Enterprise software development whole-process supervision method and system based on knowledge graph

Abstract

This invention discloses a method and system for monitoring the entire enterprise software development process based on knowledge graphs, relating to the field of software development technology. It includes analyzing the software development process to obtain information on each stage, and then obtaining a core set of process nodes based on that information. This invention achieves precise focusing of monitored nodes by constructing an initial set of process nodes and calculating core node coefficients. By using core process nodes as entities and node relationships as edges, and employing a graph database to construct an initial knowledge graph, it achieves visualization and structured representation of the topological relationships of nodes throughout the software development process. By extracting topological features such as the number of parent nodes and the number of transition steps of core process nodes, it performs two-dimensional discrete grouping and sets weights to obtain a baseline monitoring frequency. Combined with the number of lower-level nodes, it calculates the monitoring transmission coefficient, achieving differentiated and precise setting of the monitoring frequency, thus improving the continuity and efficiency of the monitoring process.

Description

Technical Field

[0001] This invention relates to the field of software development technology, specifically to a method and system for monitoring the entire enterprise software development process based on knowledge graphs. Background Technology

[0002] Enterprise software development is a complex end-to-end system encompassing requirements analysis, architecture design, coding development, testing and verification, deployment and launch, and operation and maintenance iterations. Each stage of the process is interconnected and the nodes are closely related. Oversights or deviations in any stage can lead to substandard software quality, project delays, cost overruns, or even security vulnerabilities after system launch. Enterprise software development supervision, as a core means of ensuring the smooth progress of software projects, can track the execution progress, quality compliance, and allocation of responsibilities at each process node in real time, promptly identify and issue warnings for problems such as process blockages and deviations from standards, thereby improving software development efficiency.

[0003] Currently, the supervision of enterprise software development faces several challenges, including the inability to accurately analyze the software development process and set appropriate supervision strategies based on it. Existing technologies often monitor the process at a fixed frequency without distinguishing the importance of nodes. This results in insufficient supervision of core nodes and redundant nodes consuming significant supervision resources, reducing supervision efficiency, easily overlooking key risk points, failing to accurately identify core process nodes, lacking targeted supervision, and making it difficult to trace the path of problems. Alternatively, the supervision frequency may be set directly based on the amount of input during the development process—more input means more supervision, less input means less supervision. This makes it impossible to supervise strongly related process nodes in a timely manner, reducing the efficiency and reliability of software development. Summary of the Invention

[0004] To address the aforementioned technical problems, this paper provides a knowledge graph-based method and system for monitoring the entire enterprise software development process. This technical solution solves the problems mentioned in the background, such as the inability to accurately analyze the software development process and set appropriate monitoring strategies based on the process. Existing technologies often monitor the process at a fixed frequency without distinguishing the importance of nodes. This results in insufficient monitoring of core nodes and redundant nodes consuming a large amount of monitoring resources, which reduces monitoring efficiency, easily overlooks key risk points, fails to accurately identify core process nodes, lacks targeted monitoring, and makes it difficult to trace the path of problems. Alternatively, the monitoring frequency is set directly based on the amount of input to process nodes during development—more input means more monitoring, less input means less monitoring. This makes it impossible to monitor strongly related process nodes in a timely manner, reducing the efficiency and reliability of software development.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A knowledge graph-based method for monitoring the entire enterprise software development process includes: Based on software development process analysis, information on the stages of the software development process is obtained. The stages of the software development process include the requirements analysis stage, architecture design stage, detailed design stage, coding development stage, unit testing stage, integration testing stage, system testing stage, deployment and launch stage, and operation and maintenance iteration stage. Based on the software development process phase information, obtain the core process node set; Based on the basic database of process nodes, the core process node set is extracted as the entity set of the knowledge graph; Based on the node association table, a graph database is used to construct the topology of the knowledge graph, with entities corresponding to process nodes and edges corresponding to node associations, forming an initial knowledge graph; Based on the initial knowledge graph, obtain the software development knowledge graph; Based on the software development knowledge graph, obtain the regulatory frequency information corresponding to each core process node; The entire software development process is monitored based on the monitoring frequency information corresponding to each core process node.

[0006] Preferably, obtaining the core process node set based on the software development process stage information specifically includes: Based on the software development process stage information, the specific execution steps of each process stage are broken down to obtain the development process node information. The development process nodes include requirement review nodes, architecture review nodes, code submission nodes, test case design nodes, test execution nodes, defect fixing nodes, and version release nodes. Based on the development process node information, an initial process node set is constructed; Based on the initial process node set, analyze the relationships between the nodes to obtain a node relationship table; Based on software development standards, define the node attribute information corresponding to each initial process node and construct a node attribute table; The node attribute information includes node name, stage, executing entity, output results, and quality standards; Based on the node association table, obtain the node core coefficient corresponding to each initial process node; Based on the regulatory requirements for software development, a necessity threshold should be set; Based on the node core coefficient and necessity threshold corresponding to each initial process node, determine whether the initial process node is redundant. If the node core coefficient does not exceed the necessity threshold, remove the initial process node to obtain the core process node set. The core process node set, node attribute table, and relationship table are stored together to build a basic database for process nodes.

[0007] Preferably, obtaining the node core coefficient corresponding to each initial process node based on the node association table specifically includes: Based on the node association table, the initial process nodes corresponding to the dependencies are divided into the same dataset to obtain the dependency dataset; Traverse the dependency dataset and use the total number of times each initial process node is depended on by other nodes as the node association value of that initial process node; The ratio of the node association value of each initial process node to the maximum node association value is used as the node association coefficient of that initial process node. Based on the dependency dataset, the dependency nodes corresponding to each initial process node are used as the progressive association nodes of that initial process node; Based on software development requirements, obtain the call frequency of each progressively related node to the initial process node; Based on the dependency dataset, the average call frequency is used as the execution frequency of the initial process node; The maximum value of the execution frequency and the call frequency of each initial process node is used as the node execution coefficient of that initial process node; Based on the node association table, obtain the node influence coefficient of each initial process node; Based on the node analysis requirements, set the weights corresponding to the node correlation coefficient, node execution coefficient, and node influence coefficient; Based on the node correlation coefficient, node execution coefficient, node influence coefficient and corresponding weights, the node core coefficient is obtained by weighted summation.

[0008] Preferably, obtaining the node influence coefficient of each initial process node based on the node association table specifically includes: Based on the node association table, taking any initial process node as the foundation, traverse all subsequent paths using the depth-first traversal algorithm, take all nodes on the subsequent paths as the subsequent nodes of the initial process node, take the initial process node as the feature node of the subsequent nodes, and obtain the set of subsequent nodes for each initial process node. Based on the set of subsequent nodes, obtain the preceding node information corresponding to each subsequent node; Based on the preceding node information corresponding to each subsequent node, the preceding node that has a direct or indirect connection with the feature node corresponding to the subsequent node is taken as the influence transmission node of the subsequent node. The ratio of the number of nodes whose influence is transmitted to each subsequent node to the total number of preceding nodes is used as the influence range coefficient of that subsequent node. Based on the set of subsequent nodes for each initial process node, the number of nodes with the shortest interval between each subsequent node and the initial process node is used as the influence depth coefficient of that subsequent node; Based on the set of subsequent nodes of each initial process node, the node influence coefficient of each initial process node is obtained according to the influence range coefficient and influence depth coefficient.

[0009] Preferably, the step of obtaining a software development knowledge graph based on the initial knowledge graph specifically includes: Based on the initial knowledge graph, all nodes are traversed, and core process nodes without subsequent nodes are taken as terminal nodes. Traverse the initial knowledge graph, take any core process node as the target node, take the core process node pointing to the target node as the parent node of the target node, and obtain the number of parent nodes corresponding to each core process node. The shortest distance from each core process node to all terminal nodes is calculated based on Dijkstra's algorithm, and the distance is defined as the number of flow steps of that core process node. Based on the number of parent nodes and the number of steps corresponding to each core process node, the node attributes of the initial knowledge graph are labeled to obtain the software development knowledge graph.

[0010] Preferably, obtaining the regulatory frequency information corresponding to each core process node based on the software development knowledge graph specifically includes: Based on the software development knowledge graph, and using the number of parent nodes and the number of flow steps corresponding to each core process node, a node topology feature set is obtained. Based on the node topology feature set, obtain the number of parent nodes and the number of flow steps of the core process node corresponding to the maximum and minimum values ​​of the node core coefficient; The difference between the number of parent nodes of the core process node corresponding to the maximum and minimum values ​​of the node core coefficient is used as the in-degree feature interval. The difference between the maximum and minimum number of upper-level nodes is taken as the total in-degree interval, and the ratio of the total in-degree interval to the in-degree feature interval is taken as the number of in-degree groups, with the minimum number of in-degree groups being 2. Based on the node topology feature set, and using the maximum and minimum values ​​of the number of in-degree groups and the number of superior nodes as a basis, the number of superior nodes is discretely grouped based on equally divided intervals to obtain the in-degree grouping results. The difference between the number of flow steps of the core process nodes corresponding to the maximum and minimum values ​​of the node core coefficient is used as the distance feature interval; The difference between the maximum and minimum number of circulation steps is taken as the total distance interval, and the ratio of the total distance interval to the distance feature interval is taken as the distance group number, with the minimum value of the distance group number being 2. Based on the node topology feature set, and using the maximum and minimum values ​​of the number of distance groups and the number of flow steps as a basis, the terminal distance is discretely grouped based on equally divided intervals to obtain the distance grouping results; Based on the in-degree grouping results and the distance grouping results, a topology grouping matrix is ​​constructed through two-dimensional cross-logical operations, dividing all core process nodes into several independent topology groups; Based on the in-degree grouping results and the distance grouping results, set the weights corresponding to each in-degree group and distance group; Based on the project cycle items of enterprise software development, the benchmark regulatory frequency corresponding to each independent topology group is obtained according to the weights corresponding to each in-degree group and distance group; Based on the software development knowledge graph, take any core process node as the target node, and take the core process node pointed to by the target node as the subordinate node of the target node, and obtain the number of subordinate nodes of each core process node. Based on the number of subordinate nodes of each core process node, the maximum number of subordinate nodes in each independent topology group is taken as the base number of nodes; The ratio of the number of subordinate nodes to the number of baseline nodes for each core process node in each independent topology group is used as the regulatory transmission coefficient for that core process node. The product of the regulatory transmission coefficient corresponding to each core process node and the benchmark regulatory frequency is taken as the regulatory frequency of that core process node.

[0011] Furthermore, a knowledge graph-based enterprise software development end-to-end monitoring system is proposed to implement the aforementioned monitoring methods, including: The main control module is used to construct the topology of a knowledge graph based on a node association table and a graph database. Entities correspond to process nodes, and edges correspond to node associations to form an initial knowledge graph. Based on the initial knowledge graph, a software development knowledge graph is obtained. Based on the software development knowledge graph, in-degree grouping results and distance grouping results are obtained. Based on the in-degree grouping results and distance grouping results, all core process nodes are divided into several independent topology groups, and the regulatory frequency information corresponding to each core process node is obtained. The information acquisition module is used to acquire software development process stage information based on software development process analysis, break down the specific execution links of each process stage according to the software development process stage information, acquire development process node information, analyze the relationship between nodes according to the initial process node set, acquire node relationship table, define node attribute information corresponding to each initial process node based on software development standards, and construct node attribute table. The process node evaluation module is used to obtain the node association coefficient, node execution coefficient and node influence coefficient according to the node association table, obtain the node core coefficient based on the node association coefficient, node execution coefficient, node influence coefficient and corresponding weights, and obtain the core process node set based on the node core coefficient and necessity threshold corresponding to each initial process node.

[0012] Optionally, the main control module specifically includes: The control unit is used to obtain in-degree grouping results and distance grouping results based on the software development knowledge graph, divide all core process nodes into several independent topology groups based on the in-degree grouping results and distance grouping results, and obtain the regulatory frequency information corresponding to each core process node. An information receiving unit interacts with an information acquisition module and a process node evaluation module to receive data and transmit it to a knowledge graph construction unit. The knowledge graph construction unit is used to construct the topology of the knowledge graph based on the node association table and using a graph database. Entities correspond to process nodes, and edges correspond to node associations to form an initial knowledge graph. Based on the initial knowledge graph, a software development knowledge graph is obtained.

[0013] Optionally, the information acquisition module specifically includes: The first acquisition unit is used to acquire software development process stage information based on software development process analysis, and to break down the specific execution links of each process stage according to the software development process stage information to acquire development process node information. The second acquisition unit is used to analyze the relationship between nodes based on the initial process node set, acquire a node relationship table, define the node attribute information corresponding to each initial process node based on software development standards, and construct a node attribute table.

[0014] Optionally, the process node evaluation module specifically includes: The node analysis unit is used to obtain the node association coefficient, node execution coefficient, and node influence coefficient based on the node association table. The core node evaluation unit is used to obtain the node core coefficient based on the node association coefficient, node execution coefficient, node influence coefficient and corresponding weight, and to obtain the core process node set based on the node core coefficient and necessity threshold corresponding to each initial process node.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes a knowledge graph-based method and system for monitoring the entire enterprise software development process. By constructing an initial set of process nodes and calculating the core coefficients of the nodes, precise focus on monitoring nodes is achieved. By using core process nodes as entities and node relationships as edges, and employing a graph database to construct an initial knowledge graph, visualization and structured representation of the topological relationships of nodes throughout the software development process are realized. By extracting topological features such as the number of superior nodes and the number of flow steps of core process nodes, two-dimensional discrete grouping and weighting are performed to obtain a baseline monitoring frequency. Combined with the number of subordinate nodes, the monitoring transmission coefficient is calculated, enabling differentiated and precise setting of the monitoring frequency, thereby improving the continuity and efficiency of the monitoring process. Attached Figure Description

[0016] Figure 1 This is a flowchart of a knowledge graph-based enterprise software development end-to-end monitoring method proposed in this invention. Figure 2 This invention provides a flowchart for obtaining the basic database of process nodes. Figure 3 This is a flowchart of the node core coefficient acquisition process in this invention; Figure 4 This is a flowchart of the node influence coefficient acquisition process in this invention; Figure 5 This is a flowchart of the software development knowledge graph acquisition process in this invention. Figure 6 This invention presents a structural block diagram of a knowledge graph-based enterprise software development end-to-end monitoring system. Detailed Implementation

[0017] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0018] Reference Figure 1 - Figure 5 As shown in the figure, an enterprise software development end-to-end monitoring method based on knowledge graphs in this embodiment of the invention includes: Based on software development process analysis, information on the stages of the software development process is obtained. The stages of the software development process include the requirements analysis stage, architecture design stage, detailed design stage, coding development stage, unit testing stage, integration testing stage, system testing stage, deployment and launch stage, and operation and maintenance iteration stage. Based on the software development process phase information, obtain the core process node set; Specifically, based on the software development process stage information, the core process node set is obtained, including: Based on the software development process stage information, the specific execution steps of each process stage are broken down to obtain the development process node information. The development process nodes include requirement review nodes, architecture review nodes, code submission nodes, test case design nodes, test execution nodes, defect fixing nodes, and version release nodes. Based on the development process node information, an initial process node set is constructed; Based on the initial process node set, analyze the relationships between the nodes to obtain a node relationship table; Based on software development standards, define the node attribute information corresponding to each initial process node and construct a node attribute table; The node attribute information includes node name, stage, executing entity, output results, and quality standards; Based on the node association table, obtain the node core coefficient corresponding to each initial process node; Based on the regulatory requirements for software development, a necessity threshold should be set; Based on the node core coefficient and necessity threshold corresponding to each initial process node, determine whether the initial process node is redundant. If the node core coefficient does not exceed the necessity threshold, remove the initial process node to obtain the core process node set. The core process node set, node attribute table, and relationship table are stored together to build a basic database for process nodes.

[0019] This solution precisely breaks down the specific execution steps of each stage of the software development process, extracting key development process nodes such as requirements review and code submission. This transforms the abstract development process into concrete, monitorable node units, providing a clear node carrier for subsequent targeted supervision. Based on these decomposed development process nodes, an initial process node set is constructed, achieving comprehensive coverage of key execution steps throughout the entire software development process. This includes both core business nodes and supporting nodes, providing a complete basic node pool for subsequent core node selection. By analyzing the relationships between the initial process nodes and forming a node relationship table, the solution clearly defines the logical relationships such as dependencies and connections between nodes, transforming implicit relationships between nodes into... The explicit and structured tabular data provides a logical basis for assessing the importance of nodes. Based on software development standards, the name, stage, executing entity, and output of each initial node are clearly defined, constructing a standardized node attribute table that clearly defines the execution requirements, responsible entity, and quality standards for each node. The core coefficient of each initial node is calculated through a node relationship table, transforming the multi-dimensional characteristics of nodes, such as their degree of association, execution frequency, and scope of influence, into a single quantitative indicator. Redundant nodes with insufficient core coefficients are removed based on a necessity threshold, focusing on the core process node set. This avoids wasting regulatory resources on non-critical nodes, enabling precise allocation of regulatory resources and significantly improving regulatory efficiency and targeting.

[0020] In this embodiment, the necessity threshold is 0.6. It can be understood that, based on the Pareto principle, retaining 20% ​​of the core nodes can support 80% of the core business flow. The necessity score of 0.6 is the dividing value between core nodes and redundant nodes, which can accurately screen core process nodes. When the threshold is set to 0.6, it will not cause the core nodes to be missing due to the threshold being too high, nor will it cause too many redundant nodes due to the threshold being too low. It perfectly balances the simplification of the process and the integrity of execution, and adapts to the actual needs of enterprise software development full-process supervision.

[0021] Specifically, based on the node association table, the node core coefficient corresponding to each initial process node is obtained, including: Based on the node association table, the initial process nodes corresponding to the dependencies are divided into the same dataset to obtain the dependency dataset; Traverse the dependency dataset and use the total number of times each initial process node is depended on by other nodes as the node association value of that initial process node; The ratio of the node association value of each initial process node to the maximum node association value is used as the node association coefficient of that initial process node. Based on the dependency dataset, the dependency nodes corresponding to each initial process node are used as the progressive association nodes of that initial process node; Based on software development requirements, obtain the call frequency of each progressively related node to the initial process node; Based on the dependency dataset, the average call frequency is used as the execution frequency of the initial process node; The maximum value of the execution frequency and the call frequency of each initial process node is used as the node execution coefficient of that initial process node; Based on the node association table, obtain the node influence coefficient of each initial process node; Based on the node analysis requirements, set the weights corresponding to the node correlation coefficient, node execution coefficient, and node influence coefficient; Based on the node correlation coefficient, node execution coefficient, node influence coefficient and corresponding weights, the node core coefficient is obtained by weighted summation.

[0022] This solution constructs a dependency dataset through a node association table, precisely focusing on the dependency relationships between nodes. It transforms ambiguous process dependencies into a structured dataset, clearly defining the upstream and downstream dependency boundaries between nodes. This lays a clear data foundation for the subsequent quantitative calculation of node association values ​​and execution frequencies. The node association value is obtained by counting the total number of times a node is depended on by other nodes, and then normalized to obtain the node association coefficient. This transforms the "degree of dependency" of a node into a standardized value in the 0-1 range, intuitively reflecting the core dependency status of a node in the process. This provides a unified and comparable quantitative standard for the importance of node associations. Finally, by identifying progressively related nodes and counting their call frequencies, the node execution frequency is calculated. The node execution coefficient is normalized to accurately quantify the execution activity and business call intensity of nodes in the actual development process. The node influence coefficient is introduced to quantify the scope and depth of the node's influence on subsequent processes. It supplements the importance dimension beyond "association and dependence" and "execution frequency" to make node evaluation cover the full dimensions of "degree of dependence - actual operation intensity - subsequent impact capability". The node core coefficient is obtained by weighted summation of association coefficient, execution coefficient and influence coefficient based on preset weights. Multi-dimensional features are integrated into a single quantitative indicator, realizing a comprehensive, objective and accurate quantitative evaluation of the coreness of nodes, accurately identifying key nodes with high core coefficients, and tilting regulatory resources toward core nodes.

[0023] In this embodiment, the node correlation coefficient has a weight of 0.4, the node execution coefficient has a weight of 0.25, and the node influence coefficient has a weight of 0.35. This can be understood as follows: the node correlation coefficient represents the strong dependency flow characteristics of the entire enterprise software development process. The number of times a process node is depended upon is a core prerequisite for its existence: nodes without prerequisite dependencies cannot execute, and nodes without dependencies are redundant. The degree of dependency directly determines the core position of a node in the process topology, hence it is assigned the highest weight of 0.4. The influence scope weight represents the scope of impact of node anomalies, a core dimension of software development risk management. Node failures will directly propagate to subsequent processes, causing rework, delays, quality defects, and other problems. Its importance is second only to node dependency relationships, hence it is assigned the second highest weight of 0.35. As for the execution frequency weight, the node execution frequency only reflects the execution frequency of the process and is an auxiliary quantitative indicator. It does not determine the core necessity and risk level of a node (e.g., low-frequency version release nodes are core nodes, and high-frequency document archiving nodes are secondary nodes), hence it is assigned the lowest weight of 0.25.

[0024] Specifically, based on the node association table, the node influence coefficient of each initial process node is obtained, including: Based on the node association table, taking any initial process node as the foundation, traverse all subsequent paths using the depth-first traversal algorithm, take all nodes on the subsequent paths as the subsequent nodes of the initial process node, take the initial process node as the feature node of the subsequent nodes, and obtain the set of subsequent nodes for each initial process node. Based on the set of subsequent nodes, obtain the preceding node information corresponding to each subsequent node; Based on the preceding node information corresponding to each subsequent node, the preceding node that has a direct or indirect connection with the feature node corresponding to the subsequent node is taken as the influence transmission node of the subsequent node. The ratio of the number of nodes whose influence is transmitted to each subsequent node to the total number of preceding nodes is used as the influence range coefficient of that subsequent node. Based on the set of subsequent nodes for each initial process node, the number of nodes with the shortest interval between each subsequent node and the initial process node is used as the influence depth coefficient of that subsequent node; Based on the set of subsequent nodes of each initial process node, the node influence coefficient of each initial process node is obtained according to the influence range coefficient and influence depth coefficient.

[0025] In this solution, a depth-first traversal algorithm is used to traverse all subsequent paths of the initial process node, accurately obtaining the set of subsequent nodes and completely locking all related nodes of the node to the downstream process. By using the information of the preceding nodes of the subsequent nodes, nodes with direct or indirect connections to the current node are selected to influence the transmission of influence. The ratio of the number of nodes with influence transmission to the total number of preceding nodes is used as the influence range coefficient, transforming the abstract "influence coverage" into a quantifiable numerical indicator, providing an objective standard for judging the breadth of influence. The number of nodes with the shortest interval between the subsequent node and the current node is used as the influence depth coefficient, accurately quantifying the depth of influence transmission. The fewer the number of interval nodes, the more direct the influence; the more the number of interval nodes, the more indirect but wider the influence, supplementing the quantitative dimension of influence depth, making the representation of node influence more three-dimensional. By combining the influence range coefficient (breadth) and the influence depth coefficient (depth), the node influence coefficient is obtained, integrating the "coverage range" and "transmission depth" of the node to the downstream process into a single quantitative indicator, replacing the traditional method of manual subjective judgment of the size of node influence, and realizing a comprehensive, accurate, and objective quantification of the degree of node influence.

[0026] Understandably, existing technologies often use the downstream nodes of each node as the scope of influence of the node, or the ratio of downstream nodes to total nodes to assess the scope of influence of a node. However, for each downstream node, the degree of influence of the current node on each downstream node is completely different. If the entire downstream node is used as the scope of influence, the accuracy and reliability of the assessment results will be greatly reduced.

[0027] Based on the basic database of process nodes, the core process node set is extracted as the entity set of the knowledge graph; Based on the node association table, a graph database is used to construct the topology of the knowledge graph, with entities corresponding to process nodes and edges corresponding to node associations, forming an initial knowledge graph; Based on the initial knowledge graph, obtain the software development knowledge graph; Specifically, based on the initial knowledge graph, a software development knowledge graph is obtained, including: Based on the initial knowledge graph, all nodes are traversed, and core process nodes without subsequent nodes are taken as terminal nodes. Traverse the initial knowledge graph, take any core process node as the target node, take the core process node pointing to the target node as the parent node of the target node, and obtain the number of parent nodes corresponding to each core process node. The shortest distance from each core process node to all terminal nodes is calculated based on Dijkstra's algorithm, and the distance is defined as the number of flow steps of that core process node. Based on the number of parent nodes and the number of steps corresponding to each core process node, the node attributes of the initial knowledge graph are labeled to obtain the software development knowledge graph.

[0028] In this scheme, core process nodes without subsequent nodes are selected as terminal nodes by traversing the initial knowledge graph. This clarifies the final closing stage of the entire software development process, making the "end point" of node flow clearly identifiable. This lays a clear endpoint benchmark for subsequent quantification of node flow distance and analysis of process transmission logic. By using the number of superior nodes of each core process node (the number of core nodes pointing to that node), the upstream correlation of the node is transformed into a quantifiable numerical indicator, intuitively reflecting the "dependency concentration" of the node in the process. This provides core in-degree data for subsequent node topology feature analysis. Based on Dijkstra's algorithm, the shortest distance from the core node to all terminal nodes is calculated and defined as the number of flow steps, accurately quantifying the "transmission position" of the node in the process—the fewer the flow steps, the closer the node is to the terminal and the more direct the impact; the more flow steps, the closer the node is to the source and the more profound the impact. By labeling the two core attributes of the number of parent nodes (in-degree strength) and the number of flow steps (transmission depth), the initial knowledge graph is upgraded from a basic model that "only presents the relationship between nodes" to a model that "relationships + quantified topological features". The process location, dependency strength and transmission ability of each node are clearly supported by data.

[0029] In this embodiment, the shortest distance from each core process node to all terminal nodes is calculated based on Dijkstra's algorithm as follows: Set Node To the terminal node The distance is initial value The initial distances of other nodes are set to infinity. For the first One terminal node; Based on node relationships and using a knowledge graph as a foundation, distances are iteratively updated: if edges exist... → ,but ; Get Node The minimum distance to all terminal nodes is taken as the shortest distance from that node to all terminal nodes: Based on the software development knowledge graph, obtain the regulatory frequency information corresponding to each core process node; Specifically, based on the software development knowledge graph, the monitoring frequency information corresponding to each core process node is obtained, including: Based on the software development knowledge graph, and using the number of parent nodes and the number of flow steps corresponding to each core process node, a node topology feature set is obtained. Based on the node topology feature set, obtain the number of parent nodes and the number of flow steps of the core process node corresponding to the maximum and minimum values ​​of the node core coefficient; The difference between the number of parent nodes of the core process node corresponding to the maximum and minimum values ​​of the node core coefficient is used as the in-degree feature interval. The difference between the maximum and minimum number of upper-level nodes is taken as the total in-degree interval, and the ratio of the total in-degree interval to the in-degree feature interval is taken as the number of in-degree groups, with the minimum number of in-degree groups being 2. Based on the node topology feature set, and using the maximum and minimum values ​​of the number of in-degree groups and the number of superior nodes as a basis, the number of superior nodes is discretely grouped based on equally divided intervals to obtain the in-degree grouping results. The difference between the number of flow steps of the core process nodes corresponding to the maximum and minimum values ​​of the node core coefficient is used as the distance feature interval; The difference between the maximum and minimum number of circulation steps is taken as the total distance interval, and the ratio of the total distance interval to the distance feature interval is taken as the distance group number, with the minimum value of the distance group number being 2. Based on the node topology feature set, and using the maximum and minimum values ​​of the number of distance groups and the number of flow steps as a basis, the terminal distance is discretely grouped based on equally divided intervals to obtain the distance grouping results; Based on the in-degree grouping results and the distance grouping results, a topology grouping matrix is ​​constructed through two-dimensional cross-logical operations, dividing all core process nodes into several independent topology groups; Based on the in-degree grouping results and the distance grouping results, set the weights corresponding to each in-degree group and distance group; Based on the project cycle items of enterprise software development, the benchmark regulatory frequency corresponding to each independent topology group is obtained according to the weights corresponding to each in-degree group and distance group; Based on the software development knowledge graph, take any core process node as the target node, and take the core process node pointed to by the target node as the subordinate node of the target node, and obtain the number of subordinate nodes of each core process node. Based on the number of subordinate nodes of each core process node, the maximum number of subordinate nodes in each independent topology group is taken as the base number of nodes; The ratio of the number of subordinate nodes to the number of baseline nodes for each core process node in each independent topology group is used as the regulatory transmission coefficient for that core process node. The product of the regulatory transmission coefficient corresponding to each core process node and the benchmark regulatory frequency is taken as the regulatory frequency of that core process node.

[0030] This solution uses the number of parent nodes (in-degree strength) and the number of flow steps (transmission depth) of nodes in a software development knowledge graph as the core to construct a node topology feature set. This transforms the process positioning and association characteristics of nodes into calculable quantitative indicators, providing precise feature support for subsequent scientific grouping and frequency calculation. Differential weights are set based on the importance of in-degree grouping and distance grouping. A baseline monitoring frequency is calculated by combining the enterprise's software development project cycle (e.g., longer-cycle projects require denser monitoring of intermediate nodes), ensuring that the frequency setting aligns with both node topology characteristics and actual project operation and maintenance needs. The monitoring transmission coefficient is calculated using the number of lower-level nodes (the number of downstream associated nodes), quantifying the node's ability to transmit its influence to subsequent processes—the more lower-level nodes, the larger the transmission coefficient, and the higher the monitoring frequency. This enables secondary differentiated management of nodes within the same group, directing monitoring resources towards nodes with a wider impact. Through the dynamic calculation logic of "baseline monitoring frequency × monitoring transmission coefficient," the topology characteristics, project cycle, and downstream impact of nodes are integrated to ultimately obtain a unique monitoring frequency for each core process node, improving the efficiency of monitoring resource utilization.

[0031] In this embodiment, the feature intervals of in-degree and distance are defined by the maximum and minimum values ​​of the node core coefficients. The number of groups is calculated based on the total interval (minimum value is 2). This ensures that the grouping does not miss key feature differences, nor does it waste resources due to overly fine grouping. The grouping results align with data distribution patterns, improving the adaptability of subsequent monitoring frequencies. A topology grouping matrix is ​​constructed based on in-degree grouping (upstream dependency strength) and distance grouping (downstream transmission depth). Cross-logic operations divide core process nodes into independent topology groups, allowing nodes in the same group to have similar process roles. For example, if the number of in-degree groups is 3, then the number of parent nodes is used as the basis for grouping. and Calculate the step size of the in-degree interval: Based on the range of the number of parent nodes corresponding to each core process node, the nodes are divided into high-in-degree groups. Medium-intensity group Low inclination group ; If the distance grouping number is 3, based on and Calculate the step size of the distance interval: The nodes are divided into near-terminal groups based on the range of the number of steps (i.e., distance) corresponding to each core process node. In the middle terminal group Remote terminal group ; Based on the in-degree grouping results and the distance grouping results, a topology grouping matrix M is constructed through a two-dimensional cross-logic operation, dividing all nodes into 9 independent topology groups, namely the high in-degree near-terminal group. High-in-depth terminal group High-depth remote terminal group Medium-input near-terminal group Mid-in-depth terminal group Medium-depth remote terminal group Low-intrusion near-terminal group Low-in-degree terminal group Low-input remote terminal group ; Based on enterprise standard regulatory cycle (Unit: times / month) Based on the risk transmission logic of topology groups, the upper limit of the regulatory frequency for each topology group is determined through weighted logical operations: In the formula, express The upper limit of the group's regulatory frequency, Indicates the first Weights for each in-degree group, such as high-in-degree weights. In-degree weight Low in-degree weight , Indicates the first Weights for distance-based groupings, such as near-terminal weights. Mid-terminal weight Remote terminal weight Specifically, the weights and enterprise standard regulatory cycles corresponding to each in-degree group and distance group can be set according to actual needs.

[0032] The entire software development process is monitored based on the monitoring frequency information corresponding to each core process node.

[0033] Reference Figure 6 As shown, further, combining the above-mentioned knowledge graph-based enterprise software development end-to-end monitoring method, a knowledge graph-based enterprise software development end-to-end monitoring system is proposed, including: The main control module is used to construct the topology of a knowledge graph based on a node association table and a graph database. Entities correspond to process nodes, and edges correspond to node associations to form an initial knowledge graph. Based on the initial knowledge graph, a software development knowledge graph is obtained. Based on the software development knowledge graph, in-degree grouping results and distance grouping results are obtained. Based on the in-degree grouping results and distance grouping results, all core process nodes are divided into several independent topology groups, and the regulatory frequency information corresponding to each core process node is obtained. The information acquisition module is used to acquire software development process stage information based on software development process analysis, break down the specific execution links of each process stage according to the software development process stage information, acquire development process node information, analyze the relationship between nodes according to the initial process node set, acquire node relationship table, define node attribute information corresponding to each initial process node based on software development standards, and construct node attribute table. The process node evaluation module is used to obtain the node association coefficient, node execution coefficient and node influence coefficient according to the node association table, obtain the node core coefficient based on the node association coefficient, node execution coefficient, node influence coefficient and corresponding weights, and obtain the core process node set based on the node core coefficient and necessity threshold corresponding to each initial process node.

[0034] The main control module specifically includes: The control unit is used to obtain in-degree grouping results and distance grouping results based on the software development knowledge graph, divide all core process nodes into several independent topology groups based on the in-degree grouping results and distance grouping results, and obtain the regulatory frequency information corresponding to each core process node. An information receiving unit interacts with an information acquisition module and a process node evaluation module to receive data and transmit it to a knowledge graph construction unit. The knowledge graph construction unit is used to construct the topology of the knowledge graph based on the node association table and using a graph database. Entities correspond to process nodes, and edges correspond to node associations to form an initial knowledge graph. Based on the initial knowledge graph, a software development knowledge graph is obtained.

[0035] The information acquisition module specifically includes: The first acquisition unit is used to acquire software development process stage information based on software development process analysis, and to break down the specific execution links of each process stage according to the software development process stage information to acquire development process node information. The second acquisition unit is used to analyze the relationship between nodes based on the initial process node set, acquire a node relationship table, define the node attribute information corresponding to each initial process node based on software development standards, and construct a node attribute table.

[0036] The process node evaluation module specifically includes: The node analysis unit is used to obtain the node association coefficient, node execution coefficient, and node influence coefficient based on the node association table. The core node evaluation unit is used to obtain the node core coefficient based on the node association coefficient, node execution coefficient, node influence coefficient and corresponding weight, and to obtain the core process node set based on the node core coefficient and necessity threshold corresponding to each initial process node.

[0037] In summary, the advantages of this invention are as follows: by constructing an initial set of process nodes and calculating the core node coefficients, precise focusing of regulatory nodes is achieved; by using core process nodes as entities and node relationships as edges, and employing a graph database to construct an initial knowledge graph, visualization and structured representation of the topological relationships of nodes throughout the software development process are realized; by extracting topological features such as the number of superior nodes and the number of flow steps of core process nodes, two-dimensional discrete grouping and setting weights are performed to obtain the benchmark regulatory frequency, and the regulatory transmission coefficient is calculated in combination with the number of subordinate nodes, achieving differentiated and precise setting of the regulatory frequency, thereby improving the continuity and efficiency of the regulatory process.

[0038] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A knowledge graph-based method for monitoring the entire enterprise software development process, characterized in that, include: Based on software development process analysis, information on the stages of the software development process is obtained. The stages of the software development process include the requirements analysis stage, architecture design stage, detailed design stage, coding development stage, unit testing stage, integration testing stage, system testing stage, deployment and launch stage, and operation and maintenance iteration stage. Based on the software development process phase information, obtain the core process node set; Based on the basic database of process nodes, the core process node set is extracted as the entity set of the knowledge graph; Based on the node association table, a graph database is used to construct the topology of the knowledge graph, with entities corresponding to process nodes and edges corresponding to node associations, forming an initial knowledge graph; Based on the initial knowledge graph, obtain the software development knowledge graph; Based on the software development knowledge graph, obtain the regulatory frequency information corresponding to each core process node; The entire software development process is monitored based on the monitoring frequency information corresponding to each core process node.

2. The knowledge graph-based enterprise software development process monitoring method according to claim 1, characterized in that, The step of obtaining the core process node set based on software development process stage information specifically includes: Based on the software development process stage information, the specific execution steps of each process stage are broken down to obtain the development process node information. The development process nodes include requirement review nodes, architecture review nodes, code submission nodes, test case design nodes, test execution nodes, defect fixing nodes, and version release nodes. Based on the development process node information, an initial process node set is constructed; Based on the initial process node set, analyze the relationships between the nodes to obtain a node relationship table; Based on software development standards, define the node attribute information corresponding to each initial process node and construct a node attribute table; The node attribute information includes node name, stage, executing entity, output results, and quality standards; Based on the node association table, obtain the node core coefficient corresponding to each initial process node; Based on the regulatory requirements for software development, a necessity threshold should be set; Based on the node core coefficient and necessity threshold corresponding to each initial process node, determine whether the initial process node is redundant. If the node core coefficient does not exceed the necessity threshold, remove the initial process node to obtain the core process node set. The core process node set, node attribute table, and relationship table are stored together to build a basic database for process nodes.

3. The knowledge graph-based enterprise software development process monitoring method according to claim 2, characterized in that, The step of obtaining the node core coefficient corresponding to each initial process node based on the node association table specifically includes: Based on the node association table, the initial process nodes corresponding to the dependencies are divided into the same dataset to obtain the dependency dataset; Traverse the dependency dataset and use the total number of times each initial process node is depended on by other nodes as the node association value of that initial process node; The ratio of the node association value of each initial process node to the maximum node association value is used as the node association coefficient of that initial process node. Based on the dependency dataset, the dependency nodes corresponding to each initial process node are used as the progressive association nodes of that initial process node; Based on software development requirements, obtain the call frequency of each progressively related node to the initial process node; Based on the dependency dataset, the average call frequency is used as the execution frequency of the initial process node; The maximum value of the execution frequency and the call frequency of each initial process node is used as the node execution coefficient of that initial process node; Based on the node association table, obtain the node influence coefficient of each initial process node; Based on the node analysis requirements, set the weights corresponding to the node correlation coefficient, node execution coefficient, and node influence coefficient; Based on the node association coefficient, node execution coefficient, node influence coefficient, and corresponding weights, the node core coefficient is obtained by weighted summation.

4. The knowledge graph-based enterprise software development process monitoring method according to claim 3, characterized in that, The step of obtaining the node influence coefficient of each initial process node based on the node association table specifically includes: Based on the node association table, taking any initial process node as the foundation, traverse all subsequent paths using the depth-first traversal algorithm, take all nodes on the subsequent paths as the subsequent nodes of the initial process node, take the initial process node as the feature node of the subsequent nodes, and obtain the set of subsequent nodes for each initial process node. Based on the set of subsequent nodes, obtain the preceding node information corresponding to each subsequent node; Based on the preceding node information corresponding to each subsequent node, the preceding node that has a direct or indirect connection with the feature node corresponding to the subsequent node is taken as the influence transmission node of the subsequent node. The ratio of the number of nodes whose influence is transmitted to each subsequent node to the total number of preceding nodes is used as the influence range coefficient of that subsequent node. Based on the set of subsequent nodes for each initial process node, the number of nodes with the shortest interval between each subsequent node and the initial process node is used as the influence depth coefficient of that subsequent node; Based on the set of subsequent nodes of each initial process node, the node influence coefficient of each initial process node is obtained according to the influence range coefficient and influence depth coefficient.

5. The knowledge graph-based enterprise software development process monitoring method according to claim 4, characterized in that, The step of obtaining a software development knowledge graph based on the initial knowledge graph specifically includes: Based on the initial knowledge graph, all nodes are traversed, and core process nodes without subsequent nodes are taken as terminal nodes. Traverse the initial knowledge graph, take any core process node as the target node, take the core process node pointing to the target node as the parent node of the target node, and obtain the number of parent nodes corresponding to each core process node. The shortest distance from each core process node to all terminal nodes is calculated based on Dijkstra's algorithm, and the distance is defined as the number of flow steps of that core process node. Based on the number of parent nodes and the number of steps corresponding to each core process node, the node attributes of the initial knowledge graph are labeled to obtain the software development knowledge graph.

6. The knowledge graph-based enterprise software development process monitoring method according to claim 5, characterized in that, The step of obtaining the regulatory frequency information corresponding to each core process node based on the software development knowledge graph specifically includes: Based on the software development knowledge graph, and using the number of parent nodes and the number of flow steps corresponding to each core process node, a node topology feature set is obtained. Based on the node topology feature set, obtain the number of parent nodes and the number of flow steps of the core process node corresponding to the maximum and minimum values ​​of the node core coefficient; The difference between the number of parent nodes of the core process node corresponding to the maximum and minimum values ​​of the node core coefficient is used as the in-degree feature interval. The difference between the maximum and minimum number of upper-level nodes is taken as the total in-degree interval, and the ratio of the total in-degree interval to the in-degree feature interval is taken as the number of in-degree groups, with the minimum number of in-degree groups being 2. Based on the node topology feature set, and using the maximum and minimum values ​​of the number of in-degree groups and the number of superior nodes as a basis, the number of superior nodes is discretely grouped based on equally divided intervals to obtain the in-degree grouping results. The difference between the number of flow steps of the core process nodes corresponding to the maximum and minimum values ​​of the node core coefficient is used as the distance feature interval; The difference between the maximum and minimum number of circulation steps is taken as the total distance interval, and the ratio of the total distance interval to the distance feature interval is taken as the distance group number, with the minimum value of the distance group number being 2. Based on the node topology feature set, and using the maximum and minimum values ​​of the number of distance groups and the number of flow steps as a basis, the terminal distance is discretely grouped based on equally divided intervals to obtain the distance grouping results; Based on the in-degree grouping results and the distance grouping results, a topology grouping matrix is ​​constructed through two-dimensional cross-logic operations, dividing all core process nodes into several independent topology groups; Based on the in-degree grouping results and the distance grouping results, set the weights corresponding to each in-degree group and distance group; Based on the project cycle items of enterprise software development, the benchmark regulatory frequency corresponding to each independent topology group is obtained according to the weights corresponding to each in-degree group and distance group; Based on the software development knowledge graph, take any core process node as the target node, and take the core process node pointed to by the target node as the subordinate node of the target node, and obtain the number of subordinate nodes of each core process node. Based on the number of subordinate nodes of each core process node, the maximum number of subordinate nodes in each independent topology group is taken as the baseline number of nodes. The ratio of the number of subordinate nodes to the number of baseline nodes for each core process node in each independent topology group is used as the regulatory transmission coefficient for that core process node. The product of the regulatory transmission coefficient corresponding to each core process node and the benchmark regulatory frequency is taken as the regulatory frequency of that core process node.

7. A knowledge graph-based enterprise software development end-to-end monitoring system, used to implement the monitoring method as described in any one of claims 1-6, characterized in that, include: The main control module is used to construct the topology of a knowledge graph based on a node association table and a graph database. Entities correspond to process nodes, and edges correspond to node associations to form an initial knowledge graph. Based on the initial knowledge graph, a software development knowledge graph is obtained. Based on the software development knowledge graph, in-degree grouping results and distance grouping results are obtained. Based on the in-degree grouping results and distance grouping results, all core process nodes are divided into several independent topology groups, and the regulatory frequency information corresponding to each core process node is obtained. The information acquisition module is used to acquire software development process stage information based on software development process analysis, break down the specific execution links of each process stage according to the software development process stage information, acquire development process node information, analyze the relationship between nodes according to the initial process node set, acquire node relationship table, define node attribute information corresponding to each initial process node based on software development standards, and construct node attribute table. The process node evaluation module is used to obtain the node association coefficient, node execution coefficient and node influence coefficient according to the node association table, obtain the node core coefficient based on the node association coefficient, node execution coefficient, node influence coefficient and corresponding weights, and obtain the core process node set based on the node core coefficient and necessity threshold corresponding to each initial process node.

8. A knowledge graph-based enterprise software development end-to-end monitoring system according to claim 7, characterized in that, The main control module specifically includes: The control unit is used to obtain in-degree grouping results and distance grouping results based on the software development knowledge graph, divide all core process nodes into several independent topology groups based on the in-degree grouping results and distance grouping results, and obtain the regulatory frequency information corresponding to each core process node. An information receiving unit interacts with an information acquisition module and a process node evaluation module to receive data and transmit it to a knowledge graph construction unit. The knowledge graph construction unit is used to construct the topology of the knowledge graph based on the node association table and using a graph database. Entities correspond to process nodes, and edges correspond to node associations to form an initial knowledge graph. Based on the initial knowledge graph, a software development knowledge graph is obtained.

9. A knowledge graph-based enterprise software development end-to-end monitoring system according to claim 7, characterized in that, The information acquisition module specifically includes: The first acquisition unit is used to acquire software development process stage information based on software development process analysis, and to break down the specific execution links of each process stage according to the software development process stage information to acquire development process node information. The second acquisition unit is used to analyze the relationship between nodes based on the initial process node set, acquire a node relationship table, define the node attribute information corresponding to each initial process node based on software development standards, and construct a node attribute table.

10. A knowledge graph-based enterprise software development end-to-end monitoring system according to claim 7, characterized in that, The process node evaluation module specifically includes: The node analysis unit is used to obtain the node association coefficient, node execution coefficient, and node influence coefficient based on the node association table. The core node evaluation unit is used to obtain the node core coefficient based on the node association coefficient, node execution coefficient, node influence coefficient and corresponding weight, and to obtain the core process node set based on the node core coefficient and necessity threshold corresponding to each initial process node.

Images