Computer system and production assistance method for deliverable

Abstract

This computer system retains a database which is connected to a task execution program and which stores traceability information. The traceability information is data in which task information and identification information for a deliverable are associated with identification information for input information, and the task information includes a prompt. The computer system receives identification information for changed input information, refers to the database to retrieve the traceability information including the identification information for the input information, produces a novel prompt on the basis of the task information included in the traceability information, acquires a novel deliverable by inputting the novel prompt into the task execution program, and outputs the novel deliverable to which the identification information for the deliverable included in the traceability information has been applied.

Description

Computer System and Work Product Generation Support Method Incorporation by Reference

[0001] This application claims the priority of Japanese Patent Application No. 2024-186170 filed on October 22, 2024, and incorporates its content by reference into this application.

[0002] The present invention relates to a support technology for software development using generative AI.

[0003] Software development using generative AI is known. For example, the technology described in Patent Document 1 is known.

[0004] Patent Document 1 describes that "a computer receives a summary prompt indicating an overview of a program code to be generated according to a program. The computer also inputs the received summary prompt into a natural language processing model to obtain a detailed prompt indicating detailed specification content regarding the program code. Then, the computer inputs the obtained detailed prompt into the natural language processing model to obtain a program code."

[0005] Patent No. 7564601

[0006] In recent software development, design documents and the like are frequently updated, so the source code also needs to be updated accordingly. When manually modifying the source code generated using generative AI, the workload on the user is large and difficult. When inputting a prompt generated based on the updated design document into generative AI to generate a new source code, the content of the generated source code is significantly different from the content of the source code generated from the design document before the update. That is, there is a problem that the commonality of the content of the source code accompanying the update of the design document is low.

[0007] An object of the present invention is to provide a technology for generating a work product that has commonality in content and reflects changes in input information using generative AI.

[0008] A typical example of the invention disclosed in this application is as follows: A computer system that supports the generation of deliverables in software development, comprising a processor, a storage device connected to the processor, and a network interface connected to the processor, which receives a prompt generated based on input information used to generate the deliverable as input, is communicatively connected to a task execution program that performs a task to generate the deliverable, maintains a database that stores traceability information, the traceability information is data in which identification information of the deliverable generated based on the input information and task information relating to the task performed to generate the deliverable are associated with identification information of the input information, the task information includes the prompt, the processor receives change information including identification information of the changed input information, refers to the database to search for the traceability information including identification information of the input information contained in the change information, generates a new prompt based on the task information contained in the searched traceability information, obtains a new deliverable by inputting the new prompt to the task execution program, and outputs the new deliverable with the identification information of the deliverable contained in the searched traceability information attached.

[0009] According to the present invention, it is possible to generate deliverables such as source code that have common content using generation AI. Problems, configurations, and effects other than those described above will be clarified by the following description of the embodiments.

[0010] This figure shows an example of the system configuration of Example 1. This figure shows an example of the hardware configuration of the computer constituting the software development support system of Example 1. This figure shows an example of the data structure of the traceability information DB of Example 1. This is a flowchart illustrating an example of the processing performed by the target deliverable identification unit of Example 1. This is a flowchart illustrating an example of the processing performed by the prompt generation unit of Example 1. This figure shows an example of the screen displayed by the prompt generation unit of Example 1. This is a flowchart illustrating an example of the processing performed by the task execution unit of Example 1. This is a flowchart illustrating an example of the processing performed by the traceability information generation unit of Example 1. This is a flowchart illustrating an example of the processing performed by the prompt generation unit of Example 2.

[0011] The embodiments of the present invention will be described below with reference to the drawings. However, the present invention is not to be construed as being limited to the embodiments described below. It will be readily apparent to those skilled in the art that the specific configuration can be modified without departing from the spirit or intent of the present invention.

[0012] In the configuration of the invention described below, identical or similar components or functions are denoted by the same reference numerals, and redundant descriptions are omitted.

[0013] The designations "First," "Second," "Third," etc., used in this specification are for the purpose of identifying constituent elements and do not necessarily limit their number or order.

[0014] Figure 1 shows an example of the system configuration of Example 1. Figure 2 shows an example of the hardware configuration of the computer that constitutes the software development support system 100 of Example 1.

[0015] The system consists of a software development support system 100, a version control system 101, and a version control database 102. The software development support system 100 and the version control system 101 are connected via a network (not shown). The version control system 101 and the version control database 102 are connected either directly or via the network. The network is, for example, a LAN (Local Area Network).

[0016] The version control system 101 manages the update history of deliverables in software development and the input information used to generate those deliverables.

[0017] Here, the deliverables are source code, test code, and design documents, etc. The input information consists of design documents and source code, etc.

[0018] In Example 1, when the version control system 101 receives an update to the input information, it stores the updated input information in the version control DB 102, generates change information including the changes to the input information, and transmits it to the software development support system 100. The version control system 101 registers the generated deliverables in the version control DB 102. The version control system 101 may also register the updated input information and deliverables in association with each other in the version control DB 102.

[0019] The software development support system 100 generates deliverables using generation AI. Specifically, the software development support system 100 receives change information and generates new deliverables. For example, if the input information is a design document and the deliverable is source code, source code is generated that reflects the changes in the design document to the previous source code.

[0020] Here, the generative AI is a large-scale language model (LLM) built using a large amount of text data, and is capable of performing various language processing tasks. The LLM accepts prompts containing the task content, such as questions written in natural language, understands the meaning of the task content, and generates and outputs text (outputs) that serve as the answer by executing the task.

[0021] The software development support system 100 consists of at least one computer 200. The computer 200 has a processor 201, a main memory 202, a secondary memory 203, and a network interface 204. Each hardware element is connected via a bus 205.

[0022] The processor 201 executes a program stored in the main memory 202. By executing processing according to the program, the processor 201 operates as a functional unit (module) that realizes a specific function. In the following description, when the processing is described with a functional unit as the subject, it indicates that the processor 201 is executing a program that realizes that functional unit.

[0023] The main memory 202 is a memory unit that stores the program executed by the processor 201 and the information used by the program.

[0024] The main memory 202 is also used as a work area. The secondary memory 203 is a large-capacity storage device such as an HDD (Hard Disk Drive) or SSD (Solid State Drive).

[0025] The programs and information stored in the main memory 202 may also be stored in the secondary memory 203. In this case, the processor 201 reads the programs and information from the secondary memory 203 and loads them into the main memory 202. The network interface 204 is an interface for connecting to a network.

[0026] Generally, when generating deliverables using generation AI, the commonality of the deliverable content is lost when the prompt changes. The software development support system 100 of Example 1 generates deliverables that include new content while maintaining the commonality between the content of deliverables generated based on input information before the update and the content of deliverables generated based on input information after the update.

[0027] The software development support system 100 includes a target deliverable identification unit 110, a prompt generation unit 111, a task execution unit 112, and a traceability information generation unit 113. The software development support system 100 also maintains a traceability information DB 120. Furthermore, the software development support system 100 maintains a template DB (not shown) for managing prompt templates and an LLM management DB (not shown) for managing LLMs.

[0028] The Traceability Information DB120 is a database for managing traceability information, including various information related to the generation of deliverables. The details of the data structure of the Traceability Information DB120 will be described later.

[0029] The target deliverable identification unit 110 identifies the deliverables to be updated using change information and traceability information DB 120. The target deliverable identification unit 110 generates target deliverable information for the identified deliverables and outputs it to the prompt generation unit 111 and the traceability information generation unit 113.

[0030] The prompt generation unit 111 generates prompts using the target deliverable information and the traceability information DB 120 and outputs them to the task execution unit 112. The prompt generation unit 111 also generates information about the generated prompts and the generation AI to be used and outputs it to the traceability information generation unit 113.

[0031] The task execution unit 112 executes a task to generate deliverables using generation AI and outputs the generated deliverables. The task execution unit 112 also generates generation information, including the generation of the deliverables, and outputs it to the traceability information generation unit 113. The task execution unit 112 can select the LLM to use from the LLM management DB.

[0032] The traceability information generation unit 113 generates traceability information and stores it in the traceability information DB 120.

[0033] Furthermore, regarding the functional units of the software development support system 100, multiple functional units may be combined into a single functional unit, or a single functional unit may be divided into multiple functional units according to its function.

[0034] In Example 1, the task execution unit 112 is included in the software development support system 100, but the task execution unit 112 may be located in a system other than the software development support system 100.

[0035] Figure 3 shows an example of the data structure of the traceability information DB 120 in Example 1.

[0036] The traceability information DB120 stores traceability information in a table format, for example. Each entry in the table corresponds to one piece of traceability information.

[0037] The traceability information includes date and time 301, source file 302, source revision 303, artifact 304, artifact revision 305, LLM version 306, prompt template 307, vector DB 308, modification 309, and prompt 310. However, the fields included in the traceability information are not limited to those described above. It may not include any of the fields described above, or it may include other fields.

[0038] Date and Time 301 is a field that stores the date and time when the input information was generated. Source File 302 is a field that stores the identification information of the file corresponding to the input information. Source Revision 303 is a field that stores the revision of the input information. Deliverable 304 is a field that stores the identification information of the deliverable generated based on the input information. Deliverable Revision 305 is a field that stores the revision of the deliverable.

[0039] The LLM version 306 is a field that stores the version of the LLM. The prompt template 307 is a field that stores the identification information of the prompt template. The vector DB 308 is a field that stores the external information to be referred to during task execution. The modification 309 is a field that stores the modified content of the prompt by the user. The prompt 310 is a field that stores the prompt output to the task execution unit 112.

[0040] By managing the association between input information and work products, it becomes possible to develop software that maintains the consistency of input information and work products before and after updates.

[0041] Next, the processing executed by the software development support system 100 will be described using FIGS. 4 to 8.

[0042] When the software development support system 100 receives the input of change content information, it outputs the change content information to the target work product identification unit 110 and the traceability information generation unit 113.

[0043] Hereinafter, the processing will be described by taking the case where the input information is a design document and the work product is source code as an example. In this case, the change content information includes the identification information (file name) of the design document, the generation of the design document, and the generated date and time, etc.

[0044] FIG. 4 is a flowchart for explaining an example of the processing executed by the target work product identification unit 110 of the first embodiment.

[0045] When the target work product identification unit 110 receives the input of change content information, it executes the processing described below.

[0046] The target work product identification unit 110 acquires the identification information (file name) of the changed design document from the change content information (step S101).

[0047] The target work product identification unit 110 refers to the traceability information DB 120 and searches for the traceability information corresponding to the changed design document (step S102).

[0048] Specifically, the target deliverable identification unit 110 searches the source file 302 for traceability information in which the acquired file name is set.

[0049] The target deliverable identification unit 110 determines, based on the search results, whether or not traceability information corresponding to the search result exists in the traceability information DB 120 (step S103).

[0050] If traceability information matching the searched traceability information DB 120 exists, the target deliverable identification unit 110 generates target deliverable information including a list of source code identification information stored in the deliverable 304 of the searched traceability information and design document identification information (file name), and outputs it to the prompt generation unit 111 (step S104). After that, the prompt generation unit 111 terminates processing.

[0051] Furthermore, if the source file 302 contains multiple traceability records with design document identification information, the target deliverable identification unit 110 generates the target deliverable information using the latest traceability record.

[0052] If no traceability information matching the data exists in the traceability information DB 120, the target deliverable identification unit 110 outputs a source code generation instruction to the prompt generation unit 111 (step S105). After that, the prompt generation unit 111 terminates processing.

[0053] Figure 5 is a flowchart illustrating an example of the process performed by the prompt generation unit 111 of Embodiment 1. Figure 6 is a diagram showing an example of the screen displayed by the prompt generation unit 111 of Embodiment 1.

[0054] When the prompt generation unit 111 receives input from the target deliverable identification unit 110, it executes the process described below.

[0055] The prompt generation unit 111 determines whether the received input is information about the target deliverable (step S201).

[0056] If the received input is information about the target deliverable, the prompt generation unit 111 refers to the traceability information DB 120 and searches for traceability information corresponding to the identification information of the design document included in the target deliverable (step S202).

[0057] The prompt generation unit 111 generates prompt configuration information using the retrieved traceability information (step S203).

[0058] Specifically, the prompt generation unit 111 generates prompt configuration information that includes the LLM version 306 of the retrieved traceability information, the prompt template 307, the vector DB 308, and various values ​​set in the modification 309.

[0059] The prompt generation unit 111 generates a prompt using the retrieved traceability information (step S204). After that, the prompt generation unit 111 proceeds to step S207.

[0060] Specifically, the prompt generation unit 111 retrieves a prompt template from the template DB based on the identification information of the prompt template set in the prompt template 307. The prompt generation unit 111 generates a prompt using the values ​​set in the retrieved traceability information vector DB 308 and modification 309, and the prompt template.

[0061] The generated prompt includes instructions that prioritize the changes stored in modification 309. This ensures that the output reflects the user's changes. Note that some of the prompts set in prompt 310 may be included in the new prompt.

[0062] By generating prompts based on the settings used when the task for generating the source code before the update was executed, it is possible to generate source code that includes content common to the source code before the update, as well as updated content corresponding to the update of the input information. Furthermore, by using the same LLM, the content of the source code can be reproduced more accurately.

[0063] In step S201, if the received input is a source code generation instruction, the prompt generation unit 111 generates prompt configuration information based on the default settings (step S205).

[0064] The prompt generation unit 111 generates a prompt based on the default settings (step S206). After that, the prompt generation unit 111 proceeds to step S207.

[0065] In step S207, the prompt generation unit 111 presents a prompt to the user (step S207) and waits for user input. For example, the prompt is presented via a screen 600 as shown in Figure 6. The screen 600 will now be described.

[0066] Screen 600 includes display fields 601, 602, 603, a table 604, a button 605, and a button 606.

[0067] Display field 601 is a field that displays the changes to the input information included in the change details information. Display field 602 is a field that displays the identification information of the deliverable included in the target deliverable information.

[0068] Display field 603 is the field that displays the generated prompt. The user can modify the prompt displayed in display field 603. Table 604 is the table that displays the retrieved traceability information. In Figure 6, only some fields of the traceability information are displayed.

[0069] Button 605 is an operation button for reflecting the correction of the prompt. When button 605 is pressed, a correction request including the corrected prompt is output. Button 606 is an operation button for instructing the execution of a task. When button 606 is pressed, a task execution request is output.

[0070] The prompt configuration information may also be displayed on screen 600. Furthermore, the system may be configured to accept modifications to the prompt configuration information. When a modification is received from the user, the prompt configuration information is updated.

[0071] That concludes the explanation of screen 600. Let's return to the explanation of Figure 5.

[0072] The prompt generation unit 111 determines whether the request received via the screen 600 is a request to execute a task (step S208).

[0073] If the request received via screen 600 is a modification request, the prompt generation unit 111 reflects the modification (step S209), and then waits for user input.

[0074] Specifically, the prompt generation unit 111 extracts the changes by comparing the original prompt with the modified prompt and stores them in the work area. The prompt generation unit 111 also replaces the original prompt with the modified prompt.

[0075] If the request received via screen 600 is a task execution request, the prompt generation unit 111 outputs the LLM version and prompt to the task execution unit 112 (step S210).

[0076] The prompt generation unit 111 generates task information and outputs the task information to the traceability information generation unit 113 (step S211). After that, the prompt generation unit 111 terminates its processing.

[0077] Specifically, the prompt generation unit 111 generates prompt configuration information and task information including the prompt. If the prompt has been modified, the prompt generation unit 111 includes the modification details in the task information.

[0078] Figure 7 is a flowchart illustrating an example of the process performed by the task execution unit 112 of Embodiment 1.

[0079] When the task execution unit 112 receives input for the LLM version and prompt, it executes the process described below.

[0080] The task execution unit 112 selects an LLM based on the input LLM version (step S301).

[0081] The task execution unit 112 inputs a prompt to the selected LLM and obtains the source code output by the LLM (step S302). The source code is assigned source code identification information and a generation number.

[0082] The prompt may include source code identification information, or the LLM may set the source code identification information. The prompt may also include the source code generation, or the LLM may set the source code generation.

[0083] The task execution unit 112 generates source code identification information and generation information including the generation, and outputs the generation information to the traceability information generation unit 113 (step S303). After that, the task execution unit 112 terminates processing.

[0084] Figure 8 is a flowchart illustrating an example of the process performed by the traceability information generation unit 113 in Embodiment 1.

[0085] When the traceability information generation unit 113 receives input of change details, target deliverable information, task information, and deliverable information, it executes the process described below.

[0086] The traceability information generation unit 113 generates traceability information using change information, target deliverable information, task information, and generation information (step S401).

[0087] Specifically, the traceability information generation unit 113 sets values ​​for date and time 301, source file 302, and source revision 303 based on the change information. The traceability information generation unit 113 sets a value for deliverable 304 based on the target deliverable information and sets a value for deliverable revision 305 based on the generation information. The traceability information generation unit 113 sets values ​​for LLM version 306, prompt template 307, vector DB 308, modification 309, and prompt 310 based on the task information.

[0088] The traceability information generation unit 113 stores the generated traceability information in the traceability information DB 120 (step S402). After that, the traceability information generation unit 113 terminates its processing.

[0089] As explained above, the software development support system 100 can generate deliverables that have common content and reflect changes in the input information by inputting prompts generated using traceability information into the LLM.

[0090] In Example 2, the method for generating prompts differs slightly. Below, we will describe Example 2, focusing on the differences from Example 1.

[0091] The system configuration of Example 2 is the same as that of Example 1. The hardware and software configurations of the software development support system 100 in Example 2 are the same as those of Example 1. The data structure of the traceability information DB 120 in Example 2 is the same as that of Example 1. The processing performed by the target deliverable identification unit 110, the task execution unit 112, and the traceability information generation unit 113 in Example 2 is the same as that of Example 1.

[0092] In Example 2, the processing performed by the prompt generation unit 111 is slightly different. Figure 9 is a flowchart illustrating an example of the processing performed by the prompt generation unit 111 in Example 2.

[0093] The prompt generation unit 111 generates prompt configuration information using the retrieved traceability information (step S203), and then retrieves the source code that has been modified after the source code corresponding to the traceability information was generated from the version control system 101 (step S251).

[0094] Specifically, the prompt generation unit 111 sends a query containing the values ​​of the retrieved traceability information artifact 304 and artifact revision 305 to the version control system 101. The version control system 101 searches for source code that has a generation assigned to it that is a generation later than the one included in the query. The version control system 101 sends the input information and the retrieved source code to the software development support system 100.

[0095] The prompt generation unit 111 generates a prompt using the retrieved traceability information and the modified source code (step S252).

[0096] Specifically, the prompt generation unit 111 retrieves a prompt template from the template DB based on the identification information of the prompt template set in the prompt template 307. The prompt generation unit 111 generates a prompt using the retrieved traceability information vector DB 308, the values ​​set in the modification 309, the modified source code, the input information, and the prompt template. Instructions are added to the prompt to enable few-shot learning of the relationship between the modified source code and the input information.

[0097] The other processes are the same as in Example 1.

[0098] According to Example 2, instructions can be added to the prompt to reflect modifications to a deliverable that do not involve changes to the input information into a new deliverable. This makes it possible to more accurately maintain consistency between the content of the new deliverable and the latest deliverable managed by the version control system 101.

[0099] It should be noted that the present invention is not limited to the embodiments described above, and various modifications are included. Furthermore, for example, the embodiments described above are detailed explanations of the configuration in order to clearly illustrate the present invention, and are not necessarily limited to those having all the configurations described. In addition, some of the configurations in each embodiment can be added to, deleted from, or replaced with other configurations.

[0100] Furthermore, each of the above-mentioned configurations, functions, processing units, processing means, etc., may be implemented in hardware, either partially or entirely, by designing them as integrated circuits, for example. The present invention can also be implemented by software program code that realizes the functions of the embodiment. In this case, a storage medium on which the program code is recorded is provided to a computer, and the processor of that computer reads the program code stored in the storage medium. In this case, the program code read from the storage medium itself realizes the functions of the embodiment described above, and the program code itself and the storage medium on which it is stored constitute the present invention. Examples of storage media used to supply such program code include flexible disks, CD-ROMs, DVD-ROMs, hard disks, SSDs (Solid State Drives), optical disks, magneto-optical disks, CD-Rs, magnetic tapes, non-volatile memory cards, ROMs, and the like.

[0101] Furthermore, the program code that implements the functions described in this embodiment can be implemented in a wide range of programming or scripting languages, such as assembler, C / C++, Perl, Shell, PHP, Python, and Java.

[0102] Furthermore, the program code for the software that implements the functions of the embodiment may be distributed via a network and stored in a storage means such as a computer's hard disk or memory, or in a storage medium such as a CD-RW or CD-R, and the computer's processor may read and execute the program code stored in the storage means or storage medium.

[0103] In the above-described embodiment, the control lines and information lines shown are those deemed necessary for explanation and do not necessarily represent all control lines and information lines in the actual product. All components may be interconnected.

Claims

1. A computer system for supporting the generation of deliverables in software development, comprising: a processor, a storage device connected to the processor, and a network interface connected to the processor; receiving prompts generated based on input information used to generate the deliverables as input; communicating with a task execution program that performs a task to generate the deliverables; maintaining a database for storing traceability information; the traceability information being data in which identification information of the deliverables generated based on the input information and task information relating to the task performed to generate the deliverables are associated with identification information of the input information; the task information including the prompts; the processor receiving change information including identification information of changed input information; searching the database for traceability information including identification information of the input information included in the change information; generating a new prompt based on the task information included in the retrieved traceability information; obtaining a new deliverable by inputting the new prompt to the task execution program; and outputting the new deliverable with the identification information of the deliverables included in the retrieved traceability information added.

2. A computer system according to claim 1, wherein the task execution program executes the task using a large-scale language model, the task information includes the type of the large-scale language model and the prompt input to the large-scale language model, the processor inputs the new prompt to the task execution program along with the type of the large-scale language model included in the retrieved traceability information, generates the task information including the type of the large-scale language model included in the retrieved traceability information and the new prompt, generates the traceability information by associating the identification information of the input information included in the change information, the identification information of the deliverable included in the retrieved traceability information, and the task information, and stores the generated traceability information in the database.

3. A computer system according to claim 2, wherein the processor presents an interface for displaying the new prompt to the user, includes the modification in the task information when the user requests modification of the new prompt, and generates the prompt including information about the modification when the retrieved traceability information includes the modification.

4. A computer system according to claim 2, wherein the system is accessiblely connected to a version control system that manages the change history of the input information and the deliverables, the traceability information includes the generation of the deliverables, the change information is generated by the version control system that receives the changed input information, and the processor accesses the version control system to obtain the changed input information and the deliverables of a newer generation than the generation of the deliverables included in the retrieved traceability information, and generates a new prompt that includes a learning instruction for the relationship between the changed input information and the deliverables obtained from the version control system.

5. A method for supporting the generation of deliverables in software development executed by a computer system, wherein the computer system comprises a processor, a storage device connected to the processor, and a network interface connected to the processor, and is connected in a communicative manner to a task execution program that performs a task to generate the deliverable, and maintains a database for storing traceability information, wherein the traceability information is data in which identification information of the deliverable generated based on the input information and task information relating to the task performed to generate the deliverable are associated with identification information of the input information, the task information includes the prompt, and the method for supporting the generation of the deliverable comprises: the step of the processor receiving change information including identification information of the changed input information; the step of the processor looking up the traceability information including identification information of the input information contained in the change information by referring to the database; the step of the processor generating a new prompt based on the task information contained in the retrieved traceability information; and the step of the processor obtaining a new deliverable by inputting the new prompt to the task execution program. A method for supporting the generation of an artifact, characterized by comprising the step of the processor outputting a new artifact to which identification information of the artifact included in the retrieved traceability information has been added.

Images