A testing device for internet software development

Abstract

The application relates to the technical field of Internet software development test, and particularly provides a test device for Internet software development, which comprises a network construction subsystem, a path selection subsystem and a report generation subsystem. The network construction subsystem identifies key logical nodes by calculating the difference between adjacent logical chains and combining a dynamic threshold value, and forms a multilevel logical chain network. The path selection subsystem defines a transition probability of random sampling based on the multilevel logical chain network, and generates a plurality of test paths through Monte Carlo simulation. The weight of each path is determined by the number of logical nodes covered by the path, and a plurality of paths with the highest weight are selected as preliminary test paths. The report generation subsystem constructs positive and negative example pairs in the multilevel logical chain network. The test paths are optimized so that the paths cover key nodes in the logical chain network. The test paths in the multilevel logical chain network are weighted and aggregated to generate a final test report. The application realizes automatic test path generation based on actual software running characteristics.

Description

Technical Field

[0001] This invention relates to the field of Internet software development and testing technology, and in particular to a testing device for Internet software development. Background Technology

[0002] In the field of internet software development, testing is a crucial step in ensuring software quality, functional integrity, and system stability. With the rapid development of internet technology, the complexity of software systems is increasing exponentially, with applications involving multi-module interactions, distributed architectures, and high-concurrency scenarios becoming increasingly common. However, existing testing methods and equipment demonstrate significant shortcomings when facing these complex software environments. Traditional testing methods typically rely on manually written test cases or automated testing tools based on simple rules. These methods are inadequate when dealing with non-linear logical relationships, dynamic data flows, and multi-dimensional interactions. Furthermore, existing testing equipment lacks support for complex logical derivations and cannot effectively simulate various boundary conditions and anomalies in real-world operating environments, resulting in low test coverage, high false positive rates, and potentially overlooking serious defects.

[0003] Prior art 1, Chinese Patent Application No. 202311619736.2, discloses an internet-based software development, operation, and testing management system. This system includes modules for software performance testing, software device performance evaluation, software network performance evaluation, software server processing capacity evaluation, software development and operation testing pass / fail determination, and a cloud database. While it comprehensively evaluates the target software's device performance, network performance, and server processing capacity to determine whether the target software's development and operation testing is qualified, and combines diverse testing scenarios with detailed test indicator analysis to achieve scientific and accurate testing of internet software development and operation, providing more information and guidance for internet software development and testing to ensure excellent performance under different conditions, thereby helping to improve the quality, reliability, and user satisfaction of internet software; however, the test determination is based solely on static performance indicators such as equipment, network, and server, lacking analysis of the software's internal logical structure and dynamic dependencies between modules. This may lead to redundant test paths or insufficient coverage of critical nodes.

[0004] Prior art two, Chinese patent application number 202210195299.5, discloses an internet software development system, including a software development module, a network database module, and a client. The software development module includes a software design module, a software testing module, a software modification module, and a software trial operation module. The network database module includes a software pre-storage module, a software query module, a software download module, and a software operation module. The client includes an information login module, an information filtering module, and a software download module. The software development module connects to the network database and a backend database, while the client connects to the internet. Software developed by the software development module is uploaded to the network database module, and the client accesses the network database module via the internet to select and download software. While effectively integrating with the internet can reduce errors during development, greatly improve software development efficiency, enhance the user experience, and reduce losses, the testing module relies on a fixed process and does not incorporate the dynamic logic of the software during actual operation, potentially leading to a disconnect between the test scenario and the real operating environment.

[0005] Prior art three, Chinese patent application number 202211168743.0, discloses an optimization method for a continuous integration pipeline, used to optimize existing continuous integration pipelines and solve the problems of high testing costs and low efficiency of manual debugging in existing software development processes. It includes: a method for predicting software defects using historical change characteristics of the software repository; a method for defect localization based on historical code information; and a method for defect localization using test coverage information and method call graphs. While it can predict defects in the software development process to save testing costs and locate defects after they are exposed, improving the efficiency of developers in fixing software defects and thus improving the continuous integration process to some extent, defect prediction and localization rely on historical data and do not dynamically analyze real-time runtime logic, which may lead to low efficiency in test path generation or the omission of critical defects.

[0006] Current technologies 1, 2, and 3 suffer from problems such as static test paths, insufficient coverage of key logic nodes, and a disconnect between test scenarios and actual operational logic. Therefore, this invention provides a testing device for internet software development. Summary of the Invention

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] In one aspect, the present invention provides a testing apparatus for Internet software development, comprising:

[0009] The network construction subsystem is used to extract the call relationships between functional modules by monitoring the call sequence of functional modules during software runtime, and to calculate the call probability distribution of each module; then, it uses the entropy weight method to calculate the dependency strength between modules, and constructs an initial logical chain based on the dependency strength; by calculating the difference between adjacent logical chains and combining it with dynamic thresholds, it identifies key logical nodes and forms a multi-level logical chain network.

[0010] The path selection subsystem is used to define the transition probability of random sampling based on a multi-level logical chain network and generate multiple test paths through Monte Carlo simulation. The weight of each path is determined by the number of logical nodes covered by the path, and the paths with the highest weights are selected as the initial test paths.

[0011] The report generation subsystem is used to construct positive and negative pairs in a multi-level logical chain network; optimize the test path through a loss function to ensure that the path covers the key nodes in the logical chain network; and weight and aggregate the test paths in the multi-level logical chain network to generate the final test report.

[0012] In one alternative implementation, the network construction subsystem includes:

[0013] The runtime monitoring and relationship extraction module is used to monitor the function module calling behavior of the software in real time, capture the calling order and frequency between function modules, and form a dynamic calling sequence. Based on the dynamic calling sequence, the actual interaction pattern between function modules is analyzed, the calling probability distribution of each function module is quantified, and the actual running rules inside the monitoring software are revealed.

[0014] The dependency strength modeling module is used to calculate the dependency strength between functional modules using probability distribution and entropy weight method, transforming abstract calling relationships into quantifiable logical chains. Functional modules whose dependency strength reaches the preset standard form a closely related logical chain, while dependencies that do not reach the preset standard serve as auxiliary branches, thus constructing the initial logical chain framework.

[0015] The key node identification module is used to analyze the differences between adjacent logical chains. Combined with dynamic thresholds, it filters out the core branch points of the monitoring software function corresponding to nodes with abrupt changes in differences in the initial logical chain framework, and marks them as key logical nodes. Finally, a multi-level logical network containing the trunk chain and branch chains is formed.

[0016] In one optional implementation, the operation monitoring and relationship extraction module includes:

[0017] The behavior capture and pattern extraction submodule is used to continuously observe and record the activation order and interaction frequency between various functional modules. The call order is obtained from the activation order, and the call frequency is obtained from the interaction frequency, forming a complete call time sequence record. Based on the time sequence record, the collaboration mode between functional modules is analyzed, the distribution of triggering of each functional module is calculated, and the basic characteristics of the software monitoring software operation are discovered.

[0018] The correlation quantification and structure construction submodule is used to evaluate the correlation between functional modules after understanding the probability distribution of each functional module being triggered, and to transform the collaborative relationship of functional modules into specific numerical indicators. Functional module combinations with a correlation degree exceeding the set standard are identified as strong correlations and constitute the core path of software operation; combinations with a correlation degree not exceeding the set standard are used as supplementary paths, and a relationship network that initially reflects the initial logical chain framework of the monitoring software operation is initially established.

[0019] The core element location and network improvement submodule is used to identify key turning points by comparing the differences between adjacent core paths, mark the location of key turning points, and form a hierarchical network that reveals the actual operating rules inside the monitoring software.

[0020] In one optional implementation, the dependency strength modeling module includes:

[0021] The probability distribution data preparation submodule is used to call the probability distribution data of the functional modules, extract the transition probability matrix between each functional module, and record the conditional probability of calling functional module B from any functional module A, thus forming the basic dataset for dependency strength calculation.

[0022] The information uncertainty measurement submodule is used to calculate the information entropy value of the call distribution for each functional module. The information entropy value reflects the degree of uncertainty in the flow of information from a functional module to other modules when the functional module is called. Based on the entropy value calculation results of each functional module, the results are converted into weight coefficients through normalization.

[0023] The dependency strength comprehensive calculation submodule is used to calculate the comprehensive dependency strength index between functional modules by combining the transition probability matrix and weight coefficients.

[0024] In one optional implementation, the key node identification module includes:

[0025] The input data preparation submodule is used to extract the call frequency distribution and dependency strength distribution of adjacent logical chains from the dynamic call sequence; and to compare the differences in call probability distribution, dependency strength gradient change and topological structure of the following features of the two logical chains.

[0026] The dynamic threshold adjustment submodule is used to dynamically adjust the difference determination threshold based on the overall dependency strength distribution of the current logical chain set; modules with a difference exceeding the dynamic threshold are determined to be mutation nodes.

[0027] The key logic node generation submodule is used to associate mutation nodes with module roles in the initial logic chain framework. If a mutation node belongs to multiple highly different logic chains at the same time, it is determined to be a branch convergence point; if the dependency strength of a mutation node drops sharply in a single logic chain, it is determined to be an inter-chain switching point. Nodes constitute the core branch points in the multi-level logic network, connecting the main chain and auxiliary branch chains.

[0028] In one optional implementation, the difference in the calling probability distribution of the input data preparation submodule is determined by comparing the calling probability offsets of the same functional modules in the two chains.

[0029] Dependency strength gradient change: Analyze the difference in dependency strength between identical module pairs in two chains. If the difference exceeds the dynamic threshold, it is marked as a potential difference point.

[0030] Topological structure difference: Check whether the overlapping modules of the two chains are in similar positions in their respective chains. If the positional offset is significant, increase the difference weight.

[0031] In one optional implementation, the key logic node generation submodule includes:

[0032] The multi-chain cross-validation unit is used to take the set of mutation nodes obtained from the dynamic threshold adjustment submodule as input, and backtrack the adjacent logical chain association information recorded by the input data preparation submodule. For each mutation node, it retrieves all logical chain identifiers to which it belongs and extracts the difference matrix between logical chains, including the call probability offset, dependency strength difference and topological position offset.

[0033] The single-chain strength decay analysis unit is used to switch to single-chain analysis mode for mutation nodes that have not passed multi-chain verification. The initial logical chain framework provided by the strength modeling module is activated to locate the sequence position of the mutation node in its logical chain. When the following conditions are met simultaneously, it is determined to be an inter-chain switching point:

[0034] The conflict resolution and final determination unit is used to use topological structure difference weights as decision factors when the mutated node simultaneously satisfies some features of the convergence point and the switching point. If the mutated node is in a structurally symmetrical position in most chains, the convergence point determination is strengthened. If the position distribution is discrete and accompanied by drastic fluctuations in the calling probability, the switching point determination is strengthened.

[0035] In one optional implementation, when a mutation node appears in three or more logic chains and the difference index between the logic chains all exceeds the adjusted dynamic threshold, the convergence point determination condition is triggered. At this time, the dependency strength change pattern of each chain at the mutation node is checked: if the calling probability of the node in each chain remains stable and the dependency strength distribution of the output direction shows multi-way balance, then it is confirmed that the mutation node undertakes the branch convergence function.

[0036] In one optional implementation, the path selection subsystem includes:

[0037] The logical association module is used to determine the transition probability by the dependency strength and call probability in the multi-level logical chain network. Monte Carlo simulation generates paths based on the multi-level logical chain network structure. Weight calculation depends on the key node markings in the multi-level logical chain network to form a test path optimization closed loop.

[0038] The transition probability definition module is used to define the random sampling probability of transitions between functional modules in a multi-level logical chain network, based on the functional module call probability distribution and dependency strength.

[0039] The path simulation module is used to simulate the possible paths of the monitoring software through probability sampling and generate several test paths. The weight of each test path is dynamically determined by the number of key nodes it covers, and the initial test path with the greatest impact on the core logic of the monitoring software is selected.

[0040] In one optional implementation, the report generation subsystem includes:

[0041] The path expansion module is used to expand the initial test paths generated by the path selection subsystem. Positive examples consist of path segments covering key logical nodes. The selection criteria combine the key node distribution characteristics identified by the network construction subsystem and the path weights calculated by the path selection subsystem. Negative examples come from two situations: logical chain connection regions that have not reached the dynamic threshold, or path segments with low weights in the Monte Carlo simulation.

[0042] The dual-channel feedback module extracts the path features of positive examples into multi-dimensional vectors, including dimensions such as key node coverage and neighboring node differences; the missing features of negative examples constitute the comparison samples; the optimization process iteratively adjusts the transition probability distribution of the test path, so that the positive example feature vector gradually approaches the topological centrality distribution of the logical chain network, while the negative example feature vector is pushed away from the effective coverage area.

[0043] The path optimization module is used to evaluate the node criticality scores determined by the network construction subsystem, the path weight coefficients assigned by the path selection subsystem, and the coverage gap data generated by the comparison of positive and negative examples. The optimized test path will dynamically increase the access probability of high critical nodes while maintaining the original weight distribution. The final set of paths retains the randomness characteristics of the Monte Carlo simulation.

[0044] The network construction subsystem of this invention extracts the call relationships between software functional modules through dynamic monitoring, quantifies the module dependency strength, objectively evaluates the correlation between modules using the entropy weight method, identifies key logical nodes through difference analysis and dynamic thresholds, and constructs a multi-level logical chain network that reflects the actual operating logic of the software. The path selection subsystem defines scientific transition probabilities based on the multi-level logical chain network, efficiently generates candidate test paths using the Monte Carlo method, and achieves preliminary screening of test paths through path weight evaluation, ensuring the coverage of key logical nodes by the test paths. The report generation subsystem verifies the test effectiveness using a positive and negative example pair construction method, improves the quality of test paths through loss function optimization, weighted aggregates of multi-path test results, and generates a comprehensive test report reflecting the software's logical coverage. Attached Figure Description

[0045] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0046] Figure 1 This is a block diagram of the testing device for Internet software development provided in Embodiment 1 of the present invention;

[0047] Figure 2 This is a block diagram of the network construction subsystem provided in Embodiment 2 of the present invention;

[0048] Figure 3 This is a block diagram of the path selection subsystem provided in Embodiment 3 of the present invention;

[0049] Figure 4 This is a block diagram of the report generation subsystem provided in Embodiment 11 of the present invention;

[0050] Figure 5 A block diagram of the electronic device provided by the present invention;

[0051] Figure 6 A block diagram of a computer-readable storage medium provided for this invention. Detailed Implementation

[0052] The technical solutions of the present invention will now be described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0053] Hereinafter, the terms "first," "second," etc., are used for descriptive convenience only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0054] In this invention, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a fixed mechanical connection, a detachable mechanical connection, or an integral part; or, "connection" can be a direct connection or an indirect connection through an intermediate medium. Furthermore, unless otherwise explicitly specified and limited, the term "coupling" should be interpreted broadly. For example, "coupling" can be a direct electrical connection, such as physical contact and electrical conduction between two components; it can also be understood as an electrical connection between different components in a circuit structure through physical lines capable of transmitting electrical signals, such as copper foil or wires on a printed circuit board (PCB), to transmit electrical signals; or, "coupling" can be an indirect electrical connection between two components through an intermediate medium; or, "coupling" can be an electrical connection between two components in a non-contact manner, such as an electrical connection between two components using capacitive coupling to transmit electrical signals.

[0055] In this embodiment of the invention, directional terms such as "up," "down," "left," and "right" may be defined relative to the orientation of the components shown in the accompanying drawings. It should be understood that these directional terms can be relative concepts, used for relative description and clarification, and can change accordingly depending on the orientation of the components in the accompanying drawings.

[0056] Example 1:

[0057] like Figure 1 As shown, this embodiment of the invention provides a testing device for Internet software development, comprising:

[0058] The network construction subsystem is used to extract the call relationships between functional modules by monitoring the call sequence of functional modules during software runtime, and to calculate the call probability distribution of each module; then, it uses the entropy weight method to calculate the dependency strength between modules, and constructs an initial logical chain based on the dependency strength; by calculating the difference between adjacent logical chains and combining it with dynamic thresholds, it identifies key logical nodes and forms a multi-level logical chain network.

[0059] The path selection subsystem is used to define the transition probability of random sampling based on a multi-level logical chain network and generate multiple test paths through Monte Carlo simulation. The weight of each path is determined by the number of logical nodes covered by the path, and the paths with the highest weights are selected as the initial test paths.

[0060] The report generation subsystem is used to construct positive and negative pairs in a multi-level logical chain network; optimize the test path through a loss function to ensure that the path covers the key nodes in the logical chain network; and weight and aggregate the test paths in the multi-level logical chain network to generate the final test report.

[0061] In the above embodiments, the network construction subsystem extracts the call relationships between software functional modules through dynamic monitoring, quantifies the module dependency strength, objectively evaluates the correlation between modules using the entropy weight method, identifies key logical nodes through difference analysis and dynamic thresholds, and constructs a multi-level logical chain network that reflects the actual operating logic of the software. The path selection subsystem defines scientific transition probabilities based on the multi-level logical chain network, efficiently generates candidate test paths using the Monte Carlo method, and achieves preliminary screening of test paths through path weight evaluation to ensure the coverage of key logical nodes by the test paths. The report generation subsystem verifies the test effectiveness using a positive and negative example pair construction method, improves the quality of test paths through loss function optimization, weighted aggregates of multi-path test results, and generates a comprehensive test report reflecting the software's logical coverage.

[0062] In summary, this embodiment realizes automated test path generation based on the actual operating characteristics of the software, ensures test coverage of key nodes through multi-level logical network modeling, improves test efficiency by combining probabilistic statistical methods, and finally outputs a test report with quantitative evidence, providing a reliable test verification method for software development.

[0063] Example 2:

[0064] like Figure 2 As shown, based on Embodiment 1, the network construction subsystem provided in this embodiment of the invention includes:

[0065] The runtime monitoring and relationship extraction module is used to monitor the function module calling behavior of the software in real time, capture the calling order and frequency between function modules, and form a dynamic calling sequence. Based on the dynamic calling sequence, the actual interaction pattern between function modules is analyzed, the calling probability distribution of each function module is quantified, and the actual running rules inside the monitoring software are revealed.

[0066] The dependency strength modeling module is used to calculate the dependency strength between functional modules using probability distribution and entropy weight method, transforming abstract calling relationships into quantifiable logical chains. Functional modules whose dependency strength reaches the preset standard form a closely related logical chain, while dependencies that do not reach the preset standard serve as auxiliary branches, thus constructing the initial logical chain framework.

[0067] The key node identification module is used to analyze the differences between adjacent logical chains. Combined with dynamic thresholds, it filters out the core branch points of the monitoring software function corresponding to nodes with abrupt changes in differences in the initial logical chain framework, and marks them as key logical nodes. Finally, a multi-level logical network containing the trunk chain and branch chains is formed.

[0068] In the above embodiments, this embodiment realizes the mapping and optimization from dynamic software operation behavior to a structured logical network. The specific significance is as follows: Dynamic behavior is structured by capturing the real-time calling order and frequency of functional modules, transforming discrete operation behavior into a quantifiable calling probability distribution, and establishing a dynamic data model of the software's internal logic. Logical dependencies are made explicit by using the entropy weight method to transform implicit calling relationships between modules into explicit dependency strength indicators based on the calling probability distribution, distinguishing core logic chains from auxiliary branches, and constructing an initial logic chain framework, making abstract logical relationships computable. Core nodes are focused by identifying key branch points of mutation in the logic chain through difference analysis and dynamic threshold screening, forming a multi-level network containing the main chain (core logic) and branch chains (secondary logic), clarifying the core path and potential risk points of the software function. Operation monitoring data drives dependency strength calculation; dependency strength determines the hierarchy of the logic chain, and difference and dynamic thresholds jointly extract key nodes, ultimately generating a logical network with both dynamic characteristics and structural clarity, providing accurate logical framework support for test path generation.

[0069] Example 3:

[0070] Based on Example 2, the operation monitoring and relationship extraction module provided in this embodiment of the invention includes:

[0071] The behavior capture and pattern extraction submodule is used to continuously observe and record the activation order and interaction frequency between various functional modules. The call order is obtained from the activation order, and the call frequency is obtained from the interaction frequency, forming a complete call time sequence record. Based on the time sequence record, the collaboration mode between functional modules is analyzed, the distribution of triggering of each functional module is calculated, and the basic characteristics of the software monitoring software operation are discovered.

[0072] The correlation quantification and structure construction submodule is used to evaluate the correlation between functional modules after understanding the probability distribution of each functional module being triggered, and to transform the collaborative relationship of functional modules into specific numerical indicators. Functional module combinations with a correlation degree exceeding the set standard are identified as strong correlations and constitute the core path of software operation; combinations with a correlation degree not exceeding the set standard are used as supplementary paths, and a relationship network that initially reflects the initial logical chain framework of the monitoring software operation is initially established.

[0073] The core element location and network improvement submodule is used to identify key turning points by comparing the differences between adjacent core paths, mark the location of key turning points, and form a hierarchical network that reveals the actual operating rules inside the monitoring software.

[0074] In the above embodiments, this embodiment achieves a complete revelation and structured presentation of the operational patterns of monitoring software through a hierarchical and progressive analysis method. At the basic data acquisition level, a complete runtime behavior recording system is established, enabling a quantitative description of the interaction characteristics of functional modules. At the relationship analysis level, ambiguous module collaboration relationships are transformed into quantifiable correlation indicators, distinguishing the structural hierarchy of core paths and auxiliary paths. At the network construction level, key decision nodes affecting system operation are identified, forming a multi-layered network model reflecting actual operational characteristics. The final output operational network has the following value characteristics: it fully presents the functional topology of the monitoring software, clearly marks core paths and supplementary paths, reveals the decision logic nodes within the system, and provides a visual analysis foundation for operational optimization. It achieves a complete transformation from raw operational data to a structured network model, providing a systematic analytical framework for understanding the operational mechanism of the monitoring system.

[0075] Example 4:

[0076] Based on Example 2, the dependency strength modeling module provided in this embodiment of the invention includes:

[0077] The probability distribution data preparation submodule is used to call the probability distribution data of the functional modules, extract the transition probability matrix between each functional module, and record the conditional probability of calling functional module B from any functional module A, thus forming the basic dataset for dependency strength calculation.

[0078] The information uncertainty measurement submodule is used to calculate the information entropy value of the call distribution for each functional module. The information entropy value reflects the degree of uncertainty of the flow to other modules when a functional module is called. The higher the entropy value, the more dispersed the calling targets of the module are; the lower the entropy value, the more concentrated the calling targets are. Based on the entropy value calculation results of each functional module, the results are converted into weight coefficients through normalization. Modules with high entropy values ​​are assigned smaller weights because their calling behavior is more random, while modules with low entropy values ​​are assigned larger weights because their calling behavior shows obvious regularity.

[0079] The comprehensive dependency strength calculation submodule is used to calculate the comprehensive dependency strength index between functional modules by combining the transition probability matrix and weight coefficients. For call relationships with high transition probabilities and both ends of the modules have low entropy values, a higher dependency strength score will be obtained; for call relationships with low transition probabilities or involving modules with high entropy values, a lower dependency strength score will be obtained.

[0080] In the above embodiments, this embodiment constructs an analytical framework for quantitatively evaluating the calling relationships between system functional modules through the collaborative work of three sub-modules: probability distribution data preparation, information uncertainty measurement, and comprehensive dependency strength calculation. The probability distribution data preparation sub-module provides basic data support, accurately characterizing the conditional probabilities of calls between functional modules through a transition probability matrix, laying the data foundation for analysis. The information uncertainty measurement sub-module introduces information entropy as the basis for weight adjustment, distinguishing between the regularity and randomness of calling behavior, ensuring that dependency strength calculation focuses more on highly deterministic calling relationships and reduces the interference of random calls on the overall analysis. The comprehensive dependency strength calculation sub-module combines the outputs of the previous two sub-modules to perform a weighted evaluation of the calling relationships between modules, filtering out high-probability and stable dependencies, providing objective indicators for system architecture optimization, module coupling analysis, or fault impact domain assessment. This achieves quantitative modeling of the dependencies between system functional modules, helps identify key calling chains, and assists in system design, maintenance, or refactoring decisions.

[0081] Example 5:

[0082] Based on Embodiment 2, the key node identification module provided in this embodiment of the invention includes:

[0083] The input data preparation submodule is used to extract the call frequency distribution and dependency strength distribution of adjacent logical chains from the dynamic call sequence; and to compare the differences in call probability distribution, dependency strength gradient change and topological structure of the following features of the two logical chains.

[0084] Differences in call probability distribution: Compare the offsets of the call probability of the same functional module in the two chains;

[0085] Dependency strength gradient change: Analyze the difference in dependency strength between the same module pairs in two chains. If the difference exceeds the dynamic threshold (such as the proportion of the average dependency strength within the chain), it is marked as a potential difference point.

[0086] Topology differences: Check whether the overlapping modules of the two chains are in similar positions in their respective chains (e.g., both are start nodes or end nodes). If the positional offset is significant, increase the difference weight.

[0087] The dynamic threshold adjustment submodule is used to dynamically adjust the difference determination threshold based on the overall dependency strength distribution of the current logical chain set; modules with a difference exceeding the dynamic threshold are determined to be mutation nodes.

[0088] If the difference in dependency strength between most adjacent logical chains is concentrated in the low range, the threshold is lowered to capture subtle changes.

[0089] If the system is in a highly volatile operating state (high call probability distribution entropy value), the threshold should be increased to avoid noise interference.

[0090] The key logic node generation submodule is used to associate mutation nodes with module roles in the initial logic chain framework. If a mutation node belongs to multiple highly different logic chains at the same time, it is determined to be a branch convergence point; if the dependency strength of a mutation node drops sharply in a single logic chain, it is determined to be an inter-chain switching point. Nodes constitute the core branch points in the multi-level logic network, connecting the main chain and auxiliary branch chains.

[0091] In the above embodiments, this embodiment constructs a structured logical network method based on dynamic call behavior and dependency strength analysis through the synergistic effect of three sub-modules: input data preparation, dynamic threshold adjustment, and key logical node generation. The input data preparation sub-module extracts differences in call probability distribution, dependency strength gradient changes, and topological differences between adjacent logical chains from the dynamic call sequence, providing basic data for quantitative comparison in the analysis. The dynamic threshold adjustment sub-module adaptively adjusts the difference judgment criteria according to the overall dependency strength distribution of the logical chain set, ensuring effective identification of mutation nodes under different operating conditions (such as low-fluctuation or high-entropy environments), balancing sensitivity and noise resistance. The key logical node generation sub-module distinguishes branch convergence points and inter-chain switching points by associating mutation nodes with the module roles of the initial logical chain framework, ultimately forming a multi-level logical network structure containing the trunk chain and branch chains. This achieves difference analysis and key node location of the internal logical chains of the software, providing a structured basis for system architecture optimization, core function identification, or runtime path analysis.

[0092] Example 6:

[0093] Based on Example 5, the key logic node generation submodule provided in this embodiment of the invention includes:

[0094] The multi-chain cross-validation unit is used to take the set of mutation nodes obtained from the dynamic threshold adjustment submodule as input, and backtrack the adjacent logical chain association information recorded by the input data preparation submodule. For each mutation node, it retrieves all logical chain identifiers to which it belongs and extracts the difference matrix between logical chains, including the call probability offset, dependency strength difference and topological position offset.

[0095] When a mutation node appears in three or more logical chains, and the difference index between logical chains exceeds the adjusted dynamic threshold, the convergence point determination condition is triggered. At this time, the dependency strength change pattern of each chain at the mutation node is checked: if the calling probability of the node in each chain remains stable, and the dependency strength distribution of the output direction shows multi-way balance, then it is confirmed that the mutation node undertakes the branch convergence function.

[0096] The single-chain strength decay analysis unit is used to switch to single-chain analysis mode for mutation nodes that have not passed multi-chain verification. The initial logical chain framework provided by the strength modeling module is activated to locate the sequence position of the mutation node in its logical chain. When the following conditions are met simultaneously, it is determined to be an inter-chain switching point:

[0097] The dependency strength of a node in the chain decreases by more than twice the standard deviation of the average gradient change of the chain compared to its predecessor.

[0098] The entropy value of the subsequent module call probability distribution recorded by the topology difference submodule suddenly increased by more than 50%;

[0099] The dynamic call sequence indicates that the node subsequently engaged in cross-chain call behavior;

[0100] The conflict resolution and final determination unit is used to use topological structure difference weights as decision factors when the mutated node simultaneously satisfies some features of the convergence point and the switching point. If the mutated node is in a structurally symmetrical position in most chains, the convergence point determination is strengthened. If the position distribution is discrete and accompanied by drastic fluctuations in the calling probability, the switching point determination is strengthened.

[0101] In the above embodiments, the key logic node generation submodule of this embodiment achieves accurate classification of mutated node functions and optimized construction of the logical network structure through the collaborative work of three units: multi-chain cross-validation, single-chain strength decay analysis, and conflict resolution and final determination. Accurate node function identification: The multi-chain cross-validation unit identifies nodes that maintain a stable call probability and a balanced distribution of dependency strength across multiple logical chains, clarifying their role as branch convergence points. These nodes play a core role in the aggregation and distribution of system logic. The single-chain strength decay analysis unit captures nodes with significantly decreased dependency strength accompanied by abrupt changes in call patterns, determining them as inter-chain switching points. These nodes reflect the dynamic jumping behavior of the system's logical paths. Optimized network structure construction: The conflict resolution and final determination unit resolves boundary cases in node role determination through a topology difference weight decision mechanism, ensuring the rationality of the network structure division. The final node classification results provide a reliable structural basis for constructing a multi-level logical network containing backbone chains and branch chains, enabling the network to more accurately reflect the actual operating rules of the system. The adaptive and reliable decision-making process is rigorously based on quantitative characteristics such as dynamic call sequences, dependency strength distribution, and topological differences. Multi-condition combination verification ensures the objectivity of the decision results. The application of adaptive mechanisms such as dynamic thresholds and statistical standard deviations allows the decision criteria to adapt to system characteristics under different operating conditions, improving the robustness of the analysis. Through these mechanisms, the automated identification and classification of the functions of logical nodes within the system is achieved, providing a structured network model foundation for tasks such as architecture analysis, performance optimization, and behavior prediction.

[0102] Example 7:

[0103] like Figure 3 As shown, based on Embodiment 1, the path selection subsystem provided in this embodiment of the invention includes:

[0104] The logical association module is used to determine the transition probability by the dependency strength and call probability in the multi-level logical chain network. Monte Carlo simulation generates paths based on the multi-level logical chain network structure. Weight calculation depends on the key node markings in the multi-level logical chain network to form a test path optimization closed loop.

[0105] The transition probability definition module is used to define the random sampling probability of transitions between functional modules in a multi-level logical chain network, based on the functional module call probability distribution and dependency strength.

[0106] The path simulation module is used to simulate the possible paths of the monitoring software through random sampling probability and generate several test paths. The weight of each test path is dynamically determined by the number of key nodes it covers, and the initial test path with the greatest impact on the core logic of the monitoring software is selected.

[0107] In the above embodiments, the path selection subsystem of this embodiment constructs a test path generation and optimization system based on a multi-level logical chain network through the collaborative work of the logical association module, the transition probability definition module, and the path simulation module. The dependency-driven path generation mechanism uses the logical association module to construct a quantifiable transition probability model by analyzing the dependency strength and call probability between functional modules; it dynamically calculates path weights using key node markers to form a closed-loop optimization, ensuring that the generated test paths effectively reflect the actual call relationships within the system. The probabilistic path exploration method uses the transition probability definition module to transform the call relationships between functional modules into random sampling probabilities, and the path simulation module performs probability sampling using the Monte Carlo method, enabling efficient exploration of multiple possible path combinations for the monitoring software's operation while preserving the core logical chain. The optimized screening of key node coverage uses a weight evaluation standard established by dynamically calculating the number of key nodes covered by the test paths, automatically selecting the set of test paths that have the greatest impact on the core logic of the monitoring software; this key node-based screening mechanism ensures that test resources are concentrated on verifying the system's most core functions.

[0108] Example 8:

[0109] Based on Example 7, the transition probability definition module provided in this embodiment of the invention includes:

[0110] The call probability quantization submodule is used to analyze the interaction relationships between functional modules based on a multi-level logical chain network. The dependency strength reflects the tightness between functional modules, and the call probability reflects the frequency of triggering in actual operation. Together, they form an initial correlation matrix.

[0111] The probability distribution adjustment submodule is used to normalize the dependency strength and call probability of the initial association matrix to ensure data comparability across different dimensions. Based on the hierarchical structure of the multi-level logical chain network, the probability distribution is adjusted so that the call probability of higher-level modules can more easily influence global path selection, while the dependency strength of lower-level modules can influence local transfers.

[0112] The dynamic probability mapping submodule is used to map the adjusted probability distribution onto the transition relationship between functional modules, forming a sampleable probability matrix that reflects the static call relationship. It also introduces dynamic randomness through the Monte Carlo method, so that the generated test path not only conforms to the core logic but also covers edge cases.

[0113] In the above embodiments, this embodiment transforms the dependency strength and call frequency between functional modules into calculable quantifiable relationships by constructing an initial correlation matrix, providing a data foundation for probability adjustment. This modeling method preserves the topological characteristics of the original interaction relationships. Differentiated adjustments to the probability distribution based on the network hierarchy achieve the separation of global path selection and local transfer control; high-level modules dominate the overall direction, while low-level modules refine local paths, forming a hierarchical probability distribution system. By combining static correlation relationships with dynamic random sampling through the probability matrix, the diversity of path exploration is increased while ensuring core logic coverage. The introduction of the Monte Carlo method enables the system to effectively balance the needs of key testing and edge case detection. A complete transformation process from original interaction relationships to final sampleable probabilities is achieved, providing a scientific and quantifiable probabilistic basis for test path generation.

[0114] Example 9:

[0115] Based on Example 8, the probability quantization submodule provided in this embodiment of the invention includes:

[0116] The relation transformation unit is used to transform the implicit module relationships in a multi-level logical chain network into computable structured data. Each connection line carries the dual attributes of dependency strength and invocation probability.

[0117] The convolutional processing unit is used to perform bidirectional scanning along the topological path of the multi-level logical chain network. When tracing upwards, it marks the aggregation relationship between high-level functional modules and low-level components. When probing downwards, it records the support pattern of basic functional modules for composite functions. After three-dimensional convolutional processing, an initial correlation matrix is ​​formed.

[0118] The parameter comparison unit is used to adaptively weight the co-occurrence parameters of the initial association matrix. For interaction edges with strong dependencies but low frequency, structural weight advantages are given; while for connection lines with high frequency and weak dependencies, their dynamic characteristics are preserved. Both dependency strength and call frequency are set with strong and weak standards.

[0119] In the above embodiments, the probability quantization submodule of this embodiment achieves a precise transformation from an abstract system architecture to a computable model through the collaborative operation of three core units; the relationship transformation unit establishes the mathematical expression basis for the interaction relationship between functional modules, making the implicit connections in the network topology explicit into data links with quantified attributes; the convolution processing unit expands the spatial dimension feature extraction on this basis, and captures the uplink and downlink influence relationships in the hierarchical architecture through bidirectional scanning, and its output initial correlation matrix already has a composite expression of structural and dynamic features. The parameter comparison unit finally completes the dynamic balance of key parameters, and its adaptive weighting mechanism is compatible with runtime characteristics while maintaining structural integrity, so that the final generated correlation matrix reflects both the static design constraints of the system and the dynamic mode in actual operation; the processing chain formed by the three units essentially builds a bridge connecting the system design principles and runtime behavior, providing an input benchmark that is both structurally reasonable and behaviorally realistic for probability distribution adjustment and dynamic path generation.

[0120] Example 10:

[0121] As shown in Figure 10, based on Example 8, the probability distribution adjustment submodule provided in this embodiment of the invention includes:

[0122] The feature separation unit is used to separate the two feature dimensions, dependency strength and call probability, from the initial correlation matrix and perform standardization processing on them respectively; the call probability data of high-level functional modules is adjusted by the hierarchical amplification coefficient to shift its numerical distribution towards the high-weight interval; the dependency strength of low-level functional modules undergoes local focusing processing to retain its relative weight in a specific functional domain.

[0123] The factor application unit is used to apply influence factors layer by layer starting from the top-level node. The calling probability of each high-level functional module will penetrate down along the logical chain network, and its influence will decrease exponentially with the depth of the layer. At the same time, the dependency strength of the lower-level functional modules will form a local weight pool within the scope of its directly related upper-level functional modules.

[0124] The probability adjustment unit, used for global path selection, mainly absorbs the call probabilities adjusted by high-level functional modules and forms the decision backbone after smoothing. The local transfer path depends on the weight distribution of the strength parameters of the low-level functional modules and performs regional rebalancing while maintaining the original relative relationship. The output probability distribution shows obvious hierarchical gradient characteristics. High-level parameters have a wide range of influence but a small adjustment range, while low-level parameters have a narrow range of influence but a high adjustment accuracy.

[0125] In the above embodiments, the probability distribution adjustment submodule of this embodiment achieves differentiated adjustment of parameters at different levels in a multi-layered system architecture through the collaborative processing of three units. The feature separation unit first completes the dimensional decoupling and standardization of the original parameters, establishing benchmark data for hierarchical processing. The factor application unit implements vertical parameter transfer control based on network topology, enabling the influence of higher levels to have top-down propagation characteristics while retaining the local correlation characteristics of the lower levels. The probability adjustment unit ultimately forms a dual-channel adjustment mechanism: the global path inherits the wide-range adjustment characteristics of higher-level parameters, while the local path maintains the fine-grained control capability of lower-level parameters. The output results of the entire processing flow present a strict hierarchical correspondence: the probability distribution of the higher-level modules dominates the system-level behavior pattern, and the strength parameters of the lower-level modules constrain the component-level interaction rules. The hierarchical adjustment mechanism not only ensures the stability of the overall architecture but also retains appropriate autonomy for functional units of different granularities. The final generated probability distribution simultaneously meets the macro-constraints of system design and the micro-needs of actual operation.

[0126] Example 11:

[0127] like Figure 4 As shown, based on Embodiment 1, the report generation subsystem provided in this embodiment of the invention includes:

[0128] The path expansion module is used to expand the initial test paths generated by the path selection subsystem. Positive examples consist of path segments covering key logical nodes. The selection criteria combine the key node distribution characteristics identified by the network construction subsystem and the path weights calculated by the path selection subsystem. Negative examples come from two situations: logical chain connection regions that have not reached the dynamic threshold, or path segments with low weights in the Monte Carlo simulation.

[0129] The dual-channel feedback module extracts the path features of positive examples into multi-dimensional vectors, including dimensions such as key node coverage and neighboring node differences; the missing features of negative examples constitute the comparison samples; the optimization process iteratively adjusts the transition probability distribution of the test path, so that the positive example feature vector gradually approaches the topological centrality distribution of the logical chain network, while the negative example feature vector is pushed away from the effective coverage area.

[0130] The path optimization module is used to evaluate the node criticality scores determined by the network construction subsystem, the path weight coefficients assigned by the path selection subsystem, and the coverage gap data generated by the comparison of positive and negative examples. The optimized test path will dynamically increase the access probability of high critical nodes while maintaining the original weight distribution. The final set of paths retains the randomness characteristics of the Monte Carlo simulation.

[0131] In the above embodiments, the report generation subsystem of this embodiment achieves intelligent optimization and evaluation of test paths through the collaborative action of three modules. The path expansion module first completes the initial path classification process and establishes a positive and negative example sample library based on the network features and path weights provided by the preceding subsystems. The dual-channel feedback module constructs optimization objectives based on the feature differences between positive and negative examples and guides the path evolution direction through distribution adjustments in the vector space. The path optimization module finally integrates key score, weight coefficient, and coverage gap data to improve the detection efficiency of key logic while maintaining path diversity. The entire mechanism gives the final generated test path dual characteristics: it inherits the statistical randomness of Monte Carlo simulation and strengthens the coverage capability of the core logic chain through the optimization process. The output test report not only reflects the normal operating trajectory of the system but also effectively exposes potential anomalies in deep logic chains, providing multi-dimensional detection basis including structural features and probability distributions for software quality assessment.

[0132] Figure 5 A block diagram of an exemplary electronic device suitable for implementing embodiments of the present invention is shown.

[0133] Electronic devices may include a central processing unit / microprocessor / main control chip, etc.; a storage medium coupled to the central processing unit / microprocessor / main control chip, etc., and storing computer-executable instructions therein for performing the steps of various methods of embodiments of the present invention when executed by a processor.

[0134] The central processing unit / microprocessor / main control chip, etc., may include, but are not limited to, one or more processors or microprocessors.

[0135] Storage media may include, but are not limited to, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, and computer storage media (such as hard disks, floppy disks, solid-state drives, removable disks, CD-ROMs, DVD-ROMs, Blu-ray discs, etc.).

[0136] In addition, the electronic device may include (but is not limited to) a data bus, an input / output bus / external bus / device bus, a display, and input / output devices (e.g., keyboard, mouse, speaker, etc.).

[0137] The central processing unit / microprocessor / main control chip, etc., can communicate with external devices via I / O bus through wired or wireless networks (not shown).

[0138] The storage medium may also store at least one computer-executable instruction for performing the steps of various functions and / or methods in the embodiments described herein when run by a central processing unit / microprocessor / main control chip, etc.

[0139] In one embodiment, the at least one computer-executable instruction may also be compiled into or comprise a software product, wherein one or more computer-executable instructions are executed by a processor to perform the steps of the various functions and / or methods in the embodiments described herein.

[0140] Figure 6 A schematic diagram of a computer-readable storage medium according to an embodiment of the present invention is shown.

[0141] like Figure 6 As shown, instructions, such as computer-readable instructions, are stored on a non-transitory computer-readable storage medium. When the computer-readable instructions are executed by a processor, the various methods described above can be performed. The non-transitory computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-transitory non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. For example, the non-transitory computer-readable storage medium can be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions stored on the computer-readable storage medium, the various methods described above can be performed.

[0142] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0143] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0144] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0145] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the methods of the various embodiments of this invention through a computer device (which may be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0146] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A testing device for Internet software development, characterized in that, Include: The network construction subsystem is used to extract the call relationships between functional modules by monitoring the call sequence of functional modules during software runtime, and to calculate the call probability distribution of each module; then, it uses the entropy weight method to calculate the dependency strength between modules, and constructs an initial logical chain based on the dependency strength. By calculating the difference between adjacent logical chains and combining it with dynamic thresholds, key logical nodes are identified, forming a multi-level logical chain network. The path selection subsystem is used to define the transition probability of random sampling based on a multi-level logical chain network and generate multiple test paths through Monte Carlo simulation. The weight of each path is determined by the number of logical nodes covered by the path, and the paths with the highest weights are selected as the initial test paths. The report generation subsystem is used to construct positive and negative pairs in a multi-level logical chain network; optimize the test path through a loss function to ensure that the path covers the key nodes in the logical chain network; and weight and aggregate the test paths in the multi-level logical chain network to generate the final test report. The network construction subsystem includes: The runtime monitoring and relationship extraction module is used to monitor the function module calling behavior during software runtime in real time, capture the calling order and frequency between function modules, and form a dynamic calling sequence. Based on dynamic call sequences, the actual interaction patterns between functional modules are analyzed, the call probability distribution of each functional module is quantified, and the actual operating rules inside the monitoring software are revealed. The dependency strength modeling module is used to calculate the dependency strength between functional modules using probability distribution and entropy weight method, transforming abstract calling relationships into quantifiable logical chains. Functional modules whose dependency strength reaches the preset standard form a closely related logical chain, while dependencies that do not reach the preset standard serve as auxiliary branches, thus constructing the initial logical chain framework. The key node identification module is used to analyze the differences between adjacent logical chains. Combined with dynamic thresholds, it filters out the core branch points of the monitoring software function corresponding to the nodes with abrupt changes in differences in the initial logical chain framework, and marks them as key logical nodes. Finally, a multi-level logical network containing the trunk chain and branch chains is formed. The dependency strength modeling module includes: The probability distribution data preparation submodule is used to call the probability distribution data of the functional modules, extract the transition probability matrix between each functional module, and record the conditional probability of calling functional module B from any functional module A, thus forming the basic dataset for dependency strength calculation. The information uncertainty measurement submodule is used to calculate the information entropy value of the call distribution for each functional module. The information entropy value reflects the degree of uncertainty in the flow of information from a functional module to other modules when the functional module is called. Based on the entropy value calculation results of each functional module, the results are converted into weight coefficients through normalization. The dependency strength comprehensive calculation submodule is used to calculate the comprehensive dependency strength index between functional modules by combining the transition probability matrix and weight coefficients.

2. The testing device for Internet software development as described in claim 1, characterized in that, The operation monitoring and relationship extraction module includes: The behavior capture and pattern extraction submodule is used to continuously observe and record the activation order and interaction frequency between various functional modules. The call order is obtained from the activation order, and the call frequency is obtained from the interaction frequency, forming a complete call time sequence record. Based on the time sequence record, the collaboration mode between functional modules is analyzed, the distribution of triggering of each functional module is calculated, and the basic characteristics of the software monitoring software operation are discovered. The correlation quantification and structure construction submodule is used to evaluate the correlation between functional modules after understanding the probability distribution of each functional module being triggered, and to transform the collaborative relationship of functional modules into specific numerical indicators. Functional module combinations with a correlation degree exceeding the set standard are identified as strong correlations and constitute the core path of software operation; combinations with a correlation degree not exceeding the set standard are used as supplementary paths, and a relationship network that initially reflects the initial logical chain framework of the monitoring software operation is initially established. The core element location and network improvement submodule is used to identify key turning points by comparing the differences between adjacent core paths, mark the location of key turning points, and form a hierarchical network that reveals the actual operating rules inside the monitoring software.

3. The testing device for Internet software development as described in claim 1, characterized in that, The key node identification module includes: The input data preparation submodule is used to extract the call frequency distribution and dependency strength distribution of adjacent logical chains from the dynamic call sequence; and to compare the differences in call probability distribution, dependency strength gradient change and topological structure of the following features of the two logical chains. The dynamic threshold adjustment submodule is used to dynamically adjust the difference determination threshold based on the overall dependency strength distribution of the current logical chain set; modules with a difference exceeding the dynamic threshold are determined to be mutation nodes. The key logic node generation submodule is used to associate mutation nodes with module roles in the initial logic chain framework. If a mutation node belongs to multiple highly different logic chains at the same time, it is determined to be a branch convergence point; if the dependency strength of a mutation node drops sharply in a single logic chain, it is determined to be an inter-chain switching point. Nodes constitute the core branch points in the multi-level logic network, connecting the main chain and auxiliary branch chains.

4. The testing device for Internet software development as described in claim 3, characterized in that, Differences in the calling probability distribution of the input data preparation submodule: Compare the calling probability offsets of the same functional modules in the two chains; Dependency strength gradient change: Analyze the difference in dependency strength between identical module pairs in two chains. If the difference exceeds the dynamic threshold, it is marked as a potential difference point. Topological structure difference: Check whether the overlapping modules of the two chains are in similar positions in their respective chains. If the positional offset is significant, increase the difference weight.

5. The testing device for Internet software development as described in claim 3, characterized in that, The key logic node generation submodule includes: The multi-chain cross-validation unit is used to take the set of mutation nodes obtained from the dynamic threshold adjustment submodule as input, and backtrack the adjacent logical chain association information recorded by the input data preparation submodule. For each mutation node, it retrieves all logical chain identifiers to which it belongs and extracts the difference matrix between logical chains, including the call probability offset, dependency strength difference and topological position offset. The single-chain strength decay analysis unit is used to switch to single-chain analysis mode for mutation nodes that have not passed multi-chain verification. The initial logical chain framework provided by the strength modeling module is activated to locate the sequence position of the mutation node in its logical chain. When the following conditions are met simultaneously, it is determined to be an inter-chain switching point: The conflict resolution and final determination unit is used to use the topological structure difference weight as a decision factor when the mutation node simultaneously satisfies some features of the convergence point and the switching point; if the mutation node is in a structurally symmetrical position in the majority chain, the convergence point determination is strengthened. If the location distribution is discrete and accompanied by drastic fluctuations in the call probability, then the switching point determination should be strengthened.

6. The testing device for Internet software development as described in claim 5, characterized in that, When a mutation node appears in three or more logical chains, and the difference index between the logical chains all exceeds the adjusted dynamic threshold, the convergence point determination condition is triggered. At this time, the dependency strength change pattern of each chain at the mutation node is checked: if the calling probability of the node in each chain remains stable, and the dependency strength distribution in the output direction shows multi-way balance, then it is confirmed that the mutation node undertakes the branch convergence function.

7. The testing device for Internet software development as described in claim 1, characterized in that, The route selection subsystem includes: The logical association module is used to determine the transition probability by the dependency strength and call probability in the multi-level logical chain network. Monte Carlo simulation generates paths based on the multi-level logical chain network structure. Weight calculation depends on the key node markings in the multi-level logical chain network to form a test path optimization closed loop. The transition probability definition module is used to define the random sampling probability of transitions between functional modules in a multi-level logical chain network, based on the functional module call probability distribution and dependency strength. The path simulation module is used to simulate the possible paths of the monitoring software through probability sampling and generate several test paths. The weight of each test path is dynamically determined by the number of key nodes it covers, and the initial test path with the greatest impact on the core logic of the monitoring software is selected.

8. The testing device for Internet software development as described in claim 1, characterized in that, The report generation subsystem includes: The path expansion module is used to expand the initial test paths generated by the path selection subsystem. Positive examples consist of path segments covering key logical nodes. The selection criteria combine the key node distribution characteristics identified by the network construction subsystem and the path weights calculated by the path selection subsystem. Negative examples come from two situations: logical chain connection regions that have not reached the dynamic threshold, or path segments with low weights in the Monte Carlo simulation. The dual-channel feedback module extracts the path features of positive examples into multi-dimensional vectors, including the coverage of key nodes and the difference between adjacent nodes; the missing features of negative examples constitute the comparison samples; the optimization process iteratively adjusts the transition probability distribution of the test path, so that the positive example feature vector gradually approaches the topological centrality distribution of the logical chain network, while the negative example feature vector is pushed away from the effective coverage area. The path optimization module is used to evaluate the node criticality scores determined by the network construction subsystem, the path weight coefficients assigned by the path selection subsystem, and the coverage gap data generated by the comparison of positive and negative examples. The optimized test path will dynamically increase the access probability of high critical nodes while maintaining the original weight distribution. The final set of paths retains the randomness characteristics of the Monte Carlo simulation.

Images