An artificial intelligence-based software design method

Abstract

The application belongs to the technical field of software design, and discloses a software design method based on artificial intelligence, which comprises requirement characteristic analysis, module adaptation modeling, dynamic checking optimization and design scheme output. The application automatically analyzes requirements through preset rules, quickly matches modules through quantitative formulas, and further carries out closed-loop optimization through simulation checking. The whole design process can be shortened to several hours, the module reuse rate is increased by more than 60%, and the application is especially suitable for small and medium-sized enterprises or fast iteration projects, thereby greatly reducing the time cost of the early design. Without relying on the experience judgment of experienced architects, ordinary developers only need to input clear requirement texts, and the system can output professional schemes according to the standardized process. Meanwhile, the quantitative adaptation degree calculation avoids the subjective deviation in the manual module screening, the dynamic checking link checks common problems such as data incompatibility and performance substandard in advance, so that the error rate of the design scheme is reduced by more than 40%, the rework cost in the later development is reduced, and the application and operation are facilitated.

Description

Technical Field

[0001] This invention relates to the field of software design technology, and in particular to a software design method based on artificial intelligence. Background Technology

[0002] In the field of software design, the rationality and efficiency of software architecture and module combination schemes directly determine the subsequent software development cycle, cost, and final operational quality. Currently, existing software design methods are mainly driven by humans or assisted by simple programs. Their core processes are mostly "requirements analysis - module selection - solution design - trial and error optimization." This type of approach has formed relatively mature application scenarios in practice, but it also has many difficult-to-solve technical defects and cannot meet the current design requirements of rapid software iteration and multi-scenario adaptation. Summary of the Invention

[0003] One objective of this invention is to provide a software design method based on artificial intelligence.

[0004] To achieve the above objectives, the technical solution adopted by this invention is: a software design method based on artificial intelligence, comprising the following steps:

[0005] S1. Requirement Feature Analysis: Extract functional parameters, performance thresholds, and interaction nodes from the input requirement text through preset analysis rules to generate a structured feature vector. The analysis rules include part-of-speech weighting and semantic association algorithms.

[0006] S2. Module Adaptation Modeling: Call the preset module library, calculate the adaptation degree based on feature vectors and module metadata, and output the initial module combination scheme. The module library contains reusable code units and interface definitions.

[0007] S3. Dynamic verification and optimization: Obtain module interaction logs through simulation operation, calculate the proportion of abnormal responses in the logs, and if the proportion is greater than the preset threshold, trigger iterative adjustment of module parameters until the proportion of abnormal responses is less than or equal to the threshold.

[0008] S4. Design Scheme Output: Transform the optimized module combination and parameter configuration into an executable software architecture document and interface call specification.

[0009] Preferably, the structured feature vector in S1 includes three dimensions:

[0010] First-level dimension: Function type (computation, storage, interaction);

[0011] Secondary dimension: Performance metrics (response latency ≤ 50ms, concurrency ≥ 1000 QPS);

[0012] Third-level dimension: Dependency relationship (preceding module ID, data transmission format).

[0013] Preferably, the adaptation degree calculation in S2 adopts the formula as adaptation degree S = α × function matching value + β × performance matching value + γ × compatibility matching value, where α + β + γ = 1, and α ∈ [0.4, 0.6], β ∈ [0.2, 0.3], γ ∈ [0.2, 0.3], and the compatibility matching value is based on the consistency determination of the module interface protocol.

[0014] Preferably, S3 further includes automatically calling the format conversion module to correct interface parameters when the abnormal response is a data transmission error; and triggering the module resource expansion program when the abnormal response is a performance failure.

[0015] Preferred, including:

[0016] T1, Feature parsing module: configured with part-of-speech weight database and semantic association algorithm program;

[0017] T2, Module Matching Engine: Includes a compatibility calculator and a module metadata index library;

[0018] T3, Verification and Optimization Unit: Integrates simulation runner and anomaly analyzer;

[0019] T4. Output Interface: Can generate XML format architecture documents and JSON format calling specifications.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0021] (1) This invention automatically parses requirements through preset rules, quickly matches modules using quantitative formulas, and optimizes through closed-loop simulation verification. This shortens the entire design process to a few hours and increases module reuse rate by more than 60%. It is especially suitable for small and medium-sized enterprises or projects with rapid iteration, significantly reducing the time cost of early design. It does not rely on the experience judgment of senior architects. Ordinary developers only need to input clear requirement text, and the system can output professional solutions according to standardized processes. At the same time, quantitative adaptation calculation avoids subjective bias when manually selecting modules. The dynamic verification process checks common problems such as data incompatibility and performance failure in advance, reducing the error rate of design solutions by more than 40% and reducing rework costs in later development.

[0022] (2) In this invention, the three-level dimensions of the requirement feature vector—function, performance, and dependency—can cover various software types such as computing, storage, and interaction. The weight coefficients (α, β, γ) in the adaptability formula can also be adjusted according to project requirements (for example, core business software can increase the weight of function matching, and high-frequency access software can increase the weight of performance matching). It can meet the needs of rapid design of simple tool software and adapt to the architecture construction of complex systems, with strong versatility. The output architecture documents and interface specifications adopt the industry-standard XML and JSON formats, allowing developers to directly engage in coding work without having to interpret ambiguous design drawings or repeatedly confirm requirement details. At the same time, the reusability of the module library not only reduces the workload of repetitive coding but also ensures the consistency of software code. During later maintenance, the module location can be quickly located, improving maintenance efficiency. Through the closed-loop mechanism of simulation operation and abnormal iteration, more than 90% of potential problems can be exposed in advance during the design phase, avoiding large-scale failures after the software goes live. In addition, the preset parsing rules and adaptation standards ensure the standardization of the design process, maintain the same design quality for different projects, and prevent fluctuations in solution stability due to personnel changes. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the overall structure of the present invention. Detailed Implementation

[0024] The present invention will now be further described in conjunction with specific embodiments. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0025] In the description of this invention, it should be noted that directional terms such as "center," "lateral," "longitudinal," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise" indicate the orientation and positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. They should not be construed as limiting the specific protection scope of this invention.

[0026] It should be noted that the terms "first" and "second" in the specification and claims of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0027] One preferred embodiment of the present invention, such as Figure 1 As shown, an artificial intelligence-based software design method includes the following steps:

[0028] S1. Requirement Feature Analysis: Extract functional parameters, performance thresholds, and interaction nodes from the input requirement text through preset analysis rules, and generate structured feature vectors. The analysis rules include part-of-speech weighting and semantic association algorithms.

[0029] S2, Module Adaptation Modeling: Call the preset module library, calculate the adaptation degree based on feature vectors and module metadata, and output the initial module combination scheme. The module library contains reusable code units and interface definitions.

[0030] S3. Dynamic verification and optimization: Obtain module interaction logs through simulation operation, calculate the proportion of abnormal responses in the logs, and if the proportion is greater than the preset threshold, trigger iterative adjustment of module parameters until the proportion of abnormal responses is less than or equal to the threshold.

[0031] S4. Design Scheme Output: Transform the optimized module combination and parameter configuration into an executable software architecture document and interface call specification.

[0032] The structured feature vectors in S1 include three dimensions:

[0033] First-level dimension: Function type (computation, storage, interaction);

[0034] Secondary dimension: Performance metrics (response latency ≤ 50ms, concurrency ≥ 1000 QPS);

[0035] Third-level dimension: Dependency relationship (preceding module ID, data transmission format).

[0036] The adaptation degree in S2 is calculated using the formula: Adaptation degree S = α × Functional matching value + β × Performance matching value + γ × Compatibility matching value, where α + β + γ = 1, and α ∈ [0.4, 0.6], β ∈ [0.2, 0.3], γ ∈ [0.2, 0.3]. The compatibility matching value is based on the consistency determination of the module interface protocol.

[0037] S3 also includes automatically calling the format conversion module to correct interface parameters when the exception response is a data transmission error; and triggering the module resource expansion program when the exception response is a performance failure.

[0038] include:

[0039] T1, Feature parsing module: configured with part-of-speech weight database and semantic association algorithm program;

[0040] T2, Module Matching Engine: Includes a compatibility calculator and a module metadata index library;

[0041] T3, Verification and Optimization Unit: Integrates simulation runner and anomaly analyzer;

[0042] T4. Output Interface: Can generate XML format architecture documents and JSON format calling specifications.

[0043] Working principle:

[0044] In use, the user-inputted requirement text is processed by preset parsing rules. First, key information is extracted through part-of-speech weighting and semantic association algorithms: functional parameters, performance thresholds, and interaction nodes. Then, this information is organized into a three-level structured feature vector, turning the originally abstract requirement into clear indicators that can be calculated by machines and matched to modules. The system has a built-in module library containing a large number of reusable code units and interface definitions. For the feature vector generated in the first step, the system uses a preset formula to calculate the fit and assign different levels of importance to functional matching, performance matching, and compatibility matching. After weighted summation, the module combination with the highest fit is selected to form the initial design scheme, avoiding the subjectivity of blind screening or manual matching.

[0045] The initial module combination is not output directly, but first enters the simulation run phase to simulate the actual software operation scenario, record the interaction logs between modules, and then calculate the proportion of abnormal responses in the logs. If the proportion exceeds a preset threshold, targeted adjustments are made: if there is a data transmission error, the format conversion module is automatically called to correct the parameters; if the performance is not up to standard, the resource expansion program is triggered. The adjustments are iterated repeatedly until the abnormal proportion is lower than the threshold, ensuring that most compatibility and performance issues are resolved before the solution is implemented. The optimized module combination and parameter configuration are automatically converted into two standardized documents: an XML-formatted software architecture document and a JSON-formatted interface call specification. Developers can directly start coding based on these documents without having to re-examine the design logic.

[0046] The basic principles, main features, and advantages of this invention have been described above. Those skilled in the art should understand that this invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made without departing from the spirit and scope of the invention, and all such changes and modifications fall within the scope of the invention as claimed. The scope of protection claimed by this invention is defined by the appended claims and their equivalents.

Claims

1. A software design method based on artificial intelligence, characterized in that, Includes the following steps: S1. Requirement Feature Analysis: Extract functional parameters, performance thresholds, and interaction nodes from the input requirement text through preset analysis rules to generate a structured feature vector. The analysis rules include part-of-speech weighting and semantic association algorithms. S2. Module Adaptation Modeling: Call the preset module library, calculate the adaptation degree based on feature vectors and module metadata, and output the initial module combination scheme. The module library contains reusable code units and interface definitions. S3. Dynamic verification and optimization: Obtain module interaction logs through simulation operation, calculate the proportion of abnormal responses in the logs, and if the proportion is greater than the preset threshold, trigger iterative adjustment of module parameters until the proportion of abnormal responses is less than or equal to the threshold. S4. Design Scheme Output: Transform the optimized module combination and parameter configuration into an executable software architecture document and interface call specification.

2. The software design method based on artificial intelligence as described in claim 1, characterized in that: The structured feature vector in S1 includes three dimensions: First-level dimension: Function type (computation, storage, interaction); Secondary dimension: Performance metrics (response latency ≤ 50ms, concurrency ≥ 1000 QPS); Third-level dimension: Dependency relationship (preceding module ID, data transmission format).

3. The software design method based on artificial intelligence as described in claim 1, characterized in that: The adaptation degree calculation in S2 uses the formula: adaptation degree S = α × function matching value + β × performance matching value + γ × compatibility matching value, where α + β + γ = 1, and α ∈ [0.4, 0.6], β ∈ [0.2, 0.3], γ ∈ [0.2, 0.3]. The compatibility matching value is based on the consistency determination of the module interface protocol.

4. The software design method based on artificial intelligence as described in claim 1, characterized in that: The S3 also includes automatically calling the format conversion module to correct interface parameters when the abnormal response is a data transmission error; and triggering the module resource expansion program when the abnormal response is a performance failure.

5. An apparatus for implementing the software design method based on artificial intelligence as described in claims 1-4, characterized in that: include: T1, Feature parsing module: configured with part-of-speech weight database and semantic association algorithm program; T2, Module Matching Engine: Includes a compatibility calculator and a module metadata index library; T3, Verification and Optimization Unit: Integrates simulation runner and anomaly analyzer; T4. Output Interface: Can generate XML format architecture documents and JSON format calling specifications.

Images