Code checking program and code checking device

Abstract

From the drawings and the legends of the drawings, the element names and symbols of the legends and drawings are examined. [Solution] The code inspection program causes a computer to perform the following steps: inputting a first prompt to a generation AI system 3, the first prompt including an instruction to generate a description code table 24 containing the correspondence between codes and element names from a description 23 regarding the drawing, and obtaining an answer including the specified description code table 24 from the generation AI; inputting a second prompt to the generation AI system 3, the second prompt including an instruction to generate a drawing code table 26 containing the correspondence between codes and element names from the drawing described in the description 23, and obtaining an answer including the specified drawing code table 26 from the generation AI system 3; and outputting a code determination table 27 that determines whether there are any typos in the codes from the description code table 24 and the drawing code table 26.

Description

[Technical Field]

[0001] The present invention relates to a code checking program and a code checking device. [Background technology]

[0002] With the recent development of IT (information technology), many legal tech technologies have been announced that use IT to support the creation of legal documents.In the case of patent specification preparation, many technologies have been announced that use IT to support the creation of specifications. For example, Patent Document 1 describes a document check device that can check the consistency of the use of specific character strings in a document. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2011-118861 Summary of the Invention [Problem to be solved by the invention]

[0004] Patent Document 1 discloses a code checking means for checking the consistency between descriptive terms and symbols in a document. This makes it possible to confirm the consistency between descriptive terms and symbols in the document. However, in a patent specification, it is necessary to check not only the consistency between descriptive terms and symbols in the document, but also the consistency between descriptive terms and symbols in the drawings. Patent Document 1 does not disclose how to ensure code consistency by taking into account both the drawings and the document.

[0005] Therefore, an object of the present invention is to check the element names and symbols of the description and the drawing from the drawing and the description of the drawing. [Means for solving the problem]

[0006] That is, the above-mentioned problems of the present invention are solved by the following configuration. The code inspection program of the present invention includes the steps of inputting a first prompt including an instruction to generate a description code table in which the correspondence between codes and element names is described from a description about a drawing to a generation AI, and obtaining a response including a predetermined description code table from the generation AI; and instructing the generation AI to generate a drawing code table in which the correspondence between codes and element names is described from the drawing described in the description. , and an instruction to have no element name if there is no element name a step of inputting a second prompt including the above, and obtaining an answer including a predetermined drawing code table from the generation AI; and a step of outputting a code determination table that determines whether a code is incorrect from the description code table and the drawing code table. execution It is intended to make it so.

[0007] The code inspection program of the present invention instructs a generation AI to generate a description code table in which the correspondence between codes and element names is described from a description about a drawing, and to generate a drawing code table in which the correspondence between codes and element names is described from a drawing described in the description. , and an instruction to have no element name if there is no element name a step of inputting a prompt including the above, and obtaining an answer including a predetermined description code table and a predetermined drawing code table from the generation AI; and a step of outputting a code determination table that determines whether a code is incorrect from the description code table and the drawing code table. execution It is intended to make it so.

[0008] The code inspection device of the present invention includes an explanatory code table specification unit that inputs a first prompt including an instruction to generate an explanatory code table that describes the correspondence between codes and element names from explanatory text on drawings to a generation AI and obtains a response including a predetermined explanatory code table; and an instruction to generate a drawing code table that describes the correspondence between codes and element names from drawings described in the explanatory text to the generation AI. , and an instruction to have no element name if there is no element name and a code determination table creation unit that outputs a code determination table that determines whether a code is incorrect from the description code table and the drawing code table. The other configurations will be explained in the embodiments. [Effects of the Invention]

[0009] According to the present invention, it is possible to check the element names and symbols of the description and the drawing from the drawing and the description of the drawing. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a logical configuration diagram showing a code inspection device according to a first embodiment. [Figure 2] FIG. 2 is a hardware configuration diagram showing a code inspection device. [Figure 3] 10 is a screen displayed by the code checker. [Figure 4A] 10 is a flowchart of a code check process. [Figure 4B] 10 is a flowchart of a code check process. [Figure 5] FIG. 10 is a diagram showing a request body to be sent to the generation AI system. [Figure 6] FIG. 10 is a diagram illustrating the contents of a prompt. [Figure 7] A diagram showing a chat completion object received from the generation AI system. [Figure 8] FIG. 10 is a diagram illustrating the content of a reply included in a chat completion object. [Figure 9] FIG. 10 is a diagram showing a request body to be sent to the generation AI system. [Figure 10] FIG. 10 is a diagram illustrating the contents of a prompt. [Figure 11] A diagram showing a chat completion object received from the generation AI system. [Figure 12A] FIG. 10 is a diagram illustrating the content of a reply included in a chat completion object. [Figure 12B] FIG. 10 is a diagram showing a code table of answered explanations. [Figure 13] FIG. 10 shows an image of a drawing to accompany the prompt. [Figure 14] FIG. 10 is a diagram showing a request body to be sent to the generation AI system. [Figure 15] FIG. 10 is a diagram illustrating the contents of a prompt. [Figure 16] A diagram showing a chat completion object received from the generation AI system. [Figure 17A] FIG. 10 is a diagram illustrating the content of a reply included in a chat completion object. [Figure 17B] FIG. 10 is a diagram showing the drawing code table that was answered. [Figure 18] FIG. 10 is a diagram showing a request body to be sent to the generation AI system. [Figure 19] FIG. 10 is a diagram illustrating the contents of a prompt. [Figure 20] A diagram showing a chat completion object received from the generation AI system. [Figure 21A] FIG. 10 is a diagram illustrating the content of a reply included in a chat completion object. [Figure 21B] FIG. 10 is a diagram showing a returned code determination table. [Figure 22] FIG. 10 is a logical configuration diagram showing a code inspection device according to a second embodiment. [Figure 23] 10 is a flowchart of a code check process. [Figure 24] FIG. 10 is a diagram showing a request body to be sent to the generation AI system. [Figure 25] FIG. 10 is a diagram illustrating the contents of a prompt. [Figure 26] A diagram showing a chat completion object received from the generation AI system. [Figure 27] FIG. 10 is a diagram illustrating the content of a reply included in a chat completion object. [Figure 28] 10 is a flowchart of a code inspection process that spans multiple drawings. [Figure 29] FIG. 10 is a logical configuration diagram showing a code inspection device according to a third embodiment. [Figure 30] 10 is a flowchart of a code check process. [Figure 31] FIG. 10 is a diagram showing a request body to be sent to the generation AI system. [Figure 32] FIG. 10 is a diagram illustrating the contents of a prompt. [Figure 33]A diagram showing a chat completion object received from the generation AI system. [Figure 34] FIG. 10 is a diagram illustrating the content of a reply included in a chat completion object. [Figure 35] FIG. 10 is a logical configuration diagram showing a code inspection device according to a fourth embodiment. [Figure 36] 10 is a flowchart of a code check process. [Figure 37] FIG. 10 is a diagram showing a request body to be sent to the generation AI system. [Figure 38] FIG. 10 is a diagram illustrating the contents of a prompt. [Figure 39] A diagram showing a chat completion object received from the generation AI system. [Figure 40] FIG. 10 is a diagram illustrating the content of a reply included in a chat completion object. [Figure 41] This is a diagram explaining the contents of the reply including the explanatory code table of drawing number 1. [Figure 42] This is a diagram explaining the contents of the reply including the drawing code table for drawing number 1. [Figure 43] FIG. 10 is a diagram showing an example of the explanatory text for drawing number 2. [Figure 44] This is a diagram explaining the contents of the reply including the explanatory code table of drawing number 2. [Figure 45] FIG. 10 is a diagram showing an example of a drawing image of drawing number 2. [Figure 46] This is a diagram explaining the contents of the reply including the drawing code table for drawing number 2. [Figure 47] FIG. 10 is a diagram illustrating the contents of a response including a code determination table. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 is a logical configuration diagram showing a code inspection device 1 according to the first embodiment. The code inspection device 1 includes an explanatory text identification unit 11, an explanatory text extraction unit 12, an explanatory text code table identification unit 13, an explanatory text code table extraction unit 14, a drawing code table identification unit 15, a drawing code table extraction unit 16, a code judgment table identification unit 17, and a code judgment table extraction unit 18. The code inspection device 1 is a computer, and cooperates with the generation AI system 3 to inspect the correspondence between the code and element name described in the explanatory text 23 specified by the drawing number 22 included in the natural language 21 and the code and element name described in the drawing image 25.

[0012] When the natural language sentence 21 and the drawing number 22 are input, the explanatory sentence identification unit 11 instructs the generation AI system 3 to identify the explanatory sentence 23 relating to the drawing specified by the drawing number 22 from the natural language sentence 21 using a request body 51 including a first prompt. The natural language sentence 21 includes the explanatory sentence 23 for the drawing specified by the drawing number 22.

[0013] This request body 51 is written in JSON format. The request body 51 is executed by the generation AI system 3, for example, by calling the API (Application Programming Interface) of the generation AI system 3, and a chat completion object 52 is generated. Then, the chat completion object 52 in JSON format is sent from the generation AI system 3 to the explanation specification unit 11. This chat completion object 52 includes the explanation 23 of the drawing specified by the drawing number 22.

[0014] The description extracting unit 12 extracts the description 23 of the drawing designated by the drawing number 22 from the chat completion object 52 .

[0015] When an explanatory text 23 is input, the explanatory text code table identification unit 13 instructs the generation AI system 3 to identify an explanatory text code table 24, which is a combination of the code and element name described in the explanatory text 23, using a request body 53.

[0016] The request body 53 is written in JSON format. This request body 53 is executed by the generation AI system 3, for example, by calling the API of the generation AI system 3, and a chat completion object 54 is generated. Then, the generation AI system 3 returns the chat completion object 54 in JSON format to the description code table identification unit 13. This chat completion object 54 includes the description code table 24.

[0017] The explanation code table extracting unit 14 extracts the explanation code table 24 from the chat completion object 54 .

[0018] When a drawing image 25 is input, the drawing code table identification unit 15 instructs the generation AI system 3 to identify a drawing code table 26, which is a combination of the code and element name drawn in the drawing image 25, using a request body 55.

[0019] The request body 55 is written in JSON format. This request body 55 is executed by the generation AI system 3, for example, by calling the API of the generation AI system 3, and a chat completion object 56 is generated. The generation AI system 3 then returns the chat completion object 56 in JSON format to the drawing code table identification unit 15. This chat completion object 56 includes the drawing code table 26.

[0020] The drawing code table extracting unit 16 extracts the drawing code table 26 from the chat completion object 56 .

[0021] When the description code table 24 and the drawing code table 26 are input, the code judgment table specification unit 17 instructs the generation AI system 3 to determine any code errors from the description code table 24 and the drawing code table 26 via a request body 57.

[0022] The request body 57 is written in JSON format. This request body 57 is executed by the generation AI system 3, for example, by calling the API of the generation AI system 3, and a chat completion object 58 is generated. The generation AI system 3 then returns the chat completion object 58 in JSON format to the code determination table specification unit 17. The chat completion object 58 includes the code determination table 27.

[0023] The code determination table extraction unit 18 extracts the code determination table 27 from the chat completion object 58 .

[0024] FIG. 2 is a hardware configuration diagram showing the code inspection device 1. As shown in FIG. The code checker 1 includes a CPU (Central Processing Unit) 101, a RAM (Random Access Memory) 102, and a ROM (Read Only Memory) 103.

[0025] The CPU 101 is a processing device that executes a computer program and controls the code inspection device 1. The RAM 102 is a volatile readable and writable memory that is used as a work area for the computer program by the CPU 101. The ROM 103 is a non-volatile readable-only memory that stores, for example, a BIOS (Basic I / O System).

[0026] The code inspection device 1 further includes a display unit 104, an input unit 105, a communication unit 106, and a memory unit 107. The display unit 104 is, for example, a liquid crystal display, and displays characters, figures, images, etc. The input unit 105 is, for example, a mouse, keyboard, or touch panel, and is used by the user to input information. The communication unit 106 is, for example, a network interface card, and communicates information with, for example, the generation AI system 3.

[0027] The storage unit 107 is a large-capacity storage device such as a hard disk or a solid-state drive (SSD). A code check program 1071 is stored in the storage unit 107. When the CPU 101 executes the code check program 1071, the respective functional units in FIG. 1 are realized.

[0028] FIG. 3 shows a code judgment screen 61 displayed on the display unit 104 by the code checking device 1. As shown in FIG. The code determination screen 61 includes a drawing number combo box 611, a natural language region 613, a drawing region 614, and a code determination table region 615. The code determination screen 61 is a screen for generating a code determination table by inputting natural language, a drawing number, and a drawing.

[0029] The drawing number combo box 611 is a combo box for selecting a drawing number. The natural sentence area 613 is an area for displaying the input natural sentence. For example, the user can drag and drop a natural sentence file into the natural sentence area 613, whereby the natural sentence is registered in the natural sentence area 613. The drawing area 614 is an area where the input drawing is displayed.

[0030] The code determination table area 615 is an area that displays the explanation of the drawing written in natural language and the code determination table generated from the input drawing.

[0031] 4A and 4B are flowcharts of the code determination process. First, the explanatory text identification unit 11 accepts input of natural language and drawing number (step S10). Then, the explanatory text identification unit 11 creates a prompt that specifies the natural language and drawing number to identify the explanatory text (step S11). The explanatory text identification unit 11 sends a request body including the generated prompt to the generation AI system 3 via API (step S12), and receives a chat completion object from the generation AI system 3 (step S13).

[0032] Next, the explanation extraction unit 12 extracts an explanation from the chat completion object (step S14). The description code table specification unit 13 specifies a description and creates a prompt that specifies a code table that is a combination of the code and element name described in the description (step S15).The description code table specification unit 13 sends a request body including the generated prompt to the generation AI system 3 via the API (step S16), and receives a chat completion object from the generation AI system 3 (step S17).

[0033] Next, the explanation code table extraction unit 14 acquires the explanation code table from the chat completion object (step S18).

[0034] When the drawing code table specification unit 15 receives input of a drawing (step S19), it specifies the drawing and creates a prompt to specify a code table that is a combination of the code and element name drawn on the drawing (step S20). The drawing code table specification unit 15 sends a request body including the generated prompt to the generation AI system 3 via the API (step S21), and receives a chat completion object from the generation AI system 3 (step S22).

[0035] Next, the drawing code table extraction unit 16 acquires the drawing code table from the chat completion object (step S23).

[0036] The code determination table specification unit 17 specifies the description code table 24 and the drawing code table 26, and creates a prompt to specify a code determination table that combines these code tables (step S24). The code determination table specification unit 17 sends a request body including the generated prompt to the generation AI system 3 via the API (step S25), and receives a chat completion object from the generation AI system 3 (step S26).

[0037] Next, the code determination table extraction unit 18 acquires the code determination table from the chat completion object (step S27), and the processing of FIGS. 4A and 4B ends.

[0038] FIG. 5 shows a request body 51 including a prompt to be sent to the generation AI system 3. As shown in FIG. This request body 51 is composed of a model item and a messages item. In the model item, the ID of the generation AI model is specified after a semicolon. In the messages item, a list of messages that make up the conversation so far is stored, and various message types such as text, image, and audio can be set. In the messages item, the role item and content item are written in curly brackets after a semicolon.

[0039] In the role field, enter system or user. In the contents field, enter text in the type field and enter the prompt 31 in Figure 6 in the text field.

[0040] FIG. 6 is a diagram for explaining the contents of the prompt 31. As shown in FIG. Prompt 31 contains the instruction text "Please output the description of {drawing number} from the natural language description of the invention" and "{drawing number} = Figure 2." Prompt 31 then contains the target natural language description after "#natural language." Prompt 31 causes the generation AI system 3 to output the description of Figure 2 written in the description of the invention. Note that the instruction in the prompt "from the description of the invention" may be modified depending on the structure of the natural language description.

[0041] FIG. 7 is a diagram showing the chat completion object 52 received from the generation AI system 3. The chat completion object 52 includes an id item, an object item, a created item, a model item, a choices item, and a usage item.

[0042] The id field is an identifier that uniquely identifies the chat completion object 52. The object field indicates the object type, which is always chat.completion. The created field is the Unix timestamp in seconds when the chat completion was created. The model field is the model used for the chat completion. The choices field is a list of options for the chat completion, which may be multiple. The usage field is usage statistics for the completion request.

[0043] The choices item is composed of an index item, a role item and a content item contained in a message item, and a finish_reason item. The index field is the number of the option for completing the chat. The role field included in the message field stores assistant. The content field included in the message field stores the answer 32 generated by the AI ​​generation system.

[0044] FIG. 8 is a diagram illustrating the content of the answer 32 included in the chat completion object 52. As shown in FIG. This reply 32 contains a description of the drawing designated by the drawing number. This description is shown below. FIG. 2 is a diagram showing the configuration of the terminal 4. The terminal 4 includes a communication unit 41, an input / output unit 42, a control unit 43, and a storage unit 44. The communication unit 41 is... The input / output unit 42 is... The control unit 43 is... The storage unit 44 is..."

[0045] FIG. 9 shows a request body 53 including a prompt to be sent to the generation AI system 3. As shown in FIG. This request body 53 is composed of a model item and a messages item. In the Model item, the ID of the generation AI model is specified after a semicolon. In the Messages item, a list of messages that make up the conversation so far is stored, and various message types such as text, image, and audio can be set. In the Messages item, the role item and content item are written in curly brackets after a semicolon.

[0046] In the role field, enter system or user. In the contents field, enter text in the type field and enter the prompt 33 in Fig. 10 in the text field.

[0047] FIG. 10 is a diagram for explaining the contents of the prompt 33. As shown in FIG. Prompt 33 has the following instruction text on the first line: "Extract the codes and element names from the explanatory text and output a code table for the explanatory text. Codes are written in numbers and letters. The element name is the noun phrase immediately preceding the code. For example, for Equipment 100, "100" is the code and "Equipment" is the element name. For Step S200, "S200" is the code and "Step" is the element name. If different element names correspond to the same code, please mark this as an error." The next line then contains "#explanation", which indicates the explanatory text. The following lines onwards contain the explanatory text. This allows the generative AI system 3 to extract codes and element names from the explanatory text and output them in table format.

[0048] Prompt 33, "Codes are written in numbers and letters," teaches the rules for codes to the generative AI system 3. "The element name is the noun phrase immediately before the code," teaches the rules for element names to the generative AI system 3.

[0049] "For example, if it is device 100, then "100" is the code and "device" is the element name. If it is step S200, then "S200" is the code and "step" is the element name." teaches the generation AI system 3 the code and element name using a concrete example. "If different element names correspond to the same code, please mark it as an error." teaches the generation AI system 3 about typographical errors in codes.

[0050] FIG. 11 is a diagram showing the chat completion object 54 received from the generation AI system 3. The chat completion object 54 includes an id item, an object item, a created item, a model item, a choices item, and a usage item.

[0051] The id field is an identifier that uniquely identifies the chat completion object 54. The object field indicates the object type, which is always chat.completion. The created field is the Unix timestamp in seconds when the chat completion was created. The model field is the model used for the chat completion. The choices field is a list of options for the chat completion, which may be multiple. The usage field is usage statistics for the completion request.

[0052] The choices item is composed of an index item, a role item and a content item contained in a message item, and a finish_reason item. The index field is the number of the option for completing the chat. The role field included in the message field stores assistant. The content field included in the message field stores the answer 34 generated by the AI ​​generation system.

[0053] FIG. 12A is a diagram illustrating the content of a reply included in the chat completion object 54. As shown in FIG. This answer 34 contains a description code table 24 written in Markdown notation. This description code table 24 is a code table that shows the combinations of the codes and element names described in the description 23. The text of the description code table 24 is transcribed below. | Sign | Element name | Notes | | ---- | ---- | ---- | | 4 | Device | Consistent | | 41 | Communications Department | Consistent | | 42 | I / O section | Consistent | | 43 | Control | Consistent | | 44 | Memory | Consistent |

[0054] FIG. 12B is a diagram showing the answered explanation code table 74. The explanatory text code table 74 in FIG. 12B is, for example, a Markdown-written text included in an answer 34, displayed in a Markdown viewer. The explanatory text code table 74 includes a code column, an element name column, and a remarks column. The code column stores the name of the code. The element name column stores the element name. The remarks column stores whether the use of the code is consistent.

[0055] FIG. 13 shows a drawing image 25 to be attached to the prompt. This drawing image 25 is an image of a drawing corresponding to the explanatory text written in natural language in Figure 2. This drawing image 25 is an image that explains the logical block diagram of the terminal 4.

[0056] FIG. 14 is a diagram showing a request body 55 including a prompt to be sent to the generation AI system 3. This request body 55 is composed of a model item and a messages item. In the model item, the ID of the generation AI model is specified after a semicolon. In the messages item, a list of messages that make up the conversation so far is stored, and various message types such as text, image, and audio can be set. In the messages item, the role item and content item are written in curly brackets after a semicolon.

[0057] In the role item, enter system or user. Then, in the contents item, enter text in the type item and enter prompt 35 in Figure 15 in the text item. Then, in the contents item, enter image_url in the type item and enter image data in BASE64 format or an image URL (Uniform Resource Locator) in the image_url item.

[0058] FIG. 15 is a diagram for explaining the contents of the prompt 35. As shown in FIG. Prompt 35 has the following instruction text on the first line: "Extract the symbols and element names in Japanese from the attached drawing and output a symbol table for the drawing. Symbols are numbers or letters indicated by leader lines. When each step in the flowchart or sequence diagram is assigned a symbol beginning with S, use "Step" as the element name. If there is no element name, use "No element name." If different element names correspond to the same symbol, use "Error." This causes Generative AI System 3 to extract symbols and element names from the attached drawing image and output them in table format.

[0059] "Symbols are numbers or letters indicated by lines." This describes the rules for drawings drawn on drawings. "When each step in a flowchart or sequence diagram is assigned a symbol beginning with S, the element name should be "Step." " describes the rules for symbols in flowcharts and sequence diagrams.

[0060] Patent drawings are not limited to those with the top facing upward; there are also landscape drawings where the left side faces upward. The generation AI can extract symbols and element names from landscape drawings without any special instructions. However, you can explicitly prompt the generation AI to say, "If the Japanese OCR (Optical Character Reader) processing of the attached drawing fails, rotate the drawing 90 degrees clockwise and perform Japanese OCR processing again."

[0061] "If there is no element name, please state 'No element name'" is used when there is no element name, such as a symbol on a mechanical drawing. "If different element names correspond to the same symbol, please state this as an error" is used when referring to an error in a symbol on a drawing.

[0062] FIG. 16 is a diagram showing the chat completion object 56 received from the generation AI system 3. The chat completion object 56 includes an id item, an object item, a created item, a model item, a choices item, and a usage item.

[0063] The id field is an identifier that uniquely identifies the chat completion object 56. The object field indicates the object type, which is always chat.completion. The created field is the Unix timestamp in seconds when the chat completion was created. The model field is the model used for the chat completion. The choices field is a list of options for the chat completion, which may be multiple. The usage field is usage statistics for the completion request.

[0064] The choices item is composed of an index item, a role item and a content item contained in a message item, and a finish_reason item. The index item is the number of the option for completing the chat. The role item included in the message item stores assistant. The content item included in the message item stores the answer 36 generated by the generation AI system 3.

[0065] FIG. 17A is a diagram illustrating the content of a reply included in the chat completion object 56. As shown in FIG. This answer 36 contains a drawing code table 26 written in Markdown notation. This drawing code table 26 is a code table that shows the combinations of codes and element names drawn in the drawing image 25. The text of the drawing code table 26 is transcribed below. | Sign | Element name | Notes | | ---- | ---- | ---- | | 4 | Device | Consistent | | 41 | Communications Department | Consistent | | 42 | I / O section | Consistent | | 43 | Control | Consistent | | 44 | Memory | Consistent |

[0066] FIG. 17B is a diagram showing the answered explanation code table 76. The explanatory text code table 76 in Figure 17B is, for example, a Markdown-style text included in answer 36, displayed in a Markdown viewer. This explanatory text code table 76 includes a code column, an element name column, and a remarks column. The code column stores the name of the code. The element name column stores the element name. The remarks column stores whether the use of this code is consistent.

[0067] FIG. 18 shows a request body 57 including a prompt to be sent to the generation AI system 3. As shown in FIG. This request body 57 is composed of a model item and a messages item. In the model item, the ID of the generation AI model is specified after a semicolon. In the messages item, a list of messages that make up the conversation so far is stored, and various message types such as text, image, and audio can be set. In the messages item, the role item and content item are written in curly brackets after a semicolon.

[0068] In the role field, enter system or user. In the contents field, enter text in the type field and enter the prompt 37 in Fig. 19 in the text field.

[0069] FIG. 19 is a diagram for explaining the contents of the prompt 37. As shown in FIG. Prompt 37 has the following instruction text on the first line: "Create a code determination table that combines the description code table and the drawing code table. When different element names correspond to the same code, mark it as an error. Mark codes that appear only on the drawing as errors. Mark codes that appear only in the description as errors." Based on this, the generative AI system 3 creates a code determination table that combines the description code table and the drawing code table. When different element names correspond to the same code, it marks it as an error. Mark codes that appear only on the drawing as errors. Mark codes that appear only in the description as errors.

[0070] FIG. 20 is a diagram showing the chat completion object 58 received from the generation AI system 3. This chat completion object 58 includes an id item, an object item, a created item, a model item, a choices item, and a usage item.

[0071] The id field is an identifier that uniquely identifies the chat completion object 58. The object field indicates the object type, which is always chat.completion. The created field is the Unix timestamp in seconds when the chat completion was created. The model field is the model used for the chat completion. The choices field is a list of options for the chat completion, which may be multiple. The usage field is usage statistics for the completion request.

[0072] The choices item is composed of an index item, a role item and a content item contained in a message item, and a finish_reason item. The index field is the number of the option for completing the chat. The role field included in the message field stores assistant. The content field included in the message field stores the answer 38 generated by the AI ​​generation system.

[0073] FIG. 21A is a diagram illustrating the content of a reply included in the chat completion object 58. As shown in FIG. This answer 38 contains a symbol determination table 27 written in Markdown notation. This symbol determination table 27 is a table that determines the combination of symbols and element names described in the explanatory text 23 and the symbols and element names drawn in the drawing image 25. The text of the symbol determination table 27 is transcribed below. | Symbol | Element name in description | Element name in drawing | Judgment | | ---- | ---- | ---- | ---- | | 4 | terminal | terminal | match | | 41 | Communications Department | Communications Department | Match | | 42 | Input / output section | Input / output section | Match | | 43 | Control | Control | Match | | 44 | Storage | Storage | Match |

[0074] FIG. 21B is a diagram showing the returned code determination table. The code determination table 77 in Figure 21B is a Markdown viewer display of the Markdown text included in answer 38, for example. Specifically, when you access ChatGPT4o in a browser and display the results of entering a prompt, this code determination table 77 is displayed.

[0075] This symbol determination table 77 is composed of a symbol column, a description element name column, a drawing element name column, a remarks column, and a determination column. The symbol column stores the name of the symbol. The description element name column stores the element name extracted from the description. The drawing element name column stores the element name extracted from the drawing. The remarks column stores whether the use of this symbol is consistent. The determination column stores whether the use of this symbol is a clerical error.

[0076] FIG. 22 is a logical configuration diagram showing the code inspection device 1 according to the second embodiment. The code checking device 1 includes a description code table specification unit 131, a description code table extraction unit 141, a drawing code table specification unit 15, a drawing code table extraction unit 16, and a code judgment table creation unit 171. The code checking device 1 cooperates with the generation AI system 3 to check the correspondence between the code and element name written in the description specified by the drawing number 22 included in the natural language 21 and the code and element name written in the drawing image 25.

[0077] When the natural text 21 and drawing number 22 are input, the explanatory text code table identification unit 131 instructs the generation AI system 3 to identify the explanatory text specified by the drawing number 22 from the natural text 21 using the request body 511, and to identify the explanatory text code table 24, which is a combination of the code and element name written in this explanatory text.

[0078] The request body 511 is written in JSON format. This request body 511 is executed by the generation AI system 3, for example, by calling the API of the generation AI system 3, and a chat completion object 521 is generated. Then, the generation AI system 3 returns the chat completion object 521 in JSON format to the description code table identification unit 131. This chat completion object 521 includes the description code table 24.

[0079] The explanation code table extracting unit 14 extracts the explanation code table 24 from the chat completion object 521 .

[0080] When a drawing image 25 is input, the drawing code table identification unit 15 instructs the generation AI system 3 to identify a drawing code table 26, which is a combination of the code and element name drawn in the drawing image 25, using a request body 55.

[0081] The request body 55 is written in JSON format. This request body 55 is executed by the generation AI system 3, for example, by calling the API of the generation AI system 3, and a chat completion object 56 is generated. The generation AI system 3 then returns the chat completion object 56 in JSON format to the drawing code table identification unit 15. This chat completion object 56 includes the drawing code table 26.

[0082] The drawing code table extracting unit 16 extracts the drawing code table 26 from the chat completion object 56 .

[0083] When the description code table 24 and the drawing code table 26 are input, the code determination table creation unit 171 generates the code determination table 27 from the description code table 24 and the drawing code table 26 on a rule basis.

[0084] FIG. 23 is a flowchart of the code determination process. First, the explanatory text code table identification unit 131 accepts input of natural language and drawing number (step S30). The explanatory text code table identification unit 131 specifies the natural language and drawing number and creates a prompt that identifies a code table that is a combination of symbols and element names described in the explanatory text of the drawing specified by the drawing number in the natural language (step S31). The explanatory text code table identification unit 13 sends a request body including the generated prompt to the generation AI system 3 via the API (step S32), and receives a chat completion object from the generation AI system 3 (step S33).

[0085] Next, the description code table extraction unit 141 acquires the description code table from the chat completion object (step S34).

[0086] When the drawing code table specification unit 15 receives input of a drawing (step S35), it creates a prompt that specifies the drawing and specifies a code table that is a combination of the code and element name drawn on the drawing (step S36). The drawing code table specification unit 15 sends a request body including the generated prompt to the generation AI system 3 via the API (step S37), and receives a chat completion object from the generation AI system 3 (step S38).

[0087] Next, the drawing code table extraction unit 16 acquires the drawing code table from the chat completion object (step S39).

[0088] The code determination table creating unit 171 creates a code determination table by combining the description code table 24 and the drawing code table 26 (step S40), and the processing of FIG. 23 ends.

[0089] FIG. 24 is a diagram showing a request body 511 including a prompt to be sent to the generation AI system 3. This request body 511 is composed of a model item and a messages item. In the model item, the ID of the generation AI model is specified after a semicolon. In the messages item, a list of messages that make up the conversation so far is stored, and various message types such as text, image, and audio can be set. In the messages item, the role item and content item are written in curly brackets after a semicolon.

[0090] In the role item, enter system or user. In the contents item, enter text in the type item and enter the prompt 311 in Fig. 25 in the text item.

[0091] FIG. 25 is a diagram for explaining the contents of the prompt 311. Prompt 311 contains the following instruction text: "Identify the description of {drawing number} from the description written in natural language, extract the symbols and element names from the identified description, and output a symbol table for the description. Symbols are written in numbers and letters. The element name is the noun phrase immediately preceding the symbol. For example, for device 100, "100" is the symbol and "device" is the element name. For step S200, "S200" is the symbol and "step" is the element name. If different element names correspond to the same symbol, please report this as an error." and "{drawing number} = Figure 2." The target natural language is then written after "#natural language." This causes the generative AI system 3 to output the symbol in Figure 2.

[0092] FIG. 26 is a diagram showing the chat completion object 521 received from the generation AI system 3. This chat completion object 521 includes an id item, an object item, a created item, a model item, a choices item, and a usage item.

[0093] The id field is an identifier that uniquely identifies the chat completion object 521. The object field indicates the object type, which is always chat.completion. The created field is the Unix timestamp in seconds when the chat completion was created. The model field is the model used for the chat completion. The choices field is a list of options for the chat completion, which may be multiple. The usage field is usage statistics for the completion request.

[0094] The choices item is composed of an index item, a role item and a content item contained in a message item, and a finish_reason item. The index field is the number of the option for completing the chat. The role field included in the message field stores assistant. The content field included in the message field stores the answer 321 generated by the AI ​​generation system.

[0095] FIG. 27 is a diagram illustrating the content of the reply 321 included in the chat completion object 521. As shown in FIG. This reply 321 contains a description code table 24 that combines the symbols and element names listed in the description of the drawing specified by the drawing number. The text of this description code table 24 is transcribed below. | Sign | Element name | Notes | | ---- | ---- | ---- | | 4 | Device | Consistent | | 41 | Communications Department | Consistent | | 42 | I / O section | Consistent | | 43 | Control | Consistent | | 44 | Memory | Consistent |

[0096] In patent specifications, the overall configuration of a device is often shown in Figure 1 or Figure 2, and the operation of the device is explained in subsequent drawings. In other words, the description of drawings numbered 1 to n may be written based on the reference numerals written in the drawings numbered 1 to n. Therefore, the following flowchart explains how to check the drawings and descriptions when sequentially explaining multiple drawings in the description.

[0097] FIG. 28 is a flowchart of the code determination process that spans multiple drawings. The CPU 101 of the code checking device 1 repeats the process from step S80 to step S90 from the first drawing number to the last drawing number (step S80).

[0098] The explanatory text code table identification unit 131 creates a prompt that specifies the natural text and the drawing number, and identifies a code table that is a combination of symbols and element names described in the explanatory text of the drawing specified by the drawing number in the natural text (step S81).The explanatory text code table identification unit 13 sends a request body including the generated prompt to the generation AI system 3 via the API (step S82), and receives a chat completion object from the generation AI system 3 (step S83).

[0099] Next, the explanation code table extraction unit 141 acquires the explanation code table related to the drawing with the drawing number from the chat completion object (step S84).

[0100] The drawing code table specification unit 15 specifies the drawing associated with the drawing number and creates a prompt to specify a code table that is a combination of the code and element name drawn on the drawing (step S85). The drawing code table specification unit 15 sends a request body including the generated prompt to the generation AI system 3 via the API (step S86), and receives a chat completion object from the generation AI system 3 (step S87).

[0101] Next, the drawing code table extraction unit 16 acquires the drawing code table of the drawing number from the chat completion object (step S88).

[0102] The code determination table creation unit 171 creates a code determination table that combines the explanatory code table 24 from the first drawing number to the current drawing number and the code table from the drawing code table 26 of the first drawing number to the drawing code table 26 of the current drawing number (step S89).

[0103] Then, the CPU 101 of the code checking device 1 determines whether or not the processes from step S80 to step S90 have been repeated until the end of the drawing number (step S90). If these processes have not been repeated until the end, the process returns to step S80 and repeats. If these processes have been repeated until the end, the process of FIG. 28 ends.

[0104] This makes it possible to check for errors in the correspondence between the symbols and element names in the drawings and descriptions, even if the description of the drawing with drawing number n refers to a symbol written in a drawing with a drawing number earlier than drawing number n.

[0105] FIG. 29 is a logical configuration diagram showing the code inspection device 1 according to the third embodiment. The code checking device 1 includes a code table specification unit 19, a code table extraction unit 10, and a code judgment table creation unit 171. The code checking device 1 cooperates with the generation AI system 3 to check the correspondence between the code and element name written in the explanatory text specified by the drawing number 22 included in the natural language 21 and the code and element name drawn in the drawing image 25.

[0106] When natural language 21, drawing number 22, and drawing image 25 are input, the code table identification unit 19 instructs the generation AI system 3 to identify the explanatory text specified by drawing number 22 from the natural language 21 using a request body 59, identify the explanatory text code table 24 which is a combination of the code and element name written in this explanatory text, and identify the drawing code table 26 which is a combination of the code and element name drawn in the drawing image 25.

[0107] The request body 59 is written in JSON format. This request body 59 is executed by the generation AI system 3, for example, by calling the API of the generation AI system 3, and a chat completion object 50 is generated. The generation AI system 3 then returns the chat completion object 50 in JSON format to the code table identification unit 19. This chat completion object 50 includes the description code table 24 and the drawing code table 26.

[0108] The code table extraction unit 10 extracts the description code table 24 and the drawing code table 26 from the chat completion object 521 .

[0109] When the description code table 24 and the drawing code table 26 are input, the code determination table creation unit 171 generates the code determination table 27 from the description code table 24 and the drawing code table 26 on a rule basis.

[0110] FIG. 30 is a flowchart of the code determination process. First, the code table identification unit 19 accepts input of natural language text, a drawing number, and a drawing image (step S60). The code table identification unit 19 specifies the natural language text, the drawing number, and the drawing, and identifies a code table that is a combination of the code and element name described in the description of the drawing specified by the drawing number in the natural language, and creates a prompt that identifies a code table that is a combination of the code and element name drawn in the drawing (step S51). The code table identification unit 19 sends a request body including the generated prompt to the generation AI system 3 via the API (step S52), and receives a chat completion object from the generation AI system 3 (step S53).

[0111] Next, the code table extraction unit 10 acquires the description code table and the drawing code table from the chat completion object (step S54).

[0112] The code determination table creating unit 171 creates a code determination table by combining the description code table 24 and the drawing code table 26 (step S55), and the processing of FIG. 30 ends.

[0113] FIG. 31 shows a request body 59 including a prompt to be sent to the generation AI system 3. This request body 59 is composed of a model item and a messages item. In the model item, the ID of the generation AI model is specified after a semicolon. In the messages item, a list of messages that make up the conversation so far is stored, and various message types such as text, image, and audio can be set. In the messages item, the role item and content item are written in curly brackets after a semicolon.

[0114] In the role item, enter system or user. Then, in the contents item, enter text in the type item and enter prompt 39 in Figure 31 in the text item. Then, in the contents item, enter image_url in the type item and enter image data in BASE64 format or an image URL (Uniform Resource Locator) in the image_url item.

[0115] FIG. 32 is a diagram for explaining the contents of the prompt 39. Prompt 39 includes the following instruction text: "Identify the description of {drawing number} from the natural language description, extract the symbols and element names from the identified description, and output a symbol table for the description. Symbols are written in numbers and letters. The element name is the noun phrase immediately preceding the symbol. For example, for Device 100, the symbol is "100" and the element name is "Device." For Step S200, the symbol is "S200" and the element name is "Step." If different element names correspond to the same symbol, please report this as an error."; "Extract symbols and element names in Japanese from the attached drawing and output a symbol table for the drawing. Symbols are numbers or letters indicated by callouts. If each step in the flowchart and sequence diagram is assigned a symbol beginning with S, please report the element name as "Step." If there is no element name, please report "No element name." If different element names correspond to the same symbol, please report this as an error."; and "{drawing number} = Figure 2." Then, the target natural sentence is written after "#natural sentence." As a result, the generation AI system 3 outputs the explanatory sentence code table and the drawing code table of FIG.

[0116] FIG. 33 is a diagram showing the chat completion object 50 received from the generation AI system 3. The chat completion object 50 includes an id item, an object item, a created item, a model item, a choices item, and a usage item.

[0117] The id field is an identifier that uniquely identifies the chat completion object 50. The object field indicates the object type, which is always chat.completion. The created field is the Unix timestamp in seconds when the chat completion was created. The model field is the model used for the chat completion. The choices field is a list of options for the chat completion, which may be multiple. The usage field is usage statistics for the completion request.

[0118] The choices item is composed of an index item, a role item and a content item contained in a message item, and a finish_reason item. The index field is the number of the option for completing the chat. The role field included in the message field stores assistant. The content field included in the message field stores the answer 30 generated by the AI ​​generation system.

[0119] FIG. 34 is a diagram illustrating the content of the reply 30 included in the chat completion object 50. As shown in FIG. This reply 30 contains a description code table 24 that combines the symbols and element names written in the description of the drawing specified by the drawing number, and a drawing code table 26 that combines the symbols and element names drawn in the drawing image 25. The text of this description code table 24 and drawing code table 26 is transcribed below. The explanatory code table is below: | Sign | Element name | Notes | | ---- | ---- | ---- | | 4 | Device | Consistent | | 41 | Communications Department | Consistent | | 42 | I / O section | Consistent | | 43 | Control | Consistent | | 44 | Memory | Consistent | The drawing code table is as follows: | Sign | Element name | Notes | | ---- | ---- | ---- | | 4 | Device | Consistent | | 41 | Communications Department | Consistent | | 42 | I / O section | Consistent | | 43 | Control | Consistent | | 44 | Memory | Consistent |

[0120] FIG. 35 is a logical configuration diagram showing the code checking device 1 according to the fourth embodiment. The code checking device 1 includes a code judgment table specification unit 172 and a code judgment table extraction unit 182. The code checking device 1 cooperates with the generation AI system 3 to check the correspondence between the code and element name written in the explanatory text specified by the drawing number 22 included in the natural language 21 and the code and element name drawn in the drawing image 25.

[0121] When the natural language 21, drawing number 22, and drawing image 25 are input, the code judgment table identification unit 172 instructs the generation AI system 3 to identify the explanatory text specified by the drawing number 22 from the natural language 21 using the request body 591, identify the explanatory text code table 24 which is a combination of the code and element name written in this explanatory text, identify the drawing code table 26 which is a combination of the code and element name drawn in the drawing image 25, and further identify the code judgment table 27.

[0122] The request body 591 is written in JSON format. This request body 591 is executed by the generation AI system 3, for example, by calling the API of the generation AI system 3, and a chat completion object 501 is generated. The generation AI system 3 then returns the chat completion object 501 in JSON format to the code determination table extraction unit 182. This chat completion object 501 includes the description code table 24, the drawing code table 26, and the code determination table 27.

[0123] The code determination table extraction unit 182 extracts the description code table 24 , the drawing code table 26 , and the code determination table 27 from the chat completion object 521 .

[0124] FIG. 36 is a flowchart of the code determination process. First, the code determination table specification unit 172 accepts input of natural language text, a drawing number, and a drawing image (step S60). The code determination table specification unit 172 specifies the natural language text, the drawing number, and the drawing, and specifies a code table that is a combination of the code and element name described in the description of the drawing specified by the drawing number in the natural language, specifies a code table that is a combination of the code and element name drawn in the drawing, and creates a prompt that specifies the code determination table 27 (step S61). The code determination table specification unit 172 sends a request body including the generated prompt to the generation AI system 3 via the API (step S62), and receives a chat completion object from the generation AI system 3 (step S63).

[0125] Next, the code determination table extraction unit 182 acquires the description code table, the drawing code table, and the code determination table from the chat completion object (step S64), and the processing in FIG. 36 ends.

[0126] FIG. 37 shows a request body 591 including a prompt to be sent to the generation AI system 3. This request body 591 is composed of a model item and a messages item. In the model item, the ID of the generation AI model is specified after a semicolon. In the messages item, a list of messages that make up the conversation so far is stored, and various message types such as text, image, and audio can be set. In the messages item, the role item and content item are written in curly brackets after a semicolon.

[0127] In the role item, enter system or user. Then, in the contents item, enter text in the type item and enter prompt 391 in Figure 37 in the text item. Then, in the contents item, enter image_url in the type item and enter image data in BASE64 format or an image URL (Uniform Resource Locator) in the image_url item.

[0128] FIG. 38 is a diagram for explaining the contents of the prompt 391. Prompt 391 has the instruction text " Identify the description of {Drawing Number} from the description in natural language, extract the symbols and element names from the identified description, and output a symbol table for the description. Symbols are written in numbers and letters. Element names are the noun phrase immediately preceding the symbol. For example, for Device 100, the symbol is "100" and the element name is "Device." For Step S200, the symbol is "S200" and the element name is "Step." "Extract symbols and element names in Japanese from the attached drawings and output a symbol table for the drawings. Symbols are numbers or letters indicated by callouts. When each step in a flowchart or sequence diagram is assigned a symbol beginning with S, write "Step" as the element name. If there is no element name, write "No Element Name." If the same symbol corresponds to different element names, mark it as an error. " and "Create a code determination table that combines the two code tables. When different element names correspond to the same code, mark it as an error. Codes that appear only on the drawing should be marked as an error. Codes that appear only in the description should be marked as an error." and "{Drawing number} = Figure 2" are written. Then, the target natural sentence is written after "#natural sentence". From this, the generative AI system 3 outputs the description code table, drawing code table, and code determination table.

[0129] FIG. 39 is a diagram showing the chat completion object 501 received from the generation AI system 3. This chat completion object 501 includes an id item, an object item, a created item, a model item, a choices item, and a usage item.

[0130] The id field is an identifier that uniquely identifies the chat completion object 501. The object field indicates the object type, which is always chat.completion. The created field is the Unix timestamp in seconds when the chat completion was created. The model field is the model used for the chat completion. The choices field is a list of options for the chat completion, which may be multiple. The usage field is usage statistics for the completion request.

[0131] The choices item is composed of an index item, a role item and a content item contained in a message item, and a finish_reason item. The index field is the number of the option for completing the chat. The role field included in the message field stores assistant. The content field included in the message field stores the answer 301 generated by the AI ​​generation system.

[0132] FIG. 40 is a diagram illustrating the content of the reply 301 included in the chat completion object 501. As shown in FIG. This response 301 contains a description code table 24 that combines the codes and element names written in the description of the drawing specified by the drawing number, a drawing code table 26 that combines the codes and element names drawn in the drawing image 25, and a code determination table 27. The text of this description code table 24, drawing code table 26, and code determination table 27 is transcribed below. The explanatory code table is below: | Sign | Element name | Notes | | ---- | ---- | ---- | | 4 | Device | Consistent | | 41 | Communications Department | Consistent | | 42 | I / O section | Consistent | | 43 | Control | Consistent | | 44 | Memory | Consistent | The drawing code table is as follows: | Sign | Element name | Notes | | ---- | ---- | ---- | | 4 | Device | Consistent | | 41 | Communications Department | Consistent | | 42 | I / O section | Consistent | | 43 | Control | Consistent | | 44 | Memory | Consistent | The code determination table is as follows: | Symbol | Element name in description | Element name in drawing | Judgment | | ---- | ---- | ---- | ---- | | 4 | terminal | terminal | match | | 41 | Communications Department | Communications Department | Match | | 42 | Input / output section | Input / output section | Match | | 43 | Control | Control | Match | | 44 | Storage | Storage | Match |

[0133] FIG. 41 is a diagram for explaining the contents of the reply 321 including the explanatory text code table of drawing number 1. The explanatory code table for drawing number 1 included in response 321 lists an additive manufacturing device 1, a wire W, a power supply 19, a modeling material supply unit 16, a laser oscillator 11, and a laser beam LB. The use of these codes has been determined to be consistent.

[0134] FIG. 42 is a diagram for explaining the contents of the reply 361 including the drawing code table of drawing number 1. The drawing code table for drawing number 1 included in response 361 lists an additive manufacturing apparatus 1, a wire W, a power supply 19, a modeling material supply unit 16, a laser oscillator 11, and a laser beam LB. The use of these codes has been determined to be consistent.

[0135] FIG. 43 is a diagram showing an example of the explanatory text 232 of drawing number 2. The text of description 232 is shown below: FIG. 2 is a flowchart showing an example of an operation procedure for manufacturing a model using an additive manufacturing device. First, the layered modeling apparatus 1 energizes the wire W, which is a modeling material (step S70). The layered modeling apparatus 1 energizes the wire W by causing a current to flow from the power supply 19 in accordance with a current command. The process of step S70 corresponds to a current application process. Next, the layered manufacturing apparatus 1 supplies the wire W, which is a modeling material, to the workpiece (step S71). The modeling material supply unit 16 supplies the wire W at a supply speed of the wire W in accordance with the supply command. The process of step S71 corresponds to a modeling material supply process. Thereafter, the layered manufacturing apparatus 1 outputs the laser beam LB from the laser oscillator 11 to irradiate the workpiece with the laser beam LB (step S72), and the processing in FIG. 8 ends.

[0136] FIG. 44 is a diagram for explaining the contents of the reply 322 including the explanatory text code table of drawing number 2. The explanatory text code table included in the answer 322 lists the combinations of symbols and element names included in the explanatory text 232 as a Markdown table. This code table lists the additive manufacturing apparatus 1, the wire W, step S70, the power supply 19, step S71, the modeling material supply unit 16, step S72, the laser oscillator 11, and the laser beam LB. The use of these symbols has all been determined to be consistent.

[0137] FIG. 45 is a diagram showing an example of a drawing image 252 of drawing number 2. The drawing image 252 is a flowchart explained by the explanatory text 232. By attaching this drawing image 252 and prompting the generation AI system 3 to extract symbols and element names, a drawing symbol table, which will be described later, is generated.

[0138] FIG. 46 is a diagram for explaining the contents of the reply 362 including the drawing code table of drawing number 2. The drawing code table included in the answer 362 lists the combinations of codes and element names depicted in the drawing image 252 as a Markdown table. This code table lists steps S70, S71, and S72. The use of these codes has all been determined to be consistent.

[0139] FIG. 47 is a diagram for explaining the contents of the answer 382 including the code determination tables of drawing numbers 1 and 2. The code determination table included in reply 382 is created by combining the explanatory code table for drawing numbers 1 and 2 and the drawing code table for drawing numbers 1 and 2. This code determination table shows that step S72 is not described in the explanatory text.

[0140] The configuration and effects of the present invention will be described below. [1] A procedure (steps S15 to S18) of inputting a first prompt (prompt 33) including an instruction to generate a description code table (24) containing the correspondence between symbols and element names from a description (23) related to a drawing into a generation AI (generation AI system 3), and obtaining an answer including a predetermined description code table (24) from the generation AI (generation AI system 3); A procedure (steps S20 to S23) of inputting a second prompt (prompt 35) including an instruction to generate a drawing code table (26) containing the correspondence between symbols and element names from the drawing (drawing image 25) described in the description (23) into the generation AI (generation AI system 3), and obtaining an answer including the predetermined drawing code table (26) from the generation AI; A procedure of outputting a code determination table (27) that determines whether there are any errors in the codes from the description code table (24) and the drawing code table (26) (steps S24 to S27); A code checking program that allows a computer to perform the above.

[0141] This makes it possible to check the element names and symbols of the description and drawing from the drawing and the description of the drawing.

[0142] [2] A step of inputting a third prompt (prompt 31) including an instruction to generate a description (23) of the drawing from natural language (21) and a drawing number (22) into a generation AI (generation AI system 3), and obtaining an answer including the description (23) from the generation AI (generation AI system 3); The code checking program (1071) according to claim 1 for further executing the following.

[0143] This allows the generation AI to extract explanatory text about the drawing to be judged from the entire natural text.

[0144] [3] The first prompt (prompt 33) and / or the second prompt (prompt 35) include an instruction to determine that a typographical error has occurred if the same symbol corresponds to multiple element names. The code checking program (1071) according to claim 1.

[0145] This allows for correcting typographical errors when the same symbol corresponds to multiple element names in both the drawing and the natural language.

[0146] [4] A procedure of inputting a fourth prompt (prompt 37) including an instruction to output a code determination table (27) that determines code errors from the description code table (24) and the drawing code table (26) to the generation AI (generation AI system 3), and obtaining an answer including a predetermined code determination table (27) from the generation AI (generation AI system 3); The code checking program (1071) according to claim 1 for carrying out the above.

[0147] [5] The fourth prompt (prompt 37) includes an instruction to determine that a code that exists in the drawing code table (26) but does not exist in the description code table (24) and a code that does not exist in the drawing code table (26) but exists in the description code table (24) are typographical errors. The code checking program (1071) according to claim 4 for carrying out the above.

[0148] This allows the generation AI to check the element names and symbols contained in the drawing and its description.

[0149] [6] a step of generating the code determination table (27) in which codes that exist in the drawing code table (26) but do not exist in the description code table (24) and codes that do not exist in the drawing code table (26) but exist in the description code table (24) are determined to be errors; The code checking program (1071) according to claim 4 for carrying out the above.

[0150] This allows for rule-based inspection of element names and symbols in drawings and their descriptions.

[0151] [7] A procedure of inputting a prompt (39) including an instruction to generate a description code table (24) containing correspondence between symbols and element names from a description (23) regarding a drawing, and an instruction to generate a drawing code table (26) containing correspondence between symbols and element names from the drawing described in the description (23) into the generation AI (generation AI system 3), and obtaining an answer including the predetermined description code table (24) and the predetermined drawing code table (26) from the generation AI (generation AI system 3); a step of outputting a code determination table (27) that determines whether there are any errors in the codes from the description code table (24) and the drawing code table (26); A code check program (1071) for causing a computer to carry out the above.

[0152] This makes it possible to check the element names and symbols of the description and the drawing from the natural language including the drawing and the description of the drawing.

[0153] [8] An explanatory text code table specification unit (13) that inputs a first prompt (prompt 33) including an instruction to generate an explanatory text code table (24) that describes the correspondence between symbols and element names from explanatory text about drawings to a generation AI (generation AI system 3), and obtains an answer including a predetermined explanatory text code table (24); A drawing code table identification unit (15) inputs a second prompt (prompt 35) including an instruction to generate a drawing code table (26) that describes the correspondence between symbols and element names from the drawings described in the description to the generation AI (generation AI system 3), and obtains a response including a predetermined drawing code table (26) from the generation AI (generation AI system 3); A code determination table creation unit (171) that outputs a code determination table (27) that determines errors in codes from the description code table (24) and the drawing code table (26); A code inspection device (1) comprising:

[0154] This makes it possible to check the element names and symbols of the description and drawing from the drawing and the description of the drawing.

[0155] (Variation) The present invention is not limited to the above-described embodiment, and modifications can be made without departing from the spirit of the present invention, for example, the following (a) to (f).

[0156] (a) The explanatory text code table, drawing code table, and code determination table generated by the generation AI are not limited to Markdown text, but may be, for example, comma-delimited text or tab-delimited text. (b) The generative AI system is not limited to one that operates in a cloud environment, but may also operate in an on-premise environment. (c) The description of the drawings is not limited to that of a patent specification, but may be that of a software specification or a user manual. (d) The generating AI systems that send each prompt may be the same or different, and are not limited thereto. (e) A description code table may be created from the description text on a rule-based basis, and is not limited to this. (f) All processes other than the process of generating a drawing code table from a drawing may be rule-based, and are not limited to this. [Explanation of symbols]

[0157] 1 Code checker 11 Description specification section 12 Description extraction section 13 Explanation code table identification section 14. Description code table extraction section 15 Drawing Code Table Identification Section 16 Drawing code table extraction section 17 Sign judgment table identification part 18 Sign judgment table extraction part 3. Generative AI System 21 Natural sentences 22 Drawing number 51 Request body 52 Chat Completed Object 23 Description 53 Request body 54 Chat Completed Object 55 Request body 56 Chat Completed Object 57 Request body 58 Chat Completed Object 101 CPU 102 RAM 103 ROM 104 Display section 105 Input section 106 Communications Department 107 Storage section 1071 Code Check Program 61 Sign judgment screen 611 Drawing Number Combo Box 613 Natural sentence area 614 Drawing Area 615 Sign determination table area 31 prompts 32 answers 33 prompts 34 answers 35 prompts 36 answers 37 prompts 38 answers 24 Explanatory Code Table 25 Drawing images 26 Drawing Code Table 27 Sign determination table 131 Explanation code table identification part 141 Explanation code table extraction unit 171 Code Judgment Table Creation Unit 511 request body 521 Chat Completed Object 311 prompt 321 answers 19 Code table identification part 10 Code table extraction part 59 Request body 50 Chat Completed Objects 39 prompts 30 answers 172 Sign judgment table identification part 182 Sign judgment table extraction part 591 request body 501 Chat Completed Object 391 prompt 301 answers

Claims

1. a step of inputting a first prompt including an instruction to generate a description code table containing correspondence between symbols and element names from a description of a drawing to a generation AI, and obtaining an answer including a predetermined description code table from the generation AI; a step of inputting, to the generating AI, an instruction to generate a drawing code table describing the correspondence between symbols and element names from the drawings described in the description, and a second prompt including an instruction to indicate no element name if there is no element name, and obtaining from the generating AI an answer including a predetermined drawing code table; a step of outputting a code determination table in which errors in codes are determined from the description code table and the drawing code table; A code checking program for running on a computer.

2. The second prompt further includes an instruction to set the element name to "step" when each step in the flowchart and sequence diagram is assigned a code beginning with S. The code checking program according to claim 1 .

3. a step of inputting a third prompt including an instruction to generate a description for a drawing to which the drawing number is assigned from natural language including a description of a drawing that can be specified by the drawing number and the drawing number to the generation AI, and obtaining an answer including the description from the generation AI; a step of inputting a first prompt including an instruction to generate a description code table including a correspondence between symbols and element names from a description regarding a drawing included in the answer to a generation AI, and obtaining an answer including a predetermined description code table from the generation AI; a step of inputting, to the generating AI, an instruction to generate a drawing code table describing the correspondence between symbols and element names from the drawings described in the description, and a second prompt including an instruction to indicate no element name if there is no element name, and obtaining from the generating AI an answer including a predetermined drawing code table; a step of outputting a code determination table in which errors in codes are determined from the description code table and the drawing code table; A code checking program for running on a computer.

4. the first prompt and / or the second prompt includes an instruction to determine that a typographical error has occurred if the same symbol corresponds to a plurality of element names; The code checking program according to claim 1 .

5. a step of inputting a fourth prompt including an instruction to output a code determination table obtained by determining whether a code is incorrect from the description code table and the drawing code table to the generating AI, and obtaining an answer including a predetermined code determination table from the generating AI; 2. The code checking program according to claim 1, for executing the above.

6. The fourth prompt includes an instruction to determine, as an error, a code that exists in the drawing code table but does not exist in the description code table, and a code that does not exist in the drawing code table but exists in the description code table.

6. The code checking program according to claim 5, for executing the above.

7. a step of generating the code determination table in which a code that exists in the drawing code table but does not exist in the description code table and a code that does not exist in the drawing code table but exists in the description code table are determined to be an error; 6. The code checking program according to claim 5, for executing the above.

8. a step of inputting, to a generating AI, prompts including an instruction to generate a description code table in which the correspondence between symbols and element names is described from a description regarding a drawing, an instruction to generate a drawing code table in which the correspondence between symbols and element names is described from a drawing described in the description, and an instruction to indicate no element name if there is no element name, and obtaining from the generating AI a response including a predetermined description code table and a predetermined drawing code table; a step of outputting a code determination table in which errors in codes are determined from the description code table and the drawing code table; A code checking program for running on a computer.

9. an explanatory text code table specification unit that inputs a first prompt including an instruction to generate an explanatory text code table that describes the correspondence between symbols and element names from explanatory text related to drawings to the generation AI, and obtains an answer including a predetermined explanatory text code table; a drawing code table specifying unit that inputs to the generating AI an instruction to generate a drawing code table that describes the correspondence between symbols and element names from the drawings described in the description, and an instruction to indicate that there is no element name if there is no element name, and obtains a response that includes a predetermined drawing code table from the generating AI; a code determination table creating unit that outputs a code determination table that determines whether a code is incorrectly written from the description code table and the drawing code table; A code inspection device comprising:

Images