Software requirement specification document auditing method and device and related equipment

Abstract

The invention provides a software requirement specification document auditing method and device and related equipment, and the method comprises the steps: carrying out the entry cutting of an inputted software requirement document based on a predefined requirement entry rule, and obtaining a plurality of sub-requirement entries containing a single requirement entry; determining target auditing text content based on the sub-demand items; multiple auditing standards of multiple dimensions for the target auditing text content are determined, and the relative weight of each auditing standard is determined; and according to the multiple auditing standards and the relative weight, performing multi-dimensional evaluation on the target auditing text content by utilizing a large language model to obtain a multi-dimensional evaluation result of the target auditing text content about each auditing standard. By means of the mode, the problems that manual auditing is time-consuming and depends on subjective initiative of auditing personnel are solved, refined multi-standard auditing on the requirement content is achieved, the auditing result is more accurate and reliable, and the auditing requirements of users for various complex scenes or services can be met friendly.

Description

Technical Field

[0001] This application relates to the field of document processing technology, and in particular to a method, apparatus, and related equipment for reviewing software requirement specification documents. Background Art

[0002] With the advancement of automotive intelligence, the types and volume of in-vehicle software are increasing. Software requirements specifications (SRSs) are crucial for in-vehicle software development projects. SRSs detail the software system's behavior, functionality, performance, interfaces, and other non-functional requirements, forming the foundation for software development and testing. Unclear, ambiguous, or conflicting requirements in SRSs can lead to frequent requirement changes during development. Conflicting designs can also cause incompatibilities between system functions, impacting overall software performance and stability. This ultimately leads to project delays and increased costs, while also potentially causing the product to fail to meet user expectations and negatively impact the user experience. Therefore, the review of SRSs is crucial throughout the automotive R&D lifecycle. Traditional review methods include manual inspections, automated checks based on quality process checklists, or checks based on crown tests and pre-set rules. These traditional review methods are either costly or unable to cover complex project needs or scenario requirements. In order to overcome the defects of these traditional methods, the industry has introduced large language models for auditing. However, when the industry currently introduces large language models for auditing, most of them directly use large language models to conduct a one-time audit of document content according to preset multiple standards. This leads to a decrease in the information density of the model's judgment standards, making the detection output basis unclear and the audit accuracy insufficient to meet the project development needs. Summary of the Invention

[0003] In view of this, embodiments of the present application provide a method, device, storage medium, and electronic device for reviewing a software requirement specification document to at least or partially solve the above-mentioned problems.

[0004] In a first aspect, an embodiment of the present application provides a method for reviewing a software requirements specification document, comprising:

[0005] Based on predefined requirement item rules, the entered software requirement document is divided into items to obtain multiple sub-requirement items containing a single requirement item;

[0006] Based on the sub-requirement items, determine the target review text content;

[0007] Determine multiple review criteria for multiple dimensions of text content review for the stated objectives, and determine the relative weights of each review criterion;

[0008] According to the multiple audit standards and the relative weights, a large language model is used to perform a multi-dimensional evaluation of the target audit text content to obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard.

[0009] Optionally, in an embodiment of the present application, determining the target review text content based on the sub-requirement item includes:

[0010] Determining whether a functional timing diagram exists in each of the sub-requirement items;

[0011] If there is no functional sequence diagram in the sub-requirement item, the requirement item number and requirement text content corresponding to the requirement item will be used as the target review text content;

[0012] If there is a functional timing diagram in the sub-requirement item, the functional timing diagram is converted into supplementary text content that can characterize the interaction relationship between the software business object and different software business objects, as a supplement to the item number and requirement text of the sub-requirement item, to obtain the target review text content.

[0013] Optionally, in an embodiment of the present application, determining multiple review criteria for multiple dimensions of the target review text content and determining the relative weight of each review criterion includes:

[0014] Determining, based on the determined expert experience, information on importance parameters between two different audit criteria in the multiple audit criteria of the multiple dimensions, wherein the multiple audit criteria in the multiple dimensions include at least one audit criterion of completeness, grammatical accuracy, granularity, clarity, consistency, or redundancy;

[0015] Constructing a weight judgment matrix according to the importance parameter information;

[0016] Calculate the weight value of each audit criterion according to the weight judgment matrix;

[0017] A consistency check is performed on the weight value of each audit standard, and the relative weight is determined based on the consistency check result.

[0018] Optionally, in an embodiment of the present application, the consistency check of the weight value of each audit criterion includes:

[0019] Calculate the consistency ratio value of the constructed weight judgment matrix, recorded as CR value;

[0020] CR=CI / RI,

[0021] Wherein the CI value is the consistency index, CI = (λmax-n) / (n-1), n is the matrix order of the weight judgment matrix, and RI is the consistency index of the average random matrix corresponding to the matrix order;

[0022] If the CR value is less than the preset threshold value, the weight judgment matrix is considered to be consistent.

[0023] Optionally, in an embodiment of the present application, determining multiple review standards corresponding to the target review text content includes:

[0024] Obtain user's requirement specification review requirements;

[0025] The requirement specification audit requirements are used to generate a standard inspection prompt word Prompt instruction corresponding to a single audit standard, so as to use the Prompt instruction to form multiple audit standards corresponding to the target audit text content.

[0026] Optionally, in one embodiment of the present application, the target review text content is evaluated in multiple dimensions using a large language model based on the multiple review criteria and the relative weights, to obtain a multi-dimensional evaluation result of the target review text content with respect to each review criterion, including:

[0027] Based on the relative weights and the standard prompt word Prompt instruction corresponding to a single audit standard, the target audit text is evaluated in multiple dimensions using the large language LLM model to obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard.

[0028] Optionally, in one embodiment of the present application, after obtaining the multi-dimensional evaluation results of the target review text content with respect to each review standard, the method further includes:

[0029] Based on the evaluation results of each dimension, generate an overall evaluation of the software requirement document, the overall evaluation including an evaluation score and a summary of modification suggestions;

[0030] Generate a revised requirement specification document summarizing the modification suggestions using the LLM large language model;

[0031] The evaluation score, the modification suggestion summary and the modified requirement specification document are modified and fed back.

[0032] Optionally, in an embodiment of the present application, the method further includes:

[0033] If a user responds with "Agree" to the modification, the modified requirement specification is used to update the entered software requirement document on my computer;

[0034] If a user's response to the modification feedback is "disagree", the software requirement document will be reviewed again.

[0035] In a second aspect, based on the method for reviewing a software requirements specification document described in the first aspect of the present application, an embodiment of the present application further provides an apparatus for reviewing a software requirements specification document, comprising:

[0036] The segmentation module is used to segment the input software requirement document into items based on predefined requirement item rules to obtain multiple sub-requirement items containing a single requirement item;

[0037] An extraction module, configured to determine target audit text content based on the sub-requirement items;

[0038] an analysis module for determining multiple review criteria for multiple dimensions of the target review text content and determining relative weights of the respective review criteria, wherein the multiple dimensions include at least one review criterion of completeness, grammatical accuracy, granularity, clarity, consistency, or redundancy;

[0039] The audit module is used to use a large language model to perform a multi-dimensional evaluation of the target audit text content based on the multiple audit standards and the relative weights, and obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard.

[0040] In a third aspect, an embodiment of the present application further provides a computer storage medium having computer executable instructions stored thereon, which, when executed, can execute any one of the software requirement specification document review methods described in the first aspect of the embodiment of the present application.

[0041] In a fourth aspect, an embodiment of the present application further provides an electronic device, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus;

[0042] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute any one of the software requirement specification document review methods described in the first aspect of the embodiment of the present application.

[0043] The present application provides a method, apparatus and related equipment for reviewing software requirement specification documents. Based on predefined requirement item rules, the entered software requirement document is cut into items to obtain multiple sub-requirement items containing a single requirement item; based on the sub-requirement items, the target review text content is determined; multiple review standards for multiple dimensions of the target review text content are determined, and the relative weights of each review standard are determined; based on the multiple review standards and the relative weights, the target review text content is evaluated in multiple dimensions using a large language model to obtain a multi-dimensional evaluation result of the target review text content with respect to each review standard. Through the above method, the problem that manual review is time-consuming and depends on the subjective initiative of the reviewer is solved, and a refined multi-standard review of the requirement content is achieved. The review result is more accurate and reliable, and can meet the user's review needs for various complex scenarios or businesses in a friendly manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0045] Figure 1 A schematic diagram of a workflow of a method for reviewing a software requirements specification document provided in an embodiment of the present application;

[0046] Figure 2 A schematic diagram of the structure of a software requirement specification document review device provided in an embodiment of the present application.

[0047] Figure 3 A structural diagram of an electronic device is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0048] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field should fall within the scope of protection of the embodiments of the present application.

[0049] It should be understood that the various steps described in the method embodiments of the present application can be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.

[0050] Example 1

[0051] This application embodiment provides a method for reviewing software requirement specification documents, such as Figure 1 As shown, Figure 1 A workflow diagram of a method for reviewing a software requirements specification document provided in an embodiment of the present application includes:

[0052] Step S101: Based on the predefined requirement entry rules, the entered software requirement document is itemized and cut to obtain multiple sub-requirement items containing a single requirement item. In an embodiment of the present application, the predefined requirement entry rules can be determined by the user based on the software development project. In an embodiment of the present application, no matter which cutting method is used, it is necessary to ensure that the sub-requirement items formed after cutting meet the single requirement, so as to ensure that each requirement is independent, clear, testable and feasible to the greatest extent, and avoid the emergence of ambiguous requirements, untestable requirements or requirements that exceed the current technical level or resource limitations, which affect the development of the software project.

[0053] Specifically, in an optional implementation method of the embodiment of the present application, the predefined requirement entry rules can be one or more of cutting according to the functional modules of the software to be developed, cutting according to the software requirement type, cutting according to the user role, cutting according to the requirement priority, cutting according to the business process, cutting according to the data entity, and cutting according to the interaction scenario. Among them, the functional module refers to the functional module contained in the system implemented by the software to be developed, such as the user management module, the payment module and the order management module. The way of cutting according to the functional module is more in line with the modularization idea of most software development. The multiple sub-requirement items formed after cutting can accurately and clearly reflect the requirements of the software development project, which is convenient for auditors to focus on the integrity of various functional modules. The types of requirements include functional requirements (system language switching, user identity settings, etc.) and non-functional requirements (security, performance or availability, etc.). The sub-requirement items formed after cutting according to the software requirement type can facilitate the classification and management of different types of requirements through auditing, so that the audit process pays more attention to the realization of software functions and quality constraints. Splitting by user role means splitting each role into a set of requirement items, such as car owners or other users. This splitting method starts from the user's perspective, ensuring that the needs of different user roles are described completely and clearly, so as to facilitate alignment and evaluation based on the formed sub-requirement items with the user experience perspective. Splitting by priority means splitting according to the priority of the needs, usually divided into high priority, medium priority and low priority. The sub-requirement items split in this way can pay more attention to the integrity and rationality of high-priority needs during the review, ensure the realization of core functions, and facilitate the reasonable arrangement of resources in software development. Splitting by business process means splitting the requirement specification document into pairs of sub-requirement items in the order of the software business process to be developed. The sub-requirement items formed by this splitting are more in line with the business logic of the software, which is convenient for review and evaluation from the perspective of the overall business process to ensure the integrity and consistency of the project. Splitting by data entity means splitting according to the data entity information involved in the software to be developed. Each data entity corresponds to a set of requirement items. For example, for a user management system, the requirements can be split according to data entities such as "user information", "order information" or "contract information". The requirements under "user information" include "the system should support entering user names" and "the system should support entering user contact information". This segmentation method helps to ensure the completeness and accuracy of the requirements from the perspective of data management, and facilitates the development of the system's database design and data operation functions after review. Splitting by interaction scenario means splitting according to the interaction scenario between the user and the system. Each interaction scenario corresponds to one or more sub-requirement items.For example, in interactive scenarios such as "a user opens the app for the first time," "a user performs a search operation," and "a user performs a purchase operation," this segmentation approach takes the user interaction perspective and ensures that the audited requirements content covers all possible user operation scenarios. The various predefined segmentation rules provided in this embodiment of the application can meet the user's audit requirements in different scenarios, thereby improving the efficiency and quality of the audit of requirement specification documents.

[0054] Step S102, based on the sub-requirement items, determine the target review text content. In the embodiment of the present application, the target review text content generally refers to content including functional descriptions, business rules, data input and output, data processing, data storage, interface elements, interactive processes, dependencies or interactive relationships. At this stage, the embodiment of the present application converts individual requirement items into target review text content containing the corresponding contents, so that a large language model can be used to conduct a comprehensive review of them in the future without relying on manual work, while ensuring the accuracy of the review, improving the review efficiency and reducing the review cost. Specifically, in the embodiment of the present application, the target text content exists in the form of the requirement number and specific requirement content corresponding to the sub-requirement item. The embodiment of the present application generates target review content in this form here, so that the subsequent content review process can be carried out in an itemized manner according to the requirement number, so as to ensure that the review process has better pertinence and improve the reliability of the review results.

[0055] Step S103, determine multiple audit standards for multiple dimensions of the target audit text content, and determine the relative weight of each audit standard. The relative weight described in this stage of the embodiment of the present application refers to a weight parameter determined based on the relative calculation between different audit standards determined for the document to be audited, rather than an absolute weight or a fixed weight. In the embodiment of the present application, the relative weight can better reflect the relative importance of different standards among the multiple audit standards, so that the prevention described in the embodiment of the present application can flexibly adapt to changes in software project requirements. Once the absolute weight is set, it is difficult to adjust it according to the actual situation of the software project requirements. In addition, the relative weight can also more objectively reflect the priority differences between different requirements and conform to the actual decision-making process, reducing the deviation caused by subjective absolutism and facilitating quantification and comparison. At the same time, since the relative weight supports dynamic adjustment in most cases, for example, it can be flexibly adjusted according to the progress of software project development and the actual situation, the relative weight can be adjusted to reflect this change without redefining the entire demand standard system, thereby better ensuring that the project progress completed in the software development project that has not yet been dynamically promoted can be promoted in an orderly manner.

[0056] Step S104: Based on the multiple audit standards and the relative weights, a large language model is used to conduct a multi-dimensional evaluation of the target audit text content, and a multi-dimensional evaluation result of the target audit text content with respect to each audit standard is obtained. In this stage of the embodiment of the present application, a large language model is used in combination with a multi-dimensional audit of the target audit text content based on multiple audit standards. The concept of relative weight is introduced in the process. While ensuring the flexibility of the audit, the accuracy of subsequent audits is improved in order to avoid the possibility of a decrease in the information density of the audit standards due to the direct use of multiple standards for a one-time audit. This is because this audit method utilizes a large language model under the guidance of multiple audit standards, which can better meet the actual needs of the audit task and pay differentiated attention to different demand standards, thereby achieving a more accurate evaluation of the target audit text content. Avoid the neglect of needs due to the one-sidedness of a single standard, so as to improve the learnability and objectivity of the audit.

[0057] Optionally, in one embodiment of the present application, the target audit text content is determined based on the sub-requirement items, including: judging whether there is a functional timing diagram in each of the sub-requirement items; if there is no functional timing diagram in the sub-requirement item, the requirement item number and requirement text content corresponding to the requirement item are used as the target audit text content; if there is a functional timing diagram in the sub-requirement item, the functional timing diagram is converted into supplementary text content that can characterize the interaction relationship between the software business entity object and different software business entity objects, as a supplement to the item number and requirement text content of the sub-requirement item, so as to obtain the target audit text content. In an embodiment of the present application, a functional timing diagram is a graphical data that describes the interaction process and time sequence relationship of system functions, and is used to show the interaction process between different objects (such as users, modules, components, etc.) in the system that can be implemented by the software to be developed by the system, as well as the time sequence of these interactions. It focuses on describing the dynamic behavior of system functions during operation, helping developers understand the execution process and time dependencies of system functions. It can clearly show the execution order and steps of system functions that can be achieved by the software to be developed, helping developers understand the implementation logic of functions. In the requirements analysis stage, the functional sequence diagram can help verify the integrity and rationality of the requirements; in the design stage, it can guide the system architecture design and module division. At the same time, the functional flow is presented in a graphical way, so that different roles such as developers, testers, and business personnel can understand the system functions more intuitively and reduce communication costs. At this stage, the embodiment of the present application converts the functional sequence diagram into text content to facilitate the subsequent large language model to identify and review, so as to ensure that the audit results have better accuracy.

[0058] Optionally, in one implementation of the embodiment of the present application, if there is a functional timing diagram in the sub-requirement item, the functional timing diagram is converted into supplementary text content that can characterize the interaction relationship between the software business object and different software business objects, as a supplement to the item number of the sub-requirement item and the content of the requirement text, so as to obtain the target review text content, including: using the diagram recognition capability of the multimodal large model to extract all entity objects contained in the functional timing diagram, and listing the interaction content between each entity object one by one according to the order of data transmission in the timing diagram; if the timing diagram contains conditional branches (such as selection or loop), the branch conditions and the processing content of different branches are described; the functional timing diagram is converted into text content, and used as the corresponding sub-requirement item number and the corresponding requirement text content of the index to improve the integrity of the text content of the obtained sub-requirement item, and the process is efficient, reliable and easy to implement.

[0059] Optionally, in one embodiment of the present application, the determination of multiple audit standards for multiple dimensions of the target audit text content and the determination of the relative weights of each audit standard include: determining the importance parameter information between two different audit standards in the multiple audit standards of the multiple dimensions based on the determined expert experience, wherein the multiple audit standards of the multiple dimensions include at least one audit standard of completeness, grammatical accuracy, granularity, clarity, consistency or redundancy; constructing a weight judgment matrix based on the importance parameter information; calculating the weight value of each audit standard based on the weight judgment matrix; performing a consistency check on the weight value of each audit standard, and determining the relative weight based on the consistency test result. At this stage, the embodiment of the present application judges the importance parameter information between two different standards to construct a weight judgment matrix for relative weighting to construct the relative weight of each audit standard, combining qualitative judgment and quantitative calculation, and can quantify the determined expert experience and knowledge, avoiding the ambiguity of pure qualitative analysis and making up for the deficiency caused by the lack of actual background information in pure quantitative analysis. In addition, constructing a weight judgment matrix can organize the complex audit standard system, thereby providing a clear judgment basis for subsequent audit work. Furthermore, the present embodiment uses a weight judgment matrix to calculate the weights of each audit criterion, combined with a consistency check, to reduce subjective arbitrariness. The consistency check ensures the logic and rationality of the judgment matrix, avoiding irrational weight distribution due to subjective bias by experts. This effectively improves the scientific nature, rationality, and adaptability of the audit work and reduces information redundancy in the calculation process.

[0060] Optionally, in one embodiment of the present application, the consistency check of the weight value of each audit standard includes: calculating a consistency ratio value of the constructed weight judgment matrix, recorded as a CR value;

[0061] CR = CI / RI;

[0062] Among them, the CI value is a consistency index, CI = (λmax-n) / (n-1), n is the matrix order of the weight judgment matrix, RI is the consistency index of the average random matrix corresponding to the matrix order; if the CR value is calculated and determined to be less than the preset threshold, the weight judgment matrix is considered to be consistent.

[0063] This application illustrates the consistency check of the weight value of each audit standard by way of example: for example, when a multiple-standard audit is required from the six dimensions of completeness, grammatical accuracy, granularity, clarity, consistency, and redundancy, a 6×6 weight judgment matrix can be constructed. The elements of the weight judgment matrix represent the relative importance parameter information between the two standards. A scale of 1-9 is usually used to represent the relative importance. The specific meanings are as follows: 1 indicates that the two standards are equally important, 3 indicates that one standard is slightly important, 5 indicates that one standard is obviously important, 7 indicates that one standard is strongly important, 9 indicates that one standard is extremely important, and 2 / 4 / 6 / 8 are respectively the middle values of the above judgments, such as 2 indicates that it is between equal importance and one standard is slightly important. An industry expert who is very familiar with the automotive software business to be inspected makes a judgment based on the above principles, and the following 6×6 judgment matrix will be generated. Assuming that the expert's experience determines the importance parameter information between the standards of these six dimensions and the different audit standards, the following is shown:

[0064]

[0065] After normalizing the constructed 6×6 weight judgment matrix (i.e., dividing each column of the weight judgment matrix by the sum of the columns to obtain the normalized matrix), we can obtain:

[0066]

[0067] According to the normalized matrix, the weight value of each audit standard is calculated:

[0068] Completeness: (0.382+0.474+0.313+0.348+0.375+0.333) / 6=0.371;

[0069] Grammatical accuracy: (0.127+0.158+0.188+0.174+0.188+0.222) / 6=0.176;

[0070] Particle size: (0.076+0.053+0.063+0.043+0.063+0.056) / 6=0.059;

[0071] Clarity: (0.096+0.079+0.125+0.087+0.094+0.056) / 6=0.089;

[0072] Consistency: (0.191+0.158+0.188+0.174+0.188+0.222) / 6=0.187;

[0073] Redundancy: (0.127 + 0.079 + 0.125 + 0.174 + 0.094 + 0.111) / 6 = 0.118;

[0074] In order to ensure the consistency of the judgment matrix (to avoid deviation or inconsistency in the judgment of the importance of the given standards), the consistency ratio CR value is calculated, CR = CI / RI, where the CI value is the consistency index, CI = (λmax-n) / (n-1) = (6.0-6) / (6-1) = 0, n is the matrix order; RI is the consistency index of the average random matrix, and the consistency index corresponding to the 6-order matrix is determined by looking up the table to be RI = 1.24; therefore, CR = CI / RI = 0, CR value < 0.1, it can be determined that the weight judgment matrix is consistent, and the weight value of each audit standard is credible and available. These weight values are used as the relative weights of the corresponding audit standards in the current audit stage, so as to better ensure the reliability and integrity of the audit results.

[0075] Optionally, in one embodiment of the present application, the determination of the multiple audit standards corresponding to the target audit text content includes: obtaining the user's demand specification audit requirements; using the demand specification audit requirements to generate a standard inspection prompt word Prompt instruction corresponding to a single audit standard, so as to use the Prompt instruction to constitute the multiple audit standards corresponding to the target audit text content. In an embodiment of the present application, the use of the standard inspection prompt word Prompt instruction for generating a single audit standard can clearly convey the specific requirements of the audit task to the large language model, including the scope, focus and objectives of the audit, and reduce the time spent by the large language model on understanding the task, thereby improving the audit efficiency. At the same time, through the Prompt instruction, background information related to the audit task can be provided to the model, such as by introducing contextual information to help the model better understand the background and context of the audit. In addition, the standard inspection prompt word Prompt instruction can also standardize the output format of the large language model, specify the format and style of the model output, ensure that the structure of the audit results meets expectations, and facilitate subsequent processing and analysis.

[0076] Optionally, in one embodiment of the present application, the target audit text content is evaluated in multiple dimensions using a large language model based on the multiple audit standards and the relative weights, and a multi-dimensional evaluation result of the target audit text content with respect to each audit standard is obtained, including: based on the relative weights and the standard prompt word Prompt instruction corresponding to a single audit standard, the target audit text is evaluated in multiple dimensions using a large language LLM model, and a standard prompt word Prompt instruction is used to guide the large language model to combine an audit standard and the weight first, so as to conduct a more targeted audit of the target audit text content, which can make the accuracy and quality of the required content inspection higher, and obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard. To a certain extent, the flexibility and scalability of the system are improved. At the same time, this high-precision audit of the target text content is achieved through a relatively mature large language model, which is low-cost and easy to quickly deploy and optimize performance. Compared with the use of other large language models, the LLM large language model has strong versatility and language understanding capabilities, and is highly competitive even in the face of complex audit tasks.

[0077] Optionally, in one embodiment of the present application, after obtaining the multi-dimensional evaluation results of the target audit text content with respect to each audit standard, the method further includes: generating an overall evaluation of the software requirement document based on the evaluation results of each dimension, the overall evaluation including an evaluation score and a summary of modification suggestions; using the LLM large language model to generate a modified requirement specification document for the modification suggestion summary; and modifying and feeding back the evaluation score, the modification suggestion summary, and the modified requirement specification document. At this stage, the embodiment of the present application generates an evaluation score, a modification suggestion summary, and a corresponding modified requirement document by introducing the LLM large language model in combination with the evaluation results, and feeding back this information to the user together, so that the user can intuitively, comprehensively, and clearly understand whether the system audit results and other related information are reasonable, rather than just feeding back scores or suggestions to the user. This improves the user experience.

[0078] Furthermore, in an optional implementation of the embodiment of the present application, the method further includes:

[0079] If a user responds with "acceptance" to the modification, the entered software requirement document is updated using the modified requirement specification document;

[0080] If a user's response to the modification feedback is "disagree", the software requirement document will be reviewed again.

[0081] Specifically, in an optional implementation of the present application, re-reviewing the software requirement document includes: re-executing step S103 and the subsequent method steps to re-review the software requirement document. The embodiment of the present application limits the re-review at this stage to starting from step S103, rather than re-entering the entered software requirement document into items to obtain multiple sub-requirement items containing a single requirement item, and then re-determining the relevant steps of determining the target review text content based on the sub-requirement items, so as to save review time and improve review efficiency.

[0082] At this stage, the embodiment of the present application further improves the intelligence of the system when reviewing software requirement specification documents through the above method, greatly reducing the workload of users who need to make corrections later, which can better promote the development of software projects, reduce development costs, and shorten the software development cycle.

[0083] The present application provides a method for reviewing software requirement specification documents. Based on predefined requirement item rules, the entered software requirement document is cut into items to obtain multiple sub-requirement items containing a single requirement item; based on the sub-requirement items, the target review text content is determined; multiple review standards for multiple dimensions of the target review text content are determined, and the relative weights of each review standard are determined; based on the multiple review standards and the relative weights, the target review text content is evaluated in multiple dimensions using a large language model to obtain a multi-dimensional evaluation result of the target review text content with respect to each review standard. Through the above method, the problem that manual review is time-consuming and depends on the subjective initiative of the reviewer is solved, and a refined multi-standard review of the requirement content is achieved. The review result is more accurate and reliable, and can meet the user's review needs for various complex scenarios or businesses in a friendly manner.

[0084] Example 2

[0085] Based on the software requirement specification document audit method provided in the first embodiment of the present application, the present application embodiment also provides a software requirement specification document audit device, such as Figure 2 As shown, Figure 2 A schematic diagram of the structure of a software requirements specification document review device 20 provided in an embodiment of the present application, wherein the software requirements specification document review device 20 includes:

[0086] The segmentation module 201 is used to segment the input software requirement document into items based on predefined requirement item rules to obtain multiple sub-requirement items containing a single requirement item;

[0087] Extraction module 202, for determining target review text content based on the sub-requirement items;

[0088] An analysis module 203 is configured to determine multiple audit criteria for multiple dimensions of the target audit text content and determine the relative weight of each audit criterion;

[0089] The audit module 204 is used to perform a multi-dimensional evaluation of the target audit text content using a large language model based on the multiple audit standards and the relative weights, and obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard.

[0090] Optionally, in one embodiment of the present application, the extraction module 202 is further configured to:

[0091] Determining whether a functional timing diagram exists in each of the sub-requirement items;

[0092] If there is no functional sequence diagram in the sub-requirement item, the requirement item number and requirement text content corresponding to the requirement item will be used as the target review text content;

[0093] If there is a functional timing diagram in the sub-requirement item, the functional timing diagram is converted into supplementary text content that can characterize the interaction relationship between the software business object and different software business objects, as a supplement to the item number and requirement text of the sub-requirement item, to obtain the target review text content.

[0094] Optionally, in one embodiment of the present application, the analysis module 203 is further configured to:

[0095] Determining, based on the determined expert experience, information on importance parameters between two different audit criteria in the multiple audit criteria of the multiple dimensions, wherein the multiple audit criteria in the multiple dimensions include at least one audit criterion of completeness, grammatical accuracy, granularity, clarity, consistency, or redundancy;

[0096] Constructing a weight judgment matrix according to the importance parameter information;

[0097] Calculate the weight value of each audit criterion according to the weight judgment matrix;

[0098] A consistency check is performed on the weight value of each audit standard, and the relative weight is determined based on the consistency check result.

[0099] Optionally, in one embodiment of the present application, the analysis module 203 is further configured to:

[0100] Calculate the consistency ratio value of the constructed weight judgment matrix, recorded as CR value;

[0101] CR = CI / RI;

[0102] Wherein the CI value is the consistency index, CI = (λmax-n) / (n-1), n is the matrix order of the weight judgment matrix, and RI is the consistency index of the average random matrix corresponding to the matrix order;

[0103] If the CR value is less than the preset threshold value, the weight judgment matrix is considered to be consistent.

[0104] Optionally, in one embodiment of the present application, the analysis module 203 is further configured to:

[0105] Obtain user's requirement specification review requirements;

[0106] The requirement specification audit requirements are used to generate a standard inspection prompt word Prompt instruction corresponding to a single audit standard, so as to use the Prompt instruction to form multiple audit standards corresponding to the target audit text content.

[0107] Optionally, the audit module 204 is further configured to:

[0108] Based on the relative weights and the standard prompt word Prompt instruction corresponding to a single audit standard, the target audit text is evaluated in multiple dimensions using the large language LLM model to obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard.

[0109] Optionally, in one embodiment of the present application, the device 20 also includes a summary module (not shown in the accompanying drawings), which is used to, after obtaining the multi-dimensional evaluation results of the target audit text content regarding each audit standard: generate an overall evaluation of the software requirement document based on the evaluation results of each dimension, the overall evaluation including an evaluation score and a summary of modification suggestions; generate a modified requirement specification document based on the summary of the modification opinions using the LLM large language model; and modify the evaluation score, the summary of modification suggestions and the modified requirement specification document for feedback.

[0110] Optionally, in one embodiment of the present application, the apparatus 20 further includes an update module (not shown in the drawings), which is configured to:

[0111] If a user responds with "acceptance" to the modification, the entered software requirement document is updated using the modified requirement specification document;

[0112] If a user's response to the modification feedback is "disagree", the software requirement document will be reviewed again.

[0113] The present application provides an audit device for software requirement specification documents, wherein a segmentation module is set to cut the input software requirement document into items based on predefined requirement item rules, and obtain multiple sub-requirement items containing a single requirement item; an extraction module is set to determine the target audit text content based on the sub-requirement items; an analysis module is set to determine multiple audit standards for multiple dimensions of the target audit text content, and determine the relative weight of each audit standard; an audit module is set to perform a multi-dimensional evaluation of the target audit text content using a large language model based on the multiple audit standards and the relative weights, and obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard. This device effectively solves the problem that traditional manual auditing is time-consuming and relies on the subjective initiative of the auditor, and realizes a refined multi-standard audit of the requirement content. The audit result is more accurate and reliable, and can friendly meet the user's audit needs for various complex scenarios or businesses. In addition, the structure of the device is relatively simple and easy to deploy and implement.

[0114] Example 3:

[0115] An embodiment of the present application also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the review method of any software requirement specification document as described in the first embodiment of the present application.

[0116] Example 4:

[0117] The present application also provides an electronic device, such as Figure 3 As shown, Figure 3 An embodiment of the present application provides a schematic structural diagram of an electronic device 30, which includes:

[0118] One or more processors 301, a communication interface 302, a memory 303 and a communication bus 304, wherein the processor 301, the memory 303 and the communication interface 302 communicate with each other via the communication bus 304;

[0119] Memory 303, used to store one or more programs;

[0120] When the one or more programs are executed by the one or more processors 301, the one or more processors 301 implement any one of the software requirement specification document review methods described in Example 1 of the present application.

[0121] Thus far, this application has described specific embodiments of the present subject matter. In some cases, the actions recited in the claims can be performed in a different order and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing can be advantageous.

[0122] In the 1990s, technological improvements could be clearly distinguished as either hardware improvements (for example, improvements to circuit structures like diodes, transistors, and switches) or software improvements (improvements to process flows). However, with the advancement of technology, many process flow improvements today can now be considered direct improvements to hardware circuit structures. Designers almost always create the corresponding hardware circuit structure by programming the improved process flow into the hardware circuit. Therefore, it cannot be said that a process flow improvement cannot be implemented using hardware modules. For example, a programmable logic device (PLD), such as a field programmable gate array (FPGA), is an integrated circuit whose logical function is determined by user programming. Designers can "integrate" a digital system layer onto a PLD through their own programming, eliminating the need for chip manufacturers to design and manufacture dedicated integrated circuit chips. Moreover, nowadays, instead of manually fabricating integrated circuit chips, this programming is mostly done using "logic compiler" software. This is similar to the software compiler used when developing programs. Before compilation, the original code must also be written in a specific programming language, called a hardware description language (HDL). There is not just one HDL, but many, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art will also understand that by simply programming the method flow in one of these hardware description languages and then programming it into an integrated circuit, a hardware circuit that implements the logic method flow can be easily obtained.

[0123] The controller can be implemented in any suitable manner. For example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also know that in addition to implementing the controller in a purely computer-readable program code format, the controller can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be considered as structures within the hardware component. Or even, the devices for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0124] The system layers, devices, modules, or units described in the above embodiments may be implemented by computer chips or physical devices, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0125] For the convenience of description, the above devices are described as being divided into various units according to their functions. Of course, when implementing this application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0126] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0127] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, system layers, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0128] The present application may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The present application may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communications network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0129] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system-level embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0130] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for reviewing a software requirement specification document, characterized in that: include: Based on predefined requirement item rules, the entered software requirement document is divided into items to obtain multiple sub-requirement items containing a single requirement item; Based on the sub-requirement items, determine the target review text content; Determine multiple review criteria for multiple dimensions of text content review for the stated objectives, and determine the relative weights of each review criterion; According to the multiple audit standards and the relative weights, a large language model is used to perform a multi-dimensional evaluation of the target audit text content to obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard.

2. The method for reviewing software requirement specification documents according to claim 1, characterized in that: Determining the target review text content based on the sub-requirement items includes: Determining whether a functional timing diagram exists in each of the sub-requirement items; If there is no functional sequence diagram in the sub-requirement item, the requirement item number and requirement text content corresponding to the requirement item will be used as the target review text content; If there is a functional timing diagram in the sub-requirement item, the functional timing diagram is converted into supplementary text content that can characterize the interaction relationship between the software business object and different software business objects, as a supplement to the item number and requirement text of the sub-requirement item, to obtain the target review text content.

3. The method for reviewing software requirement specification documents according to claim 1, characterized in that: The determining of multiple review criteria for multiple dimensions of the target review text content and determining the relative weight of each review criterion includes: Determining, based on the determined expert experience, information on importance parameters between two different audit criteria in the multiple audit criteria of the multiple dimensions, wherein the multiple audit criteria in the multiple dimensions include at least one audit criterion of completeness, grammatical accuracy, granularity, clarity, consistency, or redundancy; Constructing a weight judgment matrix according to the importance parameter information; Calculate the weight value of each audit criterion according to the weight judgment matrix; A consistency check is performed on the weight value of each audit standard, and the relative weight is determined based on the consistency check result.

4. The method for reviewing software requirement specification documents according to claim 3, characterized in that: The consistency check of the weight value of each audit standard includes: Calculate the consistency ratio value of the constructed weight judgment matrix, recorded as CR value; CR=CI / RI, Wherein the CI value is the consistency index, CI = (λmax-n) / (n-1), n is the matrix order of the weight judgment matrix, and RI is the consistency index of the average random matrix corresponding to the matrix order; If the CR value is less than the preset threshold value, the weight judgment matrix is considered to be consistent.

5. The method for reviewing software requirement specification documents according to claim 1, characterized in that: The determining of multiple review criteria corresponding to the target review text content includes: Obtain user's requirement specification review requirements; The requirement specification audit requirements are used to generate a standard inspection prompt word Prompt instruction corresponding to a single audit standard, so as to use the Prompt instruction to form multiple audit standards corresponding to the target audit text content.

6. The method for reviewing software requirement specification documents according to claim 5, characterized in that: The method of performing a multi-dimensional evaluation of the target audit text content using a large language model based on the multiple audit criteria and the relative weights to obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit criteria includes: Based on the relative weights and the standard prompt word Prompt instruction corresponding to a single audit standard, the target audit text is evaluated in multiple dimensions using the large language LLM model to obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard.

7. The method for reviewing software requirement specification documents according to claim 1, characterized in that: After obtaining the multi-dimensional evaluation results of the target audit text content with respect to each audit standard, the method further includes: Based on the evaluation results of each dimension, generate an overall evaluation of the software requirement document, the overall evaluation including an evaluation score and a summary of modification suggestions; Generate a revised requirement specification document summarizing the modification suggestions using the LLM large language model; The evaluation score, the modification suggestion summary and the modified requirement specification document are modified and fed back.

8. The method for reviewing software requirement specification documents according to claim 7, characterized in that: The method further comprises: If a user responds with "Agree" to the modification, the entered software requirement document is updated using the modified requirement specification document; If the user receives a reply of "disagreement" to the modification feedback, the software requirement document will be reviewed again.

9. A software requirements specification document review device, characterized in that: include: The segmentation module is used to segment the input software requirement document into items based on predefined requirement item rules to obtain multiple sub-requirement items containing a single requirement item; An extraction module, configured to determine target audit text content based on the sub-requirement items; an analysis module for determining multiple review criteria for multiple dimensions of the target review text content and determining relative weights of the respective review criteria, wherein the multiple dimensions include at least one review criterion of completeness, grammatical accuracy, granularity, clarity, consistency, or redundancy; The audit module is used to use a large language model to perform a multi-dimensional evaluation of the target audit text content based on the multiple audit standards and the relative weights, and obtain a multi-dimensional evaluation result of the target audit text content with respect to each audit standard.

10. A computer storage medium, characterized in that The computer storage medium stores computer-executable instructions, which, when executed, execute the method for reviewing a software requirement specification document according to any one of claims 1 to 8.

Images