131 results about "Graph traversal" patented technology

A system for affinity-based user acquisition is disclosed. An affinity analysis agent analyzes affinity patterns from existing customers and automatically identifies potential referrals without accessing message content. A relationship mapping agent calculates relationship strength between customers and non-customers based on communication frequency, interaction duration, timing patterns, and automatically finds the high-probability network candidates by scoring these relationships. An outreach and engagement agent creates personalized referral offers using relationship information and delivers them through optimal communication channels at ideal times. When someone accepts a referral offer, a billing integration AI agent may be initiated for automated account creation and service activation through identity verification, service selection, and account activation. Each new individual becomes part of the network. The AI communication agents analyze affinity patterns associated with the new individuals for discovering additional compatible individuals through graph traversal algorithms and centrality analysis.

The invention relates to the field of resource association analysis, and discloses a knowledge graph-based archive decision support and association analysis method, which comprises the following steps of: firstly, acquiring structured and unstructured archive data; preprocessing is carried out through data cleaning, duplicate removal and format conversion, semantic disambiguation is carried out on historical fuzzy words in unstructured data, and entity normalization processing is carried out on multi-category names of the same character; secondly, based on the preprocessed structured data, entities, relationships and attributes are extracted, and entity association evidences are supplemented through a cross-modal alignment technology; performing consistency verification, and storing the verified entity relationship network into a graph database to form an archive knowledge graph; then, in combination with a graph traversal and time sequence analysis algorithm, a hidden association path is mined, and trend prediction is carried out; and finally, generating a decision evidence chain according to an association mining result. And establishing a knowledge graph dynamic updating mechanism, and automatically reconstructing the entity association network when a file or a historical research result is newly added.

The invention relates to the field of power networks, and provides an intelligent generation method for power equipment accident recovery. The method comprises the following steps: converting key semantic elements into a structured query instruction; acquiring operation state data of the power grid equipment from a power grid monitoring system in real time, and dynamically updating the operation state data to a pre-constructed power knowledge graph; on the basis of a structured query instruction, performing graph traversal and reasoning on the updated electric power knowledge graph, and searching all reachable electric paths from an available power supply point to a target recovery object; carrying out feasibility evaluation on the reachable electrical path in combination with an operation constraint condition, and generating a candidate recovery path; performing multi-objective optimization sorting on the candidate recovery paths to generate an operation instruction sequence; and converting the operation instruction sequence into at least one form of output scheme. According to the invention, automatic generation and optimization from the natural language instruction to the intelligent recovery scheme are realized.

The invention relates to the technical field of network security, and discloses a security vulnerability AI positioning method and system based on log files. The method comprises the following steps: firstly, analyzing user feedback describing open source component security vulnerabilities to extract fault key entity information, screening related log entries based on the fault key entity information, and constructing a vulnerability propagation graph; identifying a vulnerability triggering initial point set by using a graph traversal algorithm and a graph neural network, calling a component vulnerability case library to deduce a vulnerability utilization propagation chain and calculating a matching degree with a known open source component vulnerability utilization chain feature, and obtaining a maximum probability vulnerability utilization propagation chain and an associated vulnerability triggering initial point set; and generating a root cause positioning report, and updating the knowledge base or model training data according to a user verification result. According to the method, the initial vulnerability point and the complete vulnerability propagation chain of the security vulnerability of the open source component can be accurately and efficiently positioned, and the positioning accuracy and timeliness are improved.

The invention discloses a code change influence analysis method and device, equipment and a storage medium, and relates to the technical field of software, the method comprises the following steps: performing grammar analysis on a multi-programming language code to obtain a back-end abstract syntax tree and a front-end abstract syntax tree, and converting the back-end abstract syntax tree and the front-end abstract syntax tree into a uniform syntax tree structure through a front-end and back-end calling mapping chain; a cross-language dependency graph is constructed according to the unified syntax tree structure, the cross-language dependency graph is composed of nodes and edges, the nodes represent code entities, and the edges represent dependency relations; and when the code change of the bank business is detected, traversing the cross-language dependency graph through a graph traversal algorithm to obtain a code change influence result. Due to the fact that the back-end abstract syntax tree and the front-end abstract syntax tree of the multi-programming language codes are constructed and connected in series, the perfect cross-language dependency graph is constructed, the front-end link and the rear-end link are connected, cross-language tracking can be achieved when the codes of the banking business are changed, and the influence of code change can be accurately analyzed.

The invention relates to the technical field of electrical equipment maintenance, in particular to an electrical equipment maintenance plan intelligent decision-making system based on spatial-temporal characteristics and an application method. Comprising a space-time knowledge graph construction unit; the spatial-temporal feature extraction unit is used for extracting equipment spatial distribution features, load time sequence features and maintenance resource time constraint features from the power grid operation data, constructing a multi-dimensional feature vector, and realizing feature extraction based on a graph traversal algorithm and an LSTM neural network; an intelligent decision-making unit; a main distribution network space-time cooperation unit; and a space-time risk assessment unit. According to the invention, the splicing of the main network and the distribution network CIM model is realized through the main and distribution network space-time cooperation unit, the full-voltage class fusion topological graph is generated, and the influence range of main network maintenance on the power supply path of the distribution network can be automatically identified based on a cross-voltage class topology tracking method; the mechanism effectively reduces the problem of repeated power failure or forwarding path conflict caused by insufficient coordination of the main network and the distribution network.

The invention discloses a collection, analysis and utilization linkage method for the full life cycle of vulnerability intelligence, which comprises the following steps: respectively acquiring original vulnerability intelligence of a plurality of heterogeneous data sources, and carrying out standardization processing on the original vulnerability intelligence to obtain vulnerability description information; constructing an asset knowledge graph, wherein the asset knowledge graph is used for representing IT asset objects in the enterprise and dependency and deployment relationships among the IT asset objects; performing graph traversal query on the vulnerability description information based on an asset knowledge graph, and identifying an affected asset list affected by vulnerabilities; and according to a preset response strategy, generating a safety response instruction from the affected asset list and sending the safety response instruction to the safety protection system. According to the method, the asset knowledge graph is constructed and the multi-source vulnerability information is subjected to standardization and mapping correlation analysis, so that the accurate and rapid identification of the affected assets of the power system and the automatic response strategy generation are realized, and the efficiency and the accuracy of the vulnerability management of the key information infrastructure of the power grid are effectively improved.

The invention belongs to the field of urban traffic management, and provides a hierarchical dynamic routing system and method for a computing power-energy fusion network, and the method comprises the steps: dividing a fusion network graph into a plurality of region sub-graphs, and dividing a path table and a connection point path table based on each region sub-graph; monitoring the running states of the road, the edge computing node and the charging station in real time, and predicting the road congestion intensity, the edge computing node queuing estimation information and the charging constraint; and determining a candidate path set in the partition path table and the connection point path table according to an online request of the vehicle, screening each candidate path in the candidate path set based on the road congestion intensity, the edge computing node queuing estimation information and the charging constraint, and determining an optimal path. According to the method, online retrieval is converted into rapid combination and evaluation of small candidate sets from full-graph traversal; traffic, energy and computing power are incorporated into the same decision framework, and systematicness deviation caused by evaluation of the future at present is avoided.

The invention relates to the technical field of railway four-power operation and maintenance, provides a railway four-power operation and maintenance knowledge intelligent question-answering method and system based on a large language model, and solves the problems of low logic accuracy and poor field applicability of an operation and maintenance question-answering result. The method comprises the following steps: acquiring a question and answer request containing a device identifier, position data and a question text; performing syntactic analysis on the problem text to obtain a target equipment entity and an operation intention, positioning graph nodes in the railway four-electric interlocking relation knowledge graph according to the equipment identifier, performing graph traversal, obtaining associated equipment entities, an electrical connection relation and interlocking logic information, and combining the associated equipment entities, the electrical connection relation and the interlocking logic information into a regulation constraint condition set; performing vector retrieval based on the equipment identifier and the position data to obtain an operation and maintenance knowledge fragment; and inputting the regulation constraint condition set and the operation and maintenance knowledge fragment into the large language model, and carrying out logic constraint on the generation process according to the regulation constraint condition set to generate an intelligent question and answer result text. The logic accuracy and field applicability of the operation and maintenance question and answer result are improved.

The embodiment of the invention discloses a coal mine multi-source geological data dynamic diagnosis system and method based on a knowledge graph, and the system comprises a data collection module which is used for collecting and preprocessing multi-source heterogeneous data, and obtaining a standardized database; the data fusion module is used for determining a geological knowledge map, performing multi-source data knowledge extraction on the standardized database, performing knowledge fusion and conflict resolution, obtaining a knowledge triple and updating the geological knowledge map in real time; the intelligent diagnosis and decision module is used for identifying abnormal events in the standardized database, and performing graph traversal and evidence fusion based on the geological knowledge graph to obtain a diagnosis conclusion; a geological situation is adapted to obtain a diagnosis model, and an optimized diagnosis model is obtained through an incremental learning mechanism; and the decision application module is used for performing three-dimensional visual presentation on the diagnosis conclusion and the geological knowledge map, generating a diagnosis report, updating the geological knowledge map based on the feedback data flow and optimizing the diagnosis model. According to the invention, the accuracy and real-time performance of dynamic diagnosis can be improved.

The invention relates to a transformer substation switch state monitoring system based on centralized monitoring, and relates to the technical field of power system automation, and the system comprises a data collection module which is used for integrating multi-source heterogeneous data; the state judgment module outputs the unique real-time comprehensive state of each switch according to the communication state, the alarm condition function and the multi-stage judgment logic of the original remote signaling; the outage diagnosis module is used for classifying events when the switch trips and automatically screening event sequence records (SOE) to position core diagnosis information; the influence domain pre-judgment module is used for determining all potential influenced switch sets through a graph traversal algorithm on the basis of a power grid topological structure when accident tripping occurs; and an interface generation module. According to the invention, through automatic integration and visual linkage of the operation state, the fault reason and the influence range of the switch, the monitoring intuition, the diagnosis accuracy and the emergency disposal decision-making efficiency are significantly improved.

The application discloses a crop gene function research scheme planning method based on an experimental reasoning chain and belongs to the technical field of scientific research experiment intelligent planning. In view of the problems of lack of experimental reasoning chain modeling, insufficient search and recommendation capacity and low experimental scheme planning precision in the prior art, the application constructs a gene research knowledge graph by extracting research hypothesis-experimental operation-result interpretation triplets from historical documents, receives user experimental information, generates single-step experimental recommendations based on semantic vector matching and graph traversal, simultaneously generates a multi-round experimental scheme set through multi-scenario search, standardization processing and multi-dimensional score sorting, realizes the automatic planning of the crop gene function research experimental scheme, and improves the research efficiency and scientificity.

The invention discloses a crop gene function research scheme planning method based on an experimental reasoning chain, and belongs to the technical field of scientific research experiment intelligent planning. Aiming at the problems of lack of experimental reasoning chain modeling, insufficient retrieval and recommendation ability and low experimental scheme planning precision in the prior art, the method comprises the following steps: extracting a research hypothesis-experimental operation-result interpretation triple from historical literatures to construct a gene research knowledge graph; user experiment information is received, single-step experiment recommendation is generated based on semantic vector matching and graph traversal, meanwhile, a multi-round experiment scheme set is generated through multi-situation retrieval, standardized processing and multi-dimensional scoring sorting, automatic planning of crop gene function research experiment schemes is achieved, and research efficiency and scientificity are improved.

The invention discloses a multi-source property right data verification method, and relates to the technical field of data processing. The method comprises the following steps: acquiring property right data from different data sources, and converting the property right data into a unified field according to semantics of the property right data and a preset mapping rule; after the knowledge graph is constructed based on the uniform field, the business rule vectors are coded into node attributes of the knowledge graph; and traversing nodes and edges in the knowledge graph through a preset graph traversal algorithm, and verifying the property right data according to the business rule vector to obtain a verification result. According to the method, the association relationship between the knowledge graph integration entities is constructed, and the service rule is encoded into the knowledge graph for verification, so that verification is not limited to single node data any more, cross-node verification can be performed in combination with the association information, and the verification accuracy is improved.

The invention discloses a Verilog vulnerability detection method generated based on a graph traversal rule, and the method comprises the steps: firstly receiving a Verilog source code as a target detection code, generating and fusing three code structures, and forming an attribute graph VeriPG; secondly, acquiring natural language description information of a target vulnerability, and extracting a potential vulnerability judgment condition by utilizing a judgment condition LLMs; and then generating a vulnerability detection rule of the target vulnerability based on the VeriPG and a vulnerability judgment condition drive detection rule LLMs, and verifying the generated vulnerability detection rule. And finally, designing and deploying a rule executor, and executing graph traversal on the verified vulnerability detection rule to complete Verilog vulnerability detection. According to the method, the relationship between the control logic and the data can be comprehensively described, and the Verilog security vulnerability detection method with high accuracy, high automation and high adaptability is constructed.

In one aspect, a method for modeling graphs on distributed storage, comprising: receiving graph data including vertices, edges, and properties; mapping each vertex to a vertex record in a distributed storage system; mapping each edge to an edge record in the distributed storage system; mapping each property to a property record in the distributed storage system; determining, for each vertex, whether a number of incident edges exceeds a maximum record size threshold; storing pointers to incident edge records directly within vertex records when the number of incident edges does not exceed the maximum record size threshold; utilizing an adjacency index to lookup incident edge records when the number of incident edges exceeds the maximum record size threshold; and executing graph traversal operations that navigate from vertex to vertex by accessing the appropriate edge records through either the stored pointers or the adjacency index.

The invention provides a digital main line processing method based on an industrial data atlas, a medium and a system, and belongs to the technical field of the industrial data atlas. Unified access of cross-system data is realized by constructing a multi-source data acquisition adapter, cleaning and standardization processing are performed on heterogeneous data, an integrated data set is constructed, and the data collection efficiency is improved. Data entities are abstracted as nodes, cross-system association relationships are abstracted as edges to construct an industrial data graph, a graph index optimization strategy and a hierarchical storage architecture are adopted to improve cross-system query performance, a graph adaptive evolution model is utilized to realize dynamic adjustment of a graph structure, and a cross-system deep association analysis function is realized through a graph traversal algorithm. And an access control and data security mechanism is established to complete system protection, so that the technical problem that the industrial data processing system is difficult to realize cross-system deep correlation analysis is solved.

The invention discloses a method and a system for generating a test case for covering a loop boundary based on program slices. The method comprises the following steps of: inputting a source code to be tested and a target loop line number; the method comprises the following steps: analyzing a to-be-tested source code based on an abstract syntax tree technology, and extracting key node information to construct a program dependency graph; collecting related code lines through graph traversal by taking a loop structure needing to be analyzed as a criterion, and generating a minimum executable loop slice; performing symbolic execution on the minimum executable loop slice, simulating a loop execution process, capturing condition judgment nodes, and forming a track dictionary; determining a strategy based on the trajectory dictionary and the variable state, and when the strategy is numerical search, performing parameter space exploration; when the strategy is symbolic analysis, analyzing the cyclic condition expression and solving the optimal parameter; generating candidate test cases based on the analysis result; executing the candidate test case, calculating the number of loop iterations, and determining whether an expected state is reached; and if the expected state is reached, outputting a final test case.

The invention discloses a software supply chain risk detection method and system, a medium and a product, and relates to the technical field of risk detection. The method comprises the following steps: acquiring multi-source heterogeneous data of a software supply chain, and constructing the multi-source heterogeneous data into a multi-dimensional software supply chain relation graph; querying nodes and attributes associated with the target asset node based on the multi-dimensional software supply chain relation graph to extract risk features; obtaining an initial risk score of a vulnerability node associated with the target asset node from the public vulnerability library, calculating the shortest propagation path depth between the vulnerability node and the target asset node by using a graph traversal algorithm, and adjusting the initial risk score in combination with code reachability analysis to obtain a corrected technical risk feature; and calculating a comprehensive risk score based on the corrected technical risk characteristics, and executing automatic safety examination and approval, interception or manual reexamination operation. By implementing the technical scheme, the software supply chain risk can be comprehensively and accurately detected.

The invention discloses an intelligent switch fault diagnosis method based on multilayer data reasoning, which relates to the field of switches, collects and fuses multi-source data of a switch, converts entities, attributes and relationships in a multi-source fusion data set into an RDF triple based on a predefined network operation and maintenance ontology model, and performs fault diagnosis on the RDF triple. The method comprises the steps of constructing a switch fault diagnosis knowledge graph, dynamically generating an SPARQL query statement based on real-time monitoring data, performing graph traversal and logical reasoning on the switch fault diagnosis knowledge graph, and outputting a switch fault diagnosis conclusion. According to the method, comprehensive upgrading of switch fault diagnosis from passive response to active prevention is achieved, the network operation and maintenance efficiency and decision quality are remarkably improved, equipment monitoring, topological relations and service data are integrated through a unified semantic framework, a structured knowledge graph is constructed, and fault diagnosis has semantic interpretability.

Systems and methods are directed to incorporating large language model (LLM) prompts into graph queries using graph query language and executing these graph queries. The system receives a graph query that includes a LLM prompt in the graph query language. The LLM prompt comprises an input context and a query to be answered. Based on the graph query, the system traverses a knowledge graph to obtain context data associated with the input context. The context data may be constrained by access control policies of an enterprise providing the knowledge graph. A context table may be generated using the context data based on the input context indicated in the LLM prompt. The system then invokes an LLM to determine a result by applying the query of the LLM prompt to the context data obtained from the traversal of the knowledge graph. The result is then presented on a user interface of a client device.

The invention provides a QKK product index association analysis method and system based on semantic recognition and a storage medium, and relates to the technical field of communication network operation and maintenance, and the method comprises the steps: collecting QKK index data from a plurality of heterogeneous data sources, and carrying out the association alignment; extracting a numerical semantic vector and a text semantic vector, and mapping the numerical semantic vector and the text semantic vector to a unified semantic vector space through a cross-modal semantic alignment network; constructing an initial ontology based on a domain standard, discovering new concept and new relationship candidates through clustering processing and a graph neural network, and updating the new concept and new relationship candidates to a dynamic semantic ontology library after man-machine cooperative verification; and instantiating the ontology to construct a semantic association graph, performing graph traversal on a target event to obtain an association path, and outputting a root cause analysis result by applying a causal discovery algorithm in combination with multi-evidence fusion. According to the method, the problem of semantic barrier of heterogeneous data is solved, root cause positioning of ontology dynamic evolution and causal levels is realized, and the accuracy of root cause analysis is remarkably improved.

The application discloses a social network-based insurance fraud gang identification method, system, device and medium, and relates to the technical field of computers, wherein the method comprises the following steps: constructing a social network relationship graph based on the role information and subject information in a newly added claim event and using insurance business relationships; obtaining traversal starting points for constructing a closed-loop relationship graph according to the social network relationship graph; constructing a plurality of closed-loop relationship graphs for a plurality of traversal starting points by using a first adjacent point acquisition method based on claim event relationships; and obtaining a target closed-loop relationship graph by merging the plurality of closed-loop relationship graphs by using a second adjacent point acquisition method based on claim event relationships, wherein the node information in the target closed-loop relationship graph is used to represent a target insurance fraud gang. The application is suitable for efficient investigation of fraud case identification and improves the accuracy and efficiency of fraud case identification.

A data processing method, data processing apparatus, electronic device, and storage medium are disclosed. The data processing method includes: acquiring a computation graph; traversing the computation graph sequentially according to data flow direction, branch by branch, to generate an execution sequence that satisfies data dependencies and has a defined temporal order on the computation graph. The traversal process includes at least one branch search operation, each branch search operation including: acquiring at least one branch corresponding to the branch search operation; adding at least one branch to a branch container; sequentially acquiring branches that meet processing conditions from the branch container and performing operator fusion and other processing, then generating the execution sequence portion corresponding to that branch on the computation graph. This computation graph processing method integrates the traversal of a directed acyclic graph, operator fusion, and execution sequence generation into one process, requiring only one graph traversal to complete the processing from the computation graph to the execution sequence, thus improving the processing efficiency of the computation graph.

The invention discloses a large model lightweight reasoning method and device, computer equipment and a storage medium. The method comprises the steps that a target large model is unfolded into a computational graph, and a semantic path dependency graph is constructed based on intermediate representation; defining a semantic contribution degree index, and calculating the semantic influence degree of each node in the semantic path dependency graph on final output; screening a low-influence path according to a graph traversal algorithm, and performing path pruning in combination with structure limitation to obtain a pruned model; and carrying out structure completion on the pruned model, and carrying out semantic alignment distillation by taking the original model as a teacher model to obtain a lightweight model. The embodiment of the invention relates to the field of artificial intelligence model reasoning optimization, and solves the problems that a pruning method in the prior art depends on structural information, lacks semantic perception and is easy to lose performance. The semantic graph is constructed in the middle presentation layer, task related paths are identified, redundant branches are cut off, the reasoning process is remarkably accelerated on the premise that the precision of the model is kept, and the deployment efficiency is improved.

The invention provides an intelligent early warning and root cause positioning method and system for a multi-source log and a storage medium, and the method comprises the steps: firstly, analyzing a historical log stream, generating an index dynamic baseline which is adaptive along with a period, carrying out the index extraction and deviation calculation of a real-time multi-source log stream, and carrying out the comprehensive judgment of continuous deviation and correlation deviation, thereby achieving the intelligent early warning and root cause positioning of the multi-source log. Generating an initial abnormal event carrying a context; secondly, graph traversal and deviation quantitative analysis are carried out on the event based on an association graph database, and when the calculated link deviation exceeds a threshold value, the event is upgraded to be an early warning event; then, according to a preset fault mode knowledge base, executing a plurality of root cause speculation paths containing threshold comparison according to a priority sequence, so as to quickly lock high-probability root causes; and finally, generating and displaying a structured analysis report by automatically integrating all analysis processes and results. Therefore, the early warning accuracy and the fault positioning efficiency are improved.

The invention relates to the technical field of software testing and artificial intelligence, in particular to an intelligent test case generation method and system based on a dynamic knowledge graph, and the method comprises the steps: collecting multi-source data, and constructing an initial graph; capturing modification information corresponding to system version iteration, and triggering incremental expansion of the initial atlas based on the modification information to obtain a dynamically evolved real-time atlas; obtaining a test requirement, discovering a full-chain path in the real-time map through map traversal, analyzing a coverage gap and a node risk in the full-chain path, and screening an optimal path combination based on reinforcement learning; generating a test case according to the optimal path combination and the corresponding parameter information; and obtaining a use case execution result to extract an execution index, performing defect association based on the execution index, and performing path-based defect generation algorithm optimization to adjust a graph traversal weight so as to realize closed-loop optimization. The method has the effect of efficiently coping with the dynamic change of the complex system.

The invention discloses a mass defense and mass governance efficiency optimization method based on big data analysis, and relates to the technical field of public security big data analysis, and the method comprises the steps: matching continuously updated real-time event data as a seed node in a mass defense and mass governance dynamic affair graph, deducing a high-probability event chain set through a graph traversal algorithm, and carrying out the optimization of the mass defense and mass governance efficiency; performing graph topological structure analysis on the high-probability event chain set, and identifying an intervention node list; and performing matching calculation on the ability portraits corresponding to the intervention strategy task set and the natural community set, assigning the intervention strategy task to the natural community with the highest matching degree for execution, and collecting execution feedback data to update the personnel cooperation relationship network map. According to the method, the corresponding capability portrait is created for the natural community set, so that intelligent grouping and efficiency quantitative evaluation of mass defense and mass treatment power are realized, distribution and utilization of human resources are optimized, and team cooperation capability and task execution efficiency are enhanced.

This application provides a multi-agent, multimodal collaborative method, device, equipment, and storage medium for diagnosing the condition of patients in the intensive care unit (ICU), belonging to the field of medical artificial intelligence and intelligent analysis technology for intensive care. The method includes: receiving multimodal clinical monitoring data in an ICU setting, including bedside medical images, continuous life time series, and critical care clinical text; identifying and adaptively routing the input data through a modality detection agent, distributing it to corresponding domain expert agents for pathological and physiological feature extraction; constructing a patient-centric graph structure representing the dynamic clinical condition of ICU patients through a knowledge graph agent based on the features output by each domain expert agent, this graph structure integrating structured medical knowledge of entity extraction and relational reasoning; further, inputting the constructed graph structure and original multimodal features into a collaborative agent, generating the final ICU patient condition diagnosis result (such as mortality prediction, ICU long-stay prediction, etc.) and a traceable reasoning path through graph traversal and collaborative reasoning. This application effectively breaks down data silos between different monitoring devices in the intensive care unit environment, dynamically captures cross-modal associations specific to critically ill patients, and significantly improves the accuracy of critical care clinical diagnosis and medical trust.

This invention discloses a PageRank-based agent discovery method and system. The method involves initializing the agent discovery system's agent list, comprehensive scoring weight parameters, and system configuration; associating basic information with agent skill information to form a skill candidate list; calling an API to generate semantic vectors for candidate skills and task descriptions in the skill candidate list through semantic pre-screening, calculating cosine similarity as the basic matching degree, adding a matching bonus score to the basic matching degree to obtain a semantic enhancement score, and sorting the semantic enhancement scores; calculating and weighting the PageRank score, call frequency score, call success rate score, and recent activity score; sorting the scores and returning the top M results to the user. The method models the agent's historical call records as a weighted directed graph and uses the PageRank algorithm to quantify the global importance of nodes, transforming the retrieval process into a graph traversal guided by a "importance-relevance" dual-factor approach, significantly reducing the inclusion rate of irrelevant results.