Internet software development management system and method

Abstract

The invention relates to the technical field of software development management, in particular to an internet software development management system and method. The method comprises the following steps: collecting an original function demand set and carrying out function clustering analysis; extracting and classifying demand function features and generating independent demand units; carrying out dependency analysis on the independent demand units; constructing a dependency relationship set and a demand logic structure; the method comprises the following steps: generating a demand framework matrix, carrying out organization network simulation to obtain a demand connection network, carrying out code reverse evolution according to the demand connection network, carrying out basic parameter matching to obtain evolution basic parameters, carrying out a pressure simulation test on the evolution basic parameters to generate a pressure test result, and capturing resource pressure defects based on the pressure test result. And optimizing and adjusting the evolution basic parameters to generate optimized development parameters. According to the invention, a more efficient Internet software development management method is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of software development management, and particularly to an Internet software development management system and method. Background Art

[0002] Traditional requirement management methods are difficult to effectively respond to rapidly changing market requirements. The ambiguity and uncertainty of requirements make it easy for development teams to deviate during the planning and execution processes. The lack of a systematic management process leads to resource waste and schedule delays, resulting in a reduced project success rate and difficult-to-guarantee software quality. At the same time, there is insufficient collaboration and information sharing among teams, the dependency relationships between requirements are complex, and there is a lack of a clear logical structure, causing communication barriers during the development process. The ambiguity of dependency relationships makes it difficult for developers to judge the priority and impact of requirements, thereby affecting the overall development efficiency and quality. Project review and experience accumulation are also limited, and it is difficult to form an effective knowledge management system. The evolution and maintenance of code are becoming increasingly complex, and there is a lack of effective methods for the matching of basic parameters and stress testing, unable to timely discover and solve resource pressure defects, resulting in increased later maintenance costs and greater difficulty in optimization and adjustment. The lack of a systematic Internet software development management method to support an efficient development process and resource utilization has become an urgent problem to be solved. Summary of the Invention

[0003] Based on this, it is necessary to provide an Internet software development management system and method to solve at least one of the above technical problems.

[0004] To achieve the above object, an Internet software development management method includes the following steps:

[0005] Step S1: Collect the original function requirement set; perform functional clustering analysis on the original function requirement set to obtain classified requirement function features; perform tagging processing on the classified requirement function features to generate independent requirement units;

[0006] Step S2: Perform inter-unit dependency analysis on the independent requirement units to obtain a dependency relationship set; construct a requirement logical structure based on the dependency relationship set;

[0007] Step S3: Perform virtual-real framework interweaving on the independent requirement units based on the requirement logical structure to obtain a requirement framework matrix; perform organizational network simulation according to the requirement framework matrix to generate a requirement connection network;

[0008] Step S4: Perform code reverse evolution according to the requirement connection network and perform basic parameter matching to obtain evolved basic parameters; perform stress simulation testing on the evolved basic parameters to generate stress test results;

[0009] Step S5: Capture the resource pressure defects of the evolution basic parameters based on the stress test results; optimize and adjust the evolution basic parameters according to the resource pressure defects to generate optimized development parameters.

[0010] The present invention realizes the systematicness and accuracy of requirement collection by collecting the original function requirement set and performing function clustering analysis to generate independent requirement units, promotes the effective management and analysis of function requirements, realizes the clarification of the dependency relationship between requirements by performing dependency parsing on the independent requirement units to construct a requirement logic structure, optimizes the requirement logic structure to support dynamic adjustment, generates a requirement framework matrix by interweaving virtual and real frameworks for the independent requirement units based on the requirement logic structure, and performs organizational network simulation to realize the efficient generation of the requirement framework matrix, promotes team collaboration and information sharing, realizes the improvement of code evolution efficiency by performing code reverse evolution according to the requirement connection network, performing basic parameter matching and stress simulation testing, ensures the accuracy of basic parameter matching, captures resource pressure defects based on the stress test results, optimizes the evolution basic parameters, realizes the improvement of resource utilization efficiency, reduces development costs and risks, overall improves the efficiency and quality of software development, and ensures the successful delivery and sustainable development of the project.

[0011] Preferably, step S1 includes the following steps:

[0012] Step S11: Collect the original function requirement set; perform semantic segmentation on the original function requirement set to obtain independent requirement segments;

[0013] Step S12: Extract the function features of the independent requirement segments; perform clustering analysis on the function features of each independent requirement segment to obtain classified requirement function features;

[0014] Step S13: Eliminate the redundant features of the classified requirement function features to obtain clean requirement functions;

[0015] Step S14: Perform unique label assignment on the clean requirement functions based on the classified requirement function features to generate independent requirement units.

[0016] The present invention realizes the refinement and clarification of requirement information by collecting the original function requirement set and performing semantic segmentation to generate independent requirement segments, promotes the clear expression of requirements, extracts the function features of the independent requirement segments and performs clustering analysis to obtain classified requirement function features, realizes the systematic arrangement of function features, optimizes the convenience of requirement management, eliminates redundant features to obtain clean requirement functions, improves the accuracy and operability of requirements, performs unique label assignment on the clean requirement functions based on the classified requirement function features to generate independent requirement units, realizes the traceability and effective management of requirements, enhances information sharing and understanding during team collaboration, overall improves the efficiency and quality of requirement management, and provides a solid foundation for subsequent development.

[0017] Preferably, step S2 includes the following steps:

[0018] Step S21: Hierarchically decompose the independent demand units to obtain a demand hierarchy structure; detect the lead-follow relationship of the independent demand units according to the demand hierarchy structure, and generate a unit dependency candidate set;

[0019] Step S22: Cross-validate and filter the unit dependency candidate set to obtain high-confidence dependency relationships; perform a transitive closure operation on the high-confidence dependency relationships to generate a dependency transmission chain;

[0020] Step S23: Detect the network loop of the dependency transmission chain; perform dependency analysis based on the network loop of the dependency transmission chain to obtain a set of dependency relationships;

[0021] Step S24: Construct a directed acyclic graph structure based on the set of dependency relationships, and perform demand topological sorting based on the directed acyclic graph structure to obtain a demand topological sequence; perform a priority weighting process on the demand topological sequence to generate a demand logical structure.

[0022] In the present invention, by hierarchically decomposing the independent demand units to generate a demand hierarchy structure, the clear layering and management of the demands are realized, which is convenient for the team to understand and analyze the demands. Detect the lead-follow relationship according to the demand hierarchy structure to generate a unit dependency candidate set, ensuring clear logical relationships between the demands and enhancing the accuracy of demand management. Through cross-validation and filtering, high-confidence dependency relationships are obtained, improving the credibility and effectiveness of the dependency relationships. Perform a transitive closure operation to generate a dependency transmission chain, optimizing the integrity and coherence of the dependency relationships. Detect the network loop of the dependency transmission chain to ensure that there is no dead loop in the dependency relationships between the demands. Perform dependency analysis based on the dependency transmission chain to obtain a set of dependency relationships, promoting the comprehensive analysis of the relationships between the demands. Construct a directed acyclic graph structure to ensure the rationality of the demand logical structure, perform demand topological sorting to obtain a demand topological sequence, and perform a priority weighting process to generate a demand logical structure, which is convenient for the efficient execution and resource allocation of subsequent development, and overall improves the systematicness and scientificity of demand management.

[0023] Preferably, the virtual-real framework interleaving of the independent demand units based on the demand logical structure in step S3 includes:

[0024] Perform a dot matrix conversion on the demand logical structure to obtain logical dot matrix mapping data;

[0025] Perform a semantic vector encoding mapping on the independent demand units to generate demand semantic vectors;

[0026] Calculate the similarity distance of the demand semantic vectors to obtain demand affinity data;

[0027] Perform orthogonal projection fusion on the logical dot matrix mapping data and the demand affinity data to obtain a demand position matrix;

[0028] Based on the demand position matrix, perform virtual-real boundary division to generate a virtual-real partition boundary;

[0029] Based on the virtual-real partition boundary, perform frame attribute mapping on independent demand units to obtain frame characteristic data;

[0030] According to the frame characteristic data, perform mapping rule interleaving simulation to obtain a demand frame matrix.

[0031] The present invention realizes the visualization of demand relationships by performing dot matrix conversion on the demand logical structure to generate logical dot matrix mapping data, which is convenient for analysis and understanding. The semantic vector encoding mapping of independent demand units generates demand semantic vectors, enhancing the semantic expression ability of demands. By calculating the similarity distance, demand affinity data is obtained, promoting the accurate evaluation of the relationships between demands. The orthogonal projection fusion of the logical dot matrix mapping data and the demand affinity data forms a demand position matrix, improving the spatial positioning accuracy of demands. Based on the demand position matrix, virtual-real boundary division is performed to generate a virtual-real partition boundary, clarifying the virtual and actual attributes of demands. Frame attribute mapping processes independent demand units to obtain frame characteristic data, enhancing the structured information of demands. Mapping rule interleaving simulation generates a demand frame matrix, realizing the efficient connection and integration between demands, and overall improving the systematicness and flexibility of demand management.

[0032] Preferably, the organizational network simulation according to the demand frame matrix in step S3 includes:

[0033] Evaluate the connection strength of the demand frame matrix, and based on the connection strength, allocate node weights to obtain node connection weight data;

[0034] Based on the node connection weight data, construct a multi-layer network topology structure and draw a topology diagram to generate a network topology sketch;

[0035] According to the node connection weight data, perform node clustering reconstruction on the network topology sketch to obtain reconstructed clustering nodes;

[0036] Based on the reconstructed clustering nodes, perform domain-oriented convergence mapping to obtain a functional aggregation domain;

[0037] According to the functional aggregation domain, perform organizational network simulation to generate a demand connection network.

[0038] By evaluating the connection strength of the requirements framework matrix, the present invention realizes the quantitative analysis of the relationships between requirements, ensures that the relevance of each requirement node is accurately reflected, assigns node weights based on the connection strength, clarifies the importance of different requirements, improves the scientificity and rationality of decision-making, constructs a multi-layer network topology structure and draws a topology diagram, provides an intuitive view of the interactions between requirements, facilitates the team's in-depth understanding and analysis of the relationships between requirements, optimizes the logical classification of requirements through node clustering reconstruction, improves the readability and analysis ability of the network, forms a functional aggregation domain through domain-oriented convergence mapping based on the reconstructed clustering nodes, promotes the centralized management and resource sharing of related requirements, simulates the organizational network based on the functional aggregation domain, generates a requirements connection network, clearly shows the dynamic relationships and interactions between requirements, supports the effective coordination and resource allocation in the subsequent development process, and overall improves the systematicness, flexibility and operability of requirements management, laying a solid foundation for the success of the project.

[0039] Preferably, the code reverse evolution based on the requirements connection network and the basic parameter matching in step S4 include:

[0040] Perform requirements function mapping on the requirements connection network to obtain function module decomposition data;

[0041] Expand the dependency conduction of the function module decomposition data, perform hierarchical splitting according to the conduction probability of 0.3 - 0.9, extract the deep dependency paths, and generate recursive association links;

[0042] Induce the direction of the recursive association links to obtain the main code nodes;

[0043] Perform logical reverse deduction based on the function module decomposition data and the main code nodes to generate a code mapping skeleton;

[0044] Perform structural replacement processing on the code mapping skeleton according to a preset code library to obtain an evolved code framework;

[0045] Perform reverse function derivation on the evolved code framework to generate a code evolution trajectory;

[0046] Decompose the call chain of the code evolution trajectory to obtain the parameter passing relationship;

[0047] Extract the boundary constraints of the parameter passing relationship, where the range of values that can be taken by the key variables is extracted, and the upper and lower limits are set, that is, 0.001 - 1000 for floating-point type and 1 - 5000 for integer type, to obtain a parameter constraint set;

[0048] Perform dynamic assignment derivation on the parameter constraint set to obtain a parameter transformation mapping;

[0049] Perform basic parameter matching on the parameter transformation mapping to obtain the evolution basic parameters.

[0050] Through the demand function mapping of the demand connection network, the present invention realizes the systematic decomposition of functional modules, ensures that the logical structure of demands is clearly presented, unfolds the dependency conduction of functional module decomposition data, performs hierarchical splitting based on conduction probability, extracts deep dependency paths, generates recursive association links, improves the understanding and analysis of complex dependency relationships, induces the direction of the recursive association links to help identify main code nodes, focuses on key code areas, performs logical reverse deduction based on functional module decomposition data and main code nodes to generate a code mapping skeleton, provides a basic framework for the evolution of code, optimizes the code mapping skeleton through structure replacement processing to obtain an evolved code framework, improves the maintainability and flexibility of code, generates a code evolution trajectory through reverse function derivation, helps trace the historical changes of code, disassembles the call chain to obtain parameter passing relationships, clarifies the flow path and dependencies of parameters, extracts the boundary constraints of parameter passing relationships, sets the allowable value range of key variables, ensures the validity and reasonableness of parameters, obtains a parameter transformation mapping through dynamic assignment derivation, supports the flexible adjustment and optimization of parameters, realizes the accurate acquisition of evolution basic parameters through basic parameter matching, overall improves the systematicness, reliability and efficiency of code management, and provides a solid foundation for subsequent development and maintenance.

[0051] Preferably, the pressure simulation test on the evolution basic parameters in step S4 includes:

[0052] Calculate the resource requirements for the evolution basic parameters to obtain evolution requirement resources;

[0053] Map the evolution requirement resources to resource constraint data;

[0054] Based on the resource constraint data, perform pressure scenario modeling to obtain a resource pressure test model;

[0055] According to the resource pressure test model, perform evolution simulation on the evolution basic parameters to generate a pressure test result.

[0056] Through the calculation of resource requirements for the evolution basic parameters, the present invention ensures the reasonable evaluation and prediction of resource usage, obtains evolution requirement resources, provides a basis for subsequent resource allocation, maps the evolution requirement resources to resource constraint data, clarifies the resource limitation conditions and constraints, performs pressure scenario modeling based on the resource constraint data to generate a resource pressure test model, supports the simulation and analysis of complex scenarios, performs evolution simulation on the evolution basic parameters according to the resource pressure test model to generate a pressure test result, evaluates the performance of the evolution basic parameters under different pressures, identifies potential problems and bottlenecks, overall improves the stability and reliability of the system, and provides data support and decision-making basis for subsequent optimization and adjustment.

[0057] Preferably, step S5 includes the following steps:

[0058] Step S51: Set the upper limit of resource occupancy to 70%-95% of computing resources and 80%-98% of storage resources; perform resource bottleneck mapping on the stress test results based on the upper limit of resource occupancy to obtain resource constraint conflict data;

[0059] Step S52: Perform associated load decomposition on the resource constraint conflict data and construct a resource pressure distribution matrix; perform an evolution process simulation on the evolution basic parameters to obtain a simulated evolution process;

[0060] Step S53: Perform parameter corresponding mapping on the evolution basic parameters and the simulated evolution process to generate a parameter-process correspondence relationship;

[0061] Step S54: Perform resource pressure mapping on the simulated evolution process based on the resource pressure distribution matrix to generate a pressure evolution process;

[0062] Step S55: Deduce defect parameters for the parameter-process correspondence relationship according to the pressure evolution process to obtain resource pressure defects;

[0063] Step S56: Optimize and adjust the evolution basic parameters according to the resource pressure defects to generate optimized development parameters.

[0064] The present invention ensures the maximization of resource utilization efficiency by setting the upper limit of resource occupancy to 70%-95% of computing resources and 80%-98% of storage resources, performs resource bottleneck mapping on the stress test results based on the upper limit of resource occupancy, identifies potential resource constraint conflicts, obtains resource constraint conflict data, provides a basis for subsequent analysis. The associated load decomposition of the resource constraint conflict data helps to more carefully understand the distribution and impact of resource pressure, constructs a resource pressure distribution matrix, provides a comprehensive view of the resource usage status. The evolution process simulation of the evolution basic parameters shows the performance and changes of the parameters in different scenarios. The parameter corresponding mapping generates a parameter-process correspondence relationship, clarifying the correspondence between each parameter and its evolution state. Performing resource pressure mapping on the simulated evolution process based on the resource pressure distribution matrix enhances the intuitive understanding of the pressure impact, generates a pressure evolution process. Deducing defect parameters for the parameter-process correspondence relationship according to the pressure evolution process identifies existing resource pressure defects. Optimizing and adjusting the evolution basic parameters according to the resource pressure defects improves the adaptability and efficiency of the parameters, and finally generates optimized development parameters, overall enhancing the performance and stability of the system, and providing a more reliable basis for subsequent development.

[0065] Preferably, step S55 includes the following steps:

[0066] Step S551: Perform parameter adaptability backtesting on the pressure evolution process based on the parameter-process correspondence relationship, and screen out abnormal parameters according to the execution time delay floating range of 0.05s - 3s to generate a preliminary abnormal parameter set;

[0067] Step S552: Trace the resource load of the preliminary abnormal parameter set, and screen out the parameters affected by specific resource constraints to obtain a load-sensitive parameter set;

[0068] Step S553: Calculate the bottleneck propagation path for the load-sensitive parameter set and extract key restricted parameters; extract the parameter critical tolerance for the restricted parameter mapping table, and combine the availability range of each parameter under different pressure environments, where the floating point type is 0.001 - 1000 and the integer type is 1 - 5000;

[0069] Step S554: Calculate the parameter tolerance threshold based on the parameter critical tolerance and the availability range to obtain a parameter tolerance set;

[0070] Step S555: Perform defect feature matching on the key restricted parameters based on the parameter tolerance set to obtain resource pressure defects.

[0071] The present invention performs parameter adaptability backtesting on the pressure evolution process based on the parameter-process correspondence relationship to ensure an in-depth understanding of parameter behavior, screens out abnormal parameters according to the execution time delay floating range, timely identifies potential risks, generates a preliminary abnormal parameter set, provides a basis for subsequent analysis, traces the resource load of the preliminary abnormal parameter set, screens out the parameters affected by specific resource constraints to obtain a load-sensitive parameter set, enhances the understanding of the vulnerable links of the system, calculates the bottleneck propagation path for the load-sensitive parameter set, extracts key restricted parameters, clarifies the main factors restricting performance, extracts the parameter critical tolerance for the restricted parameter mapping table, and combines the availability range of each parameter under different pressure environments to ensure the effective evaluation of parameters, calculates the parameter tolerance threshold based on the parameter critical tolerance and the availability range to obtain a parameter tolerance set, improves the grasp of parameter stability, performs defect feature matching on the key restricted parameters based on the parameter tolerance set, identifies resource pressure defects, provides specific basis for system optimization and adjustment, overall improves the reliability and performance of the system, and lays a foundation for efficient Internet software development management.

[0072] The present invention also provides an Internet software development management system for executing the Internet software development management method as described above. The Internet software development management system includes:

[0073] A requirement collection module for collecting a collection of original functional requirements; performing functional clustering analysis on the collection of original functional requirements to obtain classified requirement functional characteristics; performing tagging processing on the classified requirement functional characteristics to generate independent requirement units;

[0074] A requirements analysis module, which is used to resolve the dependencies between independent requirement units to obtain a collection of dependency relationships, and construct a requirements logic structure based on the collection of dependency relationships.

[0075] A framework modeling module, which is used to interweave virtual and real frameworks for independent requirement units based on the requirements logic structure to obtain a requirements framework matrix, and perform organizational network simulation according to the requirements framework matrix to generate a requirements connection network.

[0076] A code evolution module, which is used to perform code reverse evolution according to the requirements connection network, and perform basic parameter matching to obtain evolution basic parameters, and perform stress simulation tests on the evolution basic parameters to generate stress test results.

[0077] An optimization and regulation module, which is used to capture resource pressure defects of evolution basic parameters based on stress test results, and optimize and adjust evolution basic parameters to generate optimized development parameters.

[0078] The present invention collects an original functional requirements collection through a requirements collection module and performs functional clustering analysis to generate independent requirement units, realizes the systematicness and accuracy of requirements collection, promotes the effective management and analysis of functional requirements, resolves dependencies of independent requirement units through a requirements analysis module, constructs a requirements logic structure, realizes the clarification of dependency relationships between requirements, optimizes the requirements logic structure to support dynamic adjustment, interweaves virtual and real frameworks for independent requirement units based on the requirements logic structure through a framework modeling module to generate a requirements framework matrix, and performs organizational network simulation, realizes the efficient generation of the requirements framework matrix, promotes team collaboration and information sharing, performs code reverse evolution according to the requirements connection network through a code evolution module, performs basic parameter matching and stress simulation tests, realizes the improvement of code evolution efficiency, ensures the accuracy of basic parameter matching, captures resource pressure defects based on stress test results through an optimization and regulation module, optimizes evolution basic parameters, realizes the improvement of resource utilization efficiency, reduces development costs and risks, overall improves the efficiency and quality of software development, and ensures the successful delivery and sustainable development of projects. Brief Description of the Drawings

[0079] Figure 1 It is a schematic diagram of the step flow of an Internet software development management method.

[0080] Figure 2 It is a schematic diagram of the detailed implementation step flow of step S2.

[0081] The realization, functional features and advantages of the object of the present invention will be further described in conjunction with embodiments with reference to the drawings. Detailed Embodiments

[0082] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0083] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0084] It should be understood that although terms such as "first" and "second" may be used here to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit can be called the second unit, and similarly the second unit can be called the first unit. The term "and / or" used here includes any and all combinations of one or more of the listed related items.

[0085] To achieve the above object, please refer to Figures 1 to 2 , an Internet software development management method, including the following steps:

[0086] Step S1: Collect the original function requirement set; perform function clustering analysis on the original function requirement set to obtain the classified requirement function features; perform tagging processing on the classified requirement function features to generate independent requirement units;

[0087] Step S2: Perform inter-unit dependency parsing on the independent requirement units to obtain a set of dependency relationships; construct a requirement logic structure based on the set of dependency relationships;

[0088] Step S3: Interweave the virtual and real frameworks of the independent requirement units based on the requirement logic structure to obtain a requirement framework matrix; perform organizational network simulation according to the requirement framework matrix to generate a requirement connection network;

[0089] Step S4: Perform code reverse evolution according to the requirement connection network and perform basic parameter matching to obtain evolution basic parameters; perform stress simulation testing on the evolution basic parameters to generate stress test results;

[0090] Step S5: Capture the resource pressure defects of the evolution basic parameters based on the stress test results; optimize and adjust the evolution basic parameters according to the resource pressure defects to generate optimized development parameters.

[0091] In the present invention, by collecting the original function requirement set and performing function clustering analysis to generate independent requirement units, the systematicness and accuracy of requirement collection are realized, the effective management and analysis of function requirements are promoted. By performing dependency parsing on the independent requirement units to construct a requirement logic structure, the clarity of the dependency relationship between requirements is realized, and the requirement logic structure is optimized to support dynamic adjustment. By performing virtual and real framework interweaving on the independent requirement units based on the requirement logic structure to generate a requirement framework matrix and performing organizational network simulation, the efficient generation of the requirement framework matrix is realized, promoting team collaboration and information sharing. By performing code reverse evolution according to the requirement connection network, performing basic parameter matching and stress simulation testing, the improvement of code evolution efficiency is realized, ensuring the accuracy of basic parameter matching. By capturing resource pressure defects based on the stress test results and optimizing the evolution basic parameters, the improvement of resource utilization efficiency is realized, reducing development costs and risks, and overall improving the efficiency and quality of software development, ensuring the successful delivery and sustainable development of the project.

[0092] In an embodiment of the present invention, the method for managing Internet software development includes the following steps:

[0093] Step S1: Collect the original function requirement set; perform function clustering analysis on the original function requirement set to obtain the classified requirement function features; perform tagging processing on the classified requirement function features to generate independent requirement units;

[0094] In this embodiment, when collecting the original functional requirement set, a requirements management tool is used to collect all Internet software development requirements. The data formats collected include JSON (JavaScript Object Notation), XML (Extensible Markup Language), and YAML (YAML Ain’t Markup Language). The keywords in the requirement document are parsed to extract the names of involved functional modules, interface descriptions, performance metrics, etc., to ensure the integrity and structuring of the data. After completing the collection of the original functional requirement set, the TF-IDF (Term Frequency-Inverse Document Frequency) is used to calculate the keyword weights of the requirement text, and a hierarchical clustering algorithm is used to perform functional clustering analysis on the requirement text to obtain the classified requirement functional features. During the clustering process, the similarity threshold is set to 0.85, and the cosine similarity is used to measure the similarity between requirement texts. Requirement texts with similarity exceeding the threshold are classified into the same functional feature. After the functional feature classification is completed, a regularization method is used to perform tagging processing on the classified requirement functional features, converting them into independent requirement units in a unified format. Each independent requirement unit includes a function name, input parameters, output parameters, constraint conditions, and applicable scenarios. The function name uses the camel case naming method, and the input and output parameters are stored in JSON format with a length limit of 255 characters.

[0095] Step S2: Perform inter-unit dependency parsing on the independent requirement units to obtain a set of dependency relationships; construct a requirement logic structure based on the set of dependency relationships;

[0096] In this embodiment, when performing inter-unit dependency parsing on the independent requirement units, the depth-first search (DFS) is used to traverse the requirement units to construct an inter-unit dependency graph. First, the input-output relationships of the independent requirement units are extracted. The input parameter matching degree threshold is set to 0.9, and the dependency weights between adjacent requirement units are calculated. Dependency connections are established for requirement units with dependency weights greater than the threshold. After obtaining the set of dependency relationships, a topological sorting algorithm is used to construct the requirement logic structure to determine the execution order of the requirement units. A cyclic dependency detection mechanism is set. If a cyclic dependency is detected, the relevant units are split and reconstructed. After the requirement logic structure is constructed, the set of dependency relationships is output, and a visualization tool (such as Graphviz) is used to generate a dependency graph.

[0097] Step S3: Interweave the virtual and real frameworks for the independent requirement units based on the requirement logic structure to obtain a requirement framework matrix; perform an organizational network simulation according to the requirement framework matrix to generate a requirement connection network;

[0098] In this embodiment, when interweaving the virtual and real frameworks for the independent requirement units based on the requirement logic structure, first sort the requirement units according to the function priority, set the priority range from 1 to 10, and the larger the value, the higher the priority. After sorting, use a two-way mapping table to record the correspondence between the virtual function and the actual function, where the virtual function represents a function node that has not been specifically implemented but occupies a position in the requirement framework, and the actual function represents a function module that has completed the specific design. During the interweaving process, construct a requirement framework matrix, and use a sparse matrix to represent the mapping relationship between requirement units, where the matrix row represents the virtual function and the matrix column represents the actual function. During the interweaving process, for the unimplemented function points, use a rule-based completion method to generate placeholders to ensure the integrity of the requirement framework. After the requirement framework matrix is constructed, perform an organizational network simulation, and use a network structure based on the small-world model to generate a requirement connection network, where the node degree distribution is set to conform to the power-law distribution to simulate the collaborative relationship between requirement units.

[0099] Step S4: Perform code reverse evolution according to the requirement connection network and perform basic parameter matching to obtain evolution basic parameters; perform a stress simulation test on the evolution basic parameters to generate a stress test result;

[0100] In this embodiment, when performing code reverse evolution according to the requirement connection network, use a code parsing tool based on AST (Abstract Syntax Tree) to analyze the function call relationship of the existing code library, extract the core code fragments, and map them to the requirement units. During the basic parameter matching process, use an automatic variable binding method to align the parameter structure of the requirement units with the input and output parameters of the code fragments, and set the parameter matching similarity threshold to 0.85. After the parameter matching is completed, perform a stress simulation test, use JMeter (Apache JMeter, an open-source performance testing tool) to perform a concurrent stress test, set the number of concurrent users to 1000 - 5000, and simulate the system response in different stress environments. During the test, record key performance indicators such as CPU usage, memory occupancy, and network bandwidth, and generate a stress test result according to the stress change situation. The test result is stored in JSON format, including indicators such as test timestamp, response time, throughput, and error rate.

[0101] Step S5: Capture the resource stress defects of the evolution basic parameters based on the stress test results; optimize and adjust the evolution basic parameters according to the resource stress defects to generate optimized development parameters.

[0102] In this embodiment, when capturing the resource pressure defects of the evolution basic parameters based on the pressure test results, a time series analysis-based method is adopted to calculate the change trend of system resource occupancy, and a resource occupancy threshold is set. The upper limit of the computing resource is set to 70%-95%, and the upper limit of the storage resource is set to 80%-98%. For the resource occupancy situation exceeding the threshold, resource bottleneck mapping is performed to construct a resource pressure matrix. The K-Means clustering algorithm is used to classify the resource usage patterns to determine the high-load areas. After the resource pressure defects are identified, the evolution basic parameters are optimized and adjusted according to the resource pressure defects. The optimization goal is set to reduce the resource occupancy peak and improve the system throughput. The genetic algorithm (Genetic Algorithm, genetic optimization algorithm) is used for parameter optimization, the input parameter range is adjusted, the population size is set to 100, and the number of iterations is set to 500. The optimized parameters are verified again.

[0103] Preferably, step S1 includes the following steps:

[0104] Step S11: Collect the original function requirement set; perform semantic segmentation on the original function requirement set to obtain independent requirement segments;

[0105] Step S12: Extract the function features of the independent requirement segments; perform clustering analysis on the function features of each independent requirement segment to obtain the classified requirement function features;

[0106] Step S13: Eliminate the redundant features of the classified requirement function features to obtain the clean requirement functions;

[0107] Step S14: Based on the classified requirement function features, perform unique label assignment on the clean requirement functions to generate independent requirement units.

[0108] In this embodiment, when collecting the original functional requirement set, a text parsing tool based on natural language processing (NLP, Natural Language Processing) is used to parse the Internet software development requirement document. The parsed text formats include JSON (JavaScript Object Notation), XML (Extensible Markup Language), and Markdown (a lightweight markup language). Regular expressions are used to match key information such as functional descriptions, interface definitions, and data flows in the requirement document, and the extracted data is stored in a structured manner. The storage format adopts the key-value pattern, where the key corresponds to the requirement name and the value corresponds to the functional description text. The requirement document is batch-parsed using a multi-threaded processing method, and the maximum number of parsing tasks for each thread is set to 500. After parsing, it is classified and stored according to the source, timestamp, and version information of the requirement document, and the original functional requirement set is generated. When performing semantic segmentation on the original functional requirement set, a semantic segmentation method based on the BERT (Bidirectional Encoder Representations from Transformers) model is used to perform sentence splitting on the requirement text and identify the functional keywords in the requirement text. The minimum segmentation granularity is set to 15 characters, and the maximum segmentation granularity is set to 80 characters. The LDA (Latent Dirichlet Allocation) topic modeling algorithm is used to classify the topics of the requirement text, and the text is finely segmented based on the topic categories. After segmentation, the segmentation results are annotated, and the feature information such as the topic category, keywords, and length of each requirement segment is recorded. When extracting the functional features of independent requirement segments, the TF-IDF (Term Frequency-Inverse Document Frequency) method is used to calculate the keyword weights of each requirement segment, and the semantic similarity between requirement segments is calculated based on the Word2Vec (Word to Vector) model. The similarity calculation window size is set to 5, and the semantic vector dimension is set to 300 dimensions. After calculation, the top 10 words with the highest keyword weights are selected as functional features and stored in the functional feature vector table. When performing cluster analysis on the functional features of each independent requirement segment, the K-Means clustering algorithm is used to classify the functional features. The initial number of cluster centers is set to 10, and the Euclidean Distance is used to calculate the distance between functional features. During the clustering process,Dynamically adjust the position of the clustering center to match the functional feature points with the nearest clustering center. When the error change rate of the clustering result is lower than 0.01, stop the iteration and generate the classified demand functional features. When removing the redundant features of the classified demand functional features, use the Mutual Information method to calculate the information gain between different functional features, and remove the functional features with an information gain lower than 0.05. At the same time, use the PCA (Principal Component Analysis) method to perform dimensionality reduction on the functional features, set the cumulative contribution rate threshold of the principal components to 95%, and retain the main functional feature vectors. After the dimensionality reduction process, perform feature deduplication, calculate the similarity between functional feature vectors using the Jaccard similarity coefficient, and merge the functional features with a similarity exceeding 0.9. During the merging process, retain the functional features with higher weights and remove the redundant features. Finally, obtain the cleaning demand functions. When assigning unique labels to the cleaning demand functions based on the classified demand functional features, use the Hash Encoding method to generate unique identifiers. The unique label of each cleaning demand function consists of fields such as function name, category number, and timestamp. The label format is generated in the UUID (Universally Unique Identifier) manner and stored in the demand management database. During the unique label assignment process, use SHA-256 (Secure Hash Algorithm 256) to calculate the hash value of the functional features to ensure the uniqueness of the labels. For the demand functions with a hierarchical structure, use a tree-like label structure for management. The parent label and child label of each demand function are connected through a hash mapping relationship. After the unique label assignment is completed, generate independent demand units.

[0109] Preferably, step S2 includes the following steps:

[0110] Step S21: Decompose the independent demand units hierarchically to obtain the demand hierarchical structure; detect the lead-follow relationship of the independent demand units according to the demand hierarchical structure to generate a candidate set of unit dependencies;

[0111] Step S22: Perform cross-validation filtering on the candidate set of unit dependencies to obtain high-confidence dependency relationships; perform a transitive closure operation on the high-confidence dependency relationships to generate a dependency transmission chain;

[0112] Step S23: Detect the network loop of the dependency transmission chain; perform dependency resolution based on the network loop of the dependency transmission chain to obtain a set of dependency relationships;

[0113] Step S24: Construct a directed acyclic graph structure based on the dependency relationship set, perform requirement topological sorting based on the directed acyclic graph structure to obtain a requirement topological sequence; perform priority weighting processing on the requirement topological sequence to generate a requirement logical structure.

[0114] In this embodiment, when hierarchically deconstructing the independent requirement units, a tree parsing method is used to structurally disassemble the independent requirement units. First, syntactic analysis is performed on the functional description text of the independent requirement units. The dependency parsing method is used to identify the subject-predicate-object structure, and the action verbs and key nouns are extracted. Subsequently, based on the constituency parsing method, the subordinate relationships are further identified, and the requirements are stratified according to the functional granularity. The requirement units with the same functional category are grouped into the same level. For the requirement units with a parent-child relationship, a nested storage method is used for organization, and the storage format is in JSON (JavaScript Object Notation) format. After parsing, a requirement hierarchy structure is generated and numbered according to the hierarchical relationship. The numbering format is set as "Lx-Ny", where "L" represents the level number and "N" represents the unit serial number. When detecting the lead-follow relationship of the independent requirement units according to the requirement hierarchy structure, a directed graph modeling method is adopted. First, each independent requirement unit is mapped to a node of the directed graph, and the input and output data of each requirement unit are extracted. The data flow analysis method is used to identify the data dependency relationship. For the case where the output data matches the input data of the subsequent requirement unit, a lead-follow relationship is established, and the data transfer path is recorded. Subsequently, based on the control flow analysis method, the control dependency relationship is identified. For the requirement units with pre-execution conditions, a control dependency relationship is established, and the control conditions are stored as the weight information of the edge. All the identified dependency relationships are stored in the unit dependency candidate set. When performing cross-validation filtering on the unit dependency candidate set, the consistency check method is used to verify the logical correctness of the dependency relationship. First, the dependency paths of each requirement unit are traversed, and the logistic regression model is used to calculate the confidence level of the dependency relationship. The confidence level threshold is set to 0.8. For the dependency relationships below the threshold, manual review is performed, and the unreasonable dependency paths are deleted. Subsequently, the Bootstrapping method is used to sample and verify the dependency paths. Each time, 80% of the dependency paths are randomly selected for training, and the cross-validation score is calculated. For the dependency paths with a score below 0.75, re-evaluation is performed. Finally, the high-confidence dependency relationships are selected. When performing the transitive closure operation on the high-confidence dependency relationships, the Floyd-Warshall algorithm is used to calculate the shortest paths between all nodes, and the transitive relationship of the requirement units with indirect dependency relationships is complemented. First, the dependency relationship matrix is initialized. For the requirement units with direct dependency relationships,Mark the path weight as 1 in the matrix. For requirement units without direct dependencies, mark the path weight as infinity. Subsequently, iterate and update each intermediate node, calculate all path combinations, and update the shortest path matrix. For the calculated transfer paths, establish new dependencies and store them in the dependency transfer chain. When detecting network loops in the dependency transfer chain, use the topological sorting method to check the cycle structure of the directed graph. First, use the depth-first search (DFS) method to traverse the dependency transfer chain and record the access status of each node. When a node that has already been visited is visited again, it is judged as a loop structure and the loop path is recorded. Subsequently, perform backtracking analysis on the detected loop, calculate the execution priority of each dependency path in the loop, mark the deletion of the dependency path with the lowest priority, and store it in the network loop list. When performing dependency resolution based on the network loop of the dependency transfer chain, use the hierarchical resolution method to disassemble the loop dependencies. First, perform hierarchical analysis on the loop path, use the deepest dependency path as the parsing starting point, and check for duplicate dependency relationships. For duplicate dependency relationships, use the path pruning method to delete them. For non-duplicate dependency relationships, use the dependency merging method to simplify them, merge and store similar dependency paths, and finally form a dependency relationship collection. When constructing a directed acyclic graph structure based on the dependency relationship collection, use the topological sorting method to sort the requirement units. First, count the in-degree of all requirement units. The requirement units with an in-degree of 0 are used as the starting nodes, traverse all requirement units in sequence, and construct directed edges according to the dependency relationships. For requirement units with multiple dependency paths, use the weight calculation method to determine the main dependency path and adjust the weight of the dependency edge to ensure that the graph structure is acyclic. Finally, generate a directed acyclic graph structure. When performing requirement topological sorting based on the directed acyclic graph structure, use Kahn's algorithm for topological sorting. First, add all nodes with an in-degree of 0 to the sorting queue, and sequentially take out the requirement units in the queue, decrement the in-degree of their dependent subsequent requirement units by 1. For requirement units whose in-degree drops to 0, add them to the sorting queue until all requirement units are sorted. After sorting is completed, generate a requirement topological sequence. When performing priority weighting processing on the requirement topological sequence, use the weighted scoring method to calculate the priority of each requirement unit. First, normalize indicators such as the number of dependency paths, execution order, and resource occupancy of each requirement unit and assign different weights. Among them, the weight of the number of dependency paths is set to 0.4, and the weight of the execution order is set to 0.3,The weight of resource occupancy is set to 0.3. Subsequently, the comprehensive score is calculated for each requirement unit, and the priority ranking is carried out according to the score. Finally, the requirement logic structure is generated.

[0115] Preferably, the interweaving of the virtual and real frameworks for the independent requirement units in step S3 includes:

[0116] Perform a dot matrix conversion on the requirement logic structure to obtain logical dot matrix mapping data;

[0117] Perform a semantic vector encoding mapping on the independent requirement units to generate requirement semantic vectors;

[0118] Calculate the similarity distance of the requirement semantic vectors to obtain requirement affinity data;

[0119] Perform an orthogonal projection fusion on the logical dot matrix mapping data and the requirement affinity data to obtain a requirement position matrix;

[0120] Based on the requirement position matrix, divide the virtual and real boundaries to generate virtual and real partition boundaries;

[0121] Based on the virtual and real partition boundaries, perform a framework attribute mapping on the independent requirement units to obtain framework characteristic data;

[0122] According to the framework characteristic data, perform an interweaving simulation of the mapping rules to obtain a requirement framework matrix.

[0123] In this embodiment, when performing dot matrix conversion on the requirement logic structure, a sparse matrix modeling method is adopted. The requirement units in the requirement logic structure are regarded as the row indices of the matrix, the dependency relationships between the requirement units are regarded as the column indices, and the binary values are used to represent the relevance between the requirement units. For the requirement units with direct dependency relationships, the value 1 is filled in the corresponding positions of the matrix; for the requirement units without direct association, the value 0 is filled. Subsequently, the space coordinate mapping method is used to convert the sparse matrix into dot matrix data, and the three-dimensional coordinates (X, Y, Z) are used to represent the position of each requirement unit in the logic structure, where the X-axis represents the level of the requirement unit, the Y-axis represents the dependency breadth of the requirement unit, and the Z-axis represents the execution order of the requirement unit. After the conversion is completed, the logical dot matrix mapping data is generated. When performing semantic vector encoding mapping on the independent requirement units, the BERT (Bidirectional Encoder Representations from Transformers) model is adopted to extract semantic features from the text description of the requirement units. First, the text content of the requirement unit is tokenized, stop words are removed, and the WordPiece method is used for sub-word splitting. Subsequently, the processed text is input into the BERT model for feature extraction. The dimension of the extracted feature vector is set to 768, and the max pooling method is used to aggregate all word vectors to form the global semantic vector of a single requirement unit. Finally, the semantic vectors of all requirement units are stored in a matrix format, and the requirement semantic vector data is generated. When calculating the similarity distance of the requirement semantic vectors, the cosine similarity method is adopted to calculate the semantic similarity degree between the requirement units. First, the semantic vectors of all requirement units are normalized so that the vector norm is normalized to the unit length. Subsequently, the cosine value of the angle between any two requirement unit vectors is calculated, and the value range is set between 0 and 1, where 1 indicates completely the same and 0 indicates completely irrelevant. After the calculation is completed, the requirement affinity data is generated and stored in a symmetric matrix form, where the row and column indices of the matrix correspond to the requirement unit numbers, and the matrix element values represent the semantic similarity between two requirement units. When performing orthogonal projection fusion on the logical dot matrix mapping data and the requirement affinity data, the principal component analysis (PCA) method is adopted to reduce the dimension of the logical dot matrix data, map the three-dimensional space coordinates to a two-dimensional plane, and maintain the maximum amount of information. Subsequently, the orthogonal projection method is used to project the requirement affinity data onto the reduced-dimensional logical dot matrix plane. When projecting, the weighted inner product method is used to calculate the position of the projection point, where the coordinate values of the logical dot matrix data are set as the main reference, and the similarity value of the requirement affinity data is set as the projection weight. After the calculation is completed,The final projected positions of all demand units are stored as a two-dimensional coordinate matrix, and a demand position matrix is generated. When dividing the virtual and real boundaries based on the demand position matrix, the K-Means clustering method is used to classify the demand units. First, the value of K is set to 2 to distinguish virtual demand units from actual demand units. Subsequently, the demand position matrix is normalized so that all coordinate values are normalized to between 0 and 1, and the Euclidean Distance is used as the distance metric standard to perform K-Means clustering training on all demand units. After the training is completed, the class labels of the cluster centers are set as the virtual demand unit class and the actual demand unit class, and all demand units are classified according to their respective classes. Finally, the virtual and real partition boundaries are generated and stored as a binary marking matrix, where 0 represents virtual demand units and 1 represents actual demand units. When performing frame attribute mapping on independent demand units based on the virtual and real partition boundaries, a Multilayer Perceptron (MLP) model is used to classify and predict the frame attributes of the demand units. First, the demand semantic vector and the demand position matrix are used as input features, and the virtual and real partition boundaries are added as class labels. Subsequently, a three-layer MLP model is constructed, where the input layer is set to have a dimension of 1024, the hidden layer uses the ReLU (Rectified Linear Unit) activation function, and the number of hidden neurons is set to 512. Finally, the Softmax function is used in the output layer for frame attribute classification, and the classification categories include "business logic layer", "data storage layer", "interface call layer", and "interaction display layer". After the training is completed, the frame attributes of all demand units are predicted, and the classification results are stored. Finally, frame feature data is generated. When performing interweaving simulation of mapping rules based on the frame feature data, the Rule-Based Inference method is used to perform interweaving calculations on the frame characteristics of the demand units. First, frame mapping rules are defined, including hierarchical inheritance rules, functional dependency rules, and module nesting rules. Subsequently, all demand units are matched according to the frame feature data. For demand units that satisfy the hierarchical inheritance rules, the frame inheritance is performed in a Top-Down manner. For demand units that satisfy the functional dependency rules, the attribute mapping is performed in a Cross-Dependency manner. For demand units that satisfy the module nesting rules, the merging is performed in a Bottom-Up manner. Finally, after the calculation is completed, the interweaving mapping results of all demand units are stored in a matrix format, and a demand frame matrix is generated.,

[0124] Preferably, the organizational network simulation according to the demand frame matrix in step S3 includes:

[0125] Evaluate the connection strength of the requirements framework matrix, and allocate node weights based on the connection strength to obtain node connection weight data;

[0126] Construct a multi-layer network topology structure based on the node connection weight data, and draw a topology diagram to generate a network topology sketch;

[0127] Perform node clustering reconstruction on the network topology sketch according to the node connection weight data to obtain reconstructed clustering nodes;

[0128] Perform domain-oriented convergence mapping based on the reconstructed clustering nodes to obtain a functional aggregation domain;

[0129] Perform organizational network simulation according to the functional aggregation domain to generate a requirements connection network.

[0130] In this embodiment, when evaluating the connection strength of the requirements framework matrix, a weighted adjacency matrix is used to represent the connection relationship between requirement units. The row and column indices of the matrix correspond to the requirement unit numbers, and the matrix element values represent the association strength between requirement units. First, the dependence times are calculated based on adjacent requirement units in the requirements framework matrix. If there is a direct dependence between two requirement units, the basic weight value is set to 1. If there is an indirect dependence between two requirement units, the weight is calculated by attenuation according to the length of the dependence path, and the backpropagation mechanism is used to update the incremental weights of all requirement units on the dependence path. Subsequently, the out-degree and in-degree of each requirement unit are calculated, and the PageRank algorithm is used to normalize the global influence of all requirement units. Finally, the node connection weight data is generated. When constructing a multi-layer network topology structure based on the node connection weight data, a graph neural network (GNN) modeling method is adopted. First, the node connection weight data is used as input features, and a multi-layer graph structure is constructed. The first-layer network structure is used to represent the direct connection relationship of requirement units, the second-layer network structure is used to represent the functional dependence relationship of requirement units, and the third-layer network structure is used to represent the execution order relationship of requirement units. Subsequently, the adjacency matrix of the multi-layer network is normalized so that the weight values of all edges are normalized to between 0 and 1, and the Fruchterman-Reingold layout algorithm is used to project the multi-layer network into a two-dimensional space to ensure uniform distribution of the nodes in the topology structure, so as to generate a network topology sketch. When reconstructing the node clustering of the network topology sketch according to the node connection weight data, a hierarchical clustering method is used to group the requirement units. First, the similarity distance between all nodes is calculated, and the minimum squared error increment within the cluster is calculated based on the Ward's Linkage method. Subsequently, a tree-like hierarchical clustering structure is constructed, and an adaptive pruning method is used to set the clustering division threshold. The clustering division threshold is set as the global optimal segmentation point to maximize the average connection strength between requirement units within each cluster, and the reconstructed clustering nodes are generated. When performing domain-oriented convergence mapping based on the reconstructed clustering nodes, a principal component analysis (PCA) method is used to reduce the dimensionality of the features of the clustering nodes. First, the feature mean vector within each cluster is calculated, and the covariance matrix of all feature vectors is calculated. Subsequently, the eigenvalue decomposition of the covariance matrix is performed, and the principal components with a cumulative variance contribution rate reaching 95% are selected for mapping transformation. Finally, all clustering nodes are projected into a low-dimensional space, and the optimal functional aggregation domain is calculated based on the spatial density of the projection points.Among them, the boundary of the function aggregation domain is automatically fitted using the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, and a function aggregation domain is generated. When performing an organizational network simulation based on the function aggregation domain, the random walk method is used to simulate the dynamic relationships of demand units. First, a Markov Decision Process (MDP) is constructed based on the function aggregation domain, and a state transition probability matrix is defined. The state space consists of demand units in the function aggregation domain, and the transition probability is determined by the connection strength between demand units. Subsequently, a jump probability is introduced during the state transition process, and the Gibbs Sampling method is used to iteratively update the state transition path, enabling the entire demand connection network to form a stable convergence structure, and finally generating a demand connection network.

[0131] Preferably, in step S4, the code reverse evolution based on the demand connection network and the basic parameter matching include:

[0132] Perform a demand function mapping on the demand connection network to obtain function module decomposition data;

[0133] Expand the dependency conduction of the function module decomposition data, and perform hierarchical splitting according to a conduction probability of 0.3 - 0.9 to extract deep dependency paths and generate recursive association links;

[0134] Perform a direction induction on the recursive association links to obtain the main code nodes;

[0135] Perform a logical reverse deduction based on the function module decomposition data and the main code nodes to generate a code mapping skeleton;

[0136] Perform a structure replacement process on the code mapping skeleton according to a preset code library to obtain an evolved code framework;

[0137] Perform a reverse function derivation on the evolved code framework to generate a code evolution trajectory;

[0138] Perform a call chain disassembling on the code evolution trajectory to obtain parameter passing relationships;

[0139] Extract the boundary constraints of the parameter passing relationships, where the range of values that can be taken by the key variables is extracted, and upper and lower limits are set, namely 0.001 - 1000 for floating-point types and 1 - 5000 for integer types, to obtain a parameter constraint set;

[0140] Perform a dynamic assignment derivation on the parameter constraint set to obtain a parameter transformation mapping;

[0141] Perform basic parameter matching on the parameter transformation mapping to obtain the evolved basic parameters.

[0142] In this embodiment, when performing demand function mapping on the demand connection network, a deep neural network (DNN) is used to train the correspondence between demand units and function modules. First, the demand connection network is converted into graph-structured data, where nodes represent demand units and edges represent functional associations between demand units. Subsequently, each node is encoded with word vectors, and the self-attention mechanism is used to extract the long-range dependence relationships between nodes. On this basis, a fully connected layer is used for feature mapping, and a Softmax classifier is used to predict the functional categories of demand units. Finally, the function modules are divided according to the prediction results and stored as function module decomposition data. When performing dependency conduction expansion on the function module decomposition data, a Bayesian network modeling method is used to construct a conduction probability model. First, the function module decomposition data is topologically sorted, and the direct dependencies of each function module are calculated. Subsequently, conditional probability calculations are performed on all dependencies, where the basic conduction probability is set to 0.3 - 0.9, and the expectation maximization (EM) algorithm is used to iteratively update the probability parameters. Finally, a recursive association link for all function modules is generated. When performing direction induction on the recursive association link, a path tracing method is used to analyze the code generation direction. First, each path in the recursive association link is traced backward, and the difference between the out-degree and in-degree of each node is calculated. Among them, nodes with an in-degree greater than the out-degree are marked as dependency receiving nodes, and nodes with an out-degree greater than the in-degree are marked as dependency propagation nodes. Subsequently, the average direction vector of the dependency propagation paths is calculated, and clustering analysis is performed on all paths. Finally, the main code nodes are determined. When performing logical reverse inference based on the function module decomposition data and the main code nodes, a logistic regression model is used to model the code generation rules. First, variable definition and function call information are extracted from the main code nodes, and the transfer path of the variable is traced using backward data flow analysis. Subsequently, a code generation rule matrix is constructed, where rows represent code structures and columns represent logical relationships. Finally, Gaussian elimination is used for logical deduction, and a code mapping skeleton is generated. When performing structure replacement processing on the code mapping skeleton according to a preset code library, a template matching method is used for code replacement. First, the code mapping skeleton is parsed by an AST (abstract syntax tree), and the key node information in the code structure is extracted. Subsequently, code fragments with matching structures are searched in the code library, and the matching similarity is calculated.Among them, the matching similarity is calculated using the Jaccard similarity. Finally, structure replacement is performed based on the code snippet with the highest matching degree and stored as an evolved code framework. When performing reverse function derivation on the evolved code framework, the data dependency analysis method is used to parse the function call relationship. First, all function definitions and call positions are extracted, and a function call graph (FCG) is constructed. Subsequently, static analysis is performed on each function call path, and the transfer link of function parameters is calculated. Finally, a code evolution trajectory is generated according to the function call order and stored as time series data. When disassembling the call chain of the code evolution trajectory, the dynamic symbolic execution method is used to analyze the parameter transfer relationship. First, binary parsing is performed on the code evolution trajectory, and all function call instructions are extracted. Subsequently, the operands of each instruction are symbolized, and the constraint solving method is used to deduce the parameter transfer rules. Finally, the parameter transfer relationship is generated and stored as graph structure data. When extracting the boundary constraints of the parameter transfer relationship, the statistical data analysis method is used to calculate the range of values that variables can take. First, data sampling is performed on all variables in the parameter transfer relationship, and the maximum value, minimum value, mean value, and standard deviation of the variables are calculated. Subsequently, the variable types are classified. The value range of floating-point variables is set to 0.001 - 1000, and the value range of integer variables is set to 1 - 5000. Finally, the boundary constraint data is stored, and a parameter constraint set is generated. When performing dynamic assignment derivation on the parameter constraint set, the Monte Carlo method is used to simulate parameter transformation. First, random sampling is performed on each variable in the parameter constraint set, and N groups of random assignment data are generated. Subsequently, function calculations are performed on each group of assignment data, and the calculation results are recorded. Finally, the regression analysis method is used to fit the parameter transformation curve and store it as a parameter transformation mapping. When performing basic parameter matching on the parameter transformation mapping, the least squares optimization method is used to calculate the optimal matching parameters. First, numerical interpolation is performed on the parameter transformation mapping, and an objective function is constructed. The constraint conditions of the objective function are provided by the parameter constraint set. Subsequently, the gradient descent method is used to solve the optimal parameter combination. Finally, the evolved basic parameters are stored.,

[0143] Preferably, the pressure simulation test on the evolved basic parameters in step S4 includes:

[0144] Calculate the resource requirements for the evolution basic parameters to obtain the evolution required resources;

[0145] Map the evolution required resources to resource constraint data;

[0146] Based on the resource constraint data, conduct stress scenario modeling to obtain a resource stress test model;

[0147] According to the resource stress test model, conduct evolution simulation on the evolution basic parameters to generate stress test results.

[0148] In this embodiment, when calculating the resource requirements for the evolution basic parameters, a computing resource evaluation method is used to analyze the requirements of the CPU (Central Processing Unit), Memory, Storage, and Network. First, the computing task information in the evolution basic parameters is extracted, including compute-intensive tasks and I / O-intensive tasks, and different computing weights are set according to the task type. Subsequently, the execution time, concurrent request number, and data throughput of the tasks are measured. Among them, the CPU consumption of compute-intensive tasks is calculated according to the number of operations, and the storage consumption of I / O-intensive tasks is calculated according to the data read / write rate. Finally, combined with the benchmark performance data of the system resources, the resource requirements of various tasks are calculated, and the evolved required resources are output. When mapping the evolved required resources to resource constraint data, a normalization method is used for resource specification conversion. First, the evolved required resources are discretized. The CPU uses the unit of the number of cores, the memory uses the unit of GB, the storage uses the unit of MB / s, and the network bandwidth uses the unit of Mbps. Subsequently, according to the resource configuration information of the server cluster, the required resources are mapped to the standard resource specifications, and the resource utilization rate is calculated. The utilization rate calculation method is the ratio of the allocated resources to the total available resources. Finally, the mapped data is stored as resource constraint data. When building a stress scenario model based on the resource constraint data, a queuing theory modeling method is used to establish a resource stress test model. First, an M / M / 1 queuing model is constructed based on the task execution rate, concurrent request number, and average waiting time in the resource constraint data. The arrival rate λ (Lambda) and service rate μ (Mu) are obtained by fitting historical data. Subsequently, the traffic input curve is set according to different load scenarios, including low load (30% resource occupancy), medium load (60% resource occupancy), and high load (90% resource occupancy) scenarios, and the task processing situation is simulated under different scenarios. Finally, a resource stress test model is generated. When performing an evolution simulation on the evolution basic parameters according to the resource stress test model, a distributed stress test tool is used for simulation execution. First, a test script is generated according to the resource stress test model, and test parameters are set, including the number of concurrent users (Concurrent Users), request rate (Requests per Second), and test duration (Test Duration). Subsequently, JMeter (Apache JMeter, an open-source stress test tool) is used for distributed concurrent testing, and the CPU occupancy rate, memory usage rate, disk I / O rate, and network throughput of the system are monitored in real time. Finally, a stress test result is generated according to the test execution result.

[0149] Preferably, step S5 includes the following steps:

[0150] Step S51: Set the upper limit of resource occupancy to 70%-95% of computing resources and 80%-98% of storage resources; perform resource bottleneck mapping on the stress test results based on the upper limit of resource occupancy to obtain resource constraint conflict data;

[0151] Step S52: Perform associated load decomposition on the resource constraint conflict data and construct a resource pressure distribution matrix; perform an evolution process simulation on the evolution basic parameters to obtain a simulated evolution process;

[0152] Step S53: Perform parameter correspondence mapping on the evolution basic parameters and the simulated evolution process to generate a parameter-process correspondence relationship;

[0153] Step S54: Perform resource pressure mapping on the simulated evolution process based on the resource pressure distribution matrix to generate a pressure evolution process;

[0154] Step S55: Deduce defect parameters from the parameter-process correspondence relationship according to the pressure evolution process to obtain resource pressure defects;

[0155] Step S56: Optimize and adjust the evolution basic parameters according to the resource pressure defects to generate optimized development parameters.

[0156] In this embodiment, when setting the upper limit of resource occupancy, the available range of computing resources is limited between 70% and 95%, and the available range of storage resources is limited between 80% and 98%. First, call the historical stress test data, extract the CPU (Central Processing Unit) occupancy rate, Memory usage rate, disk IO (Input / Output) throughput, and Network traffic data under different load scenarios, and calculate the maximum resource consumption ratio. Subsequently, set the threshold interval for resource usage based on the Resource Bottleneck Theory. Finally, label the abnormal data in the stress test results based on the set upper limit of resource occupancy, collect the computing resource and storage resource data that exceed the set upper limit, and generate resource constraint conflict data. When disassembling the associated load for the resource constraint conflict data, use the Analytic Hierarchy Process,The Analytic Hierarchy Process (AHP) is used to hierarchically analyze conflict data. First, based on the computing resources, storage resources, IO throughput, and network traffic ratio involved in the conflict data, it is classified according to the resource consumption priority, and the main conflict resources are used as the first-level classification nodes. Subsequently, the load correlation of various resources is calculated, and the correlation coefficient calculation method is used to determine the dependence strength between resources, and the resources with high correlation are merged. Finally, a resource pressure distribution matrix is generated and stored in a two-dimensional matrix format. At the same time, the basic evolution parameters are simulated, and the time series analysis method is used to predict the future load change trend, and an evolution process model is fitted according to historical data. Finally, a simulated evolution process is obtained. When performing parameter correspondence mapping on the basic evolution parameters and the simulated evolution process, the hash mapping method is used to construct a parameter-process correspondence relationship. First, the key parameter items in the basic evolution parameters are extracted, including CPU allocation strategies, memory allocation modes, storage access strategies, etc., and unique identifier conversions are performed on each parameter. Subsequently, the computing nodes, storage nodes, and task scheduling strategies in the simulated evolution process are extracted, and number matching is performed on each process node. The basic evolution parameters are mapped one by one with the process nodes, and the corresponding relationship between the parameters and the processes is recorded. Finally, a parameter-process correspondence relationship is generated. When performing resource pressure mapping on the simulated evolution process based on the resource pressure distribution matrix, the matrix transformation method is used to calculate the pressure distribution of each process node. First, the resource consumption ratio of each process is extracted according to the resource pressure distribution matrix, and the resource pressure index is calculated according to indicators such as computing resources, storage resources, and network bandwidth. Subsequently, according to the task execution order in the simulated evolution process, the pressure changes in different stages are calculated, and the dynamic weight allocation method is used to adjust the resource pressure ratio. Finally, a pressure evolution process is generated. When deriving defective parameters for the parameter-process correspondence relationship according to the pressure evolution process, the differential analysis method is used to analyze the parameter change trend. First, the key inflection points in the pressure evolution process are extracted, and the proportion of nodes with sudden pressure increase is calculated. Subsequently, based on the parameter-process correspondence relationship, the main parameter items affecting the pressure change are analyzed, and the second derivative analysis method is used to calculate the influence strength of the parameter change on the pressure. The parameter items with a change amplitude exceeding the set threshold (5%-10%) are marked as defective parameters. Finally, resource pressure defect data is generated and stored as an abnormal parameter set. When optimizing and adjusting the basic evolution parameters according to the resource pressure defects, the genetic algorithm (Genetic Algorithm,GA) Optimize the calculation of defect parameters. First, extract the abnormal parameters from the resource pressure defect data, and set the optimization objective function with the reduction of resource occupancy rate as the optimization goal. Subsequently, calculate the optimal parameter combination according to the fitness function, and adaptively adjust the parameters using the crossover and mutation strategy. Finally, generate the optimized development parameters.

[0157] Preferably, step S55 includes the following steps:

[0158] Step S551: Conduct parameter adaptability backtesting on the pressure evolution process based on the parameter-process correspondence relationship, and screen out abnormal parameters according to the execution time delay floating range of 0.05s - 3s to generate a preliminary abnormal parameter set;

[0159] Step S552: Trace the resource load of the preliminary abnormal parameter set, and screen out the parameters affected by specific resource constraints to obtain the load-sensitive parameter set;

[0160] Step S553: Calculate the bottleneck propagation path of the load-sensitive parameter set and extract the key restricted parameters; extract the parameter critical tolerance from the restricted parameter mapping table, and combine the availability range of each parameter in different pressure environments, where the floating-point type is 0.001 - 1000 and the integer type is 1 - 5000;

[0161] Step S554: Calculate the parameter tolerance threshold based on the parameter critical tolerance and the availability range to obtain the parameter tolerance set;

[0162] Step S555: Match the defect characteristics of the key restricted parameters based on the parameter tolerance set to obtain the resource pressure defect.

[0163] In this embodiment, when performing parameter adaptability backtesting on the pressure evolution process based on the parameter-process correspondence relationship, first, the time series data in the pressure evolution process is extracted, and the execution delay of tasks at each time node is measured. The high-precision clock synchronization method (High Precision Clock Synchronization) is used to unify the timestamps on different servers. Subsequently, the execution delay of each task is calculated, and the floating range of the execution delay is controlled between 0.05 s and 3 s. According to this range, the parameter items with abnormal fluctuations are screened out. The Kalman Filter is used to smooth the delay change trend of the parameters and eliminate data noise. The parameters with execution delay fluctuations exceeding the set threshold are marked as abnormal parameters and stored as the preliminary abnormal parameter set. When tracing the resource load of the preliminary abnormal parameter set, the resource dependency analysis method (Resource Dependency Analysis) is used to track the occupation of computing resources, storage resources, and network resources by the parameters. First, according to the parameter-process correspondence relationship, the computing tasks corresponding to each parameter are parsed, and the load ratio of the tasks on different resources is calculated. The backtracking algorithm (Backtracking Algorithm) is used to extract the traceability path of resource occupation from the task execution log, and the key resource nodes in the traceability path are statistically analyzed. Subsequently, the parameters affected by specific resource constraints are screened out, and the K-Means Clustering method is used to group the parameters according to the load sensitivity and store them as the load-sensitive parameter set. When calculating the bottleneck propagation path for the load-sensitive parameter set, a directed acyclic graph (Directed Acyclic Graph, DAG) is used to construct the parameter influence path. First, the influence weights of the parameters in different pressure environments are extracted, and the bottleneck propagation path is constructed based on the dependency relationship of task execution. The topological sorting (TopologicalSorting) method is used to calculate the key restricted parameters in the path and store them as the restricted parameter mapping table. Subsequently, the parameter critical tolerance is extracted from the restricted parameter mapping table, and the maximum availability range of each parameter in different pressure environments is extracted. The range of floating-point parameters is set to 0.For 001 - 1000, the range of integer - type parameters is set to 1 - 5000, and the tolerance intervals of each parameter are recorded. When calculating the parameter tolerance threshold based on the critical tolerance and availability range of the parameters, the Tolerance Fitting Analysis method is used to perform regression calculations on the critical values of each parameter. First, according to the historical stress - test data, the stability index of each parameter at different resource - occupancy levels is calculated. Subsequently, the least - squares method is used to fit the parameter - change curve, and the tolerance threshold of each parameter is calculated. The tolerance thresholds are stored as a parameter tolerance set. When performing defect - feature matching on critical restricted parameters based on the parameter tolerance set, the Feature Vector Matching method is used to calculate the abnormal features of the critical restricted parameters. First, the tolerance thresholds in the parameter tolerance set are extracted, and the over - limit ratio of each parameter is calculated. The principal - component analysis (PCA) dimensionality - reduction method is used to screen out the parameters that mainly affect system stability, and the cosine similarity between these parameters and historical defect data is calculated. The parameters with a similarity exceeding the set threshold (above 0.85) are screened out and marked as resource - pressure defects.

[0164] The present invention also provides an Internet software - development management system for executing the Internet software - development management method described above. The Internet software - development management system includes:

[0165] A requirements - collection module for collecting a set of original functional requirements; performing functional - clustering analysis on the set of original functional requirements to obtain classified - requirement functional features; and performing tagging processing on the classified - requirement functional features to generate independent - requirement units.

[0166] A requirements - parsing module for performing inter - unit dependency parsing on the independent - requirement units to obtain a set of dependency relationships; and constructing a requirements - logic structure based on the set of dependency relationships.

[0167] A framework - modeling module for performing virtual - and - real framework interweaving on the independent - requirement units based on the requirements - logic structure to obtain a requirements - framework matrix; and performing organizational - network simulation according to the requirements - framework matrix to generate a requirements - connection network.

[0168] A code - evolution module for performing code reverse - evolution according to the requirements - connection network and performing basic - parameter matching to obtain evolution - basic parameters; and performing stress - simulation testing on the evolution - basic parameters to generate stress - test results.

[0169] An optimization - regulation module for capturing resource - pressure defects of the evolution - basic parameters based on the stress - test results; and performing optimization adjustment on the evolution - basic parameters according to the resource - pressure defects to generate optimized - development parameters.

[0170] Through the requirement acquisition module, the present invention acquires the original function requirement collection and conducts function clustering analysis to generate independent requirement units, realizing the systematicness and accuracy of requirement acquisition, promoting the effective management and analysis of function requirements. Through the requirement parsing module, the independent requirement units are parsed for dependencies to construct a requirement logic structure, realizing the clarification of the dependency relationships between requirements and optimizing the requirement logic structure to support dynamic adjustment. Through the framework modeling module, based on the requirement logic structure, the independent requirement units are intertwined with virtual and real frameworks to generate a requirement framework matrix and conduct organizational network simulation, realizing the efficient generation of the requirement framework matrix and promoting team collaboration and information sharing. Through the code evolution module, code reverse evolution is performed according to the requirement connection network, basic parameter matching is carried out and stress simulation tests are conducted, realizing the improvement of code evolution efficiency and ensuring the accuracy of basic parameter matching. Through the optimization and regulation module, based on the stress test results, resource pressure defects are captured and the evolution basic parameters are optimized, realizing the improvement of resource utilization efficiency, reducing development costs and risks, and overall improving the efficiency and quality of software development, ensuring the successful delivery and sustainable development of the project.

[0171] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application documents are intended to be encompassed within the present invention.

[0172] The above description is only the specific implementation manners of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for managing Internet software development, characterized in that: The following steps are involved: Step S1: Collecting a collection of original functional requirements; performing functional clustering analysis on the collection of original functional requirements to obtain classified functional characteristics of requirements; Label the functional characteristics of the classified requirements to generate independent requirement units; Step S2: performing inter-unit dependency analysis on independent demand units to obtain a dependency collection; Build the logical structure of requirements based on the collection of dependencies; Step S3: Based on the demand logic structure, the independent demand units are interwoven with virtual and real frames to obtain a demand frame matrix; the organizational network simulation is performed according to the demand frame matrix to generate a demand connection network; Step S4: according to the requirements, the network is contacted to perform reverse code evolution, and basic parameter matching is performed to obtain evolution basic parameters; a stress simulation test is performed on the evolution basic parameters to generate stress test results; Step S5: capturing resource pressure defects of the evolution basic parameters based on the pressure test results; The evolutionary basic parameters are optimized and adjusted according to resource pressure defects to generate optimized development parameters.

2. The Internet software development management method according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Collecting a collection of original functional requirements; performing semantic segmentation on the collection of original functional requirements to obtain independent requirement fragments; Step S12: extracting functional features of independent demand segments; performing cluster analysis on the functional features of each independent demand segment to obtain classified demand functional features; Step S13: Eliminate redundant features of the classified demand function features to obtain a cleaning demand function; Step S14: assigning unique labels to the cleaning demand functions based on the classified demand function characteristics to generate independent demand units.

3. The Internet software development management method according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: hierarchically deconstruct the independent demand units to obtain a demand layer structure; perform predecessor and successor relationship detection on the independent demand units according to the demand layer structure to generate a unit dependency candidate set; Step S22: cross-validate and filter the unit dependency candidate set to obtain high-confidence dependency relationships; perform transitive closure operations on the high-confidence dependency relationships to generate dependency transitive chains; Step S23: Detecting the network loop of the dependency transfer chain; performing dependency resolution based on the network loop of the dependency transfer chain to obtain a collection of dependency relationships; Step S24: construct a directed acyclic graph structure based on the dependency collection, and perform demand topology sorting based on the directed acyclic graph structure to obtain a demand topology sequence; perform priority weighting processing on the demand topology sequence to generate a demand logic structure.

4. The Internet software development management method according to claim 1, characterized in that: The step S3 of interweaving the virtual and real frames of the independent demand units based on the demand logic structure includes: Perform dot matrix conversion on the required logical structure to obtain logical dot matrix mapping data; Perform semantic vector encoding mapping on independent demand units to generate demand semantic vectors; Calculate the similarity distance of the demand semantic vector to obtain the demand affinity data; Perform orthogonal projection fusion on the logic dot matrix mapping data and demand affinity data to obtain the demand position matrix; Perform virtual-real boundary division based on the required position matrix to generate virtual-real partition boundaries; Based on the virtual and real partition boundaries, frame attribute mapping is performed on the independent demand units to obtain frame characteristic data; According to the framework feature data, the mapping rules are interwoven and simulated to obtain the demand framework matrix.

5. The Internet software development management method according to claim 1, characterized in that: The organizational network simulation according to the demand framework matrix in step S3 includes: Evaluate the connection strength of the demand framework matrix, and assign node weights based on the connection strength to obtain node connection weight data; Construct a multi-layer network topology based on node connection weight data and draw a topology diagram to generate a network topology sketch; Perform node clustering reconstruction on the network topology sketch according to the node connection weight data to obtain the reconstructed cluster nodes; Based on the reconstruction of clustering nodes, domain-to-convergence mapping is performed to obtain the functional clustering domain; Organizational network simulation is performed based on functional aggregation domains to generate demand contact networks.

6. The Internet software development management method according to claim 1, characterized in that: In step S4, contacting the network according to the requirements to perform code reverse evolution and basic parameter matching includes: Map the demand functions to the demand contact network and obtain the functional module decomposition data; The function module decomposition data is expanded by dependency conduction, and the level is split according to the conduction probability of 0.3-0.9, the deep dependency path is extracted, and the recursive association link is generated; Directionally summarize the recursive association links to obtain the main code nodes; Perform logical inference based on the functional module decomposition data and main code nodes to generate a code mapping skeleton; Perform structural replacement processing on the code mapping skeleton according to the preset code library to obtain an evolutionary code framework; Perform reverse function deduction on the evolution code framework to generate the code evolution trajectory; Disassemble the call chain of the code evolution trajectory to obtain the parameter transfer relationship; Extract the boundary constraints of the parameter transfer relationship, where the possible value range of the key variables is extracted, and the upper and lower limits are set, i.e. 0.001-1000 for floating point type and 1-5000 for integer type, to obtain the parameter constraint set; Dynamically derive the parameter constraint set to obtain the parameter transformation mapping; The parameter transformation mapping is matched with basic parameters to obtain the evolution basic parameters.

7. The Internet software development management method according to claim 1, characterized in that: The stress simulation test of the evolution basic parameters in step S4 includes: Calculate resource requirements for evolution basic parameters to obtain evolution requirement resources; Mapping evolving demand resources into resource constraint data; Perform stress scenario modeling based on resource constraint data to obtain a resource stress test model; The evolutionary basic parameters are simulated according to the resource stress test model to generate stress test results.

8. The Internet software development management method according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: setting resource occupancy upper limits to 70%-95% for computing resources and 80%-98% for storage resources; performing resource bottleneck mapping on the stress test results based on the resource occupancy upper limits to obtain resource constraint conflict data; Step S52: performing associated load decomposition on resource constraint conflict data and constructing a resource pressure distribution matrix; performing evolution process simulation on evolution basic parameters to obtain a simulated evolution process; Step S53: mapping the evolution basic parameters and the simulated evolution process to generate a parameter-process mapping relationship; Step S54: mapping the simulated evolution process with resource pressure based on the resource pressure distribution matrix to generate a pressure evolution process; Step S55: deriving defect parameters from the parameter-process correspondence relationship according to the pressure evolution process to obtain resource pressure defects; Step S56: Optimize and adjust the evolution basic parameters according to the resource pressure deficiency to generate optimized development parameters.

9. The Internet software development management method according to claim 8, characterized in that: Step S55 includes the following steps: Step S551: Perform parameter adaptability backtest on the pressure evolution process based on the parameter-process comparison relationship, and screen abnormal parameters according to the execution delay floating range of 0.05s-3s to generate a preliminary abnormal parameter set; Step S552: tracing the resource load of the preliminary abnormal parameter set, screening out the parameters affected by the specific resource constraints, and obtaining the load sensitive parameter set; Step S553: ​​perform bottleneck propagation path calculation on the load-sensitive parameter set and extract key restricted parameters; perform parameter critical tolerance extraction on the restricted parameter mapping table, combined with the availability range of each parameter under different pressure environments, where the floating point type is 0.001-1000 and the integer type is 1-5000; Step S554: Calculate the parameter tolerance threshold based on the parameter critical tolerance and the availability range to obtain a parameter tolerance set; Step S555: performing defect feature matching on key restricted parameters based on the parameter tolerance set to obtain resource pressure defects.

10. An Internet software development management system, characterized in that: Used to execute the Internet software development management method as claimed in claim 1, the Internet software development management system comprises: The demand collection module is used to collect the original functional demand collection; perform functional clustering analysis on the original functional demand collection to obtain the classified demand functional characteristics; label the classified demand functional characteristics to generate independent demand units; The demand analysis module is used to analyze the dependencies between independent demand units and obtain a collection of dependency relationships; and to build a demand logic structure based on the collection of dependency relationships; The framework modeling module is used to interweave the virtual and real frameworks of independent demand units based on the demand logic structure to obtain the demand framework matrix; the organizational network simulation is performed according to the demand framework matrix to generate the demand connection network; The code evolution module is used to contact the network for reverse code evolution according to the requirements, and to match the basic parameters to obtain the evolution basic parameters; to perform stress simulation test on the evolution basic parameters and generate stress test results; The optimization and control module is used to capture the resource pressure defects of the evolution basic parameters based on the stress test results; the evolution basic parameters are optimized and adjusted according to the resource pressure defects to generate optimized development parameters.

Images