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

Abstract

To provide a document creation support device and the like that streamline a creation process of application documents for patents and utility models.SOLUTION: A document creation support device 100 includes: a first claim draft generation unit 24 that inputs a prompt 32 including an idea 1 of an invention or plan and a creation guideline 5 of a claim into a large-scale language model 200, generates a first claim draft 6, the first claim draft generation unit 24 providing one or more reference claims 4 to the large-scale language model 200 as teacher data; a claim evaluation unit 25 that inputs a prompt 33 including the first claim draft 6 and an evaluation guideline 7 of a claim into the large-scale language model 200, and generates an evaluation result 8 for the first claim draft 6; and a second claim draft generation unit that inputs a prompt 34 including the first claim draft 6 and the evaluation result 8 into the large-scale language model 200 and generates a second claim draft 10.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a document creation support device, a document creation support method, and a program, and more particularly to a technique for supporting the creation of application documents for patents and utility models. [Background technology]

[0002] When filing a patent application or a utility model registration application, it is necessary to prepare application documents such as an application form, specification, claims (patent claims or utility model registration claims), abstract, and drawings.

[0003] These application documents are often prepared by the inventors who created the inventions. However, the process of turning the ideas of the inventions into application documents can be a time-consuming task for inventors who are not familiar with preparing application documents.

[0004] In response to this, Patent Document 1 proposes a patent document generation device that can assist in the creation of application documents. According to Patent Document 1, it is possible to automatically generate the fields of "Title of the Invention," "Technical Field to which the Invention Pertains," "Background Art," "Prior Art Documents," "Problems to be Solved by the Invention," "Effects of the Invention," and "Forms for Carrying Out the Invention," that is, the description items of the specification, based on the invention document (idea) and reference patent documents similar to the invention document. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 7086424 Summary of the Invention [Problem to be solved by the invention]

[0006] However, Patent Document 1 does not provide any specific proposal regarding automatic generation of claims (patent claims or utility model registration claims). Therefore, it would be useful to have a tool that supports the streamlining of the process of creating application documents, including claims.

[0007] The present invention has been made in consideration of the above circumstances, and has an object to provide a document creation support device and the like that improves the efficiency of the process of creating application documents for patents and utility models. [Means for solving the problem]

[0008] The first invention for solving the above-mentioned problem is a document creation support device comprising: a first claim draft generation unit that inputs an idea for an invention or device and a prompt including instructions for drafting a claim into a large-scale language model and generates a first draft claim, the first claim draft generation unit providing one or more reference claims as training data to the large-scale language model; a claim evaluation unit that inputs the first draft claim and a prompt including instructions for evaluating the claims into the large-scale language model and generates an evaluation result for the first draft claim; and a second claim draft generation unit that inputs the first draft claim and a prompt including the evaluation result into the large-scale language model and generates a second draft claim.

[0009] According to the document creation support device of the first invention, a prompt including an idea of ​​an invention or device and a method for creating a claim is input to a large-scale language model, and a first draft claim is automatically generated. At this time, one or more reference claims are provided to the large-scale language model as training data, so that an appropriate draft claim reflecting existing knowledge on the claim is generated. Furthermore, the document creation support device generates an evaluation result for the first draft claim by inputting the generated first draft claim and a prompt including a method for evaluating the claim into the large-scale language model. Then, the document creation support device generates a second draft claim by inputting the first draft claim and a prompt including the evaluation result into the large-scale language model. This makes it possible to obtain a second draft claim that is an improvement over the first draft claim. According to the present invention, even an inventor or designer who is not familiar with creating application documents can easily create a draft of the scope of a patent claim or the scope of a utility model registration, and the process of creating application documents for a patent or utility model is made more efficient.

[0010] In the first invention, the invention further comprises a database that collects and stores patent information and / or utility model information, and a reference claim acquisition unit that selects patent information and / or utility model information similar to the idea from the database and acquires claims related to the selected patent information and / or utility model information as reference claims, and the first claim proposal generation unit may provide the reference claims acquired by the reference claim acquisition unit to the large-scale language model as training data. In this way, claims of patent information and utility model information similar to the idea can be acquired from the database as reference claims and used to generate claim proposals.

[0011] The method for evaluating claims may include evaluating whether or not additional matters can be added to the claims. As a result of the evaluation, information on technical matters that can be added to the claims can be obtained, and a more substantial draft claim can be generated.

[0012] The system may further include an idea evaluation unit that inputs the idea and a prompt including an idea evaluation method into a large-scale language model to generate an evaluation result for the idea. The idea evaluation method may include evaluating the patentability of the idea. This generates an evaluation of the patentability of the idea, and provides information for determining the validity of the idea and the need for further consideration.

[0013] The system may further include a specification draft generation unit that inputs the second draft claim and a prompt including an example of a written specification into a large-scale language model to generate a draft specification. This makes it possible to automatically generate not only the claims but also the draft specification.

[0014] Furthermore, the second claim draft generation unit may provide one or more reference claims as training data to the large-scale language model when generating the second claim draft, thereby generating an appropriate claim draft (second claim draft) that reflects existing knowledge about claims.

[0015] A second invention is a document creation support method characterized in that a computer executes the following steps: inputting an idea for an invention or device, and a prompt including instructions for drafting a claim, into a large-scale language model to generate a first draft claim, in which one or more reference claims are provided to the large-scale language model as training data; inputting the first draft claim and a prompt including instructions for evaluating the claims into the large-scale language model to generate an evaluation result for the first draft claim; and inputting the first draft claim and a prompt including the evaluation result into the large-scale language model to generate a second draft claim.

[0016] A third invention is a program that causes a computer to function as a first claim draft generation unit that inputs a prompt including an idea for an invention or device and instructions for drafting a claim into a large-scale language model and generates a first claim draft, the first claim draft generation unit providing one or more reference claims as training data to the large-scale language model, a claim evaluation unit that inputs the first claim draft and a prompt including instructions for evaluating a claim into the large-scale language model and generates an evaluation result for the first claim draft, and a second claim draft generation unit that inputs the first claim draft and a prompt including the evaluation result into the large-scale language model and generates a second claim draft. Effect of the Invention

[0017] According to the present invention, a document creation support device and the like are provided that improve the efficiency of the process of creating application documents for patents and utility models. [Brief description of the drawings]

[0018] [Figure 1] FIG. 2 illustrates an example of a hardware configuration of a document creation support device. [Diagram 2] FIG. 2 is a diagram illustrating an example of a functional configuration of a document creation support device. [Diagram 3] 1 is a flowchart showing the flow of a document creation assistance method. [Figure 4] FIG. 13 is a diagram showing an example of an idea (text data) input by a document creator. [Diagram 5] FIG. 13 illustrates an example of a prompt for generating an evaluation result for an idea. [Figure 6] FIG. 13 is a diagram illustrating an example of generating an evaluation result of an idea. [Figure 7] FIG. 1 is a diagram showing an example of a reference claim. [Figure 8] FIG. 13 illustrates an example of a prompt for generating a first proposed claim. [Figure 9] FIG. 13 is a diagram showing an example of generating a first proposed claim. [Figure 10] FIG. 13 illustrates an example of a prompt for generating an evaluation result for a first proposed claim. [Figure 11] FIG. 13 is a diagram showing an example of the generation of an evaluation result of a first proposed claim. [Figure 12] FIG. 13 illustrates an example of a prompt for generating a second proposed claim. [Figure 13] FIG. 13 is a diagram showing an example of generating a second proposed claim. [Figure 14] FIG. 13 illustrates an example of a prompt for generating a draft specification. [Figure 15] FIG. 13 is a diagram showing an example of generation of a draft specification. [Figure 16] FIG. 13 is a diagram illustrating another example of the configuration of the document creation support device. [Figure 17] 1 is a flowchart showing the flow of a document creation assistance method. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

[0020] Fig. 1 is a diagram showing an example of a hardware configuration of a document creation support device 100 according to an embodiment of the present invention. For example, when a general computer is used as the document creation support device 100, the document creation support device 100 includes a control unit 101, a storage unit 102, a communication unit 103, an input unit 104, a display unit 105, a peripheral device I / F unit 106, and the like, which are connected via a bus 107, as shown in Fig. 1. Note that the configuration shown in Fig. 1 is only an example, and the document creation support device 100 can have various configurations depending on the application and purpose.

[0021] The control unit 101 has a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), etc. The CPU loads a program stored in the storage unit 102 or a recording medium such as a ROM into a work memory area on the RAM and executes the program, and drives and controls each unit connected via a bus 107 to realize each process of the document creation support device 100, which will be described later.

[0022] The ROM is a non-volatile memory that permanently stores the computer's boot program, programs such as the BIOS, data, etc. The RAM is a volatile memory that temporarily stores programs and data loaded from the storage unit 102 or a recording medium such as a ROM, and also has a work area that the control unit 101 uses to perform various processes.

[0023] The storage unit 102 is a hard disk drive (HDD), a solid state drive (SSD), a flash memory, or the like, and stores programs executed by the control unit 101, data necessary for executing the programs, an operating system (OS), and the like. In this embodiment, the storage unit 102 stores a large-scale language model 200, an intellectual property database 300, and the like. Note that the programs and data necessary for executing the programs may be stored on a network such as a cloud server. In this case, the control unit 101 downloads the programs and data necessary for executing the programs from the cloud server, or the like, before or during execution of the programs.

[0024] The communication unit 103 includes a communication interface and a communication control circuit that mediate communication of the document creation support device 100, and controls communication via a network. The network may be a local area network (LAN), a wide area network (WAN), the Internet, or the like, and may be wired or wireless.

[0025] The input unit 104 includes input devices such as a keyboard, a mouse, or a touch panel, and various operation buttons, etc. The input unit 104 transmits input data and operation instructions to the control unit 101.

[0026] The display unit 105 includes a display such as a liquid crystal panel, and displays data such as images and text on the display according to instructions from the control unit 101.

[0027] The peripheral device I / F unit 106 is a port for connecting peripheral devices, and includes short-range wireless communication such as USB and Bluetooth (registered trademark). The control unit 101 transmits and receives data to and from peripheral devices via the peripheral device I / F unit 106. For example, a printer is connected to the peripheral device I / F unit 106.

[0028] Next, the functional configuration of the document creation support device 100 will be described with reference to Fig. 2. As shown in Fig. 2, the document creation support device 100 includes, as functional units, an idea input unit 21, an idea evaluation unit 22, a reference claim acquisition unit 23, a first claim proposal generation unit 24, a claim evaluation unit 25, a second claim proposal generation unit 26, a specification proposal generation unit 27, and a document storage unit 28. These functions are realized by a control unit 101 of the document creation support device 100 reading and executing a program stored in a storage unit 102 or a ROM.

[0029] The idea input unit 21 accepts input (text input) of an idea 1 for an invention or device from a document creator (inventor or designer) via the input unit 104. The input idea 1 is stored in the storage unit 102 as text data.

[0030] The idea evaluation unit 22 inputs the prompt 31 to a large-scale language model 200 and generates an evaluation result 3 for the idea 1. In this embodiment, the large-scale language model 200 (LLM: Large Language Model) is a language model trained using a large amount of text data, and is typically GPT-3 (Generative Pre-trained Transformer 3). The large-scale language model 200 is capable of generating a sentence in response to a given prompt.

[0031] The prompt 31 includes the idea 1 input by the idea input unit 21, and the evaluation criteria 2 for the idea (FIG. 5). The evaluation criteria 2 for the idea prescribe the criteria, method, items, etc. for evaluating an idea. The evaluation criteria 2 can be set arbitrarily, and may include, for example, an evaluation of the patentability of an idea, such as "novelty," "inventive step," and "industrial applicability," as well as an evaluation of whether the technical elements included in the idea are consistent with each other and free of contradictions, such as "logical consistency."

[0032] The reference claim acquisition unit 23 selects patent information and / or utility model information similar to the idea 1 from the intellectual property database 300, and acquires the claim related to the selected patent information and / or utility model information as the reference claim 4. The intellectual property database 300 (abbreviated as IP DB) is a database that collects and accumulates patent information and / or utility model information. Each patent information (each utility model information) includes document information such as the claims, specification, and abstract of each patent (each utility model), and drawings (image data). In addition, information that identifies each patent information (each utility model information) (application number, publication number, patent number, etc.), information on natural persons and corporations involved in the patent (utility model), such as the applicant and inventor, and classification information of technical contents, such as IPC, FI, and F terms, are included.

[0033] The first claim draft generation unit 24 inputs the prompt 32 into the large-scale language model 200 to generate a first claim draft 6. The prompt 32 includes the idea 1 input by the idea input unit 21, a claim drafting procedure 5, and one or more reference claims 4 acquired by the reference claim acquisition unit 23 (FIG. 8). The claim drafting procedure 5 specifies the description rules and constraints for the claims, and can be set arbitrarily.

[0034] In this embodiment, the first claim proposal generation unit 24 provides the reference claim 4 as training data to the large-scale language model 200. Specifically, the first claim proposal generation unit 24 applies few-shot learning to the large-scale language model 200 using the reference claim 4 as training data. This enables the large-scale language model 200 to understand the writing style, etc. of the reference claim 4, and generate an appropriate claim proposal (first claim proposal 6) that reflects existing knowledge about claims.

[0035] The claim evaluation unit 25 inputs the prompt 33 into the large-scale language model 200, and generates an evaluation result 8 for the first draft claim 6. The prompt 33 includes the first draft claim 6 generated by the first draft claim generation unit 24, a claim evaluation procedure 7, etc. (FIG. 10). The claim evaluation procedure 7 prescribes the criteria, method, items, etc. for evaluating a claim. The evaluation procedure 7 can be set arbitrarily, and can include, for example, an evaluation of whether or not it is possible to include additional matters in the claim.

[0036] The second claim draft generation unit 26 inputs the prompt 34 into the large-scale language model 200 to generate a second claim draft 10. The prompt 34 includes the first claim draft 6 generated by the first claim draft generation unit 24, a claim drafting procedure 9, and an evaluation result 8 generated by the claim evaluation unit 25 (FIG. 12).

[0037] The specification draft generation unit 27 inputs the prompt 35 into the large-scale language model 200 to generate a specification draft 12. The prompt 35 includes a second claim draft 10 generated by the second claim draft generation unit 26, a specification draft example 11 (specification template), and the like (FIG. 14).

[0038] The document storage unit 28 stores the second draft claim 10 generated by the second draft claim generation unit 26 and the draft specification 12 generated by the draft specification generation unit 27 in the memory unit 102, assigning a timestamp to them as drafts of application documents generated by the document creation support device 100.

[0039] Next, a document creation support method (document creation support process) by the document creation support device 100 (computer) will be described with reference to the flowchart of Fig. 3. The control unit 101 of the document creation support device 100 executes the process of each step in Fig. 3, thereby realizing each function of the document creation support device 100 (Fig. 2).

[0040] First, the control unit 101 of the document creation support device 100 (hereinafter, simply referred to as "document creation support device 100") accepts input (text input) of an idea 1 for an invention or device from a document creator (inventor or designer) via the input unit 104 (step S1). The input idea 1 is stored in the storage unit 102 as text data.

[0041] Figure 4 shows an example of the input idea 1 (text data). In the example of Figure 4, the idea of ​​an invention (purpose of the invention, features of the invention, etc.) related to a manufacturing method of high purity and high melting point metal powder is described in free format.

[0042] Next, the document creation support device 100 inputs the idea 1 and the prompt 31 including the evaluation procedure 2 and the like to the large-scale language model 200, generates an evaluation result 3 for the idea 1, and outputs it to the display unit 105 (step S2).

[0043] FIG. 5 shows an example of a prompt 31. The prompt 31 is an instruction to evaluate a given idea 1 according to the evaluation guidelines 2. In the “### evaluation guidelines”, it is specified that the three items of “novelty”, “inventive step”, and “logical consistency” are to be evaluated as the evaluation guidelines 2, and various restrictions for the evaluation and the evaluation results are to be output in Markdown format. In the “### evaluation format”, the evaluation format at the time of output is specified. In addition, it is specified that each of the items of “novelty”, “inventive step”, and “logical consistency” is to be evaluated on a five-point scale from 1 to 5 as the evaluation results, and that any points of criticism are to be output. In the “### idea” part, the text of idea 1 (e.g., idea 1 in FIG. 4) is inserted (not shown).

[0044] Figure 6 shows an example of evaluation result 3 generated in response to prompt 31 in Figure 5. As shown in the figure, the large-scale language model 200 generates an evaluation of three items, "novelty," "inventive step," and "logical consistency," and outputs evaluation result 3 according to a specified evaluation format. This provides the document creator with information to judge the validity of his or her idea 1 and the need for further consideration.

[0045] When evaluating "novelty" and "inventive step," document information of patent information (utility model information) obtained from the intellectual property database 300 can also be included in the prompt 31. For example, the document creation support device 100 can perform a text comparison between idea 1 and document information of each patent information (each utility model information) in the intellectual property database 300 to obtain document information of patent information (utility model information) similar to idea 1, and include it in the prompt 31.

[0046] Specifically, the document creation support device 100 vectorizes the text data of the idea 1 and the text data of the specification and abstract of each patent information (utility model information) (or a document in which the specification is summarized by the large-scale language model 200) using, for example, TF-IDF (Term Frequency-Inverse Document Frequency), calculates the similarity between the vectors (such as cosine similarity), and can include in the prompt 31 the specifications and abstracts (or documents in which the specification is summarized by the large-scale language model 200) related to the top few patent information (utility model information) with high similarity. This allows the idea 1 to be evaluated more objectively using existing patent information (utility model information). Note that technical information (text) related to the idea 1 that the document creator recognizes can also be included in the prompt 31.

[0047] Note that the evaluation of idea 1 in step S2 may be performed in a subsequent step, for example, together with the evaluation of first proposed claim 6 in step S6 described below.

[0048] The document creator checks the output evaluation result 3, and if the idea is to be reconsidered (step S3; Yes), the process returns to step S1 to reconsider and re-input the idea. On the other hand, if the document creator proceeds to the preparation of a claim proposal (step S3; No), the document creation support device 100 first selects patent information and / or utility model information similar to the idea 1 from the intellectual property database 300, and obtains a claim related to the selected patent information and / or utility model information as a reference claim 4 (step S4).

[0049] For example, the document creation support device 100 refers to the intellectual property database 300, vectorizes the text data of the idea 1 and the text data of all or part of the claims (for example, only the independent claims) related to each patent information (each utility model information) using, for example, TF-IDF (Term Frequency-Inverse Document Frequency), and calculates the similarity between the vectors (such as cosine similarity). Then, the document creation support device 100 acquires all or part of the claims related to the patent information (utility model information) having a high similarity as reference claim 4. When searching (acquiring) patent information (utility model information), it is also possible to narrow down the search target by using classification information such as IPC of the patent information (utility model information). In addition, the document creation support device 100 may accept the selection of patent information (utility model information) from the document creator via the input unit 104, and acquire all or part of the claims related to the patent information (utility model information) selected by the document creator as reference claim 4.

[0050] FIG. 7 shows an example of reference claim 4. In the example of FIG. 7, two examples of reference claims 4a and 4b similar to Idea 1 are shown. Reference claim 4a is a claim related to a method for manufacturing a high melting point metal powder material, and reference claim 4b is a claim related to a method for manufacturing a high purity silicon powder.

[0051] Next, the document creation support device 100 inputs the prompt 32 including the idea 1, the claim drafting procedure 5, and the reference claim 4, etc., into the large-scale language model 200 to generate a first draft claim 6 (step S5).

[0052] FIG. 8 shows an example of the prompt 32. The prompt 32 is an instruction to create a first draft claim 6 based on the idea 1 and the reference claim 4 in accordance with the drafting guidelines 5. The drafting guidelines 5 are specified in the "### drafting guidelines" section. The text of the reference claim 4 (e.g., the reference claim 4 in FIG. 7) is inserted in the "### claim example" section, and the text of the idea 1 (e.g., the idea 1 in FIG. 4) is inserted in the "### idea" section (not shown). The drafting guidelines 5 can also include individual instructions according to the technical field (e.g., when describing the component range, describe the range and impurities contained in the idea 1, etc.).

[0053] Figure 9 is an example of a first draft claim 6 generated in accordance with prompt 32 in Figure 8. As shown in Figure 8, a claim is automatically generated that reflects the contents of idea 1. This allows even a document writer (inventor or designer) who is inexperienced in creating application documents to easily create a draft of the scope of a patent claim or the scope of a utility model registration by simply inputting idea 1.

[0054] Next, the document creation support device 100 inputs the first draft claim 6 and the prompt 33 including the claim evaluation procedure 7 and the like into the large-scale language model 200, and generates an evaluation result 8 for the first draft claim 6 (step S6).

[0055] FIG. 10 shows an example of the prompt 33. The prompt 33 is an instruction to evaluate the given first proposed claim 6 according to the evaluation guidelines 7. The “### evaluation guidelines” section specifies the evaluation guidelines 7 for the claim, and the “### evaluation format” section specifies the evaluation format for output. As the evaluation result, it is specified to generate and output items that can be added to the claim for each of the items “Possibility of adding technical information”, “Possibility of adding constituents (elements and composition ranges)”, “Possibility of adding physical property evaluation”, and “Possibility of adding application fields”. The “### proposed claim” section is inserted with the text of the first proposed claim 6 (for example, the first proposed claim 6 in FIG. 9) (not shown). Note that the evaluation guidelines 7 can also include individual instructions according to the technical field. For example, the above item “Possibility of adding constituents (elements and composition ranges)” is added as an evaluation item specific to material patents.

[0056] Fig. 11 shows an example of the evaluation result 8 generated in accordance with the prompt 33 in Fig. 10. As shown in Fig. 11, the evaluation result 8 including items that can be added to the claims is generated and output for each of the items "Possibility of adding technical information," "Possibility of adding components (elements and composition ranges)," "Possibility of adding further evaluation of physical properties," and "Possibility of adding application fields."

[0057] Next, the document creation support device 100 inputs the prompt 34, including the first draft claim 6, the claim drafting procedure 9, and the evaluation result 8, into the large-scale language model 200, generates a second draft claim 10, and outputs it to the display unit 105. The document creation support device 100 also assigns a timestamp to the generated second draft claim 10 and stores it in the memory unit 102 (step S7). By assigning a timestamp, an idea that has been "conceived" or "thought of using this document creation support device 100" can be stored as evidence.

[0058] 12 shows an example of a prompt 34. The prompt 34 is an instruction to generate a second draft claim 10 based on a given first draft claim 6 and evaluation result 8. The "### Creation Instructions" section specifies that the first draft claim 6 should be updated to reflect the evaluation result 8. In addition, the text of the first draft claim 6 (e.g., the first draft claim 6 in FIG. 9) is inserted into the "### Evaluation Result" section, and the text of the evaluation result 8 (e.g., evaluation result 8 in FIG. 11) is inserted into the "### Evaluation Result" section (not shown).

[0059] Figure 13 shows an example of a second draft claim 10 generated in accordance with prompt 34 in Figure 12. As shown in Figure 13, an improved second draft claim 10 is generated based on the first draft claim 6 (Figure 9) and its evaluation result 8 (Figure 11). The second draft claim 10 takes into consideration the technical matters that can be added, etc., contained in the evaluation result 8 (Figure 11), and is more substantial in content than the first draft claim 6.

[0060] In step S7, when generating the second draft claim 10, the document creation support device 100 can also provide a reference claim as training data to the large-scale language model 200. In that case, the text of the reference claim is inserted into the prompt 34 in Fig. 12. As the reference claim, for example, a claim related to patent information (utility model information) similar to idea 1 searched from the intellectual property database 300 can be used.

[0061] The document creator checks the output second draft claim 10, and if reconsiders (step S8; Yes), returns to step S1. On the other hand, if proceeding to create a draft specification (step S8; No), proceeds to step S9. Specifically, the document creation support device 100 inputs the second draft claim 10 and a prompt 35 including a specification draft example 11, etc., into the large-scale language model 200, generates a draft specification 12, and outputs it to the display unit 105. The document creation support device 100 also assigns a timestamp to the generated draft specification 12 and stores it in the memory unit 102 (step S9).

[0062] FIG. 14 shows an example of a prompt 35. The prompt 35 is an instruction to create a draft specification 12 based on a given second draft claim 10 in accordance with preparation example 11 (specification template). The instruction is to enter specific sentences in accordance with the content in the "〇〇" in preparation example 11. By giving such an instruction, a draft specification incorporating specific content in accordance with the specification format is generated. The text of specification preparation example 11 (specification template) is inserted in the "### preparation example" part, and the text of the second draft claim 10 (for example, the second draft claim 10 in FIG. 13) is inserted in the "### claim" part (not shown).

[0063] Figure 15 shows an example of a draft specification 12 (a part of a specification) generated in accordance with the prompt 35 in Figure 14. As shown in Figure 15, the draft specification 12 is generated by the large-scale language model 200 based on the second draft claim and the specification preparation example 11. In this way, not only the claims but also the draft specification can be automatically generated.

[0064] Through the above process, the drafts of documents (claims and specification) required for a patent (utility model) application are automatically generated. The document creator checks the drafts (second draft claim 10 and draft specification 12), and if there is any need for reconsideration (step S10; Yes), returns to step S1. On the other hand, if the drafts (second draft claim 10 and draft specification 12) are approved (step S10; No), the process ends.

[0065] As described above, according to the document creation support device 100 of the present embodiment, the prompt 32 including the idea 1 of the invention or device and the claim drafting method 5 is input to the large-scale language model 200, and the first claim draft 6 is automatically generated. At this time, one or more reference claims 4 are provided to the large-scale language model 200 as training data, so that an appropriate claim draft reflecting existing knowledge on claims is generated. Furthermore, the document creation support device 100 generates an evaluation result 8 for the first claim draft 6 by inputting the generated first claim draft 6 and a prompt 33 including the claim evaluation method 7 to the large-scale language model 200. Then, the document creation support device 100 generates a second claim draft 10 by inputting the first claim draft 6 and a prompt 34 including the evaluation result 8 to the large-scale language model 200. As a result, the second claim draft 10 which is an improvement of the first claim draft 6 can be obtained. According to this embodiment, even inventors or designers who are inexperienced in preparing application documents can easily draft patent claims or utility model registration claims, thereby improving the efficiency of the process of preparing patent and utility model application documents.

[0066] In the present embodiment, an example in which the document creation support device 100 is realized by one computer has been described, but the document creation support device 100 (document creation support system) can also be constructed using multiple computers. For example, the document creation support device 100a (document creation support system) shown in FIG. 16 is constructed by connecting a terminal 40 and a server 50 via a network 60. The terminal 40 is a terminal used by a document creator, and the server 50 is a server that generates a proposed claim and a proposed specification in response to an instruction from the document creator. In such a system configuration, the terminal 40 and the server 50 share the functions of FIG. 2. For example, the terminal 40 includes the idea input unit 21, the document storage unit 28, and the like, among the functional units in FIG. 2. Meanwhile, the server 50 includes functions such as the reference claim acquisition unit 23, the first proposed claim generation unit 24, the claim evaluation unit 25, the second proposed claim generation unit 26, and the proposed specification generation unit 27. The hardware configurations of the terminal 40 and the server 50 are the same as those of FIG. 1.

[0067] 17 is a flowchart showing the flow of the document creation support method by the document creation support device 100a (document creation support system). The control unit 101 of the terminal 40 (hereinafter simply referred to as "terminal 40") accepts input of the idea 1 from the document creator, and transmits the input idea 1 (text data) to the server 50 (step S21). When the control unit 101 of the server 50 (hereinafter simply referred to as "server 50") receives the idea 1 from the terminal 40 (step S22), it inputs the idea 1 and a prompt 31 including an evaluation method 2 for the idea, etc., into the large-scale language model 200, generates an evaluation result 3 for the idea 1, and transmits the evaluation result 3 to the terminal 40 (step S23).

[0068] The terminal 40 receives the evaluation result 3 from the server 50 and displays it on the display unit 105 (step S24). The document creator checks the evaluation result 3, and if the document creator wishes to reconsider the idea (step S25; Yes), the process returns to step S21 to reconsider and re-input the idea. On the other hand, if the document creator wishes to proceed to the creation of a proposed claim (step S25; No), the terminal 40 transmits an instruction to the server 50 to create a proposed claim (step S26).

[0069] When the server 50 receives the instruction to prepare a draft claim (step S27), it first selects patent information and / or utility model information similar to the idea 1 from the intellectual property database 300, and obtains a claim related to the selected patent information and / or utility model information as a reference claim 4 (step S28). Next, the server 50 inputs a prompt 32 including the idea 1, the claim preparation procedure 5, and the reference claim 4, etc., into the large-scale language model 200, and generates a first draft claim 6 (step S29).

[0070] Next, the server 50 inputs the first draft claim 6 and the prompt 33 including the claim evaluation procedure 7, etc., into the large-scale language model 200, and generates an evaluation result 8 for the first draft claim 6 (step S30).Then, the server 50 inputs the first draft claim 6, the prompt 34 including the claim drafting procedure 9, and the evaluation result 8, etc., into the large-scale language model 200, and generates a second draft claim 10, and transmits the generated second draft claim 10 to the terminal 40 (step S31).

[0071] The terminal 40 receives the second draft claim 10 and displays it on the display unit 105, and also assigns a timestamp to the second draft claim 10 and stores it in the memory unit 102 (step S32). The document preparer checks the second draft claim 10, and if the document preparer wishes to reconsider it (step S33; Yes), the process returns to step S21. On the other hand, if the document preparer wishes to proceed with the preparation of a draft specification (step S33; No), the terminal 40 transmits an instruction to the server 50 to prepare a draft specification (step S34).

[0072] When the server 50 receives an instruction to create a draft specification (step S35), it inputs a prompt 35 including a second draft claim 10 and a specification draft example 11, etc., into the large-scale language model 200 to generate a draft specification 12 and transmits the generated draft specification 12 to the terminal 40 (step S36).

[0073] The terminal 40 receives the draft specification 12 and displays it on the display unit 105, while also adding a timestamp to the draft specification 12 and storing it in the memory unit 102 (step S37). The document creator checks the drafts (second draft claim 10 and draft specification 12) and, if there is to be reconsideration (step S38; Yes), returns to step S21. On the other hand, if the drafts (second draft claim 10 and draft specification 12) are approved (step S38; No), the process ends.

[0074] As described above, a document creation support device (document creation support system) may be configured by the terminal 40 and the server 50. In the above example, the case where there is one terminal 40 has been described, but multiple terminals may be connected to the network 60 for each document creator. Furthermore, the division of functions between the terminal 40 and the server 50 is not limited to the above example. For example, the terminal 40 may be provided with the reference claim acquisition unit 23. In this case, the process of acquiring the reference claim 4 in step S28 is executed on the terminal 40 side, and the acquired reference claim 4 is transmitted to the server 50. Then, in step S29, the server 50 generates a first proposed claim 6 based on the idea 1, the claim drafting procedure 5, and the reference claim 4 received from the terminal 40.

[0075] Although the preferred embodiment of the present invention has been described above with reference to the accompanying drawings, the present invention is not limited to the above examples. It is clear that a person skilled in the art can come up with various modified or altered examples within the scope of the technical ideas disclosed in this application, and it is understood that these also naturally belong to the technical scope of the present invention. [Explanation of symbols]

[0076] 100: Document creation support device 100a: Document creation support device 200: Large-scale language models 300: Intellectual Property Database 1: Idea 2: How to evaluate ideas 3: Evaluation results for ideas 4:Reference claim 5: How to prepare claims 6:First claim draft 7: Claim evaluation procedure 8: Evaluation result for the first claim plan 9: How to prepare claims 10:Second claim proposal 11: Example of specification 12: Draft specification 21: Idea input section 22: Idea Evaluation Department 23: Reference Patent Acquisition Department 24:First claim generation unit 25: Claims Evaluation Department 26: Second claim draft generation unit 27: Specification draft generation section 28: Document Preservation Department 31~35: Prompt 40: Terminal 50: Server 60: Network

Claims

1. a first claim draft generation unit that inputs a prompt including an idea for an invention or device and a method for drafting a claim into a large-scale language model to generate a first claim draft, the first claim draft generation unit providing one or more reference claims as training data to the large-scale language model; a claim evaluation unit that inputs the first draft claim and a prompt including a claim evaluation method into a large-scale language model to generate an evaluation result for the first draft claim; a second claim draft generator that inputs the first draft claim and a prompt including the evaluation result into a large-scale language model to generate a second draft claim; A document creation support device comprising:

2. A database that collects and stores information on each patent and / or utility model; a reference claim acquisition unit that selects patent information and / or utility model information similar to the idea from the database, and acquires claims related to the selected patent information and / or utility model information as reference claims; 2. The document creation support apparatus according to claim 1, wherein the first claim draft generating section provides the reference claim acquired by the reference claim acquiring section as training data to the large-scale language model.

3. 2. The document creation support device according to claim 1, wherein the claim evaluation procedure includes evaluating whether or not it is possible to include additional matters in the claims.

4. 2. The document creation support device according to claim 1, further comprising an idea evaluation unit that inputs the idea and a prompt including an idea evaluation method into a large-scale language model and generates an evaluation result for the idea.

5. 5. The document creation support device according to claim 4, wherein the method for evaluating an idea includes evaluating the patentability of the idea.

6. 2. The document creation support device according to claim 1, further comprising a specification draft generation unit that inputs the second draft claim and a prompt including an example of a draft specification into a large-scale language model to generate a draft specification.

7. 2. The document creation support apparatus according to claim 1, wherein the second claim draft generation unit provides one or more reference claims as training data to the large-scale language model when generating the second claim draft.

8. The computer a step of inputting a prompt including an idea for an invention or device and a method for drafting a claim into a large-scale language model to generate a first draft claim, the step of providing one or more reference claims as training data to the large-scale language model; inputting the first draft claim and a prompt including claim evaluation instructions into a large scale language model to generate an evaluation result for the first draft claim; inputting the first draft claim and a prompt including the evaluation result into a large scale language model to generate a second draft claim; A document creation support method comprising the steps of:

9. Computer, a first claim draft generation unit for inputting a prompt including an idea for an invention or device and a method for drafting a claim into a large-scale language model to generate a first claim draft, the first claim draft generation unit providing one or more reference claims as training data to the large-scale language model; a claim evaluation unit that inputs the first draft claim and a prompt including a claim evaluation guideline into a large-scale language model to generate an evaluation result for the first draft claim; a second claim draft generator for inputting the first draft claim and a prompt including the evaluation result into a large-scale language model to generate a second draft claim; A program characterized by causing the program to function as a

Images