Program, computer, and information processing method

Abstract

To provide a program, a computer, and an information processing method that can improve convenience by automatically associating multiple pieces of document data with each other.SOLUTION: Vectorization means 26 obtains vector values, which are quantities possessing magnitude and direction within a multidimensional space, by vectorizing text extracted by text extraction means 24. Vector value storage means 28 associates the obtained vector values with received document data or text extracted from this document data and stores them in a database (storage unit 40 or database server 80). Association means 30 associates newly received document data with document data or text already stored in the database, on the basis of a vector value derived from the newly received document data and a vector value already stored in the database.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a program, a computer, and an information processing method. [Background technology]

[0002] Conventionally, various processing systems for processing data of supporting documents such as invoices have been known. For example, Patent Document 1 discloses a data processing device, a data processing method, and a program that can improve the efficiency of processing invoices. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7051168 Summary of the Invention [Problem to be solved by the invention]

[0004] If multiple document data can be automatically associated, convenience can be improved.

[0005] The present disclosure has been made in consideration of these points, and aims to provide a program, computer, and information processing method that can improve convenience by automatically associating multiple document data. [Means for solving the problem]

[0006] The program of the present disclosure is A program that causes a computer to function as a receiving means, a text extracting means, a vectorizing means, a vector value saving means, an associating means, and an outputting means, The receiving means receives document data, the text extraction means extracts text described in the document data accepted by the acceptance means; the vectorization means vectorizes the text extracted by the text extraction means to obtain a vector value, which is a quantity having a magnitude and a direction in a multidimensional space; the vector value saving means saves the vector values ​​acquired by the vectorization means in an internal or external database of the computer in association with the document data accepted by the accepting means or with text extracted from the document data; the associating means associates the document data newly accepted by the accepting means with the document data or text already stored in the database based on the vector values ​​acquired from the document data and the vector values ​​already stored in the database; The output means outputs document data or text information associated with the document data newly accepted by the accepting means.

[0007] The program of the present disclosure is The computer further functions as a document storage means; The document storage means may store association information between document data newly accepted by the acceptance means and document data or text already stored in the database, which is associated by the association means, in the database or a database separate from the database.

[0008] The program of the present disclosure is causing the computer to further function as a display instruction means; The display instruction means may send a display instruction signal to the user terminal to display the document data newly accepted by the acceptance means and the document data or text already stored in the database, which are associated by the association means, on the user terminal that sent the new document data when the acceptance means accepts the new document data.

[0009] The program of the present disclosure is causing the computer to further function as missing information detection means; The missing information detection means may detect missing information in the document data newly accepted by the acceptance means based on the document data newly accepted by the acceptance means and document data or text already stored in the database, which are associated by the association means.

[0010] In this case, the program of the present disclosure is causing the computer to further function as a display instruction means; The display instruction means may transmit a display instruction signal to the user terminal so as to display the missing information detected by the missing information detection means on the user terminal that sent the missing information when the receiving means receives new document data.

[0011] The program of the present disclosure is causing the computer to further function as missing information detection means; the receiving means further receives information regarding items of missing information in the document data when receiving the document data; The missing information detection means may detect missing information corresponding to the accepted missing information item in the document data newly accepted by the acceptance means based on the document data newly accepted by the acceptance means and the document data or text already stored in the database, which are associated by the association means.

[0012] In this case, the program of the present disclosure causing the computer to further function as a display instruction means; The display instruction means may transmit a display instruction signal to the user terminal so as to display the missing information detected by the missing information detection means on the user terminal that sent the missing information when the receiving means receives new document data.

[0013] In the program of the present disclosure, The vectorization means may input the text extracted by the text extraction means into a large-scale language model, thereby causing the large-scale language model to perform calculations on vector values.

[0014] The program of the present disclosure is causing the computer to further function as missing information detection means; The missing information detection means may determine whether information in some items of the document data newly accepted by the accepting means is missing by comparing the document data newly accepted by the accepting means with document data or text of the document data already stored in the database to which the document data is associated by the associating means.

[0015] In this case, the missing information detection means may input an instruction to a large-scale language model requesting a determination as to whether or not information is missing in some items of the document data newly accepted by the accepting means, based on the document data newly accepted by the accepting means and the document data or text of the document data already stored in the database to which the document data is associated by the associating means, thereby causing the large-scale language model to determine whether or not information is missing in some items of the document data newly accepted by the accepting means.

[0016] In addition, if the missing information detection means determines that information is missing from some items of the document data newly accepted by the accepting means, the missing information detection means may send a display instruction signal to the user terminal that was the sender when the accepting means accepted the new document data, instructing the user terminal to display information that information is missing from some items of the document data newly accepted by the accepting means.

[0017] In addition, when the missing information detection means determines that information is missing from some items of document data newly accepted by the accepting means, the missing information detection means may input an instruction to a large-scale language model to calculate the missing information of the document data newly accepted by the accepting means based on the document data newly accepted by the accepting means and the document data or text of the document data already stored in the database to which the document data is associated by the associating means, thereby causing the large-scale language model to calculate the missing information of the document data newly accepted by the accepting means.

[0018] The computer of the present disclosure includes: A computer that functions as a receiving means, a text extracting means, a vectorizing means, a vector value storing means, an associating means, and an outputting means by executing a program, The receiving means receives document data, the text extraction means extracts text described in the document data accepted by the acceptance means; the vectorization means vectorizes the text extracted by the text extraction means to obtain a vector value, which is a quantity having a magnitude and a direction in a multidimensional space; the vector value saving means saves the vector values ​​acquired by the vectorization means in an internal or external database of the computer in association with the document data accepted by the accepting means or with text extracted from the document data; the associating means associates the document data newly accepted by the accepting means with the document data or text already stored in the database based on the vector values ​​acquired from the document data and the vector values ​​already stored in the database; The output means outputs document data or text information associated with the document data newly accepted by the accepting means.

[0019] The information processing method of the present disclosure includes: An information processing method performed by a computer having a control unit, a step of the control unit receiving document data; a step of extracting text from the received document data by the control unit; a step of the control unit vectorizing the extracted text to obtain a vector value, which is a quantity having a magnitude and a direction in a multidimensional space; a step of causing the control unit to associate the acquired vector values ​​with the received document data or text extracted from the document data and store them in an internal or external database of the computer; a step of associating the newly received document data with document data or text already stored in the database based on vector values ​​acquired from the document data and vector values ​​already stored in the database by the control unit; a step of outputting document data or text information associated with the newly received document data by the control unit; The present invention is characterized by the following. [Effects of the Invention]

[0020] According to the program, computer, and information processing method disclosed herein, convenience can be improved by automatically associating multiple document data. [Brief explanation of the drawings]

[0021] [Figure 1] 1 is a diagram schematically illustrating a configuration of an information processing system according to an embodiment of the present invention. [Figure 2] 2 is a chart showing the flow of information between the components when document data is registered by a user terminal in the information processing system shown in FIG. 1. [Figure 3] 2 is a chart showing the flow of information between the components when a search for related documents is performed in the information processing system shown in FIG. 1. [Figure 4]2 is a chart showing the flow of information between the components when documents are associated in the information processing system shown in FIG. 1. [Figure 5] 4 is a flowchart showing an operation when document data is registered by a user terminal in the information processing system shown in FIG. [Figure 6] 4 is a flowchart showing the operation when searching for related documents and associating documents in the information processing system shown in FIG. 1. DETAILED DESCRIPTION OF THE INVENTION

[0022] An embodiment of the present invention will now be described with reference to the drawings. FIGS. 1 to 6 are diagrams illustrating an information processing system 1 according to the present embodiment and an information processing method using the information processing system 1. According to the information processing system 1 and information processing method according to the present embodiment, when a user, such as an employee of a client company, registers new document data on a user terminal 50, the new document data can be associated with document data already stored in a database (e.g., an external database server 80 or the internal storage unit 40 of the management server 10), thereby enabling efficient processing of the document data. In this embodiment, the document data refers to business data such as national tax-related accounting documents, such as contracts and supporting documents. Supporting documents include at least one of invoices, receipts, purchase orders, delivery notes, contracts, and approval documents. While the following description focuses primarily on cases in which supporting documents are approval documents and invoices, supporting documents other than approval documents and invoices may also be used. According to the information processing system 1 and information processing method of this embodiment, when information about the person in charge is not stated in supporting documents such as an invoice, it is difficult to search for such information about the person in charge. However, it is now possible to automatically infer the information about the person in charge from text information in the approval request form related to the invoice.

[0023] As shown in Fig. 1, an information processing system 1 according to this embodiment includes a management server (computer) 10, a user terminal 50, a management terminal 52, an operator terminal 60, a language model server 70, and a database server 80. The management server 10 is a server that can be used over a network such as the Internet. The user terminal 50, the management terminal 52, the operator terminal 60, the language model server 70, and the database server 80 are each communicably connected to the management server 10 via the network such as the Internet. Each of the components of this information processing system 1 will be described below.

[0024] The user terminal 50 includes a personal computer or the like installed in a company that wishes to receive the service according to this embodiment, and is operated by an employee of this company. Here, the service according to this embodiment includes, for example, a process of having a management company that manages the management server 10 process documents related to business operations, such as national tax-related books and documents, on behalf of the company where the user terminal 50 is installed, but is not limited to such a process.

[0025] The management terminal 52 includes a personal computer or the like installed in a company (client company) that wishes to receive the service according to this embodiment, and is operated by an administrator of this company.

[0026] The operator terminal 60 includes a personal computer or the like installed in the office of the management company that manages the management server 10, and an operator who is an employee of this management company operates the operator terminal 60 to send and receive information to the management server 10. Specifically, the operator operates the operator terminal 60 to input information for each item of a document (for example, information such as the billing date, details, amount, and payment due date of an invoice).

[0027] The language model server 70 is a server designed to be able to generate replies at high speed using, for example, OpenAI in response to requests from multiple clients (the management server 10 in this embodiment). The language model server 70 is used for large-scale natural language processing tasks, such as automatically performing machine translation, sentence generation, document classification, question answering, and the like.

[0028] The language model server 70 generally uses a neural network-based model. This model trains on a massive amount of data and then makes predictions for new inputs. The language model server 70 hosts the model, invokes the model in response to a request, and returns the results. The language model server 70 is capable of distributed processing, allowing it to respond to multiple requests simultaneously. This allows for high throughput and low latency. Additionally, it is possible to reuse a model that has been trained once. In this embodiment, the language model server 70 vectorizes (embedding) the text sent from the management server 10 to calculate vector values, which are quantities with magnitude and direction in a multidimensional space.

[0029] In addition, as the language model server 70, in addition to a server designed to be able to generate responses quickly using OpenAI, various language models such as large-scale language models (LLMs) can also be used.

[0030] The database server 80 stores document data or text extracted from the document data in association with vector values. In addition, association information for multiple pieces of document data is stored in the database server 80.

[0031] The management server 10 includes a control unit 20, a storage unit 40, and a communication unit 42. The control unit 20 is, for example, a CPU (Central Processing Unit) or the like, and functions as a reception unit 22, a text extraction unit 24, a vectorization unit 26, a vector value storage unit 28, an association unit 30, an output unit 32, a document storage unit 34, a display instruction unit 36, and a missing information detection unit 38 by executing a program stored in the storage unit 40.

[0032] The receiving means 22 receives document data from the user terminal 50, including business-related documents such as contracts and supporting documents. The text extraction means 24 extracts document text from images of the documents received by the receiving means 22 using optical character recognition. The vectorization means 26 vectorizes the text extracted by the text extraction means 24 to obtain vector values, which are quantities having magnitude and direction in multidimensional space. The vector value storage means 28 associates the vector values ​​obtained by the vectorization means 26 with the document data received by the receiving means 22 or the text extracted from the document data, and stores the associated vector values ​​in the storage unit 40 or the database server 80. The associating means 30 associates the document data newly received by the receiving means 22 with document data or text already stored in the storage unit 40 or the database server 80, based on the vector values ​​obtained from the document data and the vector values ​​already stored in the storage unit 40 or the database server 80. The output means 32 outputs information about the document data or text associated with the document data newly received by the receiving means 22.

[0033] The document storage means 34 stores the document data newly accepted by the accepting means 22 in the storage unit 40 or the database server 80. The document storage means 34 also stores in the storage unit 40 or the database server 80 association information between the document data newly accepted by the accepting means 22 and the document data or text already stored in the storage unit 40 or the database server 80, which has been associated by the associating means 30. The display instruction means 36 sends a display instruction signal to the user terminal 50 to cause the user terminal 50 to display the document data or text already stored in the storage unit 40 or the database server 80, which has been associated with the newly accepted document data. The missing information detection means 38 detects missing information in the document data newly accepted by the accepting means 22, based on the document data newly accepted by the accepting means 22 and the document data or text already stored in the storage unit 40 or the database server 80, which has been associated by the associating means 30.

[0034] The functions of each of the means 22, 24, 26, 28, 30, 32, 34, 36, and 38 will be described in detail below.

[0035] The program executed by the control unit 20 is not limited to being stored in the storage unit 40. The program executed by the control unit 20 may be a program transmitted from an external device to the control unit 20 via the communication unit 42, a program stored in a storage medium such as a USB memory attached to the management server 10, or a program stored in a server other than the management server 10 (for example, a cloud server).

[0036] The storage unit 40 is configured with, for example, a hard disk drive (HDD), random access memory (RAM), read only memory (ROM), and solid state drive (SSD), and various information is stored in the storage unit 40. The storage unit 40 is also provided with a plurality of storage areas, and document data is stored (saved) in each storage area. The storage unit 40 may also store information relating the newly received document data and the document data already stored in the storage unit 40, which are associated by the associating means 30.

[0037] The communication unit 42 connects the control unit 20 to external devices (e.g., the user terminal 50, the management terminal 52, the operator terminal 60, the language model server 70, and the database server 80) so that they can communicate with each other. The control unit 20 can send and receive signals to and from external devices via the communication unit 42.

[0038] Next, an information processing method by such information processing system 1 will be described with reference to FIGS. 2 to 6. FIG. 2 is a chart showing the flow of information between the components when document data is registered by user terminal 50 in information processing system 1 shown in FIG. 1. FIG. 3 is a chart showing the flow of information between the components when a search for related documents is performed in information processing system 1 shown in FIG. 1. FIG. 4 is a chart showing the flow of information between the components when document association is performed in information processing system 1 shown in FIG. 1. FIG. 5 is a flowchart showing the operation when document data is registered by user terminal 50 in information processing system 1 shown in FIG. 1. FIG. 6 is a flowchart showing the operation when a search for related documents and document association are performed in the information processing system shown in FIG. 1. The information processing described below is performed by control unit 20 executing a program stored in storage unit 40.

[0039] A person in charge of a company (client company) where user terminal 50 is installed accesses a predetermined website provided by the management company that manages management server 10. At this time, the person in charge of this company enters an ID and password when logging in to the website provided by management server 10. Then, the person in charge of this company uploads new document data (specifically, a document image) to the website displayed on user terminal 50. When uploading new document data to the website, if some items of the document data are missing, the person in charge of this company may input these missing information items (e.g., in the case of an invoice, items such as the recipient, contact name, and transaction number) using user terminal 50. When the document data is uploaded to the website, the document data is transmitted from user terminal 50 to management server 10, and reception unit 22 accepts the document data (step S1 in FIG. 5 ). Furthermore, reception unit 22 accepts identification information of the owner and administrator of the document data based on the ID entered when logging in to the website provided by management server 10. Furthermore, when information on missing information items of the document data is input using user terminal 50, this information is transmitted from user terminal 50 to management server 10 and accepted by reception unit 22. The document storage means 34 stores the document data newly received by the reception means 22, the identification information of the owner or administrator of the document data, and the like in the memory unit 40 or the database server 80. In another embodiment, when a management company that manages the management server 10 acts on behalf of the company where the user terminal 50 is installed to process documents, document data to be sent from a business partner of the company where the user terminal 50 is installed may be transmitted from the business partner to the management server 10. In yet another embodiment, when a management company that manages the management server 10 acts on behalf of the company where the user terminal 50 is installed to process documents, an employee of the company where the user terminal 50 is installed may forward an email of document data received from a business partner to the management server 10, or an employee of the company where the user terminal 50 is installed may forward a paper document received from a business partner to the management company that manages the management server 10.

[0040] The text extraction means 24 extracts the text of the document by optical character recognition from the image of the document accepted by the acceptance means 22 (step S2 in FIG. 5). In another aspect, the text extraction means 24 may transmit an instruction to the operator terminal 60 to input information for each item of the document data accepted by the acceptance means 22, have the operator input information for each item of the document data via the operator terminal 60, and have the information input via the operator terminal 60 be returned from the operator terminal 60 to the management server 10, thereby acquiring the text information of the document data accepted by the acceptance means 22.

[0041] The vectorization means 26 vectorizes the text extracted by the text extraction means 24 to obtain vector values, which are quantities having magnitude and direction in a multidimensional space (step S3 in FIG. 5 ). Specifically, the vectorization means 26 creates an instruction to request that the text extracted by the text extraction means 24 be vectorized to obtain vector values, and transmits the instruction to the language model server 70. As a result, the language model server 70, based on the instruction transmitted from the management server 10, vectorizes the text extracted by the text extraction means 24 to obtain vector values, which are quantities having magnitude and direction in a multidimensional space, and transmits the obtained vector values ​​back to the management server 10. In this way, the vectorization means 26 is able to obtain vector values.

[0042] The vector value saving means 28 associates the vector values ​​acquired by the vectorization means 26 with the document data accepted by the accepting means 22 or with text extracted from this document data, and saves the associated vector values ​​in the database server 80 (step S4 in FIG. 5). The vector value saving means 28 may also save this information in the memory unit 40 instead of the database server 80.

[0043] Thereafter, if the user wishes to search for documents related to the newly registered document, the user sends a search instruction for related documents to the management server 10 via the user terminal 50. Specifically, when an instruction button for searching for related documents is pressed on the user terminal 50 or a specific screen is opened on the monitor of the user terminal 50, the search instruction for related documents is transmitted from the user terminal 50 to the management server 10. When the management server 10 receives the search instruction for related documents from the user terminal 50 (step S5 in FIG. 6), the management server 10 transmits the search instruction for related documents to the database server 80. Here, if a vector value approximate to the vector value extracted from the document data newly received by the receiving means 22 is stored in the database server 80 ("YES" in step S6 in FIG. 6), information on the approximate vector value stored in the database server 80, or information on the document data or text corresponding to this vector value, is transmitted from the database server 80 to the management server 10. Note that various known methods can be used to detect vector values ​​stored in the database server 80 that approximate the vector value extracted from the document data newly received by the receiving means 22. For example, since a vector value is a quantity having a magnitude and a direction in a multidimensional space as described above, if the distance between the coordinates in the multidimensional space of a vector value extracted from document data newly accepted by the accepting unit 22 and the coordinates in the multidimensional space of a vector value stored in the database server 80 is within a predetermined threshold, the vector values ​​are determined to be similar. Furthermore, if multiple vector values ​​determined to be similar are stored in the database server 80, the vector value with the closest distance between the coordinates of the two is used. Note that, as another aspect, if multiple vector values ​​determined to be similar are stored in the database server 80, information on all vector values ​​determined to be similar, or information on the document data or text corresponding to these vector values, may be transmitted from the database server 80 to the management server 10.

[0044] Thereafter, document data corresponding to the approximate vector values ​​is transmitted from management server 10 to user terminal 50, and a document related to this document data is displayed on user terminal 50 (step S7 in FIG. 6). At this time, multiple document data corresponding to approximate vector values ​​may be transmitted from management server 10 to user terminal 50, and multiple documents may be displayed on user terminal 50. Then, when the user transmits an association instruction to management server 10 via user terminal 50 and management server 10 accepts the association instruction ("YES" in step S8 in FIG. 6), association means 30 associates the document data newly accepted by acceptance means 22 with document data or text already stored in storage unit 40 or database server 80, based on the vector values ​​acquired from this document data and the vector values ​​already stored in database server 80 (step S9 in FIG. 6). Output means 32 outputs information on the document data or text associated with the document data newly accepted by acceptance means 22. The display instruction means 36 transmits a display instruction signal to the user terminal 50 to cause the user terminal 50 to display the document data or text information output by the output means 32 and associated with the document data newly accepted by the accepting means 22. This causes the document data or text associated with the newly accepted document data to be displayed on the user terminal 50 (step S10 in FIG. 6). In this way, the operator of the user terminal 50 can check the document data and text of this document data that are related to the newly registered document data and that are already stored in the storage unit 40 or the database server 80.

[0045] As described above, when a plurality of vector values, etc., stored in database server 80 and similar to vector values ​​extracted from document data newly accepted by accepting means 22 are transmitted to management server 10, management server 10 transmits a selection instruction to user terminal 50 to select appropriate document data from the document data corresponding to each of the plurality of vector values. As a result, a plurality of document data candidates similar to the document data newly registered in management server 10 are displayed on user terminal 50. Then, when certain document data is selected by user terminal 50, information about the selected document data is transmitted from user terminal 50 to management server 10. As a result, associating means 30 associates the document data newly accepted by accepting means 22 with document data or text already stored in storage unit 40 or database server 80 based on the information about the selected document data.

[0046] The display instruction means 36 may also send a display instruction signal to the user terminal 50 to cause the user terminal 50 to display the document data or text associated with the document data newly accepted by the accepting means 22, output by the output means 32, and the newly accepted document data, side by side, for example, on a single screen. In this case, the operator of the user terminal 50 can visually compare the newly registered document data with the document data and text of the document data already stored in the storage unit 40 or the database server 80 on a single screen. This allows the operator of the user terminal 50 to visually infer the missing information even if some information items in the newly registered document data are missing by comparing them with the document data already stored in the storage unit 40 or the database server 80. The operator of the user terminal 50 then manually inputs the missing information in the newly registered document data into the user terminal 50. This causes the missing information manually input into the user terminal 50 to be transmitted to the management server 10, and the input missing information is stored in the storage unit 40 or the database server 80.

[0047] Furthermore, document storage means 34 stores document data newly transmitted to management server 10 by user terminal 50 in database server 80 or memory unit 40. At this time, document storage means 34 stores in database server 80 or memory unit 40 association information between the document data newly accepted by acceptance means 22, which has been associated by association means 30, and the document data or text already stored in database server 80 or memory unit 40 (step S11 in FIG. 6). This makes it possible for user terminal 50 or management terminal 52 to check document data managed by management server 10, by checking multiple document data in an associated state.

[0048] Furthermore, when the associating means 30 associates the document data newly accepted by the accepting means 22 with document data or text already stored in the storage unit 40 or the database server 80, the missing information detecting means 38 may automatically detect missing information in the document data newly accepted by the accepting means 22, based on the document data newly accepted by the accepting means 22 associated by the associating means 30 and the document data or text already stored in the storage unit 40 or the database server 80. For example, when the data of a request form or the data of the text of this request form is stored in the storage unit 40 or the database server 80, if the accepting means 22 newly accepts data of an invoice related to this request form and information on items such as the invoice recipient, person in charge, and transaction number is missing, the missing information detecting means 38 can detect information such as the client, person in charge, and transaction number as missing information from the data of the request form associated with this invoice or the data of the text of this request form. The missing information detected by the missing information detection means 38 is then transmitted to the user terminal 50 or the management terminal 52, and displayed on the user terminal 50 or the management terminal 52. This allows the operator operating the user terminal 50 or the administrator operating the management terminal 52 to know the missing information in the invoice data that is to be newly registered in the management server 10 by the user terminal 50. At this time, if there is an error in the missing information detected by the missing information detection means 38, the operator operating the user terminal 50 or the administrator operating the management terminal 52 can correct the missing information using the user terminal 50 or the management terminal 52.

[0049] Furthermore, as described above, when uploading new document data to a predetermined website provided by the management company that manages management server 10, if some items of the document data are missing, the operator operating user terminal 50 may input these missing information items (e.g., in the case of an invoice, items such as the recipient, contact person name, and transaction number) via user terminal 50. In this case, missing information detection means 38 detects missing information corresponding to the received missing information items in the document data newly accepted by acceptance means 22, based on the document data newly accepted by acceptance means 22 and the document data or text already stored in memory unit 40 or database server 80, which are associated by association means 30. Specifically, when the operator operating user terminal 50 registers, for example, an invoice as document data in management server 10, if it is known in advance that information on the invoice is missing in items such as the recipient, contact person name, and transaction number, the operator inputs the information on the missing information items into user terminal 50. This allows information on the missing information item to be sent from the user terminal 50 to the management server 10, and the missing information detection means 38 can detect information corresponding to this missing information item from related document data or text already stored in the memory unit 40 or database server 80.

[0050] According to the program, computer (management server 10), and information processing method of the present embodiment configured as described above, the program causes the computer to function as reception means 22, text extraction means 24, vectorization means 26, vector value storage means 28, associating means 30, and output means 32. Reception means 22 receives document data, and text extraction means 24 extracts text described in the document data received by reception means 22. Vectorization means 26 vectorizes the text extracted by text extraction means 24 to obtain vector values, which are quantities having magnitude and direction in multidimensional space, and vector value storage means 28 stores the vector values ​​obtained by vectorization means 26 in memory unit 40 or database server 80 in association with the document data received by reception means 22 or the text extracted from the document data. The associating means 30 associates the document data newly accepted by the accepting means 22 with the document data or text already stored in the storage unit 40 or the database server 80 based on the vector values ​​acquired from this document data and the vector values ​​already stored in the storage unit 40 or the database server 80, and the output means 32 outputs information on the document data or text to which the document data newly accepted by the accepting means 22 is associated. Such a program, computer, and information processing method can improve convenience by automatically associating multiple document data.

[0051] Furthermore, in the program, computer (management server 10), and information processing method of this embodiment, as described above, the program causes the computer to further function as document storage means 34, and document storage means 34 stores in storage unit 40 or database server 80 association information between document data newly accepted by acceptance means 22, which has been associated by association means 30, and document data or text already stored in storage unit 40 or database server 80. This makes it possible to manage association information between newly accepted document data and document data or text already stored in storage unit 40 or database server 80.

[0052] Furthermore, in the program, computer (management server 10), and information processing method of this embodiment, as described above, the program further causes the computer to function as display instruction means 36, and display instruction means 36 transmits a display instruction signal to user terminal 50 to cause document data newly accepted by acceptance means 22, associated by association means 30, and document data or text already stored in memory unit 40 or database server 80, to be displayed on user terminal 50, which is the sender when acceptance means 22 accepts new document data. This allows the operator operating user terminal 50 to check the document data or text associated with the newly registered document data.

[0053] Furthermore, in the program, computer (management server 10), and information processing method of the present embodiment, as described above, the program further causes the computer to function as missing information detection means 38, which detects missing information in the document data newly received by the receiving means 22 based on the document data newly received by the receiving means 22 and the document data or text already stored in the storage unit 40 or the database server 80, which are associated by the associating means 30. This makes it possible to detect missing information by referring to the document data or text already stored in the storage unit 40 or the database server 80, even if some information items in the newly received document data are missing. In this case, the program further causes the computer to function as display instruction means 36, which may send a display instruction signal to the user terminal 50, which is the source of the new document data received by the receiving means 22, to display the missing information detected by the missing information detection means 38 on the user terminal 50. In this case, the operator who registers the new document data in the management server 10 using the user terminal 50 can recognize the missing information detected by the missing information detection means 38.

[0054] Furthermore, in the program, computer (management server 10), and information processing method of the present embodiment, as described above, the program may further cause the computer to function as missing information detection means 38, and the receiving means 22 may further receive information regarding items of missing information in the document data when receiving the document data. The missing information detection means 38 may detect missing information corresponding to the items of missing information in the document data newly received by the receiving means 22, based on the document data newly received by the receiving means 22 and the document data or text already stored in the storage unit 40 or the database server 80, which have been associated by the associating means 30. This makes it possible to detect missing information corresponding to the items of missing information received by the document data by referring to the document data or text already stored in the storage unit 40 or the database server 80, even if information is missing in some items of the newly received document data. In this case, the program may further cause the computer to function as display instruction means 36, and the display instruction means 36 may send a display instruction signal to the user terminal 50, which was the source of the new document data received by the receiving means 22, to display the missing information detected by the missing information detection means 38 on the user terminal 50. In this case, the operator who registers new document data in the management server 10 using the user terminal 50 will be able to recognize the missing information detected by the missing information detection means 38 and corresponding to the item of missing information that the operator entered into the user terminal 50.

[0055] Furthermore, in the program, computer (management server 10), and information processing method of this embodiment, as described above, vectorization means 26 inputs the text extracted by text extraction means 24 into a large-scale language model (specifically, language model server 70) to cause the large-scale language model to calculate vector values. This makes it possible to accurately calculate vector values ​​corresponding to the text extracted by text extraction means 24, and therefore to more accurately associate document data newly accepted by acceptance means 22 with document data or text already stored in storage unit 40 or database server 80.

[0056] The program, computer (management server 10), and information processing method according to the present disclosure are not limited to the above-described embodiments.

[0057] For example, the vectorization means 26 is not limited to inputting the text extracted by the text extraction means 24 into a large-scale language model (specifically, the language model server 70) and causing the large-scale language model to calculate vector values. In another aspect, the vectorization means 26 may calculate vector values ​​by other means without using a large-scale language model.

[0058] Furthermore, in the above explanation, the language model server 70 as a large-scale language model is provided outside the management server 10, but the present embodiment is not limited to this. The large-scale language model may be incorporated inside the management server 10, and the vectorization means 26 may calculate vector values ​​from the text extracted by the text extraction means 24 by using the large-scale language model incorporated inside.

[0059] Furthermore, the missing information detection means 38 may be configured to determine whether or not some items of information in the document data newly accepted by the accepting means 22 are missing by comparing the document data newly accepted by the accepting means 22 with the document data or text of the document data already stored in the storage unit 40 or the database server 80 to which the document data is associated by the associating means 30. In this case, the missing information detection means 38 may input an instruction to a large-scale language model such as the language model server 70 to request a determination as to whether or not some items of information in the document data newly accepted by the accepting means 22 are missing, based on the document data newly accepted by the accepting means 22 and the document data or text of the document data already stored in the storage unit 40 or the database server 80 to which the document data is associated by the associating means 30, thereby causing the large-scale language model to determine whether or not some items of information in the document data newly accepted by the accepting means 22 are missing.

[0060] When the missing information detection means 38 determines that some items of information are missing from the document data newly accepted by the accepting means 22, a display instruction signal is transmitted from the management server 10 to the user terminal 50 to instruct the user terminal 50 to display information indicating that some items of information are missing from the document data newly accepted by the accepting means 22. This allows the operator of the user terminal 50 to recognize that some of the document data registered in the management server 10 is missing, and enables the missing information to be supplemented by the user terminal 50. When the missing information detection means 38 determines that some items of information are missing from the document data newly accepted by the accepting means 22, the missing information detection means 38 may transmit to the language model server 70 an instruction to calculate the missing information from the document data newly accepted by the accepting means 22 based on the document data newly accepted by the accepting means 22 and the document data or text of the document data already stored in the storage unit 40 or the database server 80 to which the document data is associated by the associating means 30. In this case, the language model server 70 can calculate the missing information of the document data newly received by the receiving means 22, and the calculated missing information is returned from the language model server 70 to the management server 10. The missing information returned to the management server 10 is then transmitted from the management server 10 to the user terminal 50, so that the operator of the user terminal 50 can recognize the information that is missing in some items of the document data registered in the management server 10. [Explanation of symbols]

[0061] 1. Information Processing Systems 10 Management Server 20 Control Unit 22 Reception methods 24 Text Extraction Methods 26 Vectorization Methods 28 Vector Value Storage Means 30 Association Methods 32 Output Method 34 Document storage methods 36 Display instruction means 38 Missing Information Detection Method 40 Storage section 42 Communications Department 50 User Terminals 52 Management terminal 60 Operator Terminal 70 Language Model Server 80 Database Server

Claims

1. A program that causes a computer to function as a receiving means, a text extracting means, a vectorizing means, a vector value saving means, an associating means, and an outputting means, The receiving means receives document data, the text extraction means extracts text described in the document data accepted by the acceptance means; the vectorization means vectorizes the text extracted by the text extraction means to obtain a vector value, which is a quantity having a magnitude and a direction in a multidimensional space; the vector value saving means saves the vector values ​​acquired by the vectorization means in an internal or external database of the computer in association with the document data accepted by the accepting means or with text extracted from the document data; the associating means associates the document data newly accepted by the accepting means with the document data or text already stored in the database based on the vector values ​​acquired from the document data and the vector values ​​already stored in the database; The output means outputs document data or text information associated with the document data newly accepted by the acceptance means.

2. The computer further functions as a document storage means; The program according to claim 1, wherein the document storage means stores association information between document data newly accepted by the acceptance means and document data or text already stored in the database, as associated by the association means, in the database or a database separate from the database.

3. causing the computer to further function as a display instruction means; The program of claim 1, wherein the display instruction means sends a display instruction signal to the user terminal to display the document data newly accepted by the acceptance means and the document data or text already stored in the database, which are associated by the association means, on the user terminal that sent the new document data when the acceptance means accepted the new document data.

4. causing the computer to further function as missing information detection means; The program of claim 1, wherein the missing information detection means detects missing information in the document data newly accepted by the accepting means based on the document data newly accepted by the accepting means and the document data or text already stored in the database, which are associated by the associating means.

5. causing the computer to further function as a display instruction means; 5. The program according to claim 4, wherein the display instruction means transmits a display instruction signal to the user terminal so that the missing information detected by the missing information detection means is displayed on the user terminal that is the sender when the receiving means receives new document data.

6. causing the computer to further function as missing information detection means; the receiving means further receives information regarding items of missing information in the document data when receiving the document data; The program of claim 1, wherein the missing information detection means detects missing information corresponding to the accepted missing information item in the document data newly accepted by the accepting means based on the document data newly accepted by the accepting means and the document data or text already stored in the database, which are associated by the associating means.

7. causing the computer to further function as a display instruction means; 7. The program according to claim 6, wherein the display instruction means transmits a display instruction signal to the user terminal so that the missing information detected by the missing information detection means is displayed on the user terminal that is the sender when the receiving means receives new document data.

8. 2. The program according to claim 1, wherein the vectorization means inputs the text extracted by the text extraction means into a large-scale language model, thereby causing the large-scale language model to perform a vector value calculation.

9. causing the computer to further function as missing information detection means; The program of claim 1, wherein the missing information detection means determines whether information in some items of the document data newly accepted by the accepting means is missing by comparing the document data newly accepted by the accepting means with document data or text of the document data already stored in the database to which the document data is associated by the associating means.

10. The program of claim 9, wherein the missing information detection means inputs an instruction to a large-scale language model requesting a determination as to whether or not information in some items of the document data newly accepted by the accepting means is missing, based on the document data newly accepted by the accepting means and the document data or text of the document data already stored in the database to which the document data is associated by the associating means, thereby causing the large-scale language model to determine whether or not information in some items of the document data newly accepted by the accepting means is missing.

11. 10. The program according to claim 9, wherein when the missing information detection means determines that information in some items of the document data newly accepted by the accepting means is missing, the missing information detection means transmits a display instruction signal to the user terminal that is the source of the acceptance of the new document data and that instructs the user terminal to display information that information in some items of the document data newly accepted by the accepting means is missing on the user terminal.

12. 10. The program of claim 9, wherein when the missing information detection means determines that information in some items of document data newly accepted by the accepting means is missing, the missing information detection means inputs an instruction to a large-scale language model to calculate the missing information of the document data newly accepted by the accepting means based on the document data newly accepted by the accepting means and document data or text of the document data already stored in the database to which the document data is associated by the associating means, thereby causing the large-scale language model to calculate the missing information of the document data newly accepted by the accepting means.

13. A computer that functions as a receiving means, a text extracting means, a vectorizing means, a vector value storing means, an associating means, and an outputting means by executing a program, The receiving means receives document data, the text extraction means extracts text described in the document data accepted by the acceptance means; the vectorization means vectorizes the text extracted by the text extraction means to obtain a vector value, which is a quantity having a magnitude and a direction in a multidimensional space; the vector value saving means saves the vector values ​​acquired by the vectorization means in an internal or external database of the computer in association with the document data accepted by the accepting means or with text extracted from the document data; the associating means associates the document data newly accepted by the accepting means with the document data or text already stored in the database based on the vector values ​​acquired from the document data and the vector values ​​already stored in the database; The output means outputs document data or text information associated with the document data newly accepted by the acceptance means.

14. An information processing method performed by a computer having a control unit, a step of the control unit receiving document data; a step of extracting text from the received document data by the control unit; a step of the control unit vectorizing the extracted text to obtain a vector value, which is a quantity having a magnitude and a direction in a multidimensional space; a step of causing the control unit to associate the acquired vector values ​​with the received document data or text extracted from the document data and store them in an internal or external database of the computer; a step of associating the newly received document data with document data or text already stored in the database based on vector values ​​acquired from the document data and vector values ​​already stored in the database by the control unit; a step of outputting document data or text information associated with the newly received document data by the control unit; An information processing method comprising:

Images