PROGRAM, INFORMATION PROCESSING APPARATUS, METHOD, AND SYSTEM

Abstract

Create a draft of your application document that takes into account the overall narrative of the text. [Solution] A program to be executed by a computer having a processor and a memory. The program causes the processor to execute the steps of accepting a selection of a type of subsidy, accepting input of reference information linked to each item that may be included in the selected type of subsidy application document, inputting the input reference information and an instruction sentence instructing the creation of a sentence that may be written in each item into a generative AI model and outputting the sentence from the generative AI model, and creating a draft of the subsidy application document based on the output sentence, and in the output step, the generative AI model outputs the sentence using the sentence that has already been output, and causes the generative AI model to create the sentence in a predetermined order.

Description

[Technical field]

[0001] The present disclosure relates to a program, an information processing device, a method, and a system. [Background technology]

[0002] Conventionally, technologies for creating drafts of various application documents, including subsidy application documents, have been researched and developed. For example, Patent Literature 1 discloses a technology in which a user instructs an AI model having a writing function to create an application document, and the AI ​​model outputs a draft of the application document. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7455452 specification Summary of the Invention [Problem to be solved by the invention]

[0004] However, the technology disclosed in Patent Document 1 leaves room for improvement in terms of creating a draft of an application document that takes into account the storyline of the entire text.

[0005] The objective of the present disclosure is to draft application documents that take into account the narrative nature of the entire document. [Means for solving the problem]

[0006] In order to solve the above-mentioned problems, a program according to one aspect of the present disclosure is a program to be executed by a computer having a processor and a memory. The program causes the processor to execute the steps of accepting a selection of a type of subsidy, accepting input of reference information linked to each item that may be included in the selected type of subsidy application document, inputting the input reference information and an instruction sentence instructing the creation of a sentence that may be written in each item into a generative AI model and outputting the sentence from the generative AI model, and creating a draft of the subsidy application document based on the output sentence, and in the output step, the generative AI model outputs the sentence using the output sentence, and causes the generative AI model to create the sentence in a predetermined order. Effect of the Invention

[0007] According to the present disclosure, it is possible to create a draft of a grant application document that takes into account the narrative nature of the entire text. [Brief description of the drawings]

[0008] [Figure 1] 1 is a block diagram showing an example of the overall configuration of a system 1. FIG. [Diagram 2] 2 is a block diagram showing an example of the configuration of a terminal device 10 shown in FIG. [Diagram 3] 2 is a block diagram showing an example of the configuration of a server 20 shown in FIG. 1. [Figure 4] FIG. 4 is a diagram showing the data structure of a prompt table 2021 shown in FIG. [Diagram 5] FIG. 4 is a diagram showing a data structure of a sequence table 2022 shown in FIG. [Figure 6] 13 is a flowchart showing an example of the operation of the server 20 when presenting a draft of a target document. [Figure 7] 13 is a schematic diagram showing an example of a screen of the display 141 when the selection of the type of subsidy is accepted. FIG. [Figure 8] 13 is a schematic diagram showing an example of a screen of a display 141 when input of reference information is accepted. FIG. [Figure 9] 13 is a schematic diagram showing an example of a screen of the display 141 when a draft is presented to a user. FIG. [Figure 10] FIG. 2 is a block diagram showing the basic hardware configuration of a computer 90. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, an embodiment of the present disclosure will be described with reference to the drawings. In all the drawings explaining the embodiment, the same reference numerals are given to common components, and repeated explanations are omitted. Note that the following embodiment does not unduly limit the contents of the present disclosure described in the claims. In addition, not all components shown in the embodiment are necessarily essential components of the present disclosure. In addition, each figure is a schematic diagram and is not necessarily illustrated strictly.

[0010] [1. Overview] The system according to this embodiment accepts a user's selection of a type of subsidy and input of reference information linked to each item that may be included in a subsidy application document. The system according to this embodiment inputs a data set for each item, consisting of the input reference information, a first instruction statement instructing the creation of a sentence that may be written in a specific item, and an output sentence, in a specific order to an LLM, and outputs a sentence that may be written in each item from the LLM. The system according to this embodiment inputs the output sentence for each item and a second instruction statement instructing the creation of a draft of the subsidy application document to an LLM, and outputs the draft from the LLM.

[0011] [2. Overall system configuration] Fig. 1 is a block diagram showing an example of the overall configuration of system 1. System 1 is a system for providing a subsidy application document creation service (hereinafter, document creation service). System 1 shown in Fig. 1 includes, for example, a terminal device 10, a server 20, and an AI system 30. The terminal device 10, the server 20, and the AI ​​system 30 are communicatively connected via, for example, a network 80.

[0012] A subsidy application document is a document submitted when applying for various subsidies to a public institution or a private organization. There are various types of subsidy application documents submitted to a public institution or a private organization, such as an application form, a business plan, a budget plan, a certified copy of a corporate registry, and a list of officers. In this specification, unless otherwise specified, a subsidy application document refers to a business plan, and a document creation service refers to a service for creating a business plan. However, it goes without saying that the subsidy application document in this specification covers various types of documents other than a business plan, such as a budget plan. In addition, the subsidy application document in this specification includes both an electronic input application type and a document submission application type using document creation software or the like. However, the subsidy application document in this specification does not exclude application types other than these two types of application types.

[0013] 1 shows an example in which the system 1 includes one terminal device 10, but for example, the system 1 may include two or more terminal devices 10. In FIG. 1, an example in which the system 1 includes one server 20, but for example, a collection of multiple devices may be one server 20. The manner in which multiple functions required to realize the server 20 are allocated to one or multiple pieces of hardware can be determined appropriately depending on the processing capacity of each piece of hardware and / or the specifications required for the server 20, etc.

[0014] 1 shows an example in which the system 1 includes one AI system 30, but the system 1 may include two or more AI systems 30. Also, while FIG. 1 shows an example in which the AI ​​system 30 is independent from the server 20, the server 20 may include the functions of the AI ​​system 30. In other words, the server 20 may store a generative AI model (described in detail below) included in the AI ​​system 30.

[0015] The terminal device 10 is, for example, an information processing device operated by a user. The user is, for example, a user of a document creation service, that is, an applicant for a subsidy. The terminal device 10 is realized by, for example, a mobile terminal such as a smartphone or a tablet. In this embodiment, the terminal device 10 is a smartphone. The terminal device 10 may also be realized by, for example, a stationary PC (Personal Computer), a laptop PC, or the like.

[0016] The terminal device 10 includes a communication IF (Interface) 12, an input device 13, an output device 14, a memory 15, a storage 16, and a processor 19. The input device 13 is a device for receiving an input operation from a user (for example, a touch panel, a touch pad, a pointing device such as a mouse, a keyboard, etc.). The output device 14 is a device for presenting information to a user (a display, a speaker, etc.). In this embodiment, the terminal device 10 includes a touch panel in which the input device 13 and the output device 14 are integrated.

[0017] The server 20 is, for example, an information processing device for managing / operating a document creation service, and is an information processing device realized by a computer connected to a network 80. As shown in Fig. 1, the server 20 includes a communication IF 22, an input / output IF 23, a memory 25, a storage 26, and a processor 29. The input / output IF 23 functions as an interface for an input device for receiving input operations from a manager / operator of the document creation service (hereinafter abbreviated as manager / operator) and an output device for outputting information to the manager / operator.

[0018] The AI ​​system 30 is, for example, a system in which a generative AI model is constructed. The generative AI model is, for example, a natural language model. The natural language model is an artificial intelligence model used in the field of natural language processing (NLP), and by learning text data, it is possible to understand patterns of language used by humans and effectively perform natural language generation (NLG) tasks.

[0019] The AI ​​system 30 includes, for example, one or more LLMs (large-scale language models) as generative AI models. LLMs are natural language models constructed by learning large-scale text data, and are used in many NLG tasks such as generating responses to specific questions, automatically generating sentences, and summarizing text. Examples of LLMs include the following:

[0020] ·OpenAI: GPT-4o Google: Gemini 1.5 Flash ·Anthropic: Claude 3.5 Sonnet

[0021] Note that the generative AI model included in the AI ​​system 30 does not have to be an LLM. The AI ​​system 30 may include, for example, one or more NLP models. The NLP model is a natural language model that is not an LLM, such as a feedforward neural network or a custom model designed for a specific task or domain. For example, the AI ​​system 30 may also include one or more multimodal generative AI models. In this case, the multimodal generative AI model may be input with, for example, a template (format) for a grant application document, image data (e.g., scanned data) of a document including reference information, or the like.

[0022] The AI ​​system 30 inputs the text data transmitted from the server 20 to the LLM, and causes the LLM to output an answer based on the input text data. The AI ​​system 30 transmits the answer output from the LLM to the server 20.

[0023] Each information processing device, such as the terminal device 10, the server 20, and the AI ​​system 30, is configured by a computer 90 (see FIG. 10) equipped with an arithmetic unit and a storage device. The basic hardware configuration of the computer 90 and the basic functional configuration of the computer 90 realized by the basic hardware configuration will be described later. Note that for each of the terminal device 10, the server 20, and the AI ​​system 30, descriptions that overlap with the basic hardware configuration of the computer 90 and the basic functional configuration of the computer will be omitted.

[0024] <2.1 Terminal device configuration> Fig. 2 is a block diagram showing an example of the configuration of the terminal device 10 shown in Fig. 1. As shown in Fig. 2, the terminal device 10 includes a communication unit 120, an input device 13, an output device 14, an audio processing unit 170, a microphone 171, a speaker 172, a camera 160, a position information sensor 150, an acceleration sensor 155, a storage unit 180, and a control unit 190. The blocks included in the terminal device 10 are electrically connected to each other, for example, by a bus or the like.

[0025] The communication unit 120 performs processes such as modulation and demodulation processes for communication between the terminal device 10 and an external device (for example, the server 20). The communication unit 120 performs transmission processes on signals generated by the control unit 190 and transmits the signals to the external device. The communication unit 120 performs reception processes on signals received from the external device and outputs the signals to the control unit 190.

[0026] The input device 13 is a device for a user to input instructions or information. The input device 13 is realized, for example, by a touch-sensitive device 131 in which an instruction is input by touching an operation surface. When the terminal device 10 is a PC or the like, the input device 13 may be realized by a reader, a keyboard, a mouse, or the like. The input device 13 converts an instruction input by a user into an electrical signal and outputs it to the control unit 190. The input device 13 may include, for example, a receiving port that receives an electrical signal input from an external input device.

[0027] The output device 14 is a device for presenting information to a user. The output device 14 is realized, for example, by a display 141. The display 141 displays various information according to the control of the control unit 190. The display 141 is realized, for example, by an LCD (Liquid Crystal Display) or an organic EL (Electro-Luminescence) display.

[0028] The audio processing unit 170 performs, for example, digital-analog conversion processing of an audio signal. The audio processing unit 170 converts a signal provided from the microphone 171 into a digital signal and provides the converted signal to the control unit 190. The audio processing unit 170 also provides the audio signal to the speaker 172. The audio processing unit 170 is realized, for example, by a processor for audio processing. The microphone 171 accepts audio input and provides an audio signal corresponding to the audio input to the audio processing unit 170. The speaker 172 converts the audio signal provided from the audio processing unit 170 into audio and outputs the audio to the outside of the terminal device 10.

[0029] Camera 160 is an imaging device that captures an image using visible light. In other words, camera 160 is a device that receives visible light with a light receiving element and outputs image data as an image capture signal. Camera 160 captures an image of a subject in a certain direction and within a certain image capture range relative to terminal device 10, and outputs image data as the image capture result. If camera 160 has a function that allows the image capture range, or more precisely, the angle of view, to be adjustable, camera 160 also outputs information regarding this angle of view. Such a function is known as a zoom function.

[0030] The position information sensor 150 is a sensor that detects the position of the terminal device 10, and is generally a GNSS device, for example, a GPS module. The GPS module is a receiving device used in a satellite positioning system. In the satellite positioning system, signals are received from at least three or four satellites, and the current position of the terminal device 10 equipped with the GPS module is detected as a coordinate value based on the received signals. The position information sensor 150 may detect the current position of the terminal device 10 from the position of a wireless base station to which the terminal device 10 is connected via the communication unit 120.

[0031] The acceleration sensor 155 is a sensor that detects the acceleration given to the terminal device 10. Preferably, the acceleration sensor 155 has a function of detecting the inclination about each axis (X-axis, Y-axis, Z-axis) of a three-dimensional coordinate system with the position of the terminal device 10 as the origin. The acceleration sensor 155 having such a function can detect the attitude of the terminal device 10, that is, the direction with respect to the X-axis, Y-axis, and Z-axis, by detecting the gravitational acceleration of the gravitational force with respect to the earth.

[0032] 1, and stores data and programs used by the terminal device 10. The programs include application programs such as a web browser application.

[0033] The control unit 190 is realized, for example, by the processor 19 reading a program stored in the storage unit 180 and executing instructions included in the program. The control unit 190 controls the operation of the terminal device 10. The control unit 190 performs the functions of an operation reception unit 191, a transmission / reception unit 192, and a presentation control unit 193 by operating according to the read program.

[0034] The operation reception unit 191 performs a process for receiving an instruction or information input from the input device 13. Specifically, the operation reception unit 191 receives an instruction or information input from the touch-sensitive device 131. The transmission / reception unit 192 performs a process for the terminal device 10 to transmit and receive data between the terminal device 10 and an external device in accordance with a communication protocol. Specifically, the transmission / reception unit 192 transmits an instruction or information input from a user to the server 20. The transmission / reception unit 192 receives information transmitted from the server 20. The presentation control unit 193 controls the output device 14 to present various information to the user.

[0035] <2.2 Server configuration> Fig. 3 is a block diagram showing an example of the configuration of the server 20 shown in Fig. 1. As shown in Fig. 3, the server 20 performs the functions of a communication unit 201, a storage unit 202, and a control unit 203.

[0036] The communication unit 201 performs processing for the server 20 to communicate with an external device (e.g., the terminal device 10). The storage unit 202 is realized by the memory 25 and the storage 26, and stores data and programs used by the server 20. The programs include application programs such as a web browser application. The storage unit 202 stores, for example, a prompt table 2021 and an app 2023.

[0037] The prompt table 2021 is a table that stores prompt information. The prompt information is information about an instruction sentence that is an input sentence to the AI ​​system 30, specifically, the LLM, and includes, for example, first prompt information and second prompt information.

[0038] The first prompt information is information related to a first instruction sentence that instructs the creation of a candidate sentence. The candidate sentence is a sentence that may be written in a target item of a target document. The target document is a subsidy application document of a type selected by a user. The target item is an item that may be included in the target document. The first instruction sentence is an example of an instruction sentence according to one embodiment of the present disclosure. The second prompt information is information related to a second instruction sentence that instructs the creation of a target document. Details of the prompt table 2021 will be described later.

[0039] Here, the target items may include, for example, required items designated in advance according to the type of the target document, as well as added points. Required items include, for example, information about the applicant (user), an outline of the new business, the purpose of using the subsidy, and a breakdown of expenses to which the subsidy will be applied. Added points are items that are written in the subsidy application document according to the subsidy application guidelines. Added points include, for example, in the application guidelines, (i) items that are required to be written, (ii) items that can be construed as being recommended to be written, and (iii) items that are recommended to be written based on the opinions / know-how of multiple experienced reviewers.

[0040] Compared to required items, the content of the score items has a higher degree of freedom. Therefore, by including score items in the items that can be included in the target document, content / expressions that take into account the story of the entire target document can be written in the score items. This makes it easier to create a draft in which all candidate sentences have a story as a whole.

[0041] The application 2023 is an application for managing the use of the document creation service by the user. The application 2023 runs, for example, in the background of other applications installed in the server 20 and monitors the processes executed by the user. The user can access the application 2023 in the server 20 by using a web browser application installed in the terminal device 10.

[0042] The server 20 may, for example, grasp and manage the usage status of the application 2023 and execute a predetermined analysis process. The application 2023 may be installed in the terminal device 10 and stored in the storage unit 180, for example.

[0043] The control unit 203 is realized by the processor 29 reading a program stored in the storage unit 202 and executing instructions included in the program. The control unit 203 controls the operation of the server 20. The control unit 203 operates according to the loaded program to fulfill the functions of a reception control module 2031, a transmission control module 2032, a presentation control module 2033, and a document creation processing module 2034.

[0044] The reception control module 2031 controls the process in which the server 20 receives a signal from an external device in accordance with a communication protocol. The transmission control module 2032 controls the process in which the server 20 transmits a signal to an external device in accordance with a communication protocol. The presentation control module 2033 controls the process in which various information is presented to the user.

[0045] The document creation processing module 2034, for example, inputs the reference information and the first instruction sentence to the LLM and outputs a candidate sentence from the LLM. The document creation processing module 2034, for example, stores the candidate sentence output from the LLM in a memory (not shown) in the control unit 203 or the storage unit 202.

[0046] Reference information is information linked to each target item, and is information that is referenced when creating candidate sentences. For example, if the item is information about an applicant, and the user is a corporation, the reference information would be the corporation name, corporation number, representative's name, summary of existing business, capital, etc. For example, if the item is a summary of a new business, the reference information would be the name of the new business, the purpose of the new business, specific implementation / activity details (including details of products / services to be provided), funding sources other than subsidies, business schedule, details of expected results / evaluation criteria, etc.

[0047] In this embodiment, one piece of reference information is associated with one target item. However, for example, one target item may be associated with multiple pieces of reference information, or one piece of reference information may be associated with multiple target items.

[0048] The document creation processing module 2034, for example, uses the output candidate sentence to output the candidate sentence from the LLM. In this embodiment, the document creation processing module 2034, for example, inputs the output candidate sentence in addition to the reference information and the first directive to the LLM to output the candidate sentence from the LLM.

[0049] The document creation processing module 2034, for example, causes the LLM to create candidate sentences in a predetermined order. In this embodiment, the LLM creates candidate sentences in order of the degree to which the contents of the target items have a large effect on the contents of other target items.

[0050] The order of creating candidate sentences in this embodiment will be described below using an example in which the user is a corporation. In this example, the target items are the item "information about the applicant," the item "purpose of use of subsidy," the item "specific implementation / activity content," and the item "expected effect."

[0051] First, the content of the item "Information about the applicant" generally has the greatest influence on the content of the other three target items among the four target items. Therefore, the LLM first creates a candidate text that can be written in the item "Information about the applicant". In this case, since there is no candidate text that has already been output, the data input to the LLM is the reference information linked to the item "Information about the applicant" and the first instruction text that corresponds to the item "Information about the applicant".

[0052] Next, the content of the item "Purpose of use of subsidy" generally has the greatest influence on the content of the other two target items among the remaining three target items. Therefore, the LLM creates candidate sentences that can be written in the item "Purpose of use of subsidy" secondly. In this case, the data input to the LLM will be the reference information linked to the item "Purpose of use of subsidy", the first instruction sentence corresponding to the item "Purpose of use of subsidy", and the candidate sentences that can be written in the output item "Information on the applicant".

[0053] Next, the contents of the remaining item "Specific implementation / activity details" generally have a greater impact on the contents of the other target items than the remaining item "Expected effects". Therefore, the LLM creates a third candidate sentence that can be written in the item "Specific implementation / activity details". In this case, the data input to the LLM are the reference information linked to the item "Specific implementation / activity details", the first instruction sentence corresponding to the item "Specific implementation / activity details", and the candidate sentence that can be written in the output item "Intended use of grant funds". Here, the input data to the LLM may include the candidate sentence that can be written in the output item "Information on the applicant".

[0054] Finally, the LLM creates fourth candidate sentences that may be written in the "Expected Effects" field. In this case, the data input to the LLM are the reference information linked to the "Expected Effects" field, the first instruction corresponding to the "Expected Effects" field, and candidate sentences that may be written in the output field "Specific Implementation / Activities." Here, the input data to the LLM may include at least one of candidate sentences that may be written in the output field "Information about the Applicant" and candidate sentences that may be written in the output field "Intended Use of Grants."

[0055] The method of determining the degree of influence on the contents of other target items is not limited to the above example. For example, the content of the reference information may be taken into consideration in the above determination method. Specifically, even if it is common to prepare candidate sentences that can be written in the item "Information on the applicant" before candidate sentences that can be written in the item "Purpose of use of the subsidy", if the content of the reference information linked to the item "Information on the applicant" is insufficient, while the reference information linked to the item "Purpose of use of the subsidy" is detailed enough to affect the candidate sentences that can be written in the item "Information on the applicant", the LLM may prepare the candidate sentences that can be written in "Purpose of use of the subsidy" first.

[0056] Furthermore, the order in which the candidate sentences are created is merely an example. As long as all of the candidate sentences that can be written in the target document have a story-like quality as a whole, the order in which the candidate sentences are created may be in any order. For example, the LLM may create candidate sentences according to the general arrangement of each target item in the target document. That is, the LLM may create candidate sentences for each target item in order, for example, from the target item that is generally located at the top to the target item that is generally located further down.

[0057] The document creation processing module 2034 creates a draft of the target document (hereinafter, abbreviated as "draft") based on, for example, the candidate sentences output from the LLM. The document creation processing module 2034 may, for example, cause the LLM to output the draft. In addition, for example, the document creation processing module 2034 may obtain the candidate sentences output from the LLM and create the draft itself. In this embodiment, the document creation processing module 2034 inputs, for example, the candidate sentences and the second instruction sentence to the LLM and causes the LLM to output the draft.

[0058] [3 Data Structure] 4 and 5 are diagrams showing the data structures of tables stored in the server 20. Note that Fig. 4 and Fig. 5 are merely examples and do not exclude data that is not listed. In addition, even data that is listed in the same table may be stored in separate storage areas in the storage unit 202.

[0059] Fig. 4 is a diagram showing the data structure of the prompt table 2021. The prompt table 2021 shown in Fig. 4 is a table having columns of prompt data, item name, reference information, and type name, with a prompt ID as a key.

[0060] The "prompt ID" item is an item that stores an identifier for uniquely identifying a prompt. The "prompt data" item is an item that stores, for example, text data of the first instruction statement or the second instruction statement that is partially blank. The "prompt data" item may store, for example, reference information (path) for a prompt data file stored in another memory area. The "item name" item is an item that stores the name of each item listed in the subsidy application document. The "reference information" item is an item that stores reference information linked to the target item of the name stored in the "item name" item. The "type name" item is an item that stores the name of the type of subsidy.

[0061] The text data of the first instruction sentence stored in the item "prompt data" is, for example, as follows: "Please refer to XXX (reference information and output candidate sentences are completed) and create the sentence (candidate sentence) to be written in item △△△ (target item is written)."

[0062] The text data of the second instruction sentence stored in the item "prompt data" is, for example, as follows: "Please create a draft of the grant application document ▽▽▽ (the type of the target document is listed) that includes all of the sentences △△△ (blank part: candidate sentence will be completed) to be entered in item XXX, ···, and sentences ☆☆☆ (blank part: candidate sentence will be completed) to be entered in item □□□."

[0063] When the document creation processing module 2034 creates a draft by itself, the text data of the second instruction sentence is not necessary and is not stored in the item "prompt data". In this case, the storage unit 202 may store, for example, an application document table in which a format (or reference information (path) to the format) for each type of subsidy application document is stored. The document creation processing module 2034 may create a draft by, for example, reading a format corresponding to the target document from the application document table or the like and writing candidate sentences output from the LLM in the format. Alternatively, the contents of the application document table may be incorporated into the prompt table 2021.

[0064] In addition, in this embodiment, the first instruction stored in the item "prompt data" is in a format in which reference information and output candidate sentences are incorporated into text data, and the second instruction stored in the item "prompt data" is in a format in which candidate sentences are incorporated into text data, but this is not limited to this case. For example, each of the text data of the first instruction and the second instruction stored in the item "prompt data" may be, for example, a complete instruction without any blanks. In this case, for example, the reference information and output candidate sentences may be input into the LLM separately from the first instruction, and the candidate sentences may be input into the LLM separately from the second instruction.

[0065] Also, for example, at least one of the first instruction statement and the second instruction statement may not be stored in the prompt table 2021. In this case, for example, the operation receiving unit 191 may receive an input of at least one of the first instruction statement and the second instruction statement, so that the document creation processing module 2034 acquires at least one of the first instruction statement and the second instruction statement. When the operation receiving unit 191 receives an input of both the first instruction statement and the second instruction statement, the prompt table 2021 is not necessary.

[0066] Fig. 5 is a diagram showing the data structure of the order table 2022. The order table 2022 shown in Fig. 5 is a table having columns of type name and order, with type ID as a key.

[0067] The "Type ID" field stores an identifier for uniquely identifying the type of subsidy. The "Type Name" field stores the name of the type of subsidy. The "Order" field stores the names of each item listed in the subsidy application document based on the order in which the LLM creates candidate sentences.

[0068] In the example of FIG. 5, the item "order" includes, for example, items "1st" to "7th." Item "1st" stores the name of the item for which, among all the items, the LLM will create a candidate sentence first (first). Item "2nd" stores the name of the item for which, among all the items, the LLM will create a candidate sentence second. Following this rule, the names of the applicable items are stored up to item "7th."

[0069] For example, in a column in a specific order, the names of multiple items may be stored in that column. That is, depending on the subsidy document, candidate sentences for multiple items may be created in the same order. For example, in the example of FIG. 5, the name of item 12 may be stored in addition to the name of item 11 in the item "1st" corresponding to type 1, and the names of items 45 to 47 may be stored in addition to the name of item 44 in the item "4th" corresponding to type 4.

[0070] In this embodiment, it is assumed that all of the items whose names are stored in the "Order" item are included in the draft. Also, in this embodiment, it is assumed that the names of the items stored in each of the "1st" to "7th" items are determined, for example, by an administrator / manager judging the degree to which the contents of a given item affect the contents of other items, and are determined based on the results of that judgment. However, the storage destination for the names of each item may also be determined according to the general arrangement of each item in subsidy application documents of each type.

[0071] [4 actions] An example of the operation of the server 20 when a draft of a target document is presented will be described below. Fig. 6 is a flowchart showing an example of the operation of the server 20 when a draft of a target document is presented.

[0072] In step S11 shown in FIG. 6, the server 20 accepts the selection of the type of subsidy (step of accepting selection).

[0073] Specifically, for example, the operation receiving unit 191 receives a selection operation of a subsidy type by the user. The transmitting / receiving unit 192 transmits, for example, type information on the type of subsidy selected by the user to the server 20. The reception control module 2031 receives, for example, the type information transmitted from the terminal device 10. In this way, the server 20 receives the selection of the subsidy type by the user.

[0074] An example of a screen on display 141 when the selection of the type of subsidy is accepted will be described below with reference to Fig. 7. Fig. 7 is a schematic diagram showing an example of a screen on display 141 when the selection of the type of subsidy is accepted.

[0075] 7, for example, a selection screen 1411 is displayed on the display screen of the display 141. The selection screen 1411 is a user interface for accepting a selection of a type of subsidy by a user. On the selection screen 1411, for example, three types of selection buttons 1413 are displayed in an area 1412.

[0076] That is, the operation receiving unit 191 receives, for example, a presentation request for the selection screen 1411 from a user, and transmits it to the server 20. For example, the presentation control module 2033 receives the presentation request, and transmits information for displaying the selection screen 1411 to the terminal device 10. For example, the presentation control unit 193 receives the information for displaying the selection screen 1411, and causes the selection screen 1411 to be displayed on the display screen of the display 141.

[0077] 7, for example, when the user taps one of the three types of selection buttons 1413, the type of subsidy corresponding to the tapped selection button 1413 is selected. That is, the operation acceptance unit 191, for example, accepts the tap operation of the selection button 1413 by the user, and transmits type information to the server 20. The reception control module 2031 receives, for example, the type information transmitted from the terminal device 10. This determines the type of the target document.

[0078] 7 is merely an example. Various variations are assumed for the display contents and display modes of the selection screen 1411. For example, the number of selection buttons 1413 increases or decreases depending on the number of types of subsidies selectable in the system 1.

[0079] In step S12, the server 20 accepts an input of reference information linked to each target item of the selected target document (input accepting step).

[0080] Specifically, for example, the operation reception unit 191 receives an input operation of reference information by the user. The transmission / reception unit 192 transmits, for example, the reference information to the server 20. For example, the reception control module 2031 receives the reference information transmitted from the terminal device 10, and stores the reference information in the prompt table 2021 in the item "reference information" of the type corresponding to the type information. In this way, the server 20 receives the reference information input by the user, and links the reference information to the target item. The operation reception unit 191 and the reception control module 2031 execute the process of step S12 until the server 20 receives all of the reference information for each target item.

[0081] In this embodiment, it is assumed that it is determined which target item the reference information is linked to when the server 20 accepts the input of the reference information. In other words, the user is assumed to understand in advance what kind of reference information should be input for each target item, and then to perform the input operation of the reference information in accordance with that understanding.

[0082] However, the timing at which it is determined which target item the reference information is linked to is not limited to the example of this embodiment. For example, the document creation processing module 2034 may determine the target item to which the reference information is linked by performing natural language analysis (rule-based) on the reference information. Alternatively, the document creation processing module 2034 may input the reference information to a trained machine learning model (not shown) and cause the machine learning model to output the target item to which the reference information is linked.

[0083] In this way, by determining the target items to which the reference information is linked using natural language analysis or a trained machine learning model, the user does not need to understand in advance what kind of reference information should be input for each target item. In addition, the user does not need to input reference information that includes nuances that clarify the relationship with the target items, or input reference information in a form that has an appropriate sentence structure in relation to the target items. Therefore, the user can create a draft with a small amount of reference information. Not only that, the user can create a draft without even understanding what the required target items are, and no literacy regarding subsidies is required. These things allow the user to create a draft more easily, further improving user convenience.

[0084] An example of a screen of display 141 when input of reference information is accepted will be described below with reference to Fig. 8. Fig. 8 is a schematic diagram showing an example of a screen of display 141 when input of reference information is accepted.

[0085] 8, for example, an input screen 1421 is displayed on the display screen of the display 141. The input screen 1421 is a user interface for accepting input of reference information by a user. For example, four types of items 1422 are displayed on the input screen 1421 as target items of a target document. In addition, for example, a question box 1423 is displayed directly below each item 1422 on the input screen 1421.

[0086] That is, the operation receiving unit 191 receives, for example, a presentation request for the input screen 1421 from a user, and transmits it to the server 20. For example, the presentation control module 2033 receives the presentation request, and transmits information for displaying the input screen 1421 to the terminal device 10. For example, the presentation control unit 193 receives information for displaying the input screen 1421, and causes the input screen 1421 to be displayed on the display screen of the display 141.

[0087] 8, for example, the user inputs reference information linked to each of the four types of items 1422 into the corresponding question box 1423. That is, the operation receiving unit 191 transmits the reference information to the server 20 by, for example, receiving an input operation of the reference information into the question box 1423 by the user. The reception control module 2031 receives the reference information transmitted from, for example, the terminal device 10. This completes the reception of the input of the reference information.

[0088] 8 is merely an example. Various variations are assumed for the display contents and display modes of the input screen 1421. For example, depending on the type of subsidy selected by the user, some or all of the contents of the items 1422 change, and the number of items 1422 increases or decreases.

[0089] In step S13, the server 20 inputs the input reference information and the first instruction sentence into the LLM and causes the LLM to output candidate sentences. At this time, the server 20 causes the LLM to output candidate sentences using the candidate sentences that have already been output, and causes the LLM to create candidate sentences for each target item in a predetermined order (the above is the output step).

[0090] Specifically, for example, the document creation processing module 2034 transmits reference information for each target item, a first instruction sentence corresponding to each target item, and the output candidate sentences to the AI ​​system 30 in a predetermined order.

[0091] That is, the document creation processing module 2034, for example, refers to the order table 2022 to identify the creation order of the candidate sentence for each target item. The document creation processing module 2034, for example, reads out the first instruction sentence corresponding to the target item (hereinafter, creation item) that has become the creation order of the candidate sentence from the prompt table 2021 according to the identified creation order. In addition, the document creation processing module 2034, for example, reads out the output candidate sentence from the memory of the control unit 203 or the storage unit 202.

[0092] At this time, there is no particular limitation on the content and number of candidate sentences that the document creation processing module 2034 reads out. In other words, there is no particular limitation on the content and number of candidate sentences that have been output and are input to the LLM. The document creation processing module 2034 may, for example, read out the first candidate sentence output from the LLM and the candidate sentence output from the LLM one time before the creation order. Alternatively, for example, the document creation processing module 2034 may read out only the first candidate sentence output from the LLM regardless of the creation order. Alternatively, for example, the document creation processing module 2034 may read out only the candidate sentence output from the LLM one time before the creation order. Alternatively, the document creation processing module 2034 may read out all candidate sentences output from the LLM up to one time before the creation order.

[0093] The document creation processing module 2034, for example, completes the blank portion of the first instruction sentence based on the reference information linked to the creation item and the output candidate sentence. The document creation processing module 2034 transmits the first instruction sentence completed using the reference information and the output candidate sentence to the AI ​​system 30.

[0094] The AI ​​system 30, for example, inputs the first instruction sentence (including reference information and output candidate sentences) received from the server 20 to the LLM and causes the LLM to output candidate sentences for the creation item. The AI ​​system 30 transmits the candidate sentences for the creation item output from the LLM to the server 20. The reception control module 2031 receives, for example, the candidate sentences for the creation item transmitted from the AI ​​system 30.

[0095] The document creation processing module 2034 and the AI ​​system 30 execute the above-mentioned series of processes in step S13, for example, until the reception control module 2031 receives all of the candidate sentences for each target item. When the reception control module 2031 receives all of the candidate sentences for each target item, the server 20 ends the process of outputting the candidate sentences from the LLM.

[0096] In this operation example, the document creation processing module 2034 identifies the creation order by referring to the order table 2022, but is not limited to this example. The document creation processing module 2034 may identify the creation order by using the LLM instead of the order table 2022, for example.

[0097] That is, the document creation processing module 2034 may input, for example, a part or all of a data set of each target item and reference information linked to the target item to the LLM together with an instruction statement instructing the specification of the creation order. The LLM may, for example, analyze the contents of each of the input target items and reference information to determine the creation order and output the determination result. Here, the LLM may execute the above-mentioned analysis process based on, for example, the degree to which the contents of the target items affect the contents of other target items. The document creation processing module 2034 may specify the creation order by, for example, acquiring the determination result output from the LLM. In this case, the order table 2022 is not necessary. In addition, the LLM that determines the creation order may be the same as or different from the LLM that outputs the candidate sentences.

[0098] Also, for example, in this operation example, the document creation processing module 2034 inputs the output candidate sentence to the LLM, but is not limited to this example. For example, the document creation processing module 2034 may exclude the output candidate sentence from the input data to the LLM.

[0099] In this case, for example, a first instruction statement including content specifying an output candidate sentence to be referred to in creating a candidate sentence may be stored in the prompt table 2021. In this case, the text data of the first instruction statement may be, for example, the following content. "Please create the sentence (candidate sentence) to be written in item △△△ (where the target item is written) by referring to XXX (blank part: reference information will be added) and the sentence output one time ago (candidate sentence that has already been output)." "Please create a sentence (candidate sentence) to be written in item △△△ (target item is written) by referring to XXX (blank part: reference information will be added) and all the sentences that have been output so far (output candidate sentences)."

[0100] Also, for example, in this operation example, the document creation processing module 2034 inputs all of the first instruction sentences corresponding to each target item into one LLM, but this is not limited to this example. For example, the document creation processing module 2034 may input the first instruction sentences corresponding to each target item separately into multiple LLMs. That is, the AI ​​system 30 may include multiple LLMs that output candidate sentences.

[0101] In this case, for example, an LLM specialized for each target item may be constructed. For example, this LLM may be constructed using sentences written in a predetermined target item in the past as learning data so as to improve compatibility with the predetermined target item. For example, the document creation processing module 2034 may input a first instruction sentence corresponding to the predetermined target item into the LLM specialized for the predetermined target item.

[0102] In an example in which the AI ​​system 30 includes multiple LLMs, the multiple LLMs may be structured, for example, for each type of grant.

[0103] In step S14, the server 20 creates a draft based on the candidate sentences output from the LLM (creating step).

[0104] Specifically, for example, the document creation processing module 2034 reads the second instruction sentence from the prompt table 2021 and completes the blank part of the second instruction sentence based on the candidate sentence of each target item. The document creation processing module 2034 transmits the completed second instruction sentence using the candidate sentence of each target item to the AI ​​system 30.

[0105] The AI ​​system 30, for example, inputs the second instruction sentence (including candidate sentences for each target item) received from the server 20 into the LLM and causes the LLM to output a draft. In this operation example, the draft output from the LLM includes each target item and all candidate sentences for each target item. The AI ​​system 30 transmits the draft output from the LLM to the server 20. The reception control module 2031 receives, for example, the draft transmitted from the AI ​​system 30. When the reception control module 2031 receives the draft, the server 20 ends the process of creating the draft.

[0106] In addition, the LLM that outputs the draft may be the same as the LLM that outputs the candidate sentences, or it may be different. If the LLM that outputs the draft is different from the LLM that outputs the candidate sentences, the LLM that outputs the draft may be constructed for each type of grant, for example.

[0107] Also, for example, the draft output from the LLM may not include a part of each target item and a part of the candidate sentences for each target item. That is, when the LLM determines that at least one of the input target items and the input candidate sentences for each target item contains some inappropriate content or some insufficient content, the LLM may output a draft that does not include the target item and the candidate sentences for the target item.

[0108] In an example where the LLM outputs a draft that does not include the target item and some of the candidate sentences for that target item, the LLM may, for example, determine that if the target item is a required item, it must be included in the draft along with the candidate sentences, and if the target item is a point-adding item, it may determine that some or all of the candidate sentences are not included in the draft depending on the content, etc. of the target item. This makes it possible to create a draft that does not contain unnatural sentences or unnecessary sentences in the point-adding items while avoiding omissions in the description, such as failure to include required items. Therefore, it is possible to create a draft that ensures a storyline while reliably meeting the application requirements.

[0109] In step S15, the server 20 presents the created draft to the user.

[0110] Specifically, for example, the presentation control module 2033 presents to the user the draft created by the document creation processing module 2034. The transmission control module 2032 transmits to the terminal device 10, for example, information for displaying the draft.

[0111] The transmitting / receiving unit 192 receives, for example, information for displaying the draft from the server 20. The presentation control unit 193 causes the display 141 to display the draft on its display screen when, for example, a presentation request is received from a user.

[0112] When the output device 14 is a printer, the presentation control module 2033 may, for example, control the presentation control unit 193 to print out a paper medium on which the contents of the draft are printed from the output device 14. In addition, it is not essential that the presentation control module 2033 presents the draft.

[0113] Hereinafter, a screen example of the display 141 when the draft is presented to the user will be described with reference to Fig. 9. Fig. 9 is a schematic diagram showing a screen example of the display 141 when the draft is presented to the user.

[0114] 9, for example, a draft 1431 is displayed in an area 14311 on the display screen of the display 141. In the draft 1431, for example, four types of items 1432 are described as target items. In addition, in the draft 1431, for example, candidate sentences 1433 are described directly below each item 1432. Furthermore, a confirm button 1434 is displayed in an area 14312 on the display screen.

[0115] That is, the operation receiving unit 191, for example, receives a presentation request for the draft 1431 from a user and transmits it to the server 20. The presentation control module 2033, for example, receives the presentation request and transmits information for displaying the draft 1431 to the terminal device 10. The presentation control unit 193, for example, receives the information for displaying the draft 1431 and causes the draft 1431 to be displayed on the display screen of the display 141.

[0116] Furthermore, the operation receiving unit 191 transmits a confirmation request to the server 20, for example, by receiving a tap operation of the confirm button 1434 by the user. The confirmation request is a request to confirm the content of the target document with the content of the draft 1431. For example, by receiving the confirmation request, the document creation processing module 2034 stores the draft 1431 in the storage unit 202 as a finalized version of the target document.

[0117] 9 is merely an example. Various variations are assumed for the display content and display mode of the draft 1431. In addition, for example, the presentation control unit 193 may not display the confirmation button 1434.

[0118] [5 Summary] As described above, in this embodiment, the reception control module 2031 receives, for example, type information transmitted from the terminal device 10. The reception control module 2031 receives, for example, reference information transmitted from the terminal device 10 and stores it in the item "reference information" of the prompt table 2021. The document creation processing module 2034 refers to the order table 2022 to identify the creation order of the candidate sentences of each target item. The document creation processing module 2034 transmits the first instruction sentence completed using the reference information linked to the creation item and the output candidate sentence to the AI ​​system 30. The AI ​​system 30 inputs the received first instruction sentence to the LLM, outputs the candidate sentences of the creation item from the LLM, and transmits them to the server 20. The reception control module 2031 receives the candidate sentences of the creation item transmitted from the AI ​​system 30. The document creation processing module 2034 transmits the second instruction sentence completed using the candidate sentences of each target item to the AI ​​system 30. The AI ​​system 30 inputs the received second instruction sentence into the LLM, and causes the LLM to output a draft, which is then transmitted to the server 20. The reception control module 2031 receives the draft transmitted from the AI ​​system 30.

[0119] As a result, the candidate sentences written in the draft will be the contents of the reference information plus the candidate sentences that have already been output. Also, the candidate sentences written in the draft are created in an order according to the degree of influence that a given candidate sentence has on the other candidate sentences. Therefore, it is possible to create a draft in which all the candidate sentences have a story-like quality as a whole.

[0120] [6. Modifications] The system 1 may adopt various aspects other than the example of this embodiment. Below, some examples of modified examples of the system 1 will be described.

[0121] <6.1 Output of suggested revisions for candidate sentences> The server 20 may, for example, accept corrections to the output candidate sentences. Then, the server 20 may, for example, input the input corrections to the LLM and cause the LLM to output a proposed correction to the candidate sentences.

[0122] Specifically, for example, the operation receiving unit 191 may receive an input operation of the correction content of the candidate sentence by the user, and transmit correction information indicating the correction content to the server 20.

[0123] Here, it is assumed that, as a precondition for the user to perform the above-mentioned input operation, for example, a draft output from the LLM is displayed on the display screen of the display 141 before the above-mentioned input operation (see FIG. 9). It is also assumed that the user visually confirms the draft displayed on the display screen before the above-mentioned input operation. Note that in this situation, for example, a part or all of the candidate sentences output from the LLM may be displayed on the display screen instead of the draft.

[0124] The reception control module 2031 may receive, for example, correction information transmitted from the terminal device 10. The document creation processing module 2034 may create, for example, an instruction sentence for instructing correction of the candidate sentence based on the correction information, and transmit the correction information and the instruction sentence to the AI ​​system 30. At this time, the document creation processing module 2034 may further transmit, for example, at least one of the candidate sentence to be corrected and reference information corresponding to the candidate sentence to be corrected to the AI ​​system 30.

[0125] The AI ​​system 30 may, for example, input the correction information and the instruction sentence into the LLM and output a correction proposal for the candidate sentence to be corrected from the LLM. In this case, the LLM that outputs the correction proposal for the candidate sentence to be corrected may be, for example, the same as the LLM that outputs the candidate sentence, or may be different. The AI ​​system 30 may, for example, transmit the correction proposal output from the LLM to the server 20. The reception control module 2031 may, for example, receive the correction proposal transmitted from the AI ​​system 30.

[0126] In this way, the server 20 outputs the proposed revisions to the candidate sentences from the LLM, thereby reducing the man-hours required for the user to revise the draft. In addition, the user is less required to adjust the revision content during the revision work, taking into consideration how the revision work will affect the contents of the candidate sentences other than the target of revision. This allows the user to efficiently proceed with the revision work of the draft. Furthermore, the user can extract preferable descriptions from both the original text of the candidate sentence and the proposed revisions, and edit the candidate sentences themselves.

[0127] In a variation in which the server 20 outputs a proposed revision to a candidate sentence from the LLM, the server 20 may, for example, present the candidate sentence and the proposed revision to the candidate sentence side by side.

[0128] Specifically, for example, the presentation control module 2033 may present the original text of the candidate sentence to be corrected and the proposed revision side by side to the user. The transmission control module 2032 may transmit, for example, information for displaying the candidate sentence to be corrected and information for displaying the proposed revision to the terminal device 10.

[0129] The transmitting / receiving unit 192 may receive, for example, information for displaying the correction target and information for displaying the correction suggestion from the server 20. For example, when the presentation control unit 193 receives a presentation request from a user, it may display the original text of the candidate sentence to be corrected and the correction suggestion side by side on the display screen of the display 141.

[0130] Here, there is no particular limitation on the arrangement of the original text and the proposed revision of the candidate text to be revised. For example, the original text and the proposed revision may be arranged horizontally on the page, or vertically on the page. However, it is preferable that the presentation control unit 193 displays the original text and the proposed revision close to each other to an extent that the user can easily compare the original text and the proposed revision.

[0131] In this way, the server 20 presents the original text of the candidate sentence to be corrected and the proposed revision side by side, so that the user can compare the two. This makes it easy for the user to decide whether to select the original text or the proposed revision, or whether to edit the proposed revision, and allows the user to efficiently proceed with the work of correcting the draft.

[0132] In a variant in which the server 20 outputs suggested revisions to a candidate sentence from the LLM, the server 20 may, for example, when the revision content affects the content of another candidate sentence, output suggested revisions to the other candidate sentence from the LLM.

[0133] Specifically, for example, the document creation processing module 2034 may sequentially read out the candidate sentences already output from the memory or storage unit 202 of the control unit 203, and compare them with the proposed revision of the candidate sentence to be revised to determine whether the revision content of the proposed revision affects the content of the other candidate sentences. When it is determined that the revision content of the proposed revision affects the content of the other candidate sentences, the document creation processing module 2034 may create an instruction sentence to instruct the revision of the other candidate sentences based on the proposed revision and the other candidate sentences affected by the proposed revision, for example. The document creation processing module 2034 may transmit, for example, the proposed revision, the other candidate sentences affected by the proposed revision, and the instruction sentence to the AI ​​system 30. Note that the document creation processing module 2034 may also transmit, for example, reference information corresponding to the other candidate sentences affected by the proposed revision to the AI ​​system 30.

[0134] The AI ​​system 30 may, for example, input the proposed revision, other candidate sentences affected by the proposed revision, and an instruction sentence to an LLM, and output the proposed revision of the other candidate sentence from the LLM. In this case, the LLM that outputs the proposed revision of the other candidate sentence may be, for example, the same as or different from the LLM that outputs the candidate sentence. The AI ​​system 30 may, for example, transmit the proposed revision of the other candidate sentence output from the LLM to the server 20. The reception control module 2031 may, for example, receive the proposed revision of the other candidate sentence transmitted from the AI ​​system 30.

[0135] In this way, the server 20 receives the correction content of the correction target from the LLM and outputs the correction proposals for the other candidate sentences. This reduces the need for the user to adjust the correction content while taking into consideration how the correction work will affect the contents of the candidate sentences other than the correction target. Therefore, the user can proceed with the correction work of the draft more efficiently.

[0136] The document creation processing module 2034 may, for example, cause the LLM to determine whether the correction content of the proposed correction affects the content of other candidate sentences. That is, the document creation processing module 2034 may, for example, sequentially read out the candidate sentences that have already been output from the memory of the control unit 203 or the storage unit 202, input them to the LLM together with the proposed correction of the candidate sentence to be corrected, and cause the LLM to output the judgment result. Here, the LLM that outputs the judgment result may be the same as or different from the LLM that outputs the candidate sentences.

[0137] <6.2 Other Modifications> First, for example, in the process of outputting candidate sentences from the LLM, the server 20 may further input real-time search information related to a predetermined item to the LLM.

[0138] Real-time search information (hereinafter, latest search information) is, for example, the latest search information related to the type of subsidy selected by the user. The latest search information may be searched, for example, by the user or by a third party other than the user. The content of the latest search information itself may be any information related to the type of subsidy selected by the user, regardless of the degree of relevance.

[0139] Specifically, for example, the operation accepting unit 191 may accept an input operation of the latest search information by the user, and transmit the latest search information to the server 20. Alternatively, the operation accepting unit 191 may store the latest search information input by the user in a memory or storage unit 180 (not shown) in the control unit 190. In this case, when the transmitting / receiving unit 192 accepts a request to transmit the latest search information by the user, for example, the transmitting / receiving unit 192 may read the latest search information from the memory or storage unit 180 and transmit it to the server 20.

[0140] The reception control module 2031 may, for example, receive search information transmitted from the terminal device 10. The document creation processing module 2034 may, for example, transmit the reference information, the output candidate sentence, the first instruction sentence, and the search information to the AI ​​system 30.

[0141] The AI ​​system 30 may, for example, input the reference information, the output candidate sentence, the first instruction sentence, and the search information to the LLM and cause the LLM to output the candidate sentence. The processes of the server 20 and the AI ​​system 30 after the LLM outputs the candidate sentence are the same as those in the example of this embodiment.

[0142] In this way, the server 20 outputs candidate sentences that take into account the content of the latest search information from the LLM, making it possible to create a draft that reflects the latest trends, latest research results, etc., related to the type of subsidy selected by the user.

[0143] The server 20 may then create a draft, for example, in a format compatible with the application format for each type of grant.

[0144] Specifically, for example, the AI ​​system 30 may cause the LLM to output a draft in a format compatible with application formats for all other types of grants that the system 1 can handle. Each process of the server 20 and the AI ​​system 30 after the LLM outputs the draft is similar to the example of this embodiment.

[0145] In this way, the server 20 outputs the draft from the LLM in a format compatible with all other types of subsidy application forms that the system 1 can handle. This allows the contents and format of the draft to be reflected directly in the format of other subsidy application documents (e.g., copy and paste). Therefore, once a user creates a draft for one type of subsidy application document, the user does not need to create other types of drafts from scratch, which greatly simplifies the work when creating multiple types of drafts. This improves the user's convenience in creating subsidy application documents. In addition, it is possible to greatly reduce the amount of work required by the user when creating subsidy application documents other than the target document.

[0146] Next, the server 20 may, for example, input the edited candidate sentence to an LLM and cause the LLM to output a determination result as to whether or not the edited sentence satisfies a predetermined condition.

[0147] Specifically, for example, the operation accepting unit 191 may accept an editing operation of a candidate sentence by a user, and transmit the edited candidate sentence to the server 20. Alternatively, the operation accepting unit 191 may store the candidate sentence edited by the user in a memory or storage unit 180 (not shown) in the control unit 190. In this case, when the transmitting / receiving unit 192 accepts a request to transmit the candidate sentence edited by the user, for example, it may read out the edited candidate sentence from the memory or storage unit 180 and transmit it to the server 20.

[0148] The transmission control module 2032 may, for example, transmit the edited candidate sentence received from the terminal device 10 to the AI ​​system 30. The AI ​​system 30 may, for example, input the edited candidate sentence to an LLM and cause the LLM to output a judgment result as to whether or not the edited candidate sentence satisfies a predetermined condition. Here, the LLM that outputs the judgment result may be the same as or different from the LLM that outputs the candidate sentence, for example.

[0149] There is no particular limitation on the predetermined condition. The predetermined condition may be, for example, that the edited candidate text is in accordance with the solicitation requirements for the subsidy application that is the subject of the target document. The predetermined condition may be, for example, that the edited candidate text matches the content of a past adoption case of the subsidy that is the subject of the target document. The predetermined condition may be, for example, that the edited candidate text always includes a candidate text for a required item. If the target item includes a point-adding item, the predetermined condition may be that the edited candidate text always includes a candidate text for the point-adding item. Alternatively, the predetermined condition may be set for each target item, may be set only for the required items, or may be set only for the point-adding items.

[0150] The AI ​​system 30 may, for example, transmit the judgment result output from the LLM to the server 20. The reception control module 2031 may, for example, receive the judgment result transmitted from the AI ​​system 30. The presentation control module 2033 may, for example, present the judgment result to the user.

[0151] In this way, the server 20 inputs the edited candidate sentences to the LLM and causes the LLM to output the judgment result as to whether the edited candidate sentences satisfy a predetermined condition. This allows the user to further edit the edited candidate sentences in accordance with the judgment result, for example, by visually checking the judgment result displayed on the display screen of the display 141. Therefore, a draft can be created in which all the candidate sentences as a whole have a more story-like quality.

[0152] Note that the server 20 may output the judgment result to, for example, a machine learning model other than the generative AI model, instead of the LLM. Also, for example, the document creation processing module 2034 may perform the above-mentioned judgment process instead of the LLM.

[0153] Next, the server 20 may, for example, accept a change in the type of grant, and may create a draft of a grant application document for the changed type based on, for example, the reference information.

[0154] Specifically, for example, the operation receiving unit 191 may receive an operation by the user to change the type of subsidy, and transmit to the server 20 post-change type information indicating the type of subsidy after the change.

[0155] The reception control module 2031 may receive, for example, the changed type information transmitted from the terminal device 10. The document creation processing module 2034 may read, for example, from the prompt table 2021, a part or all of the reference information used when creating the draft. The document creation processing module 2034 may read, for example, a part or all of the candidate sentences included in the draft from the memory of the control unit 203 or the storage unit 202. The document creation processing module 2034 may create, for example, an instruction statement instructing the creation of a draft of the subsidy application document related to the changed type (hereinafter, the changed draft). The document creation processing module 2034 may transmit, for example, the reference information, the candidate sentences, and the instruction statement to the AI ​​system 30.

[0156] The AI ​​system 30 may, for example, input reference information, candidate sentences, and instruction sentences to the LLM and output the modified draft from the LLM. Here, the LLM that outputs the modified draft may be the same as or different from the LLM that outputs the candidate sentences. Each process of the server 20 and the AI ​​system 30 after the LLM outputs the modified draft is the same as the example of this embodiment. Note that, instead of the LLM, for example, the document creation processing module 2034 may create the modified draft. Also, inputting the candidate sentences into the LLM is not essential in creating the modified draft, and at least the reference information may be input into the LLM.

[0157] In this way, the server 20 creates a new draft after the change by simply accepting a change in the type of subsidy, using the information used to create the draft. This eliminates the need for the user to input new reference information when the type of subsidy is changed, and can greatly simplify and streamline the work of creating a new draft after the type is changed.

[0158] Next, the AI ​​system 30 may include, for example, a generative AI model that is trained using a subsidy application document created by a user as training data as a generative AI model that outputs candidate sentences. This generative AI model may be, for example, an LLM, or a generative AI model other than an LLM, such as a multimodal generative AI model. However, in this modified example, it is assumed that the user who created the subsidy application document has given permission to use the subsidy application document as training data.

[0159] The generative AI model of this modified example is a model specialized for, for example, the characteristics of the user's writing style, the characteristics of the user's sentence expression / phrasing, vocabulary selection, and / or sentence structure. Therefore, the user can create a dofrat with the user's desired writing style, etc., by uploading, for example, a business plan or the like that the user created in the past and using it as learning data. This can reduce the effort of the user's correction work.

[0160] The server 20 may then, for example, present the created draft to the user and to parties with whom the user has a business relationship.

[0161] The concept of a related party includes all people who have a business relationship with the user, regardless of the degree of the relationship. Examples of related parties include employees of the user's company, employees of other companies that are planning the project for which the user is applying for a subsidy together with the user, and external experts (consulting companies, lawyers, etc.) that are affiliated with the user's company.

[0162] Specifically, for example, the presentation control module 2033 may present the draft to the user and related parties. The transmission control module 2032 may transmit information for displaying the draft to the terminal device 10 and related parties' terminal devices (not shown: hereinafter, related terminal devices). Subsequent processes of the transmission / reception unit 192 and the presentation control unit 193 are similar to those in the example of this embodiment.

[0163] The transmitting / receiving unit of the associated terminal device may receive, for example, information for displaying the draft from the server 20. The presentation control unit of the associated terminal device may display the draft on the display screen of the display of the associated terminal device when receiving, for example, a presentation request from a user. Note that the presentation control module 2033 may, for example, differentiate the presentation mode of the draft for the user from the display mode of the draft for related parties.

[0164] In this way, the server 20 presents the draft to the related parties, which makes it easy for the user and the related parties to share / discuss information about the draft. Therefore, the user can revise / edit the draft taking into account the opinions, provided information, etc. of the related parties.

[0165] [7 Basic Computer Hardware Configuration] 10 is a block diagram showing the basic hardware configuration of a computer 90. The computer 90 includes at least a processor 901, a main storage device 902, an auxiliary storage device 903, and a communication IF 991 (interface). These are electrically connected to each other by a communication bus 921.

[0166] The processor 901 is hardware for executing an instruction set described in a program, and is composed of an arithmetic unit, a register, a peripheral circuit, and the like.

[0167] The main memory device 902 is for temporarily storing programs, data to be processed by the programs, etc. For example, it is a volatile memory such as a DRAM (Dynamic Random Access Memory).

[0168] The auxiliary storage device 903 is a storage device for saving data and programs, such as a flash memory, a hard disk drive (HDD), a magneto-optical disk, a CD-ROM, a DVD-ROM, or a semiconductor memory.

[0169] The communication IF 991 is an interface for inputting and outputting signals for communicating with other computers via a network using a wired or wireless communication standard.

[0170] The network is composed of the Internet, a LAN, various mobile communication systems constructed by wireless base stations, etc. For example, the network includes 3G, 4G, 5G mobile communication systems, LTE (Long Term Evolution), wireless networks that can connect to the Internet via a specified access point (e.g., Wi-Fi (registered trademark)), etc. In the case of wireless connection, communication protocols include, for example, Z-Wave (registered trademark), ZigBee (registered trademark), Bluetooth (registered trademark), etc. In the case of wired connection, the network also includes a network that is directly connected by a USB (Universal Serial Bus) cable or the like.

[0171] It should be noted that the computer 90 can be virtually realized by distributing all or part of each hardware configuration among multiple computers 90 and connecting them together via a network. In this way, the computer 90 is a concept that includes not only a computer 90 housed in a single housing or case, but also a virtualized computer system.

[0172] [8 Basic Functional Configuration of Computer 90] A description will now be given of the functional configuration of a computer realized by the basic hardware configuration (FIG. 10) of a computer 90. The computer comprises at least the functional units of a control unit, a storage unit, and a communication unit.

[0173] The functional units of the computer 90 can also be realized by distributing all or part of the functional units among multiple computers 90 connected to each other via a network. The computer 90 is a concept that includes not only a single computer 90 but also a virtualized computer system.

[0174] The control unit is realized by the processor 901 reading out various programs stored in the auxiliary storage device 903, expanding the programs in the main storage device 902, and executing processes according to the programs. The control unit can realize functional units that perform various information processing depending on the type of program. In this way, the computer is realized as an information processing device that performs information processing.

[0175] The storage unit is realized by a main storage device 902 and an auxiliary storage device 903. The storage unit stores data, various programs, and various databases. Furthermore, the processor 901 can secure a storage area corresponding to the storage unit in the main storage device 902 or the auxiliary storage device 903 in accordance with a program. Furthermore, the control unit can cause the processor 901 to execute processes of adding, updating, and deleting data stored in the storage unit in accordance with the various programs.

[0176] The term database refers to a relational database, which is used to manage sets of data called masters and tables in a tabular format structurally defined by rows and columns, by associating them with each other. In a database, a table is called a table or master, a column in a table is called a column, and a row in a table is called a record. In a relational database, relationships between tables and masters can be set and associated.

[0177] Usually, a column that serves as a primary key for uniquely identifying a record is set in each table and each master, but setting a primary key in a column is not essential. The control unit can cause the processor 901 to add, delete, or update records in a specific table or master stored in the storage unit according to various programs.

[0178] Furthermore, by storing data, various programs, and various databases in the storage unit, it can be considered that the information processing device and information processing system according to the present disclosure have been manufactured.

[0179] In addition, the database and master in this disclosure may include any data structure (such as a list, a dictionary, an associative array, or an object) in which information is structurally defined. The data structure also includes data that can be considered as a data structure by combining data with a function, class, method, or the like written in any programming language.

[0180] The communication unit is realized by the communication IF 991. The communication unit realizes a function of communicating with other computers 90 via a network. The communication unit can receive information transmitted from other computers 90 and input the information to the control unit. The control unit can cause the processor 901 to execute information processing on the received information in accordance with various programs. In addition, the communication unit can transmit information output from the control unit to other computers 90.

[0181] In addition, the above-mentioned configurations, functions, processing units, processing means, etc. may be realized in part or in whole by hardware, for example, by designing them as integrated circuits. The present invention can also be realized by software program code that realizes the functions of the embodiments. In this case, a storage medium on which the program code is recorded is provided to a computer, and a processor included in the computer reads the program code stored in the storage medium. In this case, the program code itself read from the storage medium realizes the functions of the above-mentioned embodiments, and the program code itself and the storage medium storing it constitute the present invention. Examples of storage media for supplying such program code include flexible disks, CD-ROMs, DVD-ROMs, hard disks, SSDs, optical disks, magneto-optical disks, CD-Rs, magnetic tapes, non-volatile memory cards, ROMs, etc.

[0182] Furthermore, the program code for realizing the functions described in this embodiment can be implemented in a wide range of program or script languages, such as assembler, C / C++, perl, Shell, PHP, Java (registered trademark), and the like.

[0183] Furthermore, the program code of the software that realizes the functions of the embodiments may be distributed over a network and stored in a storage means such as a computer's hard disk or memory, or in a storage medium such as a CD-RW or CD-R, and the processor of the computer may read and execute the program code stored in the storage means or storage medium.

[0184] The functions performed by the components described herein may be implemented in circuitry or processing circuitry, including general purpose processors, application specific processors, integrated circuits, ASICs (Application Specific Integrated Circuits), a CPU (a Central Processing Unit), conventional circuits, and / or combinations thereof, programmed to perform the functions described. Processors include transistors and other circuits and are considered to be circuitry or processing circuitry. A processor may be a programmed processor that executes a program stored in a memory.

[0185] In this specification, a circuitry, unit, or means is hardware that is programmed to realize or performs the described functions, which may be any hardware disclosed in this specification or any hardware known to be programmed to realize or perform the described functions.

[0186] If the hardware is a processor considered to be a type of circuitry, the circuitry, means, or unit is a combination of the hardware and software used to configure the hardware and / or processor.

[0187] Although several embodiments of the present disclosure have been described above, these embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and modifications are within the scope of the invention and its equivalents as described in the claims, as well as the scope and spirit of the invention.

[0188] [9 Notes] The matters described in the above embodiments will be supplemented below.

[0189] <Appendix 1> A program to be executed on a computer having a processor and a memory, the program causing the processor to execute the following steps: accepting a selection of a type of subsidy; accepting input of reference information linked to each item that may be included in the selected type of subsidy application document; inputting the input reference information and instructions for creating sentences that may be written in each item into a generative AI model and outputting the sentences from the generative AI model; and creating a draft of the subsidy application document based on the output sentences.In the output step, the program causes the generative AI model to output sentences using the sentences that have already been output, and causes the generative AI model to create the sentences in a predetermined order.

[0190] <Appendix 2> In the output step, the generative AI model creates sentences in order of the items whose contents have the greatest influence on the contents of other items (the program described in Appendix 1).

[0191] <Appendix 3> A program described in (Appendix 1) or (Appendix 2), which causes a processor to execute a step of accepting corrections to the output sentence, and in the output step, inputs the input corrections to a generative AI model and outputs a suggested correction to the sentence from the generative AI model.

[0192] <Appendix 4> A program described in any one of (Appendix 1) to (Appendix 3) that causes a processor to execute a step of presenting the output sentence and the suggested revisions side by side.

[0193] <Appendix 5> In the output step, if the input correction content affects the content of other sentences, a correction suggestion for the other sentences is output from the generative AI model (a program described in Appendix 3).

[0194] <Appendix 6> In the output step, real-time search information related to a specified item is input into a generative AI model, the program being described in any one of (Appendix 1) to (Appendix 5).

[0195] <Appendix 7> In the creation step, a program described in any one of (Appendix 1) to (Appendix 6) creates a draft in a format compatible with the application format for each type of subsidy.

[0196] <Appendix 8> Each item is a program described in any of (Appendix 1) to (Appendix 7), including bonus items according to the subsidy application requirements.

[0197] <Appendix 9> A program described in any one of (Appendix 1) to (Appendix 8), which causes a processor to execute a step of inputting edited text into a generative AI model and outputting a judgment result from the generative AI model as to whether the edited text satisfies specified conditions.

[0198] <Appendix 10> A program described in any of (Appendix 1) to (Appendix 9) that causes the processor to execute a step of accepting a change in type, and in the creation step, creates a draft of a subsidy application document of the changed type based on the reference information.

[0199] <Appendix 11> The generative AI model is trained using subsidy application documents created by users as training data. A program described in any one of (Appendix 1) to (Appendix 10).

[0200] <Appendix 12> A program described in any one of (Appendix 1) to (Appendix 11) that causes a processor to execute a step of presenting the created draft to the user and related parties having a business relationship with the user.

[0201] <Appendix 13> An information processing device comprising a control unit and a storage unit, the control unit executing all steps in a program according to any one of (Supplementary Note 1) to (Supplementary Note 12).

[0202] <Appendix 14> A method executed by a computer having a processor and a memory, wherein the processor executes all the steps of a program according to any one of (Appendix 1) to (Appendix 12).

[0203] <Appendix 15> A system comprising means for executing all steps in the program described in any one of (Appendix 1) to (Appendix 12). [Explanation of symbols]

[0204] 1. System 10...Terminal device 120…Communications Department 13...Input device 14...Output device 15…Memory 16…Storage 19…Processor 20…Server 22…Communication IF 23... Input / Output IF 25…Memory 26…Storage 29…Processor 30…AI system

Claims

1. A program to be executed by a computer having a processor and a memory, The program causes the processor to: accepting a selection of a type of grant; A step of accepting input of reference information linked to each item that may be included in the selected type of grant application document; A step of inputting the input reference information and an instruction sentence instructing the creation of a sentence that can be written in each of the items into a generative AI model and outputting the sentence from the generative AI model; and creating a draft of the grant application document based on the output text. In the output step, the program causes the generative AI model to output the sentence using the sentence that has already been output, and causes the generative AI model to create the sentence in order of the items whose contents have the greatest influence on the contents of other items.

2. having the processor execute a step of accepting corrections to the outputted text; The program according to claim 1 , wherein in the output step, the inputted correction content is inputted into the generative AI model, and a correction proposal for the sentence is outputted from the generative AI model.

3. 3. The program according to claim 2, further comprising a step of displaying the outputted sentence and the proposed revision side by side.

4. The program according to claim 2, wherein in the output step, if the input correction content affects the content of another sentence, a correction proposal for the other sentence is output from the generative AI model.

5. The program according to claim 1 , wherein in the output step, real-time search information related to a specified item is input to the generating AI model.

6. 2. The program according to claim 1, wherein in said creating step, said draft is created in a format compatible with application formats for each type of said subsidy.

7. The program according to claim 1 , wherein the items include points that correspond to the application requirements for the subsidy.

8. The program according to claim 1, which causes the processor to execute a step of inputting the edited sentence into the generative AI model and outputting a judgment result from the generative AI model as to whether the edited sentence satisfies a predetermined condition.

9. having the processor perform the step of accepting a change in type; 2. The program of claim 1, wherein the creating step creates the draft of the grant application document of the modified type based on the reference information.

10. The program according to claim 1 , wherein the generative AI model is trained using the subsidy application document created by a user as training data.

11. 2. The program according to claim 1, further comprising a step of causing the processor to execute a step of presenting the created draft to the user and to a person who has a business relationship with the user.

12. 12. An information processing apparatus comprising: a control unit; and a storage unit, the control unit executing all steps of the program according to claim 1.

13. A method implemented on a computer having a processor and a memory, the processor executing all the steps of a program according to any one of claims 1 to 11.

14. A system comprising means for executing all the steps of the program according to any one of claims 1 to 11.

Images