transmits the first summary request to the LLM server 17. The control unit 30 may transmit the first summary request to either the first LLM server 17A or the second LLM server 17B.

[0092] The control unit 30 transmits the description text and the first pointer information 31A to the LLM server 17 together with the first summary request. The first pointer information 31A includes pointer information for generating a summary text by selecting words from the description text. The first pointer information 31A includes pointer information for generating a summary text with the order of the description text adjusted. The first pointer information 31A includes pointer information for generating a summary text adjusted for each specified item.

[0093] In this way, the control unit 30 generates a prompt that includes the first summary request, the description text, and the first pointer information 31A and instructs to summarize the selection and order of the description text, and transmits it to the LLM server 17.

[0094] On the other hand, when the LLM server 17 receives the first summary request, it inputs the description text and the first pointer information 31A into the summary model. As a result, in response to the reception of the first summary request, the LLM server 17 generates a summary text by selecting from the description text based on the first pointer information 31A and adjusting the order of the description text for each specified item. This summary text corresponds to an example of the first summary text. The LLM server 17 transmits the summary result including the summary text to the information processing device 11.

[0095] In step S13, the control unit 30 executes the first summary text acquisition process. In the first summary text acquisition process, the control unit 30 acquires the summary text by receiving the summary result from the LLM server 17. In this way, the control unit 30 acquires the summary text based on the first pointer information 31A. In this case, the control unit 30 acquires the summary text summarized from the description text in the first language regarding the travel service in the first country based on the first pointer information 31A. The control unit 30 stores the acquired summary text in the storage unit 31.

[0096] In this way, the control unit 30 obtains a summary sentence that summarizes the description text by transmitting the description text and the pointer information to the summary model of the generative artificial intelligence. That is, the control unit 30 obtains a processed sentence obtained by processing the original text by transmitting the original text and the pointer information to the processing model of the generative artificial intelligence.

[0097] The control unit 30 executes the processes of steps S11 to S13 as the first-stage process. In the first-stage process, the control unit 30 obtains a summary sentence that is selected from the description text based on the first pointer information 31A and whose order of the description text is adjusted for each specified item.

[0098] In step S14, the control unit 30 executes the second pointer information acquisition process. In the second pointer information acquisition process, the control unit 30 obtains the second pointer information 31B by reading the second pointer information 31B from the storage unit 31.

[0099] In this way, the control unit 30 obtains the pointer information regarding the summary of the description text from the storage unit 31. That is, the control unit 30 obtains the pointer information regarding the processing of the original text from the storage unit 31. In step S15, the control unit 30 executes the second summary request transmission process. In the second summary request transmission process, the control unit 30 transmits the second summary request to the LLM server 17. The control unit 30 may transmit the second summary request to either the first LLM server 17A or the second LLM server 17B.

[0100] The control unit 30 transmits the first summary sentence and the second pointer information 31B to the LLM server 17 together with the second summary request. The second pointer information 31B includes pointer information for generating a summary sentence that adjusts the sentence expression of the description text.

[0101] In this way, the control unit 30 generates a prompt that includes the second summary request, the first summary sentence, and the second pointer information 31B and instructs to summarize the sentence expression of the description text, and transmits it to the LLM server 17.

[0102] On the other hand, when the LLM server 17 receives the second summarization request, it inputs the first summary text and the second pointer information 31B into the summarization model. As a result, in response to the reception of the second summarization request, the LLM server 17 generates a summary text with the sentence expression of the explanatory text adjusted based on the second pointer information 31B. This summary text corresponds to an example of the second summary text. The LLM server 17 transmits a summary result including the summary text to the information processing device 11.

[0103] In step S16, the control unit 30 executes the second summary text acquisition process. In the second summary text acquisition process, the control unit 30 acquires the summary text by receiving the summary result from the LLM server 17. Thus, the control unit 30 acquires the second summary text based on the second pointer information 31B. In this case, the control unit 30 acquires a summary text summarized from the explanatory text in the first language regarding the travel service in the first country based on the second pointer information 31B. The control unit 30 stores the acquired summary text in the storage unit 31.

[0104] In this way, the control unit 30 acquires a summary text obtained by summarizing the explanatory text by transmitting the explanatory text and the pointer information to the summarization model of the generative artificial intelligence. That is, the control unit 30 acquires a processed text obtained by processing the original text by transmitting the original text and the pointer information to the processing model of the generative artificial intelligence.

[0105] The control unit 30 executes the processes of steps S14 to S16 as the second-stage process. In the second-stage process, the control unit 30 acquires a summary text with the sentence expression of the explanatory text adjusted based on the second pointer information 31B.

[0106] In step S17, the control unit 30 executes the translation request transmission process. In the translation request transmission process, the control unit 30 transmits a translation request to the LLM server 17. The control unit 30 may transmit the translation request to either the first LLM server 17A or the second LLM server 17B.

[0107] In addition to the translation request, the control unit 30 transmits the summary text and information indicating the language after translation to the LLM server 17. In this way, the control unit 30 includes the translation request, the summary text, and information indicating the language after translation, generates a prompt instructing to translate from the summary text into a specified language, and transmits it to the LLM server 17.

[0108] On the other hand, when the LLM server 17 receives a translation request, it inputs the summary text and the language after translation into the translation model. As a result, the LLM server 17 generates a translated text obtained by translating the summary text into a specified language in response to the reception of the translation request. The LLM server 17 transmits a translation result including the translated text to the information processing apparatus 11.

[0109] In step S18, the control unit 30 executes a translated text acquisition process. In the translated text acquisition process, the control unit 30 acquires the translated text by receiving the translation result from the LLM server 17. In this case, the control unit 30 acquires a translated text obtained by translating the summary text in the first language regarding the travel service in the first country into the second language. That is, the control unit 30 acquires the summary text in the second language regarding the travel service in the first country as the translated text. The control unit 30 stores the acquired translated text in the storage unit 31.

[0110] In this way, the control unit 30 executes the processes in steps S17 and S18 as the third-stage process. In the third-stage process, the control unit 30 acquires a translated text obtained by translating the summary text into a different language.

[0111] In step S19, the control unit 30 executes an information providing process. In the information providing process, the control unit 30 transmits the acquired translated text to the second travel management server 12B. As a result, the second travel management server 12B controls the received translated text to be viewable. In particular, the second travel management server 12B controls the received translated text to be viewable for each specified item. That is, the control unit 30 provides the summary text for each specified item to the intended users of the transaction target. In particular, the control unit 30 provides the summary text translated into different languages as the translated text to the intended users of the transaction target.

[0112] <Evaluation control process> Next, with reference to FIG. 9, the evaluation control process will be described. The evaluation control process is a process executed by the control unit 30 when the evaluation condition is satisfied. The evaluation condition may be satisfied by the instruction of the operator of the information processing system 10. The evaluation condition may be satisfied when the summary sentence of the explanatory text is obtained.

[0113] In step S20, the control unit 30 executes an evaluation information acquisition process. In the evaluation information acquisition process, the control unit 30 reads out the explanatory text, the pointer information, and the summary sentence for evaluation from the storage unit 31 as evaluation information. Thereby, the control unit 30 acquires the explanatory text, the pointer information, and the summary sentence for evaluation.

[0114] In step S21, the control unit 30 executes a first evaluation request transmission process. In the first evaluation request transmission process, the control unit 30 transmits a first evaluation request to the first LLM server 17A. The first evaluation request is an evaluation request to be transmitted to the first LLM server 17A.

[0115] In addition to the first evaluation request, the control unit 30 transmits the explanatory text, the pointer information, and the summary sentence acquired as evaluation information to the first LLM server 17A. In particular, for the explanatory text and the summary sentence, the control unit 30 transmits to the first LLM server 17A the information for evaluating which of the appropriate evaluation result, the inappropriate evaluation result, and the invalid information it is for each pointer item of the pointer information.

[0116] In this way, the control unit 30 generates a prompt that includes the first evaluation request, the explanatory text, the pointer information, and the summary sentence, and instructs to perform an evaluation of the summary sentence summarized from the explanatory text based on the pointer information, and transmits it to the first LLM server 17A.

[0117] On the one hand, when the first LLM server 17A receives the first evaluation request, it inputs the description text, the pointer information, and the summary text into the evaluation model. Thereby, the first LLM server 17A generates an evaluation result for the summary text summarized from the description text based on the pointer information in response to the reception of the first evaluation request. In particular, the first LLM server 17A generates an evaluation result for the summary text summarized from the description text based on the pointer information for each pointer item in response to the reception of the first evaluation request. The first LLM server 17A transmits the evaluation result to the information processing device 11.

[0118] In step S22, the control unit 30 executes the first evaluation result acquisition process. In the first evaluation result acquisition process, the control unit 30 acquires the first evaluation result by receiving the evaluation result from the first LLM server 17A. In this way, the control unit 30 acquires the evaluation result for the summary text summarized from the description text based on the pointer information.

[0119] The control unit 30 stores the acquired evaluation result in the storage unit 31 as the evaluation result information 31C. In particular, the control unit 30 calculates an evaluation value by multiplying the evaluation result by an evaluation coefficient. The control unit 30 stores the calculated evaluation value in the storage unit 31 as the evaluation result information 31C.

[0120] In step S23, the control unit 30 executes the second evaluation request transmission process. In the second evaluation request transmission process, the control unit 30 transmits the second evaluation request to the second LLM server 17B. The second evaluation request is an evaluation request to be transmitted to the second LLM server 17B.

[0121] In addition to the second evaluation request, the control unit 30 transmits the description text, the pointer information, and the summary text acquired as evaluation information to the second LLM server 17B. In particular, for the description text and the summary text, the control unit 30 transmits to the second LLM server 17B the information for evaluating which of the appropriate evaluation result, the inappropriate evaluation result, and the invalid information it is for each pointer item of the pointer information. The information transmitted to the first LLM server 17A and the information transmitted to the second LLM server 17B are the same information.

[0122] In this way, the control unit 30 generates a prompt that includes the second evaluation request, the description text, the pointer information, and the summary text, and instructs to perform an evaluation on the summary text summarized from the description text based on the pointer information, and transmits it to the second LLM server 17B.

[0123] On the other hand, when the second LLM server 17B receives the second evaluation request, it inputs the description text, the pointer information, and the summary text into the evaluation model. Thereby, the second LLM server 17B generates an evaluation result for the summary text summarized from the description text based on the pointer information in response to the reception of the second evaluation request. In particular, the second LLM server 17B generates an evaluation result for the summary text summarized from the description text based on the pointer information for each pointer item in response to the reception of the second evaluation request. The second LLM server 17B transmits the evaluation result to the information processing apparatus 11.

[0124] In step S24, the control unit 30 executes the second evaluation result acquisition process. In the second evaluation result acquisition process, the control unit 30 acquires the second evaluation result by receiving the evaluation result from the second LLM server 17B. In this way, the control unit 30 acquires the evaluation result for the summary text summarized from the description text based on the pointer information.

[0125] The control unit 30 stores the acquired evaluation result in the storage unit 31 as evaluation result information 31C. In particular, the control unit 30 calculates an evaluation value by multiplying the evaluation result by an evaluation coefficient. The control unit 30 stores the calculated evaluation value in the storage unit 31 as evaluation result information 31C.

[0126] In step S25, the control unit 30 executes an evaluation process. In the evaluation process, the control unit 30 performs an evaluation based on the evaluation result information 31C stored in the storage unit 31. The control unit 30 evaluates the evaluation result information 31C stored in the storage unit 31 for each pointer item. In particular, the control unit 30 performs an evaluation on the summary text acquired based on the pointer information by transmitting the description text, the pointer information, and the summary text to the evaluation model.

[0127] As a specific example, the control unit 30 performs an evaluation on the selection of the summary text based on the evaluation results of the pointer items associated with maintenance and deletion as the pointer types. In particular, the control unit 30 performs an evaluation as to whether the information is to be left as a summary text in the description text based on the evaluation results of the pointer items associated with maintenance as the pointer type. The control unit 30 performs an evaluation as to whether the information is not to be left as a summary text in the description text based on the evaluation results of the pointer items associated with deletion as the pointer type.

[0128] When appropriate evaluation results and inappropriate evaluation results are associated as the evaluation results of the pointer items associated with maintenance as the pointer type, the control unit 30 makes the information to be evaluated as to whether it is to be left as a summary text in the description text. When invalid information is associated as the evaluation results of the pointer items associated with maintenance as the pointer type, the control unit 30 determines that the information to be left as a summary text in the description text is not included. In this case, the control unit 30 does not make the information to be evaluated as to whether it is to be left as a summary text in the description text.

[0129] When appropriate evaluation results and inappropriate evaluation results are associated as the evaluation results of the pointer items associated with deletion as the pointer type, the control unit 30 makes the information to be evaluated as to whether it is not to be left as a summary text in the description text. When invalid information is associated as the evaluation results of the pointer items associated with deletion as the pointer type, the control unit 30 determines that the information not to be left as a summary text in the description text is not included. In this case, the control unit 30 does not make the information to be evaluated as to whether it is not to be left as a summary text in the description text.

[0130] The control unit 30 performs an evaluation on the order of the summary text based on the evaluation results of the pointer items associated with the order as the pointer type. The control unit 30 performs an evaluation on the text expression of the summary text based on the evaluation results of the pointer items associated with the text expression as the pointer type.

[0131] Also, the control unit 30 performs a first evaluation of the summary text by transmitting the explanatory text, the pointer information, and the summary text to the first evaluation model of the first LLM server 17A. The control unit 30 performs a second evaluation of the summary text by transmitting the explanatory text, the pointer information, and the summary text to the second evaluation model of the second LLM server 17B. Then, the control unit 30 evaluates the summary text by comparing the first evaluation result and the second evaluation result.

[0132] In step S26, the control unit 30 determines whether or not the pointer correction condition is satisfied. The pointer correction condition is satisfied based on the evaluation result. Specifically, the pointer correction condition is satisfied based on the evaluation value.

[0133] The pointer correction condition may be satisfied based on the first evaluation result and the second evaluation result. The pointer correction condition may be satisfied when there is a pointer item in which both the evaluation value corresponding to the first evaluation result and the evaluation value corresponding to the second evaluation result are less than or equal to a specified value. The pointer correction condition may be satisfied when there is a pointer item in which at least one of the evaluation value corresponding to the first evaluation result and the evaluation value corresponding to the second evaluation result is less than or equal to a specified value. The pointer correction condition may be satisfied when both the comprehensive evaluation value corresponding to the first evaluation result and the comprehensive evaluation value corresponding to the second evaluation result are less than or equal to a specified value. The pointer correction condition may be satisfied when at least one of the comprehensive evaluation value corresponding to the first evaluation result and the comprehensive evaluation value corresponding to the second evaluation result is less than or equal to a specified value.

[0134] If the control unit 30 determines that the pointer correction condition is not satisfied, the process proceeds to step S29. If the control unit 30 determines that the pointer correction condition is satisfied, the process proceeds to step S27.

[0135] In step S27, the control unit 30 executes a corrected pointer information generation process. In the corrected pointer information generation process, the control unit 30 generates corrected pointer information for correcting the pointer information. In particular, the control unit 30 may generate, as the corrected pointer information, information indicating a pointer item in which both the evaluation value corresponding to the first evaluation result and the evaluation value corresponding to the second evaluation result are less than or equal to a specified value.

[0136] In step S28, the control unit 30 executes a correction guideline information notification process. In the correction guideline information notification process, the control unit 30 notifies the generated correction guideline information. To give a specific example, the control unit 30 may notify the correction guideline information to the operation terminal device managed by the operator of the information processing system 10.

[0137] In step S29, the control unit 30 determines whether there is a correction instruction. The correction instruction may be input by the operator of the information processing system 10. The correction instruction may include a correction target and correction content for correcting the guideline information. When the control unit 30 determines that there is no correction instruction, it ends the evaluation control process. When the control unit 30 determines that there is a correction instruction, it transfers the process to step S30.

[0138] In step S30, the control unit 30 executes a guideline information correction process. In the guideline information correction process, the control unit 30 corrects the guideline information based on the correction instruction. The control unit 30 updates the guideline information according to the correction instruction by storing the corrected guideline information in the storage unit 31. In this way, the control unit 30 corrects the guideline information regarding the summary of the description based on the evaluated result.

[0139] <Actions and Effects of the First Embodiment> The actions and effects of the first embodiment will be described. (1-1) The control unit 30 acquires the description text input by the administrator of the transaction target. The control unit 30 acquires the guideline information regarding the summary of the description text from the storage unit 31. The control unit 30 transmits the description text and the guideline information to the summary model of the generative artificial intelligence to acquire the summary text obtained by summarizing the description text. Therefore, by transmitting the guideline information to the summary model of the generative artificial intelligence in addition to the description text input by the administrator, the possibility of obtaining an appropriate summary text along with the operation of the information processing system 10 can be increased. Accordingly, the accuracy of summarizing from the description text input by the administrator can be improved.

[0140] (1-2) The pointer information includes information regarding the selection of the description text. Therefore, by sending the pointer information in addition to the description text input by the administrator to the summarization model of the artificial intelligence that generates the pointer information, it is possible to increase the likelihood of obtaining an appropriate summary text regarding the selection of the description text during the operation of the information processing system 10. Thus, the accuracy of summarizing from the description text input by the administrator can be improved.

[0141] (1-3) The pointer information includes information regarding the order of the description text. Therefore, by sending the pointer information in addition to the description text input by the administrator to the summarization model of the artificial intelligence that generates the pointer information, it is possible to increase the likelihood of obtaining an appropriate summary text regarding the order of the description text during the operation of the information processing system 10. Thus, the accuracy of summarizing from the description text input by the administrator can be improved.

[0142] (1-4) The pointer information includes information regarding the sentence expression of the description text. Therefore, by sending the pointer information in addition to the description text input by the administrator to the summarization model of the artificial intelligence that generates the pointer information, it is possible to increase the likelihood of obtaining an appropriate summary text regarding the sentence expression of the description text during the operation of the information processing system 10. Thus, the accuracy of summarizing from the description text input by the administrator can be improved.

[0143] (1-5) The control unit 30 provides the summary text so that it can be viewed by the intended users of the transaction target for each specified item. Therefore, it is possible to provide the summary text with improved accuracy of summarizing from the description text input by the administrator to the intended users of the transaction target.

[0144] (1-6) The control unit 30 provides the translated text obtained by translating the summary text into different languages so that it can be viewed by the intended users of the transaction target. Therefore, it is possible to improve the accuracy of summarizing from the description text input by the administrator and provide the summary text translated into the desired language to the intended users of the transaction target.

[0145] (1-7) Conventionally, it has been desired to appropriately evaluate whether a desired summary text can be obtained in the operation of the information processing system 10. Therefore, the control unit 30 evaluates the summary text based on the guideline information by transmitting the explanatory text, the guideline information, and the summary text to the evaluation model of the generative artificial intelligence. For this reason, by transmitting the explanatory text, the guideline information, and the summary text to the evaluation model of the generative artificial intelligence, it is possible to evaluate whether an appropriate summary text can be obtained in the operation of the information processing system 10. Based on this, it is possible to make a judgment based on the evaluation result of whether an appropriate summary text can be obtained in the operation of the information processing system 10. And it is possible to increase the possibility of obtaining an appropriate summary text in the operation of the information processing system 10. Therefore, the accuracy of summarizing from the explanatory text input by the administrator can be improved.

[0146] (1-8) The control unit 30 evaluates the selection of the summary text. For this reason, by transmitting the explanatory text, the guideline information, and the summary text to the evaluation model of the generative artificial intelligence, it is possible to evaluate whether an appropriate summary text can be obtained for the selection of the summary text in the operation of the information processing system 10. Based on this, it is possible to make a judgment based on the evaluation result of whether an appropriate summary text can be obtained for the selection of the summary text in the operation of the information processing system 10. Therefore, the accuracy of summarizing from the explanatory text input by the administrator can be improved.

[0147] (1-9) When the explanatory text does not contain information to be left as the summary text, the control unit 30 does not regard it as an evaluation target for whether it is information to be left as the summary text in the explanatory text. For this reason, by selecting the evaluation target without regarding unnecessary information as the evaluation target, it is possible to improve the accuracy of the evaluation of whether an appropriate summary text can be obtained in the operation of the information processing system 10. Therefore, the accuracy of summarizing from the explanatory text input by the administrator can be improved.

[0148] (1-10) If the description text does not contain information that is not to be left as a summary sentence, the control unit 30 does not make it an object of evaluation as information that is not to be left as a summary sentence in the description text. Therefore, by selecting the evaluation object without making unnecessary information an evaluation object, it is possible to improve the accuracy of evaluating whether an appropriate summary sentence can be obtained in connection with the operation of the information processing system 10. Accordingly, it is possible to improve the accuracy of summarizing from the description text input by the administrator.

[0149] (1-11) The control unit 30 evaluates the order of the summary sentences. Therefore, by transmitting the description text, the pointer information, and the summary sentences to the evaluation model of the generative artificial intelligence, it is possible to evaluate whether an appropriate summary sentence has been obtained regarding the order of the summary sentences in connection with the operation of the information processing system 10. Thereby, it is possible to make a judgment based on the evaluation result as to whether an appropriate summary sentence has been obtained regarding the order of the summary sentences in connection with the operation of the information processing system 10. Accordingly, it is possible to improve the accuracy of summarizing from the description text input by the administrator.

[0150] (1-12) The control unit 30 evaluates the sentence expression of the summary sentences. Therefore, by transmitting the description text, the pointer information, and the summary sentences to the evaluation model of the generative artificial intelligence, it is possible to evaluate whether an appropriate summary sentence has been obtained regarding the sentence expression of the summary sentences in connection with the operation of the information processing system 10. Thereby, it is possible to make a judgment based on the evaluation result as to whether an appropriate summary sentence has been obtained regarding the sentence expression of the summary sentences in connection with the operation of the information processing system 10. Accordingly, it is possible to improve the accuracy of summarizing from the description text input by the administrator.

[0151] (1-13) The control unit 30 evaluates the summary sentences based on the evaluation value obtained by calculating the evaluation result from the evaluation model and the evaluation coefficient. Thereby, it is possible to weight the evaluation result by the evaluation coefficient. Accordingly, it is possible to improve the accuracy of summarizing from the description text input by the administrator.

[0152] (1-14) The control unit 30 performs a first evaluation of the summary text by transmitting the description text, the pointer information, and the summary text to the first evaluation model of the generative artificial intelligence. The control unit 30 performs a second evaluation of the summary text by transmitting the description text, the pointer information, and the summary text to the second evaluation model of the generative artificial intelligence. The control unit 30 evaluates the summary text by comparing the first evaluation and the second evaluation. Therefore, it is possible to determine based on both the evaluation result by the first evaluation model and the evaluation result by the second evaluation model whether an appropriate summary text is being obtained in connection with the operation of the information processing system 10. Accordingly, the accuracy of summarizing from the description text input by the administrator can be improved.

[0153] (1-15) The control unit 30 corrects the pointer information based on the evaluation result. Therefore, based on the evaluation result of whether an appropriate summary text is being obtained in connection with the operation of the information processing system 10, the pointer information itself can be corrected. Accordingly, the accuracy of summarizing from the description text input by the administrator can be improved.

[0154] [Second Embodiment] Next, the second embodiment will be described. In the following description, the same components as those in the already described embodiments are denoted by the same reference numerals, and the overlapping descriptions are omitted or simplified.

[0155] In the second embodiment, the control unit 30 executes summary control processing for the first LLM server 17A. That is, the control unit 30 obtains a summary text by outputting the description text and the pointer information to the first processing model. In this way, the summary text obtained from the first processing model corresponds to an example of the first processed text.

[0156] Also, the control unit 30 executes summary control processing for the second LLM server 17B. That is, the control unit 30 obtains a summary text by outputting the description text and the pointer information to the second processing model. In this way, the summary text obtained from the second processing model corresponds to an example of the second processed text.

[0157] Then, the control unit 30 executes evaluation control processing for the first LLM server 17A. In this case, in step S21, the control unit 30 outputs the summary text, the original text, and the pointer information obtained from the first processing model to the first LLM server 17A, and in step S22, obtains the evaluation result.

[0158] In step S23, the control unit 30 outputs the summary text, the original text, and the pointer information obtained from the second processing model to the first LLM server 17A, and in step S24, obtains the evaluation result.

[0159] In this way, the control unit 30 transmits the explanatory text, the pointer information, the summary text obtained from the first processing model, and the summary text obtained from the second processing model to the evaluation model. Thereby, the control unit 30 evaluates the summary text obtained from the first processing model and the summary text obtained from the second processing model.

[0160] In particular, in step S25, the control unit 30 compares the evaluation value for the summary text obtained from the first processing model with the evaluation value for the summary text obtained from the second processing model. In this way, the control unit 30 can evaluate which of the summary text obtained from the first processing model and the summary text obtained from the second processing model is appropriate based on the comparison result.

[0161] <Actions and Effects of the Second Embodiment> The actions and effects of the second embodiment will be described. (2-1) By transmitting the explanatory text, the pointer information, the summary text obtained from the first processing model, and the summary text obtained from the second processing model to the evaluation model, the control unit 30 evaluates the summary text from each processing model. Therefore, by comparing the evaluation values for the summary texts obtained from multiple processing models, it is possible to evaluate the summary texts obtained from multiple processing models. Thereby, the processing accuracy by multiple processing models can be evaluated. Therefore, the accuracy of summarizing from the explanatory text input by the administrator can be improved.

[0162] [Modification Example] This embodiment can be implemented with the following modifications. This embodiment and the following modification examples can be implemented in combination with each other within a technically non - conflicting range.

[0163] · When the translation condition is satisfied, the control unit 30 may obtain a translated text from the abstract text. The translation condition may be satisfied according to an instruction from the operator of the information processing system 10. The translation condition may be satisfied according to the type of the server that provides the abstract text. The control unit 30 does not necessarily have to obtain a translated text from the abstract text.

[0164] · The control unit 30 may execute the first - stage processing and the second - stage processing as processing with three or more stages. Taking a specific example, the control unit 30 may obtain a first summary text selected from the description text by transmitting the description text and the first pointer information 31A to the LLM server 17. The control unit 30 may obtain a second summary text with the arrangement order of the description text adjusted by transmitting the first summary text and the first pointer information 31A to the LLM server 17. The control unit 30 may obtain a third summary text with the sentence expression of the description text adjusted as the summary text by transmitting the second summary text and the second pointer information 31B to the LLM server 17. The control unit 30 may execute the multi - stage processing in an arbitrary execution order. As the multi - stage processing, after obtaining the first summary text selected from the description text, the control unit 30 may obtain a second summary text with the arrangement order of the first summary text adjusted. In this way, the control unit 30 may execute the multi - stage processing in an arbitrary execution order. By inputting various information into the learning model in multiple stages, the accuracy of the information output from the learning model can be improved. Also, when inputting various information into the learning model in multiple stages, if the amount of information output from the learning model is less than the amount of information input to the learning model, the amount of information input to the learning model in the next stage can be reduced.

[0165] ·The control unit 30 may execute the first-stage process and the second-stage process as a single process. To give a specific example, the control unit 30 may obtain a summary sentence summarized from the description text based on the pointer information by sending the description text, the first pointer information 31A, and the second pointer information 31B to the LLM server 17. Thereby, the number of times of input to the learning model can be reduced.

[0166] ·The control unit 30 may obtain an evaluation result from communication with the second LLM server 17B for the summary sentence obtained through communication with the first LLM server 17A. The control unit 30 may obtain an evaluation result from communication with the first LLM server 17A for the summary sentence obtained through communication with the second LLM server 17B. That is, the control unit 30 may use different LLM servers 17 for obtaining the summary sentence and for obtaining the evaluation result.

[0167] · The evaluation coefficient may be set based on the action history of the prospective traders with respect to the description text and the abstract text. That is, the evaluation coefficient may be based on the action history of one or more prospective traders. Also, a high evaluation coefficient may be set for the specified items with a high value indicating the action history of one or more prospective traders. The action history may include impressions (IMP: impression). Impressions indicate the number of times the specified items in the description text and the abstract text are displayed. The action history may include the click-through rate (CTR: Click Through Rate). The click-through rate indicates the ratio of the number of clicks to the number of times the specified items in the description text and the abstract text are displayed. The action history may include conversions (CV: Conversion). Conversions indicate the number of times the target action is taken starting from the specified items in the description text and the abstract text. Conversions include the conclusion of transactions in travel services and the purchase of travel-related goods or services. Conversions include the conclusion of transactions in e-commerce services and the purchase of goods or services in e-commerce. To give a specific example, a high evaluation coefficient may be set for the specified items with a high number of impressions. A high evaluation coefficient may be set for the specified items with a high click-through rate. A high evaluation coefficient may be set for the specified items with a high conversion rate.

[0168] · Based on the evaluation result, the control unit 30 may select the LLM server 17 with a high evaluation result as the server for obtaining the abstract text. That is, based on the evaluation result, the control unit 30 may select a summary model with a high evaluation result.

[0169] · The control unit 30 may leave the text that does not correspond to maintenance and deletion as the pointer type as it is. The control unit 30 may delete the text that does not correspond to maintenance and deletion as the pointer type without leaving it.

[0170] ·The control unit 30 may perform an evaluation on at least one of the selection of the explanatory text, the order of the explanatory text, and the sentence expression of the explanatory text. The control unit 30 may execute either one of the evaluation of whether the information is to be left as a summary text in the explanatory text and the evaluation of whether the information is not to be left as a summary text in the explanatory text, and it is not necessary to execute the other one. The control unit 30 may perform an evaluation on a content different from the selection of the explanatory text, the order of the explanatory text, and the sentence expression of the explanatory text.

[0171] ·The control unit 30 may execute, as a process of summarizing from the explanatory text, at least one of the selection of the explanatory text, the order of the explanatory text, and the sentence expression of the explanatory text. The control unit 30 may execute either one of the process of leaving the explanatory text as a summary text and the process of not leaving the explanatory text as a summary text, and it is not necessary to execute the other one. The control unit 30 may execute a process of summarizing from the explanatory text for a content different from the selection of the explanatory text, the order of the explanatory text, and the sentence expression of the explanatory text.

[0172] ·The information processing system 10 may provide a service different from the travel service. To give a specific example, the information processing system 10 may provide an e-commerce trading service. In this case, the explanatory text of the transaction target may be an explanatory text regarding the product traded in the e-commerce trading service. The information processing system 10 may provide a video distribution service. In this case, the explanatory text of the transaction target may be an explanatory text regarding the video traded in the video distribution service. The information processing system 10 may provide a plurality of services. The information processing system 10 does not necessarily have to provide a travel service. That is, the information processing system 10 may be a system that provides a service to be a transaction target. The information processing system 10 itself may be a system that does not provide a service to be a transaction target.

[0173] ·The first travel management server 12A and the second travel management server 12B may be configured as an integrated travel management server 12. The travel management server 12 may include a third travel management server different from the first travel management server 12A and the second travel management server 12B.

[0174] · The first LLM server 17A and the second LLM server 17B may be configured as an integrated LLM server 17. The LLM server 17 may include a third LLM server different from the first LLM server 17A and the second LLM server 17B.

[0175] · The LLM server 17 may manage at least one of a summarization model, a translation model, and an evaluation model as an integrated learning model. The LLM server 17 may include a learning model other than the summarization model, the translation model, and the evaluation model.

[0176] · The information processing device 11 may have a configuration of another server. To give a specific example, the information processing device 11 may include part or all of the configuration related to the travel management server 12. The information processing device 11 may include part or all of the configuration related to the LLM server 17.

[0177] · The information processing device 11 may be composed of a plurality of servers. To give a specific example, the plurality of servers may include a server for obtaining a summary from a description based on pointer information. The plurality of servers may include a server for obtaining a translation from the summary. The plurality of servers may include a server for evaluating the summary summarized from the description based on pointer information. The administrators of the plurality of servers may be different or the same.

[0178] · The information processing device 11 may obtain something other than the description, as long as it obtains the original text. The information processing device 11 may obtain something other than the summary to be summarized from the original text, as long as it obtains a processed text processed from the original text. The pointer information may be information other than the information related to the summary of the original text, as long as it is information related to the processing of the original text. That is, the information processing device 11 may be configured to obtain a processed text of the original text by inputting the original text and the pointer information related to the processing of the original text into a processing model.

[0179] · The information processing apparatus 11 may acquire processed information including a processed image based on original information including an original image and pointer information. In this case, the pointer information may include information indicating a pointer to maintain the image as the processed information of the original information. The pointer information may include information indicating a pointer to delete the image as the processed information of the original information.

[0180] · As used herein, the expression "at least any one" means one or more of the desired options. As an example, as used herein, the expression "at least any one" means only one option or both of the two options if the number of options is two. As another example, as used herein, the expression "at least any one" means only one option or any combination of two or more options if the number of options is three or more.

[0181] [Appendix] The following describes the technical ideas grasped from the above-described embodiments and modification examples. (A) An information processing apparatus includes an original text acquisition unit that acquires an original text input by an administrator, a pointer information acquisition unit that acquires pointer information regarding the processing of the original text acquired by the original text acquisition unit from a storage unit, a processed text acquisition unit that acquires a processed text obtained by processing the original text acquired by the original text acquisition unit by transmitting the original text acquired by the original text acquisition unit and the pointer information acquired by the pointer information acquisition unit to a processing model, and an evaluation unit that evaluates the processed text acquired by the processed text acquisition unit by transmitting the original text acquired by the original text acquisition unit, the pointer information acquired by the pointer information acquisition unit, and the processed text acquired by the processed text acquisition unit to an evaluation model.

[0182] (B) In the information processing apparatus according to (A), the pointer information includes information regarding the selection of the original text acquired by the original text acquisition unit, and the evaluation unit evaluates the selection of the processed text acquired by the processed text acquisition unit.

[0183] In the information processing apparatus according to (C)(B), when the information to be left as a processed sentence is not included in the original text acquired by the original text acquisition unit, the evaluation unit does not make it an evaluation target as to whether it is information to be left as a processed sentence in the original text acquired by the original text acquisition unit.

[0184] In the information processing apparatus according to (D)(B) or (C), when the information not to be left as a processed sentence is not included in the original text acquired by the original text acquisition unit, the evaluation unit does not make it an evaluation target as to whether it is information not to be left as a processed sentence in the original text acquired by the original text acquisition unit.

[0185] In the information processing apparatus according to any one of (E)(A) to (D), the pointer information includes information regarding the order of the original text acquired by the original text acquisition unit, and the evaluation unit evaluates the order of the processed sentences acquired by the processed sentence acquisition unit.

[0186] In the information processing apparatus according to any one of (F)(A) to (E), the pointer information includes information regarding the sentence expression of the original text acquired by the original text acquisition unit, and the evaluation unit evaluates the sentence expression of the processed sentences acquired by the processed sentence acquisition unit.

[0187] In the information processing apparatus according to any one of (G)(A) to (F), the evaluation unit evaluates the processed sentences acquired by the processed sentence acquisition unit based on an evaluation value obtained by calculating an evaluation result from an evaluation model and an evaluation coefficient.

[0188] In the information processing apparatus according to (H)(G), the evaluation unit evaluates the processed sentences acquired by the processed sentence acquisition unit based on an evaluation value obtained by calculating an evaluation result from an evaluation model and an evaluation coefficient based on the action history of the transaction planner.

[0189] In the information processing apparatus according to any one of (I)(A) to (H), the evaluation unit transmits the original text acquired by the original text acquisition unit, the pointer information acquired by the pointer information acquisition unit, the first processed text acquired by the processed text acquisition unit from the first processing model, and the second processed text acquired by the processed text acquisition unit from the second processing model to an evaluation model, thereby comparing the evaluation of the first processed text acquired by the processed text acquisition unit with the evaluation of the second processed text acquired by the processed text acquisition unit, and evaluating the first processed text and the second processed text acquired by the processed text acquisition unit.

[0190] The information processing apparatus according to any one of (J)(A) to (I) includes a pointer correction unit that corrects pointer information regarding the processing of the original text acquired by the original text acquisition unit based on the result evaluated by the evaluation unit.

[0191] An information processing system includes an original text acquisition unit that acquires an original text input by an administrator, a pointer information acquisition unit that acquires pointer information regarding the processing of the original text acquired by the original text acquisition unit from a storage unit, a processed text acquisition unit that acquires a processed text obtained by processing the original text acquired by the original text acquisition unit by transmitting the original text acquired by the original text acquisition unit and the pointer information acquired by the pointer information acquisition unit to a processing model, and an evaluation unit that evaluates the processed text acquired by the processed text acquisition unit by transmitting the original text acquired by the original text acquisition unit, the pointer information acquired by the pointer information acquisition unit, and the processed text acquired by the processed text acquisition unit to an evaluation model.

[0192] An information processing method is executed by one or more computers to acquire an original text input by an administrator, acquire pointer information regarding the processing of the original text from a storage unit, acquire a processed text obtained by processing the original text by transmitting the original text and the pointer information to a processing model, and evaluate the processed text by transmitting the original text, the pointer information, and the processed text to an evaluation model.

[0193] (M) The program causes one or more computers to acquire the original text input by an administrator, acquire the guideline information regarding the processing of the original text from the storage unit, acquire the processed text obtained by processing the original text by transmitting the original text and the guideline information to a processing model, and perform an evaluation of the processed text by transmitting the original text, the guideline information, and the processed text to an evaluation model.

Explanation of Signs

[0194] 10… Information processing system, 11… Information processing apparatus, 12… Travel management server, 12A… First travel management server, 12B… Second travel management server, 17… Large language model server, 17A… First large language model server, 17B… Second large language model server, 18… Terminal device, 30… Control unit, 30A… Explanation text acquisition unit (original text acquisition unit), 30B… Guideline information acquisition unit, 30E… Summary text acquisition unit (processed text acquisition unit), 30G… Evaluation unit, 30H… Information providing unit, 30I… Guideline correction unit, 31… Storage unit, 31A… First guideline information, 31B… Second guideline information, 31C… Evaluation result information.

Claims

1. An original text acquisition unit that acquires the original text input by the administrator, A guideline information acquisition unit that acquires guideline information regarding the processing of the original text acquired by the original text acquisition unit from the storage unit, A processed text acquisition unit that acquires a processed text obtained by processing the original text acquired by the original text acquisition unit by transmitting the original text acquired by the original text acquisition unit and the guideline information acquired by the guideline information acquisition unit to a processing model, An evaluation unit that evaluates the processed text acquired by the processed text acquisition unit by transmitting the original text acquired by the original text acquisition unit, the guideline information acquired by the guideline information acquisition unit, and the processed text acquired by the processed text acquisition unit to an evaluation model, An information processing apparatus comprising:

2. The information processing apparatus according to claim 1, wherein the guideline information includes information regarding the selection of the original text acquired by the original text acquisition unit, and the evaluation unit evaluates the selection of the processed text acquired by the processed text acquisition unit. An information processing apparatus.

3. The information processing apparatus according to claim 2, wherein the evaluation unit does not evaluate whether the information to be left as a processed text in the original text acquired by the original text acquisition unit when the original text acquired by the original text acquisition unit does not include information to be left as a processed text. An information processing apparatus.

4. The information processing apparatus according to claim 2, wherein the evaluation unit does not evaluate whether the information not to be left as a processed text in the original text acquired by the original text acquisition unit when the original text acquired by the original text acquisition unit does not include information not to be left as a processed text. An information processing apparatus.

5. The information processing apparatus according to claim 1, wherein the guideline information includes information regarding the order of the original text acquired by the original text acquisition unit, and the evaluation unit evaluates the order of the processed text acquired by the processed text acquisition unit. An information processing apparatus.

6. The information processing apparatus according to claim 1, wherein the guideline information includes information regarding the sentence expression of the original text acquired by the original text acquisition unit, and the evaluation unit evaluates the sentence expression of the processed text acquired by the processed text acquisition unit. An information processing apparatus.

7. In the information processing apparatus according to any one of claims 1 to 6, the evaluation unit evaluates the processed text acquired by the processed text acquisition unit based on an evaluation value obtained by calculating an evaluation result from an evaluation model and an evaluation coefficient. An information processing apparatus.

8. The information processing apparatus according to claim 7, The evaluation unit performs an evaluation on the processed sentence acquired by the processed sentence acquisition unit based on an evaluation value obtained by calculating an evaluation result from an evaluation model and an evaluation coefficient based on the behavior history of the transaction planner. An information processing apparatus.

9. In the information processing apparatus according to any one of Claims 1 to 6, the evaluation unit transmits the original text acquired by the original text acquisition unit, the guideline information acquired by the guideline information acquisition unit, the first processed sentence acquired by the processed sentence acquisition unit from the first processing model, and the second processed sentence acquired by the processed sentence acquisition unit from the second processing model to an evaluation model, compares the evaluation of the first processed sentence acquired by the processed sentence acquisition unit with the evaluation of the second processed sentence acquired by the processed sentence acquisition unit, and performs an evaluation on the first processed sentence and the second processed sentence acquired by the processed sentence acquisition unit. An information processing apparatus.

10. In the information processing apparatus according to any one of Claims 1 to 6, a guideline correction unit that corrects guideline information regarding the processing of the original text acquired by the original text acquisition unit based on the result evaluated by the evaluation unit is provided. An information processing apparatus.

11. An original text acquisition unit that acquires an original text input by an administrator; a guideline information acquisition unit that acquires guideline information regarding the processing of the original text acquired by the original text acquisition unit from a storage unit; a processed sentence acquisition unit that transmits the original text acquired by the original text acquisition unit and the guideline information acquired by the guideline information acquisition unit to a processing model, and thereby acquires a processed sentence obtained by processing the original text acquired by the original text acquisition unit; an evaluation unit that transmits the original text acquired by the original text acquisition unit, the guideline information acquired by the guideline information acquisition unit, and the processed sentence acquired by the processed sentence acquisition unit to an evaluation model, and thereby performs an evaluation on the processed sentence acquired by the processed sentence acquisition unit; An information processing system comprising the above.

12. One or more computers acquire an original text input by an administrator; acquire guideline information regarding the processing of the original text from a storage unit; transmit the original text and the guideline information to a processing model, and thereby acquire a processed sentence obtained by processing the original text; transmit the original text, the guideline information, and the processed sentence to an evaluation model, and thereby perform an evaluation on the processed sentence; An information processing method for executing the above.

13. One or more computers are caused to acquire an original text input by an administrator; acquire guideline information regarding the processing of the original text from a storage unit; By sending the original text and the pointer information to a processing model, obtaining a processed sentence obtained by processing the original text; By sending the original text, the pointer information, and the processed sentence to an evaluation model, evaluating the processed sentence; A program that causes the above to be executed.

Images