Document creation support apparatus, document creation support method, and program

Abstract

To improve the efficiency in creating a document with high accuracy.SOLUTION: A document creation support apparatus 100 is provided with a retrieval unit 121 for retrieving a related document according to a character string included in a document creation instruction inputted by a document creator, a generation unit 122 for generating a draft document based on the document creation instruction and the retrieved related document, and an extraction unit 123 for extracting confirmation items by a document confirmer in the draft document based on contents of the draft document generated by the document creator from the draft document.SELECTED DRAWING: Figure 9

Description

[Technical field]

[0001] The present disclosure relates to a document creation support device, a document creation support method, and a program. [Background technology]

[0002] The contents of documents handled in the government are very important, and accuracy is required. Therefore, the process from document drafting to approval takes a lot of time and effort. In particular, the process from drafting to approval is mainly manual and visually checked by humans, which takes time and effort and is inefficient.

[0003] Meanwhile, in recent years, as described in Patent Document 1, documents have been automatically generated using large-scale language models. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2023-018624 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, in the case of administrative documents as described above, the problem occurs that automatic generation of documents using language models cannot sufficiently satisfy the accuracy. In addition, the process of approving the drafted document is necessary, which is still time-consuming and inefficient. In addition, the above-mentioned problem occurs not only in administrative documents but also in the creation of any document that requires accuracy.

[0006] Therefore, one of the objectives of the present disclosure is to provide a document creation support device that can solve the above-mentioned problem of low efficiency when creating highly accurate documents. [Means for solving the problem]

[0007] A document creation support device according to an embodiment of the present disclosure includes: a search unit for searching for documents related to a character string included in a document creation command input by a document creator; a generating unit for generating a draft document based on the document creation command and the retrieved related documents; an extraction unit that extracts confirmation items to be confirmed by a document confirmer from the draft document based on the contents of the draft document generated by the document creator from the draft document; Equipped with The structure is as follows. Further, a document creation support method according to an embodiment of the present disclosure includes: Searching for related documents according to a character string included in a document creation command input by a document creator; generating a draft document based on the document creation command and the retrieved related documents; extracting confirmation items to be confirmed by a document confirmer from the draft document based on the contents of the draft document generated by the document creator from the draft document; Document creation assistance methods. The structure is as follows. In addition, a program according to an embodiment of the present disclosure includes: Searching for related documents according to a character string included in a document creation command input by a document creator; generating a draft document based on the document creation command and the retrieved related documents; extracting confirmation items to be confirmed by a document confirmer from the draft document based on the contents of the draft document generated by the document creator from the draft document; Have a computer carry out the process, The structure is as follows. Effect of the Invention

[0008] By being configured as described above, the present disclosure can improve the efficiency of creating highly accurate documents. [Brief description of the drawings]

[0009] [Figure 1] 1 is a block diagram showing a configuration of a document creation support device according to the present disclosure. [Diagram 2] FIG. 1 is a diagram showing an example of a document generated using a document creation support device according to the present disclosure. [Diagram 3] FIG. 2 is a diagram illustrating an example of information stored in a document creation support device according to the present disclosure. [Figure 4] FIG. 2 is a diagram illustrating an example of information generated by the document creation support device according to the present disclosure. [Diagram 5] FIG. 2 is a diagram illustrating an example of information generated by the document creation support device according to the present disclosure. [Figure 6] FIG. 2 is a diagram illustrating an example of information generated by the document creation support device according to the present disclosure. [Figure 7] FIG. 2 is a sequence diagram showing a processing operation of the document creation support system according to the present disclosure. [Figure 8] 1 is a block diagram showing a hardware configuration of a document creation support device according to the present disclosure. [Figure 9] 1 is a block diagram showing a configuration of a document creation support device according to the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] <First embodiment> A first embodiment of the present disclosure will be described with reference to the drawings. Note that the drawings may be relevant to any embodiment.

[0011] [composition] The document creation support system in this embodiment is used to support document creation activities including document creation work by a document creator who is a person who creates a document, and document approval work by a document approver who is a person who checks and approves the document. Here, the document created and approved in this embodiment is, for example, a document that expresses the contents of cases such as plans, policies, and applications for projects carried out by the government. One example of the document is a "draft document" that expresses the contents of a case such as a "new park construction plan" as shown in FIG. 2, and such a document includes character strings and sentences that express the contents such as "background of drafting", "planned construction site", and "outline of the plan". However, the document whose creation is supported by the document creation support system may be a document of any content and is not limited to being a government document.

[0012] 1, the document creation support system includes a document creation support device 10, a creator terminal 20, and an approver terminal 30. The creator terminal 20 and the approver terminal 30 are configured as information processing devices such as personal computers, and are connected to the document creation support device 10 via a network. Each component will be described below.

[0013] The creator terminal 20 has a function of first transmitting a document creation command input by the document creator to the document creation support device 10, as described later. The creator terminal 20 also has a function of displaying a draft document generated by the document creation support device 10 in response to the document creation command on a display device, generating a draft document in response to input of correction data from the document creator for the draft document, and requesting approval from the approver terminal 30 via the document creation support device 10, as described later. The creator terminal 20 also has a function of displaying a reason for sending back a draft document sent back by a document approver on a display device, creating a draft document corrected in response to the reason for sending back, and requesting the document creation support device 10 to generate a correction proposal in response to the reason for sending back. The functions of the creator terminal 20 will be described later together with the functions of the document creation support device 10.

[0014] As described below, the approver terminal 30 has a function of first displaying on a display device a draft document generated by a document creator and a checklist generated by the document creation support device 10, and accepting a confirmation result and a request for return of the draft document by a document approver. For example, the approver terminal 30 accepts a specification of a character string portion in the draft document that is determined by the document approver to require correction or review. The approver terminal 30 then returns the draft document together with the specified character string portion to the document creation support device 10. The approver terminal 30 also has a function of performing approval processing if there is no problem with the checked draft document. The functions of the approver terminal 30 will be described later together with the functions of the document creation support device 10.

[0015] The document creation support device 10 is composed of one or more information processing devices each having a calculation device and a storage device. As shown in FIG. 1, the document creation support device 10 includes a document search unit 11, a document generation assistance unit 12, a checklist generation unit 13, a return reason presentation unit 14, and a revision proposal unit 15. The functions of the document search unit 11, the document generation assistance unit 12, the checklist generation unit 13, the return reason presentation unit 14, and the revision proposal unit 15 can be realized by the calculation device executing a program for realizing each function stored in the storage device. The document creation support device 10 also includes a related document storage unit 16. The related document storage unit 16 is composed of a storage device. Each component of the document creation support device 10 will be described in detail below.

[0016] The document search unit 11 (search unit) accepts a document creation command input by the document creator from the creator terminal 20. At this time, the document creation command includes a keyword consisting of a character string related to the case for which the document is to be created. For example, when planning to create an administrative document such as that shown in Fig. 2, a document command such as "Create a document related to a new park construction plan" or "Create a plan for constructing a new park" is input, and "new park construction" or "new park" is included as a keyword.

[0017] Then, the document search unit 11 searches for related documents corresponding to the keywords included in the document creation command from among the related documents stored in the related document storage unit 16. Here, a database of administrative documents used in past cases is first built in the related document storage unit 16. At this time, as shown in FIG. 3, the administrative document database is set with items such as "document number," "date," "sender," "destination," "subject," "department," "budget," and "expected completion date" that identify administrative documents, and administrative documents corresponding to each "document number" are stored. That is, in the example of FIG. 3, the related document storage unit 16 stores administrative documents such as "Central Library Renovation Plan" and "New Public Swimming Pool Construction Plan" that have been used in the past.

[0018] The related document storage unit 16 also stores related documents including information that can be referred to as materials in the corresponding business when creating an administrative document. For example, a template describing the basic contents of an administrative document corresponding to a case may be stored as the related document. The related document may also include information managed by the administration, such as statistical information on the population and land of a municipality governed by the administration. The related document may also include documents including information on laws and regulations, basic knowledge, etc. related to the case, such as architectural plans and urban plans. The related document may also include documents including information on basic knowledge, precautions, etc. in documents that correspond to a preset character string and include the character string. The related document storage unit 16 may also store related documents other than the above-mentioned contents.

[0019] Then, the document search unit 11 searches for related documents corresponding to the keyword included in the document creation command from related documents such as past administrative documents and laws and regulations stored in the above-mentioned related document storage unit 16. For example, in the above example, in response to a keyword such as "construction of a new park," plans, which are past administrative documents used when constructing parks and other facilities, documents of laws and regulations related to the construction of parks and facilities, and documents such as basic knowledge and cautions related to parks and construction are searched for, and the searched related documents are extracted from the related document storage unit 16.

[0020] The document generation assistant unit 12 (generation unit) assists in creating an administrative document in response to the document creation command from the document creator described above. Specifically, the document generation assistant unit 12 generates a draft document of an administrative document based on the document creation command from the document creator and the related documents searched as described above. As an example, the document generation assistant unit 12 analyzes the structure of past administrative documents that are the searched related documents, and uses the administrative documents while making corrections corresponding to the keywords included in the document creation command according to the structure to generate a draft document. As another example, the document generation assistant unit 12 uses a template corresponding to a case example that is the searched related document, and makes corrections corresponding to the statistical information and keywords of the local government that are the searched related documents to generate a draft document.

[0021] Then, the document generation assistant unit 12 outputs the generated draft document to be displayed on the display device of the creator terminal 20. At this time, the document generation assistant unit 12 associates information on the retrieved related documents with the generated draft document and outputs the document to be displayed on the display device of the creator terminal 20. For example, if the retrieved related documents are past administrative documents, the title, date, document number, etc. of the administrative document may be associated with the draft document, or the administrative document itself may be associated with the draft document. Also, if the related documents are materials such as statistical information, the source of the documents may be associated with the draft document, and if the retrieved related documents are documents such as laws and regulations or basic knowledge, the provisions of the related laws and regulations or basic knowledge may be associated with the draft document.

[0022] Furthermore, when outputting the generated draft document for display on the display device of the creator terminal 20, the document generation assistant unit 12 may output the generated draft document with a reliability attached thereto. At this time, the document generation assistant unit 12 calculates the similarity between the contents of the searched related document and the contents of the generated draft document as the reliability. For example, the document generation assistant unit 12 calculates a higher reliability as the contents of the draft document are more similar to the contents of the searched related document. As an example, the document generation assistant unit 12 calculates the similarity based on the distance between words included in the related document and the draft document or the distance between the documents, and calculates it as the reliability. However, any method of calculating the reliability may be used.

[0023] After the draft document is displayed on the creator terminal 20 in the above manner, the document creator creates a draft document by referring to the draft document. The document creator may use the draft document as it is, or may modify the character string in the draft document to create the draft document. At this time, the document creator can modify the draft document by referring to related documents associated with the draft document or by referring to the reliability. The document creator may also create an attachment to be attached to the draft document by referring to the draft document and related documents. Then, the document creator requests the approver terminal 30 to approve the created draft document from the creator terminal 20 via the document creation support device 10. Note that, when requesting approval, the creator terminal 20 transmits the related documents associated with the draft document and the created attachment document to the document creation support device 10 in association with the draft document, and requests approval.

[0024] The checklist generation unit 13 (extraction unit) accepts a request for approval of a draft document from the creator terminal 20 to the approver terminal 30, and transmits the draft document to the approver terminal 30 to request approval. At this time, in order to support the approval work of the document approver, the checklist generation unit 13 extracts check items (confirmation items) in the draft document based on the contents of the draft document, and generates a checklist, which is a list of such check items. As the checklist data, as shown in Fig. 4, a "check item number" that identifies the check item, a "check item" that indicates the content of the check item, and "details and precautions" that indicate the content to be checked and points to note are set.

[0025] Specifically, when a related document is associated with the above-mentioned draft document, the checklist generating unit 13 generates a checklist by referring to the related document. For example, when additional information on revisions such as past revision history or revision comments is added to the related document, the relevant part of the draft document corresponding to the part where the additional information is added is extracted as a check item. As an example, when a revision history or a comment is added to a "background" item in a past related document, the "background" item of the draft document is extracted as a "check item", and a sentence of "details and points to note" is generated based on the comment added to the related document. Note that, when basic knowledge or notes for a predetermined character string are included in the related document, the checklist generating unit 13 may extract a character string in the draft document corresponding to the character string in the related document as a check item, and generate a checklist by using the basic knowledge or notes corresponding to the character string as a note.

[0026] As another example, the checklist generating unit 13 extracts a checklist according to a character string in the draft document. For example, when a predetermined character string such as "budget", "planned location", or a numerical value is present in the draft document, the checklist generating unit 13 may extract the corresponding part of the character string as a "check item". When a sentence representing basic knowledge or points of caution is prepared in advance in association with a predetermined character string, the checklist generating unit 13 may set the sentence as a "details / points of caution" sentence. The checklist generating unit 13 may extract a checklist by any method.

[0027] Then, the checklist generation unit 13 outputs the generated checklist to be displayed on the display device of the approver terminal 30 together with the draft document. At this time, the checklist generation unit 13 may transmit related documents and attached materials to the approver terminal 30 together with the draft document and the checklist. After the draft document and the checklist are displayed on the approver terminal 30 in this manner, the document approver can check the draft document by referring to the checklist. At this time, the document approver can also check the draft document by referring to the related documents associated with the draft document and the attached attached materials together with the checklist.

[0028] If the document approver checks the draft document and finds no defects, the document approver performs a process to approve the draft document at the approver terminal 30. On the other hand, if the draft document is defective, the document approver performs a process to specify the defective part at the approver terminal 30 and performs a process to send back the draft document. For example, the document approver performs an input to specify the defective part (point of criticism) by selecting a character string corresponding to the defect in the draft document at the approver terminal 30, and also inputs the content of the defect using a character string explaining the content of the defect, and performs a process to send back the draft document. For example, if the detailed address is not written in the "planned construction site" field in the draft document, the document approver specifies the character string "planned construction site" as the defective part, inputs "detailed address unknown" as the content of the defect, and performs a send back operation. The approver terminal 30 then generates a return document with deficiency items and deficiency details indicating the deficiencies in the draft document, and transmits the return document and the draft document to the document creation support device 10, requesting return to the creator terminal 20. At this time, the approver terminal 30 associates the return document and the draft document with related documents and attachments, transmits them to the document creation support device 10, and requests return. Note that the document approver may simply point out the deficiencies in the draft document and does not necessarily need to input the deficiency details. In this case, the return document may simply have the deficiencies added to it.

[0029] The sendback reason presentation unit 14 (reason generation unit) receives a sendback request from the approver terminal 30 as described above, and generates a sendback reason (reason for indication) that indicates the reason and background for the deficiency in the deficiency part and the deficiency content contained in the sendback document. At this time, the sendback reason presentation unit 14 generates a sendback reason according to the deficiency part and the deficiency content contained in the sendback document, for example, the content of the character string of the deficiency part and the deficiency content. As data for the sendback reason, as shown in Fig. 5, "deficiency item" indicating the item corresponding to the deficiency part, "deficiency content" indicating the content of the deficiency, "deficiency reason" indicating the reason for the deficiency, and "laws and provisions" indicating the laws and provisions related to the content of the deficiency are set.

[0030] Specifically, when a related document is associated with the draft document requested to be sent back as described above, the sendback reason presenting unit 14 generates a deficiency reason corresponding to the deficiency item and the deficiency content included in the sent back document by referring to the related document. For example, when additional information related to the correction such as a correction history, a correction comment, or a related article is added to the past related document, the sendback reason presenting unit 14 generates a deficiency reason and a statute / article for the deficiency item corresponding to the location where the additional information is added from the content of the additional information. As an example, when a specific address or map is added as a correction history to the "planned construction site" item in the past related document, or a reason for the correction, a statute / article is added as a comment, the sendback reason presenting unit 14 generates a deficiency reason such as "detailed address and map are required" and a statute / article such as "Article 12 of the City Planning Act" according to the content of the correction history and comment, as shown in the first line of FIG. 5. Note that, when basic knowledge or cautions about a predetermined character string are included in the related document, the sendback reason presenting unit 14 may generate a deficiency reason for the deficiency item in the draft document using the basic knowledge or cautions.

[0031] As another example, the remand reason presenting unit 14 generates a deficiency reason according to the character string of the deficiency part included in the remand document and the deficiency content. For example, when the character string of the deficiency part in the draft document is a preset character string such as "planned construction site", "budget", or "period", the remand reason presenting unit 14 may generate a preset deficiency reason or law / provision corresponding to such character string, such as "·· detailed address and map required", "specific breakdown required", "detailed schedule required", "·· Law Article·", etc., as a deficiency reason corresponding to each deficiency part, as shown in FIG. 5. The remand reason presenting unit 14 may generate a deficiency reason by detecting missing data or insufficient data for the character string of the deficiency part, or data inconsistency with other parts. The remand reason presenting unit 14 may generate a deficiency reason or law / provision by any method.

[0032] The reason for return presentation unit 14 then outputs the generated reason for return to be displayed on the display device of the preparer terminal 20 together with the draft document. At this time, the reason for return presentation unit 14 associates related documents and attached materials with the draft document and transmits them to the preparer terminal 20. After the draft document and the reason for return are displayed on the preparer terminal 20 in this manner, the document preparer can refer to the reason for return on the preparer terminal 20 and revise the draft document. At this time, the document preparer can also refer to the related documents and attached materials associated with the draft document along with the reason for return and revise the draft document.

[0033] Furthermore, the document creator can also request the document creation support device 10 from the creator terminal 20 to create a proposed revision of the draft document based on the reason for return. At this time, the creator terminal 20 transmits the draft document together with the reason for return to the document creation support device 10, and requests a proposed revision of the draft document. Note that the creator terminal 20 may also transmit related documents and attached materials associated with the draft document to the document creation support device 10.

[0034] The revision proposal unit 15 creates a revision proposal based on the reason for return in response to a request for a revision proposal for a draft document from the document creator. As data for the revision proposal, as shown in Fig. 6, a "revise proposal" indicating the proposed content of the revision is set in correspondence with a "deficient item" indicating the item subject to the deficiency and a "deficient content" indicating the content of the deficiency.

[0035] Specifically, when a related document is associated with the draft document, the correction suggestion unit 15 generates a correction proposal corresponding to the deficiency reason included in the reason for remand by referring to the related document. For example, when additional information related to correction such as a correction history, a correction comment, or a related clause is added to a past related document, a correction proposal corresponding to the deficiency reason of the deficiency item or the deficiency content corresponding to the location where the additional information is added is generated from the content of the additional information. As an example, when information such as a specific address, a map, and even ownership is added as a correction history to the "planned construction site" item in the past related document, or when such correction content is added as a comment, a correction proposal as shown in the first line of FIG. 6 is generated as a correction proposal for the deficiency reason as shown in the first line of FIG. 5 based on the content of the correction history and the comment.

[0036] As another example, the correction suggestion unit 15 may generate a correction method that can eliminate the content pointed out in the reason for remand as a correction suggestion. For example, when necessary information is pointed out in the reason for remand, the correction suggestion unit 15 may generate a correction suggestion that instructs the addition of such information. Note that the generation of the correction suggestion by the correction suggestion unit 15 may be performed by any method.

[0037] Then, the revision proposal unit 15 outputs the generated revision proposal to be displayed on the display device of the creator terminal 20 together with the draft document. At this time, the revision proposal unit 15 associates related documents and attached materials with the draft document and transmits them to the creator terminal 20. After the draft document and the revision proposal are displayed on the creator terminal 20 in this manner, the document creator can revise the draft document by referring to the revision proposal on the creator terminal 20. At this time, the document creator can also revise the draft document by referring to the related documents and attached materials associated with the draft document together with the revision proposal.

[0038] Then, after the document creator has revised the draft document, in the same manner as described above, the document creator terminal 20 requests approval of the draft document from the approver terminal 30 via the document creation support device 10. In response to this, in the same manner as described above, the document creation support device 10 creates a checklist for the draft document and sends an approval request to the approver terminal 30 together with the checklist, or, if the document is sent back from the approver terminal 30, generates a reason for the send back in response to the send back and sends the document back to the creator terminal 20 again.

[0039] Here, the functions of the document creation support device 10 described above, namely, generation of a draft document by the document generation assistant unit 12, generation of a checklist by the checklist generation unit 13, generation of a reason for remand by the reason for remand presenting unit 14, and generation of revision suggestions by the revision suggesting unit 15, may be performed using a generation AI (Artificial Intelligence) such as a language model. In other words, the document generation assistant unit 12 and the like are configured to generate a draft document, a checklist, a reason for remand, revision suggestions, and the like, using a generation AI that is a language model generated by machine learning a database of related documents and the like generated in the past. An example of a configuration in which the document generation assistant unit 12 and the like perform output using a generation AI will be described below.

[0040] First, a definition of generative AI (language model) will be described. A language model is a model that learns the relationship between words in a sentence and generates a related string related to a target string from the target string. By using a language model that has learned sentences and sentences of various contexts, it is possible to generate a related string with appropriate content related to the target string. For example, a case where a language model is used in question answering will be described. The language model accepts an input of a question such as "What kind of country is Japan?" as a target string. The language model generates a string such as "Japan is an island country in the northern hemisphere..." as an answer to the question. Note that the learning method of the language model is not particularly limited, but as an example, it may be a model that has been learned to output at least one sentence including an input string.

[0041] As a specific example, the language model is a Generative Pre-Training-Transformer (GPT) that outputs a sentence including an input string by predicting a string that is likely to follow the input string. Other examples of language models include Text-to-Text Transfer Transformer (T5), Bidirectional Encoder Representations from Transformers (BERT), Robustly optimized BERT approach (RoBERTa), and Efficiently Learning an Encoder that Classifies Token Replacements Accurately (ELECTRA).

[0042] Furthermore, the character string generated by the language model is not limited to a natural language. For example, the language model may output an artificial language (such as a program source code) for a character string input in a natural language. For example, the language model accepts an input question such as "How can I retrieve data containing a specific character string from a database?" as a target character string. The language model may output a program source code for performing database processing. Alternatively, the language model may output a natural language corresponding to a character string input in an artificial language. Furthermore, the content generated by the language model is not limited to a character string. For example, the language model may generate image data, video data, audio data, or other data formats corresponding to the input character string.

[0043] Here, as an example, a processing method in which the above-mentioned document generation assistant unit 12 generates a draft document using a generation AI will be described. Here, a processing method in which the generation AI generates a draft document using a past administrative document that is a related document will be described.

[0044] First, the generation AI is assumed to have an acquisition unit, an extraction unit, a rewriting unit, a generation unit, an output control unit, a calculation unit, and a judgment unit, which are configured in a calculation unit of an information processing device.

[0045] The acquisition unit of the generation AI acquires data to be input to a language model, such as a query input by a user. Here, the acquisition unit acquires a character string of a document creation command as data input by a document creator. Hereinafter, the character string acquired by the acquisition unit is also referred to as a first character string. The acquisition unit stores the acquired first character string in a storage unit.

[0046] The extraction unit of the generation AI extracts documents related to the character string. As an example, the extraction unit may extract documents related to the first character string (such as administrative documents created in the past) from documents stored in the storage unit. In the above example, the extraction unit extracts past administrative documents searched according to the keyword as documents related to the document creation command.

[0047] As another example, the extraction unit may extract documents related to the first character string from documents stored in a database connected to the information processing device via a network. The extraction unit stores the extracted documents in the storage unit in association with the first character string. Note that the method by which the extraction unit extracts documents related to the character string is not limited. For example, the extraction unit may extract documents including the character string. Also, the extraction unit may extract documents related to the character string using an existing search engine.

[0048] The rewriting unit of the generation AI rewrites the character string. As an example, the rewriting unit rewrites the first character string using the document extracted by the extraction unit. The character string rewritten by the rewriting unit is also referred to as the second character string. The rewriting unit associates the second character string with the first character string and stores it in the storage unit. The method by which the rewriting unit rewrites the character string is not limited. As an example, the rewriting unit may generate the second character string by adding a character string described in the document to the first character string. As another example, the rewriting unit may generate the second character string by adding a character string described in the document and a character string indicating an additional instruction to the first character string. As yet another example, the second character string may be generated by rewriting the first character string to a character string indicating an additional instruction and then adding a character string described in the document. In the above example, the rewriting unit may generate the second character string by adding a document in a past administrative document to the document creation command.

[0049] The generation unit of the generation AI generates a result of performing natural language processing on a character string. As an example, the generation unit generates a result of performing natural language processing on a second character string. As described above, the generation unit may use a language model that receives a character string as input and outputs a result of performing natural language processing on the character string. That is, the generation unit may input the second character string into the language model and generate a result output from the language model as a result of performing natural language processing on the second character string. The result generated by the generation unit is also referred to as a first result. The generation unit stores the generated first result in the storage unit in association with the second character string.

[0050] The output control unit outputs a result in which information identifying the document extracted by the extraction unit is added to the first result generated by the generation unit. The output control unit may add information identifying a portion of the documents extracted by the extraction unit to the first result generated by the generation unit. Examples of information identifying a document include the document name, author name, and publication date and time, as well as a URL (Uniform Resource Locator) indicating the location where the document is stored. The result output by the output control unit is also referred to as a second result.

[0051] Furthermore, the output control unit may output the reliability calculated by the calculation unit described later in addition to the second result. With this configuration, the output control unit can present the reliability of the second result to the user.

[0052] The output control unit may be configured to output the second result when a later-described determination unit determines that the reliability calculated by the calculation unit exceeds a threshold value. With this configuration, the output control unit can output the second result with high reliability.

[0053] The calculation unit calculates the reliability of the result. As an example, the calculation unit calculates the reliability of the first result generated by the generation unit. The calculation unit stores the calculated reliability in the storage unit in association with the first result. With this configuration, the calculation unit can grasp the degree to which the first result is reliable.

[0054] The method by which the calculation unit calculates the reliability of the first result is not limited. The calculation unit may calculate the reliability of the first result using an existing technique such as a factuality analysis, which is a technique for analyzing whether an event actually occurred. As an example, the calculation unit may calculate the reliability of the first result using the first result and the document extracted by the extraction unit. Examples of the method by which the calculation unit calculates the reliability of the first result using the first result and the document extracted by the extraction unit include (a) a technique based on a word distance, (b) a technique based on a document distance, or (c) a technique based on a learning model.

[0055] When using a method based on word distance, the calculation unit calculates the reliability of the first result based on the word distance between each word included in the first result and each word included in the document. Specifically, the calculation unit first calculates the word distance for each combination between each word included in the first result and each word included in the document. The calculation unit calculates the reliability of the first result such that the shorter the calculated word distance, the higher the reliability of the first result.

[0056] When using a method based on document distance, the calculation unit calculates the reliability of the first result based on the document distance between a sentence included in the first result and a sentence included in the document. Specifically, the calculation unit first calculates the document distance between a sentence included in the first result and a sentence included in the document. The calculation unit calculates the reliability of the first result such that the shorter the calculated document distance, the higher the reliability of the first result.

[0057] When using a method based on a learning model, the calculation unit uses a learning model that has been trained by machine learning to input two sentences and output the similarity between the two sentences. In this case, the calculation unit inputs the sentence included in the first result and the sentence included in the document to the learning model. Then, the calculation unit calculates the reliability of the first result such that the higher the similarity output from the learning model, the higher the reliability of the first result.

[0058] When the extraction unit extracts a plurality of documents, the calculation unit may calculate the reliability of the first result for each of the plurality of documents. Then, the calculation unit may calculate an arithmetic mean value of the calculated reliability as the reliability of the first result.

[0059] As another example, the calculation unit may calculate the reliability using the first result, the document extracted by the extraction unit, and the character string acquired by the acquisition unit. For example, the calculation unit may calculate the reliability by referring to a result of inputting the first result, the document extracted by the extraction unit, and the character string acquired by the acquisition unit into a language model used by the generation unit. As an example, the calculation unit acquires an index from the language model indicating how correct the first result referring to the document is as a response to the first character string. The calculation unit calculates the reliability of the first result such that the reliability of the first result is higher as the output result is more positive (for example, the larger the value of the index). With this configuration, the calculation unit can suitably calculate the reliability of the first result using the language model that generated the first result.

[0060] The calculation unit may input the first result, the document extracted by the extraction unit, and the character string acquired by the acquisition unit to the language model multiple times. In this case, the calculation unit may score (likelihood, positive frequency, etc.) the results output from the language model for each of the multiple inputs, and calculate the reliability of the first result such that the higher the score, the higher the reliability of the first result.

[0061] The calculation unit may input, instead of the character string acquired by the acquisition unit, a character string instructing the user to answer whether the first result is correct or not in consideration of the document specified by the information added by the output control unit, to the language model. Even in this case, the calculation unit calculates the reliability of the first result such that the reliability of the first result is higher as the output result is more positive.

[0062] The determination unit determines whether the value exceeds a threshold. As an example, the determination unit determines whether the reliability calculated by the calculation unit exceeds a threshold. The determination unit associates the determination result with the reliability and stores it in the storage unit.

[0063] In this way, the information processing device constructing the generation AI performs natural language processing using a language model on a second character string obtained by rewriting the first character string using a document related to the acquired first character string. Therefore, the information processing device can prevent the language model from performing natural language processing by referring to documents whose contents are not known to be correct, such as documents with low credibility, documents of unknown authenticity, and documents of unknown origin, and can generate answers by referring to only highly reliable sentences. Therefore, the information processing device can improve the reliability of answers generated using the language model.

[0064] The information processing device also outputs a result of the natural language processing performed by the language model on the second character string, with information identifying the document related to the first character string added. Therefore, the information processing device can present the result of the natural language processing and information on the document referenced in the natural language processing to the user. Therefore, the information processing device can present the reliability of the natural language processing performed using the language model to the user.

[0065] As described above, the generation of a checklist by the checklist generation unit 13 can also be realized by using a generation AI. In this case, the generation AI is trained in machine learning to output a checklist in response to input of a draft document and related documents. As described above, the generation of a reason for remand by the reason for remand presentation unit 14 can also be realized by using a generation AI. In this case, the generation AI is trained in machine learning to output a reason for remand in response to input of deficiencies in the draft document and related documents. As described above, the generation of a revision proposal by the revision proposal unit 15 can also be realized by using a generation AI. In this case, the generation AI is trained in machine learning to output a revision proposal in response to input of deficiencies in the draft document, the content of the deficiencies, the reasons for the deficiencies, and related documents.

[0066] [Operation] Next, the operation of the document creation support device 10 will be described with reference to the drawings.

[0067] First, a document creator such as an administrative official decides to draft a new administrative document and inputs a document creation command from the creator terminal 20 (step S1 in FIG. 7). At this time, the document creation command from the creator terminal 20 contains keywords consisting of character strings related to the case for which the document is to be created. For example, when planning to create an administrative document such as that shown in FIG. 2, a document command such as "Create a document related to a new park construction plan" or "Create a plan for constructing a new park" is input, and among these, "construction of a new park" and "new park" are included as keywords.

[0068] Next, the document search unit 11 of the document creation support device 10, which has received the document creation command from the creator terminal 20, searches for related documents corresponding to the keywords included in the document creation command from related documents consisting of past cases and materials stored in the related document storage unit 16 (step S2 in FIG. 7). For example, the document search unit 11 searches for administrative documents such as "Central Library Renovation Plan" and "New Public Swimming Pool Construction Plan" that have been used in the past, according to the keywords, from related documents such as those shown in FIG.

[0069] Next, the document creation assistant unit 12 of the document creation support device 10 generates a draft document of the administrative document to be created based on the document creation command and the searched related documents (step S3 in FIG. 7). At this time, the document creation assistant unit 12 generates a draft document corresponding to the document creation command by using past administrative documents or templates that are the searched related documents. In addition, the document creation assistant unit 12 may input the document creation command and the searched related documents to a generation AI generated in advance by machine learning, obtain a draft document output from the generation AI, and generate the draft document.

[0070] At this time, the document generation assistant unit 12 may assign a reliability to the generated draft document and output it (step S4 in FIG. 7). For example, the document generation assistant unit 12 calculates the similarity between the contents of the searched related documents and the contents of the generated draft document as the reliability.

[0071] Then, the document generation assistant unit 12 outputs the generated draft document to be displayed on the display device of the creator terminal 20 (step S5 in FIG. 7). At this time, the document generation assistant unit 12 outputs the generated draft document to be displayed on the display device of the creator terminal 20, associating the generated draft document with information on the searched related documents (exhibits, laws and regulations, statistical materials, etc.) and the reliability described above.

[0072] After the draft document is displayed on the preparer terminal 20 in the above manner, the document preparer creates a draft document by referring to the draft document (step S6 in FIG. 7). At this time, the document preparer can create a draft document by modifying the draft document by referring to related documents associated with the draft document or by referring to the reliability. Then, the document preparer requests approval of the created draft document from the preparer terminal 20 to the approver terminal 30 via the document creation support device 10 (step S6 in FIG. 7). At this time, when requesting approval, the preparer terminal 20 transmits the related documents associated with the draft document and the created attached documents to the document creation support device 10 in association with the draft document, and requests approval.

[0073] Next, the checklist generation unit 13 of the document creation support device 10 accepts an approval request for the draft document from the creator terminal 20 to the approver terminal 30, and generates a checklist for the draft document based on the contents of the draft document (step S7 in FIG. 7). At this time, the checklist generation unit 13 may generate a checklist based on additional information related to revisions, such as past revision history and revision comments, that are added to related documents associated with the draft document, or may generate a checklist by extracting check items that are set in advance according to character strings in the draft document. In addition, the checklist generation unit 13 may input the draft document and related documents to a generation AI generated in advance by machine learning, obtain a checklist output from the generation AI, and generate the checklist.

[0074] Then, the checklist generation unit 13 outputs the generated checklist to be displayed on the display device of the approver terminal 30 together with the draft document (step S8 in FIG. 7). Thereafter, the document approver checks the draft document by referring to the checklist. At this time, the document approver can also check the draft document by referring to the related documents associated with the draft document and the attached materials together with the checklist.

[0075] If the document approver checks the draft document and finds no defects, the document approver performs a process to approve the draft document at the approver terminal 30. On the other hand, if the draft document is defective, the document approver performs a process to point out the defective part at the approver terminal 30 and performs a process to return the draft document (step S9 in FIG. 7). For example, the document approver inputs the defective part (point of point of point) by specifying a character string corresponding to the defect in the draft document at the approver terminal 30, and also inputs the defect content in the form of a statement explaining the content of the defect, and performs a process to return the draft document. As a result, the approver terminal 30 generates a return document consisting of the defective part and the defect content of the draft document, and transmits it to the document creation support device 10 together with the draft document, requesting the send back of the draft document to the creator terminal 20.

[0076] The remand reason presenting unit 14 of the document creation support device 10, which has received a remand request from the approver terminal 30, generates a remand reason indicating the reason and background of the deficiency in the deficiency part and the deficiency content in the draft document that is the remand document (step S10 in FIG. 7). At this time, the remand reason presenting unit 14, for example, refers to the related documents of the draft document that is the target of remand, and generates a remand reason including the deficiency reason of the deficiency item corresponding to the part where the additional information is added, and laws and provisions, from the contents of the additional information related to the correction, such as the past correction history, correction comments, and related provisions, that are added to the related documents. In addition, the remand reason presenting unit 14 may input the remand document and related documents to a generation AI generated in advance by machine learning, and obtain the remand reason output from the generation AI to generate the remand reason.

[0077] Then, the sendback reason presenting unit 14 outputs the sendback reason including the generated deficiency reason and laws and provisions to be displayed on the display device of the creator terminal 20 together with the draft document (step S10 in FIG. 7). As a result, the document creator refers to the sendback reason at the creator terminal 20 and modifies the draft document. At this time, the document creator may request the document creation support device 10 from the creator terminal 20 to propose a modification of the draft document based on the sendback reason (step S11 in FIG. 7). In response to this, the modification suggesting unit 15 of the document creation support device 10 generates a modification proposal corresponding to the deficiency reason included in the sendback reason (step S12 in FIG. 7). For example, the modification suggesting unit 15 generates a modification proposal corresponding to the deficiency reason by referring to additional information related to the modification, such as past modification history, modification comments, and related provisions, in the related documents associated with the draft document. Then, the modification suggesting unit 15 transmits the generated modification proposal to the creator terminal 20.

[0078] Thereafter, the document creator modifies the draft document by referring to the proposed revisions at the creator terminal 20, and requests approval of the draft document (step S13 in FIG. 7). In this manner, when the document creator requests approval of the draft document, the processes from step S6 in FIG. 7 onwards are repeated. Then, when the document approver confirms that there are no problems with the draft document, approval processing is carried out at the approver terminal 30, and the document is officially adopted.

[0079] As described above, the document creation support device 10 in this embodiment first searches for related documents in response to a document creation command from a document creator, and generates a draft document based on the related documents. In addition, the document creation support device 10 creates a checklist for the document approver to check the draft document generated by the document creator from the draft document. This allows the document creator to easily create the draft document, and the document approver to easily check the draft document by referring to the checklist. As a result, even for documents that require accuracy, such as administrative documents, the process from drafting to settlement can be made more efficient.

[0080] In addition, the document creation support device 10 also generates a reason for sending back a document when the document approver sends it back. This reduces the work required for the document approver to send back a document, and also allows the document creator to easily recognize the reason for sending back a document. As a result, the document creation process can be made more efficient.

[0081] In addition, when a draft document is created, the document creation support device 10 provides the document creator with information on related documents and the reliability of the draft document. This makes it easier for the document creator to create a draft document, and further improves the efficiency of document creation.

[0082] <Second embodiment> Next, a second embodiment of the present disclosure will be described with reference to the drawings. In this embodiment, an outline of the configuration of the document creation support device described in the above embodiment is shown. Note that Figures 8 to 9 are diagrams for explaining the configuration, and these drawings may be related to any of the embodiments.

[0083] First, the hardware configuration of the document creation support device 100 will be described with reference to Fig. 8. The document creation support device 100 is configured as a general information processing device, and is equipped with the following hardware configuration, for example. ·CPU(Central Processing Unit)101(Arithmetic unit) ROM (Read Only Memory) 102 (storage device) RAM (Random Access Memory) 103 (storage device) · Programs 104 loaded into RAM 103 A storage device 105 for storing a program group 104 A drive device 106 that reads and writes data from a storage medium 110 external to the information processing device A communication interface 107 for connecting to a communication network 111 outside the information processing device Input / output interface 108 for inputting and outputting data Bus 109 connecting each component

[0084] 8 shows an example of the hardware configuration of the information processing device which is the document creation support device 100, and the hardware configuration of the information processing device is not limited to the above-mentioned case. For example, the information processing device may be configured with a part of the above-mentioned configuration, such as not having the drive device 106. Furthermore, the information processing device may use a GPU (Graphic Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating point number Processing Unit), a PPU (Physics Processing Unit), a TPU (Tensor Processing Unit), a quantum processor, a microcontroller, or a combination of these, instead of the above-mentioned CPU.

[0085] The document creation support device 100 can be equipped with a search unit 121, a generation unit 122, and an extraction unit 123 shown in Fig. 9 by the CPU 101 acquiring the program group 104 and executing it. The program group 104 is stored in the storage device 105 or the ROM 102 in advance, for example, and the CPU 101 loads the program group 104 into the RAM 103 and executes it as necessary. The program group 104 may be supplied to the CPU 101 via the communication network 111, or may be stored in the storage medium 110 in advance, and the drive device 106 may read out the program and supply it to the CPU 101. However, the above-mentioned search unit 121, generation unit 122, and extraction unit 123 may be constructed of dedicated electronic circuits for realizing such means.

[0086] The search unit 121 searches for related documents corresponding to a character string included in a document creation command input by a document creator. The generation unit 122 generates a draft document based on the document creation command and the searched related documents. Furthermore, the extraction unit 123 extracts confirmation items to be confirmed by a document checker from the draft document based on the contents of the draft document generated by the document creator from the draft document.

[0087] With the above-described configuration, the present disclosure allows a draft document to be generated based on related documents searched in response to a document creation command from a document creator, and the document creator can easily create a more accurate draft document using the draft document. Also, a checklist for the draft document is created, and a document reviewer can easily check the draft document by referring to the checklist. As a result, the process from drafting to settlement can be made more efficient, even for documents that require accuracy, such as administrative documents.

[0088] At least one or more of the functions of the above-mentioned search unit 121, generation unit 122, and extraction unit 123 may be executed by an information processing device installed and connected anywhere on the network, that is, they may be executed by so-called cloud computing.

[0089] The above-mentioned program can be stored and provided to a computer using various types of non-transitory computer readable media. The non-transitory computer readable medium includes various types of tangible storage media. Examples of the non-transitory computer readable medium include magnetic recording media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (e.g., magneto-optical disks), CD-ROM (Read Only Memory), CD-R, CD-R / W, and semiconductor memory (e.g., mask ROM, PROM (Programmable ROM), EPROM (Erasable PROM), flash ROM, RAM (Random Access Memory)). The program may be provided to a computer by various types of transitory computer readable media. Examples of the transitory computer readable medium include electric signals, optical signals, and electromagnetic waves. The transitory computer readable medium can provide the program to a computer via a wired communication path such as an electric wire and an optical fiber, or via a wireless communication path.

[0090] Although the present disclosure has been described above with reference to the above-mentioned embodiments, the present disclosure is not limited to the above-mentioned embodiments. Various modifications that can be understood by a person skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. Furthermore, each of the above-mentioned embodiments can be appropriately combined with other embodiments.

[0091] <Additional Notes> A part or all of the above-described embodiments may be described as follows: The configurations of the document creation support device, document creation support method, and program according to the present disclosure are outlined below. However, the present disclosure is not limited to the following configurations. (Appendix 1) a search unit for searching for documents related to a character string included in a document creation command input by a document creator; a generating unit for generating a draft document based on the document creation command and the retrieved related documents; an extraction unit that extracts confirmation items to be confirmed by a document confirmer from the draft document based on the contents of the draft document generated by the document creator from the draft document; A document creation support device comprising: (Appendix 2) 2. A document creation support device according to claim 1, a reason generating unit for generating a reason for a point pointed out in the draft document by the document checker in accordance with the extracted confirmation items, based on the content of the point pointed out, Document creation support device. (Appendix 3) 3. A document creation support device according to claim 2, a revision suggestion unit for generating a revision suggestion for the draft document based on the reason for the indication; Document creation support device. (Appendix 4) 2. A document creation support device according to claim 1, The search unit searches for the related documents, which are documents used in past cases, based on the character string; the generating unit generates the draft document based on the contents of the retrieved related documents and the document creation command. Document creation support device. (Appendix 5) 2. A document creation support device according to claim 1, The search unit searches for the related documents, which are documents used in past cases, based on the character string; the extraction unit extracts the confirmation items from the draft document based on additional information on amendments to the related documents contained in the searched related documents and on the contents of the draft document; Document creation support device. (Appendix 6) 3. A document creation support device according to claim 2, The search unit searches for the related documents, which are documents used in past cases, based on the character string; the reason generating unit generates the reason for the indication based on additional information regarding the correction of the related document contained in the searched related document and the content of the indicated portion. Document creation support device. (Appendix 7) 2. A document creation support device according to claim 1, the generation unit outputs information on the searched related documents to the document creator in association with the draft document. Document creation support device. (Appendix 8) 2. A document creation support device according to claim 1, the generation unit calculates a reliability of the draft document based on the contents of the searched related documents and the contents of the draft document, and outputs the reliability to the document creator in association with the draft document. Document creation support device. (Appendix 9) Searching for related documents according to a character string included in a document creation command input by a document creator; generating a draft document based on the document creation command and the retrieved related documents; extracting confirmation items to be confirmed by a document confirmer from the draft document based on the contents of the draft document generated by the document creator from the draft document; Document creation assistance methods. (Appendix 10) 10. The document creation support method according to claim 9, further comprising: generating a reason for a point pointed out in the draft document by the document checker in accordance with the extracted confirmation items, based on the content of the point pointed out; Document creation assistance methods. (Appendix 11) 11. The document creation support method according to claim 10, further comprising: generating a revision proposal for the draft document based on the reason for the indication; Document creation assistance methods. (Appendix 12) 10. The document creation support method according to claim 9, further comprising: Searching for related documents that are documents used in past cases based on the character string; generating the draft document based on the contents of the retrieved related documents and the document creation instructions; Document creation assistance methods. (Appendix 13) 10. The document creation support method according to claim 9, further comprising: Searching for related documents that are documents used in past cases based on the character string; extracting the confirmation items from the draft document based on additional information on the correction of the related document contained in the retrieved related document and the contents of the draft document; Document creation assistance methods. (Appendix 14) 11. The document creation support method according to claim 10, further comprising: Searching for related documents that are documents used in past cases based on the character string; generating the reason for the indication based on additional information regarding the correction of the related document contained in the searched related document and the content of the indicated portion; Document creation assistance methods. (Appendix 15) 10. The document creation support method according to claim 9, further comprising: outputting information about the retrieved related documents to the document creator in association with the draft document; Document creation assistance methods. (Appendix 16) 10. The document creation support method according to claim 9, further comprising: calculating a reliability of the draft document based on the contents of the retrieved related documents and the contents of the draft document, and outputting the reliability to the document creator in association with the draft document; Document creation assistance methods. (Appendix 17) Searching for related documents according to a character string included in a document creation command input by a document creator; generating a draft document based on the document creation command and the retrieved related documents; extracting confirmation items to be confirmed by a document confirmer from the draft document based on the contents of the draft document generated by the document creator from the draft document; A program that causes a computer to carry out processing. [Explanation of symbols]

[0092] 10 Document creation support device 11 Document Search Section 12 Document Generation Assistant 13 Checklist Generation Unit 14 Reasons for remand presentation section 15 Revision Proposal Department 16 Related Document Storage Unit 20 Creator Terminal 30 Approver terminal 100 Document creation support device 101 CPU 102 ROM 103 RAM 104 Programs 105 Storage device 106 Drive device 107 Communication Interface 108 Input / Output Interface 109 Bus 110 Storage medium 111 Communication Network 121 Search Department 122 Generation part 123 Extraction part

Claims

1. a search unit for searching for documents related to a character string included in a document creation command input by a document creator; a generating unit for generating a draft document based on the document creation command and the retrieved related documents; an extraction unit that extracts confirmation items to be confirmed by a document confirmer from the draft document based on the contents of the draft document generated by the document creator from the draft document; A document creation support device comprising:

2. 2. The document creation support device according to claim 1, a reason generating unit for generating a reason for a point pointed out in the draft document by the document checker in accordance with the extracted confirmation items, based on the content of the point pointed out, Document creation support device.

3. 3. The document creation support device according to claim 2, a revision suggestion unit for generating a revision suggestion for the draft document based on the reason for the indication; Document creation support device.

4. 2. The document creation support device according to claim 1, The search unit searches for the related documents, which are documents used in past cases, based on the character string; the generating unit generates the draft document based on the contents of the retrieved related documents and the document creation command. Document creation support device.

5. 2. The document creation support device according to claim 1, The search unit searches for the related documents, which are documents used in past cases, based on the character string; the extraction unit extracts the confirmation items from the draft document based on additional information on amendments to the related documents contained in the searched related documents and on the contents of the draft document; Document creation support device.

6. 3. The document creation support device according to claim 2, The search unit searches for the related documents, which are documents used in past cases, based on the character string; the reason generating unit generates the reason for the indication based on additional information regarding the correction of the related document contained in the searched related document and the content of the indicated portion. Document creation support device.

7. 2. The document creation support device according to claim 1, the generation unit outputs information on the searched related documents to the document creator in association with the draft document. Document creation support device.

8. 2. The document creation support device according to claim 1, the generation unit calculates a reliability of the draft document based on the contents of the searched related documents and the contents of the draft document, and outputs the reliability to the document creator in association with the draft document. Document creation support device.

9. Searching for related documents according to a character string included in a document creation command input by a document creator; generating a draft document based on the document creation command and the retrieved related documents; extracting confirmation items to be confirmed by a document confirmer from the draft document based on the contents of the draft document generated by the document creator from the draft document; Document creation assistance methods.

10. Searching for related documents according to a character string included in a document creation command input by a document creator; generating a draft document based on the document creation command and the retrieved related documents; extracting confirmation items to be confirmed by a document confirmer from the draft document based on the contents of the draft document generated by the document creator from the draft document; A program that causes a computer to carry out processing.

Images