A General Static Analysis Method for Quality Characteristics Based on Knowledge Graphs

By using a knowledge graph-based approach, the objects to be analyzed are transformed into a unified representation and associated with a quality characteristic knowledge graph to generate an analysis context graph. This solves the problems of inconsistent object representation and difficulty in rule reuse in existing technologies, and enables unified organization and updating of quality characteristic analysis results for different object types.

CN123117291APending Publication Date: 2026-11-17TIANHANG CHANGYING (JIANGSU) TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611577808.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-13
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing static analysis products require the creation of separate object models and reconfiguration of analysis rules when dealing with different types of objects to be analyzed. This results in inconsistent representation of the objects and makes it difficult to reuse rules across object types.

Method used

By using a knowledge graph-based approach, the objects to be analyzed are transformed into a unified static intermediate representation, an object instance subgraph is constructed, and semantic mapping is performed with the quality characteristic knowledge graph to generate an analysis context graph. Candidate analysis paths are generated according to preset path constraints, rule matching and evidence extraction are performed, and risk assessment and result fusion are conducted.

Benefits of technology

It enables unified organization and updating of quality characteristic analysis results for different types of objects to be analyzed, solves the problems of inconsistent representation of analysis objects and difficulty in reusing rules, and improves the relevance and accuracy of analysis results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117291A_ABST
    Figure CN123117291A_ABST
Patent Text Reader

Abstract

This invention relates to the field of software static analysis and quality evaluation technology, and discloses a general static analysis method for quality characteristics based on knowledge graphs. The method includes the following steps: constructing a quality characteristic knowledge graph; S1 processing the objects to be analyzed; S2 associating knowledge graphs; S3 generating analysis paths; S4 rule matching; S5 risk assessment and result fusion; S6 generating analysis results; and incremental analysis and knowledge updating. By adopting a technical solution that converts different types of objects to be analyzed into a unified static intermediate representation, extracts object entities, entity attributes, and static relationships, constructs an object instance subgraph, and then semantically maps the object instance subgraph to the quality characteristic knowledge graph, this method achieves the technical effect of performing static analysis of quality characteristics on different types of objects based on a unified object representation. It solves the problems of inconsistent object representation methods and difficulty in reusing analysis rules across object types.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of software static analysis and quality evaluation technology, specifically to a general static analysis method for quality characteristics based on knowledge graphs. Background Technology

[0002] As the structures of software systems, information systems, and other digital products become increasingly complex, static analysis of product quality characteristics is widely used to examine source code, configuration files, interface definitions, data models, design models, and related documents before product launch. Existing static quality characteristic analysis products typically analyze the structure, attributes, and dependencies of the object under analysis, and, based on pre-configured analysis rules, assess quality characteristics such as reliability, security, maintainability, compatibility, and performance efficiency to identify potential quality issues. Because different types of objects under analysis differ in data format, composition, and semantic expression, static analysis products are usually configured with parsing tools, object models, and analysis rules corresponding to the object type.

[0003] However, existing static analysis products typically use independent object representation methods for different types of objects to be analyzed. When a static analysis product needs to handle new object types or needs to perform unified quality characteristic analysis on multiple object types, it usually needs to establish object models for each object type and reconfigure or rewrite existing analysis rules. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a general static analysis method for quality characteristics based on knowledge graphs, which solves the problems of inconsistent representation of analysis objects and difficulty in reusing analysis rules across object types.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a general static analysis method for quality characteristics based on knowledge graphs, comprising the following steps: S1, Object processing: Obtain the object to be analyzed and the target quality characteristic configuration, parse and standardize the object to be analyzed, generate a static intermediate representation, extract the object entity, entity attribute and static association relationship from the static intermediate representation, and construct the object instance subgraph; S2, Knowledge Graph Association: Semantically map the object instance subgraph constructed in S1 to the pre-constructed quality characteristic knowledge graph, filter quality characteristic nodes, measurement index nodes and analysis rule nodes according to the target quality characteristic configuration, and associate the filtered nodes with the object instance subgraph to form an analysis context graph; S3, Analysis Path Generation: Based on the analysis context graph formed in S2, starting with the target quality characteristic node, perform graph path search according to the correlation between graph nodes and preset path constraints to generate candidate analysis paths; S4, Rule Matching: Based on the candidate analysis path generated in S3, obtain the rule conditions of the corresponding analysis rule, match the object entity, entity attribute and static association relationship, obtain the rule hit result, extract the corresponding static analysis evidence and generate a static analysis evidence chain; S5, Risk Assessment and Result Fusion: Based on the rule hit results and static analysis evidence chain obtained in S4, determine the quality risk value and quality problem level, and merge, resolve conflicts and fuse across quality characteristics of the rule hit results to obtain the fusion analysis results; S6, Analysis Result Generation: Generate a static analysis report based on the fusion analysis results obtained in S5 and their associated data.

[0006] By adopting the above technical solution, different types of objects to be analyzed are converted into a unified static intermediate representation, and the entities and their relationships in the objects to be analyzed are represented by object instance subgraphs. Then, the object instance subgraphs are associated with the quality characteristic knowledge graph. The analysis rules and their applicable objects are determined based on graph path search. The static analysis evidence chain is generated by combining the rule matching results, and the analysis results are risk-assessed and fused. Therefore, it is possible to perform quality characteristic analysis on different types of objects to be analyzed according to a unified data structure, and form the correlation analysis results between quality characteristics, analysis rules, object entities and static analysis evidence.

[0007] Preferably, in step S1, the processing of the object to be analyzed includes: The corresponding parsing adapter is called according to the object type of the object to be analyzed to extract the structural information, dependency information and association information of the object to be analyzed, and convert it into a static intermediate representation using a unified data structure. Based on the object model in the quality characteristic knowledge graph, object entities, entity attributes, and static associations are identified from the static intermediate representation; The object entity is used as an instance node, the static association relationship is used as an instance edge, and the entity attributes are written into the corresponding instance node to construct an object instance subgraph.

[0008] By adopting the above technical solution, the structural information, dependency information and association information in the object to be analyzed are converted into a unified data structure by calling the corresponding parsing adapter for different object types. According to the object model, the object entity, entity attribute and static association relationship are represented as instance node, node attribute and instance edge respectively. Therefore, the differences in the original format and semantic expression of different objects to be analyzed can be eliminated, and the object instance data with consistent structure can be provided for subsequent semantic mapping, path search and rule matching.

[0009] Preferably, in step S2, the knowledge graph association includes: Extract semantic features of object entities in the object instance subgraph and semantic features of object model nodes in the quality characteristic knowledge graph; The semantic features of the object entity are matched with the semantic features of the object model node to obtain a semantic matching result; Based on the semantic matching results and preset mapping constraints, establish a mapping relationship between object entities and object model nodes; Based on the target quality characteristic configuration, the corresponding quality characteristic nodes, metric nodes, and analysis rule nodes are extracted from the quality characteristic knowledge graph, and the extracted nodes, the mapping relationship, and the object instance subgraph are combined to form an analysis context graph.

[0010] By adopting the above technical solution, since a mapping relationship is established based on the semantic features of the object entity and the object model node, and the quality characteristic node, measurement index node and analysis rule node corresponding to the current analysis task are extracted according to the target quality characteristics, it is possible to associate the instance data of the object to be analyzed with the quality characteristic knowledge, and exclude the graph nodes that are not related to the current target quality characteristics from the analysis context, thus forming an analysis context graph corresponding to the current static analysis task.

[0011] Preferably, in step S3, the analysis path generation includes: Based on the target quality characteristic nodes, determine the associated metric nodes, and based on the metric nodes, determine the corresponding analysis rule nodes; Based on the applicable conditions of the analysis rule nodes, the object entity nodes in the analysis context graph are filtered; According to the preset node type order, relationship direction, path depth and loop access restrictions, a graph path search is performed from the target quality characteristic node to the filtered object entity node; Graph paths that meet the preset path integrity conditions are identified as candidate analysis paths.

[0012] By adopting the above technical solution, the search direction is determined according to the association relationship between quality characteristic nodes, measurement index nodes, analysis rule nodes and object entity nodes. The graph path search process is constrained by node type order, relationship direction, path depth and cycle access restriction. Therefore, graph paths that do not meet the application conditions of analysis rules or path integrity conditions can be excluded, and candidate association paths between target quality characteristics, measurement indexes, analysis rules and object entities can be determined.

[0013] Preferably, in step S4, rule matching includes: Obtain the rule prerequisites, rule judgment conditions, and evidence extraction conditions from the analysis rule nodes in the candidate analysis path; Based on the aforementioned rule prerequisites, filter the object entities associated with the candidate analysis paths; The rules are used to determine the filtered object entities, entity attributes, and static relationships. When the rule determination conditions are met, a rule hit result is generated, and the corresponding static analysis evidence is extracted according to the evidence extraction conditions. Based on the relationships between nodes in the candidate analysis path and the static analysis evidence, a static analysis evidence chain is generated.

[0014] By adopting the above technical solution, since object entities are first screened according to the preconditions of the rules, and then the object entities, entity attributes and static relationships are matched according to the judgment conditions of the rules, and static analysis evidence is obtained according to the evidence extraction conditions, the rule hit results can be associated with their corresponding analysis objects, judgment conditions and evidence information, and a static analysis evidence chain consisting of target quality characteristics, analysis rules, object entities and static analysis evidence can be formed.

[0015] Preferably, in step S5, the risk assessment and result fusion includes: The quality risk value is determined based on the rule attributes of the analysis rules, the evidence attributes of the static analysis evidence, and the correlation between quality characteristic nodes. The quality risk value is compared with the preset risk level conditions to determine the quality problem level. Merge the rule matching results that meet the preset merging conditions; Rule hit results with inconsistent judgment results are processed according to preset conflict resolution conditions; Based on the correlation between the rule-hitting results and nodes of different quality characteristics, cross-quality characteristic fusion is performed to form fusion analysis results.

[0016] By adopting the above technical solution, the quality risk value is determined by combining the relationship between the analysis rule attributes, static analysis evidence attributes and quality characteristics, and the rule hit results of duplication, conflict and multiple quality characteristics are processed. Therefore, multiple rule hit results can be converted into fusion analysis results with quality problem level and quality characteristic relationship, reducing the situation of multiple duplicate analysis results or inconsistent analysis results corresponding to the same quality problem.

[0017] Preferably, in step S6, the generation of analysis results includes: Configure a result identifier for the fusion analysis results; Establish the association between the result identifier and the object instance node, quality characteristic node, analysis rule node, quality evaluation data, and static analysis evidence chain; Based on the preset report data structure, the fusion analysis results and their associated data are organized to generate a static analysis report.

[0018] By adopting the above technical solution, since the fusion analysis results are associated with object instance nodes, quality characteristic nodes, analysis rule nodes, quality evaluation data and static analysis evidence chain through result identification, and the associated data is organized according to the preset report data structure, the correspondence between the analysis results and their object data, rule data, quality evaluation data and evidence data can be maintained in the static analysis report.

[0019] Preferably, before step S1, a quality characteristic knowledge graph construction is further included, the quality characteristic knowledge graph construction including: Establish nodes for quality characteristics, metrics, analysis rules, object models, and evidence models; Establish graph edges between the quality characteristic nodes, metric nodes, analysis rule nodes, object model nodes, and evidence model nodes, and configure the relationship type and relationship direction for the graph edges; Entity alignment is performed on graph nodes from different sources that meet preset semantic matching conditions, and node fusion is performed based on the entity alignment results; Configure version information for the graph nodes and graph edges.

[0020] By adopting the above technical solution, since different types of graph nodes are used to represent quality characteristics, metrics, analysis rules, object models and evidence models respectively, and graph edges with relation types and relation directions are used to describe the association relationships between various types of nodes, while entity alignment and node fusion are performed on graph nodes from different sources, a unified quality characteristic knowledge graph can be formed, and different update states of graph nodes and graph edges can be distinguished by version information.

[0021] Preferably, after step S6, incremental static analysis is further included, the incremental static analysis including: Obtain change information of the object to be analyzed, and determine the object entity, entity attributes and static associations that have changed based on the change information; Based on the relationships between the changed object entities in the object instance subgraph, determine the affected object entities and the affected analysis paths; Update the object instance subgraph based on the changed object entities, entity attributes, and static relationships; Re-execute S2 through S6 for the affected object entities and affected analysis paths, and update the static analysis report based on the analysis results obtained from the re-execution.

[0022] By adopting the above technical solution, since the changed object entities and static relationships are determined based on the change information of the object to be analyzed, and the affected object entities and affected analysis paths are determined based on the relationships in the object instance subgraph, the scope of the re-executed analysis can be limited to the changed object entities and their associated analysis paths, and the static analysis report can be updated based on the re-obtained analysis results.

[0023] Preferably, step S6 further includes analysis feedback and knowledge update, wherein the analysis feedback and knowledge update include: Obtain the analysis feedback information corresponding to the static analysis results, and establish the association between the analysis feedback information and the corresponding analysis rule nodes, object entity nodes and static analysis evidence chain; Adjust the rule attributes or risk assessment parameters of the corresponding analysis rules based on the analysis feedback information; Based on the analysis feedback information, extract object features or relationships not covered by existing analysis rules, and generate corresponding candidate rules and candidate evidence patterns; The candidate rules and candidate evidence patterns are verified, and the candidate rules and candidate evidence patterns that meet the preset verification conditions are written into the quality characteristic knowledge graph.

[0024] By adopting the above technical solution, the analysis feedback information is associated with the analysis rule nodes, object entity nodes, and static analysis evidence chain. The rule attributes or risk assessment parameters are adjusted according to the analysis feedback information. At the same time, object features or relationships not covered by existing analysis rules are extracted to generate candidate rules and candidate evidence patterns. Therefore, the rule data, evidence patterns, and related parameters in the quality characteristic knowledge graph can be updated according to the analysis feedback information.

[0025] This invention provides a general static analysis method for quality characteristics based on knowledge graphs. It has the following beneficial effects: 1. This invention achieves the technical effect of static analysis of quality characteristics of different types of objects based on a unified object representation by adopting a technical solution that converts different types of objects to be analyzed into a unified static intermediate representation, extracts object entities, entity attributes and static associations and constructs object instance subgraphs, and then performs semantic mapping between the object instance subgraphs and the quality characteristic knowledge graph. Compared with the existing technical solution of establishing analysis models and analysis rules for different object types, this invention solves the problems of inconsistent object representation and difficulty in reusing analysis rules across object types.

[0026] 2. This invention employs a technical solution that involves configuring and extracting quality characteristic nodes, measurement index nodes, and analysis rule nodes based on target quality characteristics to form an analysis context graph. It then generates candidate analysis paths according to preset path constraints and performs rule matching and evidence extraction on the object entities associated with these candidate analysis paths. This achieves the technical effect of establishing the relationship between target quality characteristics, measurement indexes, analysis rules, object entities, and static analysis evidence. Compared to existing technologies that directly output judgment results based on mutually independent static analysis rules, this invention solves the problems of unclear relationships between analysis rules and quality characteristics, and the lack of corresponding judgment paths and evidence associations in the analysis results.

[0027] 3. This invention employs a risk assessment method based on the correlation between rule hit results, static analysis evidence, and quality characteristics. It merges rule hit results, resolves conflicts, and integrates cross-quality characteristics. Simultaneously, it updates the object instance subgraph based on changes to the object under analysis and re-analyzes the affected analysis paths. This achieves the technical effect of uniformly organizing the correlation analysis results and updating the static analysis results based on object changes. Compared with the existing technology that outputs each rule detection result separately and re-executes the entire static analysis process after the object under analysis changes, this invention solves the problems of lack of unified processing of duplicate and conflicting results and difficulty in determining the analysis scope due to local changes in the object. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the method steps of the present invention; Figure 2 This is a schematic diagram of the quality characteristic knowledge graph structure of the present invention; Figure 3 This is a schematic diagram illustrating the processing of the object to be analyzed in this invention; Figure 4 This is a schematic diagram illustrating the knowledge graph association and analysis path generation of the present invention; Figure 5 This is a schematic diagram illustrating the rule matching, result fusion, and updating of the present invention. Detailed Implementation

[0029] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] According to the appendix Figure 1 As shown, this embodiment of the invention provides a general static analysis method for quality characteristics based on knowledge graphs, including the following steps: According to the appendix Figure 2 As shown, the first step is to construct a knowledge graph of quality characteristics, including: Establish nodes for quality characteristics, metrics, analysis rules, object models, and evidence models; Establish graph edges between quality characteristic nodes, metric nodes, analysis rule nodes, object model nodes, and evidence model nodes, and configure the relationship type and direction for the graph edges; Entity alignment is performed on graph nodes from different sources that meet preset semantic matching conditions, and node fusion is performed based on the entity alignment results; Configure version information for graph nodes and graph edges; Specifically, the process involves acquiring quality standard documents, quality evaluation specifications, static analysis rule data, object model data, historical analysis data, and preset configuration data; structuring the acquired data; and establishing corresponding types of graph nodes based on the data content.

[0031] The Quality Characteristic node configures the quality characteristic identifier, quality characteristic name, quality characteristic level, and scope of application. The Measurement Index node configures the index identifier, index name, index type, measurement object, and evaluation conditions. The Analysis Rule node configures the rule identifier, applicable objects, applicable technical environment, rule prerequisites, rule judgment conditions, evidence extraction conditions, rule severity, and rule confidence level. The Object Model node configures the object type, object attribute model, object relationship model, and applicable technical environment. The Evidence Model node configures the evidence type, evidence source, evidence location representation method, evidence elements, and evidence integrity conditions.

[0032] According to the appendix Figure 3As shown, graph edges are established based on the semantic relationships between different graph nodes. The relationship types of graph edges include at least one of the following: inclusion relationship, measurement relationship, applicability relationship, support relationship, influence relationship, and propagation relationship. Graph edges between quality characteristic nodes and measurement index nodes record the correspondence between quality characteristics and measurement indices. Graph edges between measurement index nodes and analysis rule nodes record the correspondence between measurement indices and analysis rules. Graph edges between analysis rule nodes and object model nodes record the applicable object types of the analysis rules. Graph edges between analysis rule nodes and evidence model nodes record the evidence types and evidence extraction requirements corresponding to the analysis rules. Graph edges between quality characteristic nodes record the inclusion relationship, influence relationship, or propagation relationship between different quality characteristics.

[0033] For graph nodes from different sources, node name features, node attribute features, node adjacency features, and node context features are extracted respectively. The matching degree of the graph nodes to be compared in terms of node name, node attribute, node adjacency, and node context is determined, and they are combined according to the corresponding matching weights to obtain the comprehensive semantic matching degree between the graph nodes to be compared.

[0034] The matching weights corresponding to node name features, node attribute features, node adjacency features, and node context features are determined based on the labeled aligned and unaligned node pairs. Specifically, multiple sets of candidate matching weights are set, and the comprehensive semantic matching degree between graph nodes is calculated based on each set of candidate matching weights. The calculation results are then compared with the labeled node alignment results, and the matching weights are determined based on the comparison results.

[0035] The entity alignment threshold is determined based on the labeled aligned node pairs and unaligned node pairs. Multiple candidate entity alignment thresholds are set, and entity alignment results are generated based on each candidate entity alignment threshold. The entity alignment threshold is then determined based on the degree of consistency between the entity alignment results and the labeled results.

[0036] When the overall semantic matching degree reaches the entity alignment threshold, the corresponding graph node is identified as the node to be merged. A unified node identifier is configured for the node to be merged, and the corresponding node attributes are merged. For node attributes with consistent values, the unified attribute value is retained; for node attributes with inconsistent values, the attribute value, data source, applicable conditions, and update time are recorded respectively, and the currently valid attribute is determined according to the data source priority, applicable scope, and update time.

[0037] Configure version identifiers, data sources, creation times, update times, and validity status for graph nodes and edges. When a graph node or edge changes, a new version record is generated, and the original version is recorded as a historical version.

[0038] In this scheme, the matching weight and entity alignment threshold corresponding to the node features are preferably determined based on the labeled node pairs that correspond to the current map node type and data source.

[0039] After the quality characteristic knowledge graph is constructed, the object model nodes serve as the model basis for identifying object entities in step S1, while the quality characteristic nodes, measurement index nodes, analysis rule nodes, and corresponding graph edges serve as the knowledge data for forming the analysis context graph in step S2.

[0040] S1, Object processing: Obtain the configuration of the object to be analyzed and the target quality characteristics, parse and standardize the object to be analyzed, generate a static intermediate representation, extract the object entity, entity attribute and static association from the static intermediate representation, and construct the object instance subgraph; Furthermore, in S1, the processing of the object to be analyzed includes: Based on the object type of the object to be analyzed, the corresponding parsing adapter is called to extract the structural information, dependency information and association information of the object to be analyzed, and then converted into a static intermediate representation using a unified data structure. Based on the object model in the quality characteristic knowledge graph, object entities, entity attributes, and static relationships are identified from the static intermediate representation; By treating object entities as instance nodes, static relationships as instance edges, and writing entity attributes into the corresponding instance nodes, an object instance subgraph is constructed. Specifically, a parsing adapter registry is pre-established, which records the correspondence between object types, file formats, technical environments, and parsing adapters.

[0041] After obtaining the object to be analyzed, the object type of the object to be analyzed is determined based on the file extension, content characteristics, protocol identifier or object description information of the object to be analyzed, and the corresponding parsing adapter is called from the parsing adapter registry.

[0042] The parsing adapter performs lexical, syntactic, structural, or semantic analysis on the object to be analyzed, extracting the constituent units, constituent unit attributes, the source location of the constituent units, and the static relationships between the constituent units.

[0043] The parsing results output by different parsing adapters are converted into a static intermediate representation using a unified data structure. The static intermediate representation includes at least component data, component attribute data, source location data, and static relationship data.

[0044] Based on the object type, object attribute pattern, and object relationship pattern defined by the object model nodes in the quality characteristic knowledge graph, the constituent units in the static intermediate representation are matched with the object model nodes, and the object entities are identified based on the matching results.

[0045] For each identified object entity, an entity identifier is generated, and the object type, entity name, entity attributes, source location, and technical environment are recorded. The entity identifier is generated based on at least one of the object identifier to be analyzed, the object hierarchy path, the object name, and the source location. When multiple object entities have the same name, they are distinguished based on the object hierarchy path, the object to which they belong, or the source location.

[0046] Object entities are treated as object instance nodes, and the relationships between object entities (including containment, call, dependency, reference, inheritance, read-write, and configuration associations) are treated as object instance edges. Entity attributes are written to the corresponding object instance nodes to construct an object instance subgraph. For each object instance edge, the starting object instance node, ending object instance node, relationship type, relationship direction, source location, and version information are recorded.

[0047] When the object to be analyzed includes multiple object types, the parsing order is determined according to the dependencies between different object types, the corresponding parsing adapters are called in sequence, and the object instance data output by each parsing adapter is merged by entity identifier, object hierarchy path or static association relationship.

[0048] The object instance subgraph output in step S1 serves as the input for the semantic mapping between object entities and object model nodes in step S2.

[0049] S2, Knowledge Graph Association: Semantically map the object instance subgraph constructed in S1 to the pre-constructed quality characteristic knowledge graph. Based on the target quality characteristic configuration, filter quality characteristic nodes, measurement index nodes, and analysis rule nodes. Associate the filtered nodes with the object instance subgraph to form an analysis context graph. Furthermore, in S2, knowledge graph associations include: Extract semantic features of object entities in the object instance subgraph and semantic features of object model nodes in the quality characteristic knowledge graph; The semantic features of the object entity are matched with the semantic features of the object model nodes to obtain the semantic matching result; Based on the semantic matching results and preset mapping constraints, establish the mapping relationship between object entities and object model nodes; Based on the target quality characteristic configuration, the corresponding quality characteristic nodes, measurement index nodes and analysis rule nodes are extracted from the quality characteristic knowledge graph, and the extracted nodes, mapping relationships and object instance subgraphs are combined to form an analysis context graph. Specifically, the object instance subgraph constructed in step S1 is read, and the semantic features corresponding to the object entity's object type, object name, object attributes, adjacency relationship, and technical environment are extracted. The object model nodes in the quality characteristic knowledge graph are read, and the semantic features corresponding to the object model node's object type description, object name description, object attribute pattern, object relationship pattern, and applicable technical environment are extracted.

[0050] The matching degree between the object entity and the object model node is determined separately in terms of object type, object name, object attributes, adjacency relationship and technical environment. The matching degree of each item is combined according to the corresponding mapping weight to obtain the comprehensive mapping matching degree between the object entity and the object model node.

[0051] The mapping weights corresponding to object type, object name, object attributes, adjacency relationships, and technical environment are determined based on the already labeled object entity and object model node mapping data. Specifically, multiple sets of candidate mapping weights are set, and mapping results between object entities and object model nodes are generated based on each set of candidate mapping weights. The mapping weights are then determined based on the degree of consistency between the mapping results and the already labeled mapping results.

[0052] The mapping threshold is determined based on the mapping data between labeled object entities and object model nodes. Multiple candidate mapping thresholds are set, and mapping relationships are established based on each candidate threshold. The mapping threshold is then determined based on the consistency between the established mapping relationship and the labeled mapping relationship.

[0053] When the overall mapping matching degree reaches the mapping threshold and satisfies object type constraints, attribute constraints, relationship constraints, and technical environment constraints, a mapping edge is established between the object instance node and the object model node corresponding to the object entity. When the overall mapping matching degree reaches the mapping threshold but does not meet the object type constraints, no mapping edge is established. When the overall mapping matching degree does not reach the mapping threshold, but the object type, key attributes, and technical environment meet the preset mandatory mapping conditions, a mapping edge is established according to the preset mapping rules, and a mandatory mapping identifier and mapping basis are configured for the mapping edge.

[0054] Based on the configuration of the target quality characteristics, the target quality characteristic nodes are located in the quality characteristic knowledge graph, and quality sub-characteristic nodes, measurement index nodes, analysis rule nodes, object model nodes and corresponding graph edges are extracted along the inclusion relationship, measurement relationship and applicability relationship to form the target knowledge subgraph.

[0055] The object instance subgraph, target knowledge subgraph, and mapping edges between object instance nodes and object model nodes generated in step S1 are combined to form the analysis context graph.

[0056] In this scheme, the mapping weights and mapping thresholds are preferably determined based on the labeled mapping data corresponding to the current object type, object model type, and technical environment.

[0057] The analysis context graph formed in step S2 serves as the data basis for the graph path search in step S3, according to the appendix. Figure 4 As shown.

[0058] S3, Analysis Path Generation: Based on the analysis context graph formed in S2, starting with the target quality characteristic node, perform graph path search according to the correlation between graph nodes and preset path constraints to generate candidate analysis paths; Furthermore, in S3, the analysis path generation includes: Based on the target quality characteristic nodes, determine the associated metric nodes, and based on the metric nodes, determine the corresponding analysis rule nodes; Based on the applicable conditions of the analysis rule nodes, the object entity nodes in the analysis context graph are filtered; According to the preset node type order, relationship direction, path depth and loop access restrictions, a graph path search is performed from the target quality characteristic node to the filtered object entity node; Graph paths that meet the preset path integrity conditions are identified as candidate analysis paths; Specifically, the target quality characteristic node is read from the analysis context graph formed in step S2, the associated metric index node is found along the metric relationship, and the analysis rule node corresponding to the metric index node is found along the applicability relationship.

[0059] The system reads the applicable object type, applicable technical environment, applicable object level, and rule prerequisites recorded in the analysis rule node, and determines the corresponding object model node based on the applicable relationship between the analysis rule node and the object model node. Using the mapping edge established in step S2, the system locates the object instance nodes in the object instance subgraph from the object model node, and filters the object instance nodes based on the applicable object type, applicable technical environment, applicable object level, and rule prerequisites.

[0060] The search begins with the target quality characteristic node and ends with the filtered object instance nodes. Graph path search is performed according to node type order, relationship direction, maximum path depth, necessary node set, and loop access constraints. Each candidate analysis path must include at least the target quality characteristic node, metric node, analysis rule node, object model node, and object instance node. When a quality sub-characteristic node exists between the target quality characteristic node and the metric node, these sub-characteristic nodes are written into the candidate analysis path in the actual connection order.

[0061] Candidate analysis paths are prioritized based on the strength of graph relationships, the mapping matching degree between object instance nodes and object model nodes, the applicability of analysis rules to object entities, path completeness, and path length.

[0062] The path search parameters include maximum path depth, allowed number of repeated visits, the number of candidate analysis paths to retain for each target quality characteristic node, and path priority parameters. These parameters are determined based on the number of nodes in the quality characteristic knowledge graph, the number of graph edges, the average node degree, the path distribution corresponding to the target quality characteristic, and the labeled candidate analysis paths.

[0063] Specifically, multiple sets of candidate path search parameters are set to generate candidate analysis paths. The generated candidate analysis paths are then compared with the labeled candidate analysis paths. The path search parameters are determined based on path coverage, the number of irrelevant paths, the number of rules executed, and the path search time. When multiple sets of candidate path search parameters all meet the preset path evaluation conditions, the parameter set with the smaller maximum path depth or the smaller number of candidate analysis paths is selected first.

[0064] In this scheme, the path search parameters are preferably determined based on the labeled candidate analysis paths that correspond to the current target quality characteristics and the version of the quality characteristic knowledge graph.

[0065] The candidate analysis path generated in step S3 serves as the input for step S4 to read the analysis rule conditions and determine the instance node of the object to be judged.

[0066] S4, Rule Matching: Based on the candidate analysis paths generated in S3, obtain the rule conditions of the corresponding analysis rules, match the object entities, entity attributes, and static relationships to obtain rule hit results, extract the corresponding static analysis evidence, and generate a static analysis evidence chain, according to the appendix... Figure 5 As shown,; Furthermore, in S4, rule matching includes: Obtain the rule prerequisites, rule judgment conditions, and evidence extraction conditions from the analysis rule nodes in the candidate analysis path; Filter the object entities associated with candidate analysis paths based on the rule prerequisites; The rules are used to determine the filtered object entities, entity attributes, and static relationships. When the rule determination conditions are met, a rule hit result is generated, and the corresponding static analysis evidence is extracted according to the evidence extraction conditions. Based on the relationships between nodes in the candidate analysis path and the static analysis evidence, a static analysis evidence chain is generated. Specifically, the candidate analysis paths generated in step S3 are read, and the rule preconditions, rule judgment conditions, and evidence extraction conditions are obtained from the analysis rule nodes in the candidate analysis paths. Rule preconditions include at least one of the following: object type conditions, technical environment conditions, attribute existence conditions, and relationship existence conditions. Rule judgment conditions include at least one of the following: entity attribute conditions, entity quantity conditions, relationship existence conditions, path existence conditions, dependency level conditions, and threshold comparison conditions.

[0067] First, determine whether the object entities associated with the candidate analysis path meet the rule preconditions. If the rule preconditions are not met, stop executing the corresponding analysis rule for the current object entity; if the rule preconditions are met, execute the rule judgment for the object entity, entity attributes, and static association relationships.

[0068] When the rule determination conditions are not met, the corresponding candidate analysis path is marked as a missed path; when the rule determination conditions are met, a rule hit result is generated, and static analysis evidence is extracted from the static intermediate representation or object instance subgraph according to the evidence extraction conditions. Static analysis evidence includes at least one of the following: evidence object, evidence attribute, evidence relationship, and evidence location.

[0069] Based on the evidence elements and evidence integrity conditions recorded in the evidence model nodes, the evidence integrity of statically analyzed evidence is determined. The parameters for determining evidence integrity include the weights corresponding to different evidence elements and evidence integrity thresholds.

[0070] The weights corresponding to different evidentiary elements are determined based on the degree to which each evidentiary element distinguishes the rule-based judgment result and the confirmed static analysis evidence.

[0071] Multiple sets of evidence element weights are set to determine the completeness of static analysis evidence, and the evidence element weights are determined based on the degree of consistency between the completeness of evidence and the confirmed state of evidence completeness.

[0072] The evidence completeness threshold is determined based on the confirmed complete evidence and incomplete evidence. Multiple candidate evidence completeness thresholds are set to classify static analysis evidence, and the evidence completeness threshold is determined based on the consistency between the classification results and the confirmed results.

[0073] When the completeness of static analysis evidence reaches the evidence completeness threshold, the corresponding rule hit result is marked as a complete evidence result; when the completeness of static analysis evidence is below the evidence completeness threshold, the corresponding rule hit result is marked as an evidence result to be reviewed.

[0074] Based on the node order in the candidate analysis path, the target quality characteristic node, metric node, analysis rule node, object model node, and object instance node are connected, and static analysis evidence is associated with the object instance node or analysis rule node to form a static analysis evidence chain.

[0075] In this scheme, the weight of evidence elements and the threshold for evidence completeness are preferably determined based on the confirmed static analysis evidence corresponding to the current analysis rules and evidence model.

[0076] The rule hit results and static analysis evidence chain output in step S4 serve as inputs for determining the quality risk value and fusing the execution results in step S5.

[0077] S5, Risk Assessment and Result Fusion: Based on the rule hit results and static analysis evidence chain obtained in S4, determine the quality risk value and quality problem level, and merge, resolve conflicts and fuse across quality characteristics of the rule hit results to obtain the fusion analysis results; Furthermore, in S5, risk assessment and results integration include: The quality risk value is determined based on the rule attributes of the analysis rules, the evidence attributes of the static analysis evidence, and the correlation between quality characteristic nodes. The quality risk value is compared with the preset risk level conditions to determine the level of quality problem. Merge the rule matching results that meet the preset merging conditions; Rule hit results with inconsistent judgment results are processed according to preset conflict resolution conditions; Based on the correlation between the rule-hitting results and nodes of different quality characteristics, cross-quality characteristic fusion is performed to form fusion analysis results; Specifically, the rule hit results and static analysis evidence chain output in step S4 are read, the rule severity and rule confidence are obtained from the rule nodes, the evidence completeness is obtained from the static analysis evidence, and the degree of influence of quality characteristic propagation is determined based on the influence relationship, propagation direction and propagation level between quality characteristic nodes.

[0078] Analysis rules For object entities The degree of propagation of the quality characteristics corresponding to the quality problem is determined by the following formula: ; in, Representation Analysis Rules For object entities The extent to which the quality characteristics corresponding to the quality problems are propagated and affected; This indicates the number of other quality characteristic nodes that are associated with the propagation of quality problems. Indicates the relationship between the current quality characteristic node and the first node. The strength of the relationship between nodes with associated quality characteristics; This represents the influence coefficient corresponding to the direction of propagation; Indicates the current quality characteristic node up to the th The number of graph edges between nodes with associated quality characteristics.

[0079] in, and The values ​​range from 0 to 1. It is a non-negative integer.

[0080] when At that time, it was stipulated that: .

[0081] The strength of the relationship is determined based on the relationship type between quality characteristic nodes, data source, historical co-occurrence, and confirmed propagation results.

[0082] The propagation direction influence coefficient is determined based on the direction and type of the propagation relationship and the confirmed propagation results.

[0083] The quality risk value is determined according to the following formula: ; in, Representation Analysis Rules For object entities Quality risk value; Representation Analysis Rules The severity of the rules; Representation Analysis Rules Rule confidence; Representation Analysis Rules For object entities The corresponding static analysis of evidence completeness; Indicates the degree of influence of quality characteristics propagation; , , and These represent the weights corresponding to the severity of the rule, the confidence level of the rule, the completeness of the evidence, and the degree of impact of its dissemination.

[0084] Each weight is a non-negative number and satisfies: ; The risk assessment weights are determined based on the confirmed rule hit results and the corresponding quality problem levels.

[0085] Specifically, multiple sets of candidate risk assessment weights are set, quality risk values ​​are calculated for each set, and the quality problem level determined based on the quality risk value is compared with the confirmed quality problem level. The risk assessment weight is determined based on the comparison result.

[0086] The risk level threshold is determined based on the confirmed quality problem level and the corresponding quality risk value.

[0087] Multiple candidate risk level threshold combinations are set up to classify quality risk values ​​into different levels, and the risk level threshold is determined based on the consistency between the classification results and the confirmed quality problem levels.

[0088] Let the threshold values ​​for the first risk level, the second risk level, and the third risk level be respectively... , and ,satisfy: ; When the quality risk value is less than the first risk level threshold, it is determined to be a first quality problem level; when the quality risk value reaches the first risk level threshold but is less than the second risk level threshold, it is determined to be a second quality problem level; when the quality risk value reaches the second risk level threshold but is less than the third risk level threshold, it is determined to be a third quality problem level; when the quality risk value reaches the third risk level threshold, it is determined to be a fourth quality problem level.

[0089] When merging rule hit results, compare the object entities, evidence locations, static association paths, and root causes of problems corresponding to different rule hit results.

[0090] When the object entities are the same, the evidence locations meet the preset overlap conditions, and the root cause of the problem meets the preset similarity conditions, the corresponding rule hit results will be merged.

[0091] For rule-based hit results that result in inconsistent judgments for the same object entity, conflict resolution is carried out based on rule priority, rule confidence, evidence completeness, and the degree of adaptation to the technical environment.

[0092] Conflict resolution parameters are determined based on the confirmed conflict resolution results. Multiple sets of candidate conflict resolution parameters are set, conflict resolution results are generated for each set, and the final conflict resolution parameters are determined based on the degree of consistency between the conflict resolution results and the confirmed results.

[0093] Based on the direct and propagation relationships between the rule-hitting results and nodes with different quality characteristics, the analysis results of related entities or root causes of the same problem are combined to form a fusion analysis result.

[0094] In this scheme, the risk assessment weights, risk level thresholds, and conflict resolution parameters are preferably determined based on the confirmed analysis results corresponding to the current target quality characteristics and analysis rule types.

[0095] The fusion analysis results and their associated data generated in step S5 serve as the data source for generating the static analysis report in step S6.

[0096] S6, Analysis Result Generation: Generate a static analysis report based on the fusion analysis results and related data obtained in S5; Furthermore, in S6, the analysis results are generated including: Configure result labels for the fusion analysis results; Establish the association between result identifiers and object instance nodes, quality characteristic nodes, analysis rule nodes, quality evaluation data, and static analysis evidence chains; Based on the preset report data structure, the fusion analysis results and their related data are organized to generate a static analysis report; Specifically, a result identifier is configured for each fusion analysis result obtained in step S5.

[0097] The result identifier is generated based on the identifier of the object to be analyzed, the object entity identifier, the main analysis rule identifier, and the analysis batch identifier, or it can be generated through a preset unique identifier generation method.

[0098] The result identifier is used as an index for the fusion analysis results, and the association between the result identifier and object instance nodes, quality characteristic nodes, analysis rule nodes, quality evaluation data, and static analysis evidence chain is established.

[0099] Quality assessment data includes quality risk values, quality problem levels, and cross-quality characteristic fusion data.

[0100] The preset report data structure includes at least one of the following: result index area, object data area, quality evaluation data area, rule data area, and evidence data area.

[0101] The results index area records the result identifier, and the object data area records the object instance nodes and object instance edge data associated with the result identifier. The quality evaluation data area records the quality risk value, quality problem level, and associated quality characteristic data, and the rule data area records the corresponding analysis rule data. The evidence data area records the static analysis evidence chain and corresponding evidence location data. According to the preset report data structure, the fusion analysis results and their associated data are organized to generate a static analysis report. The report data structure is determined according to the report output format, the data type of the static analysis results, and the report's target audience. Version information is configured for the report data structure. The generated static analysis report serves as the original analysis result in incremental static analysis and as a feedback associated object in analysis feedback and knowledge updates.

[0102] S6 is followed by incremental static analysis, which includes: Obtain change information for the object to be analyzed, and determine the object entity, entity attributes, and static relationships that have changed based on the change information; Based on the relationships between the changed object entities in the object instance subgraph, determine the affected object entities and the affected analysis paths; Update the object instance subgraph based on the changed object entities, entity attributes, and static relationships; Re-execute S2 through S6 for the affected object entities and affected analysis paths, and update the static analysis report based on the analysis results obtained from the re-execution.

[0103] Specifically, after completing step S6, versioning of the object to be analyzed is recorded. When a change is detected in the object to be analyzed, the current version is compared with the previous version to determine the newly added, deleted, or modified object entities, entity attributes, and static relationships.

[0104] The object entity that changes is identified as the directly changed object entity, and an association search is performed in the object instance subgraph along the containment relationship, call relationship, dependency relationship, reference relationship and data relationship, starting from the directly changed object entity.

[0105] object entity The degree of impact of the change is determined by the following formula: ; in, Representing object entities The extent of the impact of the changes; This indicates a direct change to the object entity. The number of associated paths between them; Indicates the first The influence weight corresponding to the relationship type in each associated path; Indicates the first The strength of association along each associated path; Indicates the first The number of object instance edges contained in a single associated path.

[0106] in, and The values ​​range from 0 to 1. It is a non-negative integer.

[0107] When there is no direct change from object entity to object entity When associating paths, the object entity The impact of the change is set to 0.

[0108] The influence weights and association strengths of different relationship types are determined based on historical change data and confirmed affected entities.

[0109] The impact threshold of the change is determined based on the confirmed affected and unaffected entities in the historical change data; Set multiple sets of candidate relationship types to influence weights, determine the association strength parameters and change impact thresholds, identify the affected object entities respectively, and determine the corresponding parameters based on the consistency between the determined results and the confirmed results in the historical change data.

[0110] The maximum path depth for association search is determined based on the number of nodes in the object instance subgraph, the average node degree, the historical change propagation range, and the association search time.

[0111] When multiple sets of candidate parameters meet the preset change impact assessment conditions, the parameter set with a smaller associated search range or a smaller number of affected entities is selected first.

[0112] When the impact of a change on an object entity reaches the change impact threshold, the object entity is identified as an affected object entity, and the candidate analysis paths associated with the affected object entity are identified as affected analysis paths.

[0113] Update the object instance subgraph based on the change information.

[0114] For newly added object entities or static relationships, add corresponding object instance nodes or object instance edges to the object instance subgraph.

[0115] For deleted object entities or static associations, mark the corresponding object instance nodes or object instance edges as invalid, or delete them from the object instance subgraph of the current version.

[0116] For modified object entities, entity attributes, or static relationships, update the corresponding object instance node attributes or object instance edge attributes, and record the version information before and after the change.

[0117] For the affected object entities and the affected analysis paths, re-execute steps S2 to S6, and compare the newly obtained fusion analysis results with the corresponding results in the original static analysis report. Based on the comparison results, add, delete, or update the result data in the static analysis report.

[0118] In this scheme, the relationship type influence weight, the parameter for determining the association strength, the maximum path depth, and the threshold for the impact of changes are preferably determined based on historical change data corresponding to the type of object to be analyzed.

[0119] S6 and later also include analysis feedback and knowledge updates, which include: Obtain the analysis feedback information corresponding to the static analysis results, and establish the association between the analysis feedback information and the corresponding analysis rule nodes, object entity nodes, and static analysis evidence chain; Adjust the rule attributes or risk assessment parameters of the corresponding analysis rules based on the analysis feedback information; Based on the analysis feedback information, extract object features or relationships that are not covered by existing analysis rules, and generate corresponding candidate rules and candidate evidence patterns; The candidate rules and candidate evidence patterns are validated, and the candidate rules and candidate evidence patterns that meet the preset validation conditions are written into the quality characteristic knowledge graph.

[0120] Specifically, the analysis feedback information and its corresponding result identifier are obtained, the static analysis report is queried based on the result identifier, the analysis rule nodes, object instance nodes and static analysis evidence chains associated with the analysis feedback information are determined, and the corresponding feedback associations are established.

[0121] Adjust the rule confidence level, applicable conditions, judgment conditions, evidence extraction conditions, or risk assessment parameters of the analysis rules based on the analysis feedback information, and record the parameter information and version information before and after the adjustment.

[0122] For analysis feedback information not covered by existing analysis rules, extract the corresponding object type, entity attribute or static relationship, generate candidate rules, and generate candidate evidence patterns based on the corresponding evidence information.

[0123] Candidate rules and candidate evidence patterns are validated using historical object instance data and analysis feedback information. When a candidate rule meets the preset support and reliability conditions, and a candidate evidence pattern meets the preset evidence integrity conditions, the candidate rule and candidate evidence pattern are written into the quality characteristic knowledge graph and the corresponding version information is configured.

[0124] The preset support conditions, preset reliability conditions, and preset evidence integrity conditions are determined based on historical analysis results and analysis feedback information.

[0125] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A general static analysis method for quality characteristics based on knowledge graphs, characterized in that, Includes the following steps: S1, Object processing: Obtain the object to be analyzed and the target quality characteristic configuration, parse and standardize the object to be analyzed, generate a static intermediate representation, extract the object entity, entity attribute and static association relationship from the static intermediate representation, and construct the object instance subgraph; S2, Knowledge Graph Association: Semantically map the object instance subgraph constructed in S1 to the pre-constructed quality characteristic knowledge graph, filter quality characteristic nodes, measurement index nodes and analysis rule nodes according to the target quality characteristic configuration, and associate the filtered nodes with the object instance subgraph to form an analysis context graph; S3, Analysis Path Generation: Based on the analysis context graph formed in S2, starting with the target quality characteristic node, perform graph path search according to the correlation between graph nodes and preset path constraints to generate candidate analysis paths; S4, Rule Matching: Based on the candidate analysis path generated in S3, obtain the rule conditions of the corresponding analysis rule, match the object entity, entity attribute and static association relationship, obtain the rule hit result, extract the corresponding static analysis evidence and generate a static analysis evidence chain; S5, Risk Assessment and Result Fusion: Based on the rule hit results and static analysis evidence chain obtained in S4, determine the quality risk value and quality problem level, and merge, resolve conflicts and fuse across quality characteristics of the rule hit results to obtain the fusion analysis results; S6, Analysis Result Generation: Generate a static analysis report based on the fusion analysis results obtained in S5 and their associated data.

2. The general static analysis method for quality characteristics based on knowledge graphs according to claim 1, characterized in that, In step S1, the processing of the object to be analyzed includes: The corresponding parsing adapter is called according to the object type of the object to be analyzed to extract the structural information, dependency information and association information of the object to be analyzed, and convert it into a static intermediate representation using a unified data structure. Based on the object model in the quality characteristic knowledge graph, object entities, entity attributes, and static associations are identified from the static intermediate representation; The object entity is used as an instance node, the static association relationship is used as an instance edge, and the entity attributes are written into the corresponding instance node to construct an object instance subgraph.

3. The general static analysis method for quality characteristics based on knowledge graphs according to claim 1, characterized in that, In S2, the knowledge graph association includes: Extract semantic features of object entities in the object instance subgraph and semantic features of object model nodes in the quality characteristic knowledge graph; The semantic features of the object entity are matched with the semantic features of the object model node to obtain a semantic matching result; Based on the semantic matching results and preset mapping constraints, establish a mapping relationship between object entities and object model nodes; Based on the target quality characteristic configuration, the corresponding quality characteristic nodes, metric nodes, and analysis rule nodes are extracted from the quality characteristic knowledge graph, and the extracted nodes, the mapping relationship, and the object instance subgraph are combined to form an analysis context graph.

4. The general static analysis method for quality characteristics based on knowledge graphs according to claim 1, characterized in that, In step S3, the analysis path generation includes: Based on the target quality characteristic nodes, determine the associated metric nodes, and based on the metric nodes, determine the corresponding analysis rule nodes; Based on the applicable conditions of the analysis rule nodes, the object entity nodes in the analysis context graph are filtered; According to the preset node type order, relationship direction, path depth and loop access restrictions, a graph path search is performed from the target quality characteristic node to the filtered object entity node; Graph paths that meet the preset path integrity conditions are identified as candidate analysis paths.

5. The general static analysis method for quality characteristics based on knowledge graphs according to claim 1, characterized in that, In S4, rule matching includes: Obtain the rule prerequisites, rule judgment conditions, and evidence extraction conditions from the analysis rule nodes in the candidate analysis path; Based on the aforementioned rule prerequisites, filter the object entities associated with the candidate analysis paths; The rules are used to determine the filtered object entities, entity attributes, and static relationships. When the rule determination conditions are met, a rule hit result is generated, and the corresponding static analysis evidence is extracted according to the evidence extraction conditions. Based on the relationships between nodes in the candidate analysis path and the static analysis evidence, a static analysis evidence chain is generated.

6. The general static analysis method for quality characteristics based on knowledge graphs according to claim 1, characterized in that, In S5, risk assessment and result fusion include: The quality risk value is determined based on the rule attributes of the analysis rules, the evidence attributes of the static analysis evidence, and the correlation between quality characteristic nodes. The quality risk value is compared with the preset risk level conditions to determine the quality problem level. Merge the rule matching results that meet the preset merging conditions; Rule hit results with inconsistent judgment results are processed according to preset conflict resolution conditions; Based on the correlation between the rule-hitting results and nodes of different quality characteristics, cross-quality characteristic fusion is performed to form fusion analysis results.

7. The general static analysis method for quality characteristics based on knowledge graphs according to claim 1, characterized in that, In step S6, the analysis results are generated including: Configure a result identifier for the fusion analysis results; Establish the association between the result identifier and the object instance node, quality characteristic node, analysis rule node, quality evaluation data, and static analysis evidence chain; Based on the preset report data structure, the fusion analysis results and their associated data are organized to generate a static analysis report.

8. The general static analysis method for quality characteristics based on knowledge graphs according to claim 1, characterized in that, Before step S1, a quality characteristic knowledge graph construction is included, which comprises: Establish nodes for quality characteristics, metrics, analysis rules, object models, and evidence models; Establish graph edges between the quality characteristic nodes, metric nodes, analysis rule nodes, object model nodes, and evidence model nodes, and configure the relationship type and relationship direction for the graph edges; Entity alignment is performed on graph nodes from different sources that meet preset semantic matching conditions, and node fusion is performed based on the entity alignment results; Configure version information for the graph nodes and graph edges.

9. The general static analysis method for quality characteristics based on knowledge graphs according to claim 1, characterized in that, Following S6, incremental static analysis is further included, which comprises: Obtain change information of the object to be analyzed, and determine the object entity, entity attributes and static associations that have changed based on the change information; Based on the relationships between the changed object entities in the object instance subgraph, determine the affected object entities and the affected analysis paths; Update the object instance subgraph based on the changed object entities, entity attributes, and static relationships; Re-execute S2 through S6 for the affected object entities and affected analysis paths, and update the static analysis report based on the analysis results obtained from the re-execution.

10. The general static analysis method for quality characteristics based on knowledge graphs according to claim 1, characterized in that, The process following S6 also includes analysis feedback and knowledge updates, which include: Obtain the analysis feedback information corresponding to the static analysis results, and establish the association between the analysis feedback information and the corresponding analysis rule nodes, object entity nodes and static analysis evidence chain; Adjust the rule attributes or risk assessment parameters of the corresponding analysis rules based on the analysis feedback information; Based on the analysis feedback information, extract object features or relationships not covered by existing analysis rules, and generate corresponding candidate rules and candidate evidence patterns; The candidate rules and candidate evidence patterns are verified, and the candidate rules and candidate evidence patterns that meet the preset verification conditions are written into the quality characteristic knowledge graph.