CN-122019602-A - Performance management method, device and medium based on knowledge graph

CN122019602ACN 122019602 ACN122019602 ACN 122019602ACN-122019602-A

Abstract

The application provides a performance management method based on a knowledge graph, which comprises the steps of preliminarily integrating multi-source heterogeneous data to obtain a target breeding event message, converting the discrete target breeding event message into structural graph data by utilizing a talent breeding knowledge graph, and mining implicit relations which are not explicitly expressed in the message through semantic complementation and context association to avoid omission of breeding workload. And then converting the mapped drawing event data into drawing event vectors based on the graph neural network model, thereby quantifying the complex drawing work of teachers, being suitable for business logic of different business scenes and improving the flexibility of the performance management method. Finally, matching the guiding event vector with a preset guiding target rule base, thereby realizing intelligent accounting of the guiding target completion degree, improving the accounting precision and efficiency of performance, supporting flexible adjustment of guiding rules, adjusting and expanding, and improving the suitability of the performance management method.

Inventors

JIANG WEI
SONG XIAOMENG
ZHANG LEYU
PANG LI
YANG XIAOYING
SONG YUNYUN
WANG XING
LU CHENXI

Assignees

广州科技贸易职业学院

Dates

Publication Date: 20260512
Application Date: 20251223

Claims (10)

1. The performance management method based on the knowledge graph is characterized by comprising the following steps of: Acquiring a target breeding event message for representing breeding workload of a teacher to be calculated; Based on a preset talent education knowledge graph, carrying out context association and semantic complementation on the target education event message to generate mapped education event data; Generating a drawing event vector of the teacher to be calculated based on the graph neural network model and the mapped drawing event data; And calculating the drawing target completion degree of the teacher to be calculated based on a preset drawing target rule base and the drawing event vector, so as to finish talent drawing performance calculation of the teacher based on the drawing target completion degree.
2. The performance management method of claim 1, wherein the generating mapped educated event data based on the preset talent educated knowledge graph by performing context correlation and semantic completion on the target educated event message comprises: Based on the predefined entity type and the predefined relation type in the talent promotion knowledge graph, entity identification and relation extraction are carried out on the target promotion event message, so that a target promotion event entity and a target promotion event relation are obtained; fusing the target education event entity to a corresponding node in the talent education knowledge graph; Metadata used for describing inherent characteristics of the target promotion event entity in the target promotion event message is used as node attributes to be added to corresponding nodes in the talent promotion knowledge graph; Metadata used for describing the relation characteristics of the target educated events in the target educated event message is used as an edge attribute to be attached to the corresponding edge in the talent educated knowledge graph; based on the display relationship of the fused talent introduction knowledge graph, carrying out implicit relationship reasoning on the target introduction event message; And constructing the context association of the target eduction event message based on the talent eduction knowledge map fused with the implicit relation, and obtaining the mapped eduction event data based on the context association.
3. The performance management method as recited in claim 2, wherein said performing implicit relationship reasoning on said target promotion event message based on the display relationship of the merged talent promotion knowledge graph comprises: Constructing at least one candidate relation chain starting from a teacher node to be calculated based on a display relation path template corresponding to the display relation, wherein the teacher node to be calculated is a teacher node in the target introduction event message; filtering candidate relation chains with the opposite confidence level lower than a preset confidence level threshold based on a multisource confidence level fusion strategy to obtain filtered candidate relation chains; And calculating the complement probability of each missing link in the candidate relation chain based on the graph embedding model and the candidate relation chain, and generating the implicit relation based on the missing links of which the complement probability exceeds a preset probability threshold.
4. The performance management method of claim 2, wherein the constructing the contextual relevance of the target educated event message based on the talent educated knowledge-graph fused with the implicit relationship comprises: Adding the implicit relation as a new triplet to the talent introduction knowledge graph; Taking a to-be-calculated teacher node corresponding to the to-be-calculated teacher as a center, and extracting a first-degree association subgraph corresponding to a first-degree neighbor node from the added talent introduction knowledge graph, wherein the first-degree neighbor node has a direct relationship with the to-be-calculated teacher node; Based on the first-degree association subgraph, extracting a second-degree association subgraph corresponding to a second-degree neighbor node from the added talent introduction knowledge graph, wherein the second-degree neighbor node has an indirect relationship with the teacher node to be calculated; and integrating the first-degree association subgraph and the second-degree association subgraph to obtain the context association.
5. The performance management method as set forth in claim 4, wherein the extracting a first-degree associative subgraph corresponding to a first-degree neighbor node from the added talent introduction knowledge graph centering on a to-be-calculated teacher node corresponding to the to-be-calculated teacher includes: Starting from the teacher node to be calculated, traversing all outgoing edges and incoming edges of the teacher node to be calculated, and determining one-degree neighbor nodes directly connected with the teacher node to be calculated; Extracting edges of the teacher node to be calculated and all the one-degree neighbor nodes to obtain a direct relation set, wherein the direct relation comprises a guiding relation, a participation relation, a membership relation and an introduction relation; And combining the teacher node to be calculated, the first-degree neighbor node set and the direct relation set to construct the first-degree association subgraph.
6. The performance management method according to claim 4, wherein the extracting the second-degree associative subgraph corresponding to the second-degree neighbor node from the added talent introduction knowledge graph based on the first-degree associative subgraph includes: starting from each neighbor node in the first-degree neighbor node set, traversing the outgoing edge and the incoming edge of each neighbor node, and determining a second-degree neighbor node indirectly connected with the teacher node to be calculated; obtaining edges connecting the first-degree neighbor node set and the second-degree neighbor node set to obtain an indirect relation set, wherein the indirect relation is used for representing indirect contribution of the teacher to be calculated; The indirect relation comprises papers published by students, projects participated by the students, projects responsible for introducing talents and expenses of projects to account; and combining the first-degree association subgraph, the second-degree neighbor node set and the indirect relation set to construct the second-degree association subgraph.
7. The performance management method of claim 1, wherein the generating the training event vector for the teacher to be calculated based on the graph neural network model and the mapped training event data comprises: converting the mapped educated event data into a graph structure comprising preset educated entity nodes and preset educated event edges, wherein the educated entity nodes comprise teacher nodes, event nodes, student nodes and project nodes, and the educated event edges comprise guiding relations, participation relations and membership relations; Initializing the characteristics of each node, wherein the characteristics of the teacher node comprise ages, job titles, disciplines and historical performance scores, the characteristics of the event nodes comprise event types, event levels, occurrence events and continuous events, the characteristics of the student nodes comprise entrance years, professional directions and culture levels, and the characteristics of the project nodes comprise project levels, expense amounts and research periods; Based on a graph rolling network or a graph annotation meaning network, carrying out neighbor node feature aggregation and feature update on the initialized nodes to obtain node vectorization representation of each node, wherein the node vectorization representation is used for vectorizing self features and local neighborhood information of each node; And extracting the teacher characteristic vector of the teacher to be calculated and the event characteristic vector used for representing the guiding workload of the teacher to be calculated from the node vectorization representation of each node as the guiding event vector.
8. The performance management method according to any one of claims 1-7, wherein the calculating the fertility target completion of the teacher to be calculated based on a preset fertility target rule base and the fertility event vector includes: Coding each rule in the eduction target rule base as a rule feature vector, and calculating the matching degree of the eduction event vector and each rule feature vector; determining the number of completed education events of the teacher to be calculated based on a preset matching degree threshold and each matching degree; and calculating the completion degree of the seeding target based on the total number of the seeding events and the completion number of the seeding events.
9. A computer device comprising a processor, a memory, and a knowledge-based performance management program stored on the memory and executable by the processor, wherein the knowledge-based performance management program, when executed by the processor, implements the steps of the knowledge-based performance management method of any of claims 1-8.
10. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a knowledge-based performance management program, wherein the knowledge-based performance management program, when executed by a processor, implements the steps of the knowledge-based performance management method according to any one of claims 1 to 8.

Description

Performance management method, device and medium based on knowledge graph Technical Field The present application relates to the field of data processing, and in particular, to a performance management method, apparatus, and medium based on a knowledge graph. Background The traditional talent guidance performance management of universities depends on a manual reporting and periodic assessment mode, and the data verification of the method excessively depends on manual auditing, and performance accounting logic is solidified and rigidified, so that the method is difficult to adapt to dynamic rule adjustment and real-time decision scenes. The manual filling data is easy to be interfered by subjective factors, the filling efficiency and the accuracy are low, the execution process of the performance rules depends on manual experience judgment, the rule analysis precision is low, the expansibility is weak, and the real-time processing requirement of a large-scale high concurrent service scene cannot be met. Therefore, how to improve the accounting efficiency and the accounting accuracy of the performance becomes a technical problem to be solved urgently at present. Disclosure of Invention The application mainly aims to provide a performance management method, computer equipment and a computer readable storage medium based on a knowledge graph, aiming at improving the accounting efficiency and the accounting precision of performance. The application provides a performance management method based on a knowledge graph, which comprises the following steps of obtaining a target eduction event message used for representing eduction workload of a teacher to be verified, carrying out context association and semantic complementation on the target eduction event message based on a preset talent eduction knowledge graph to generate mapped eduction event data, generating eduction event vectors of the teacher to be verified based on a graph neural network model and the mapped eduction event data, and calculating eduction target completion degree of the teacher to be verified based on a preset eduction target rule base and the eduction event vectors to complete talent eduction performance verification of the teacher based on the eduction target completion degree. In addition, to achieve the above object, the present application further provides a computer device, where the computer device includes a processor, a memory, and a performance management program based on a knowledge graph stored in the memory and executable by the processor, where the performance management program based on a knowledge graph, when executed by the processor, implements the steps of the performance management method based on a knowledge graph as described above. In addition, in order to achieve the above object, the present application further provides a computer readable storage medium, on which a performance management program based on a knowledge graph is stored, wherein the performance management program based on the knowledge graph realizes the steps of the performance management method based on the knowledge graph when being executed by a processor. The application provides a performance management method based on a knowledge graph, which is characterized in that multi-source heterogeneous data are preliminarily integrated to obtain a target breeding event message, the discrete target breeding event message is converted into structural graph data by utilizing a talent breeding knowledge graph, and an implicit relation which is not explicitly expressed in the message is mined through semantic complementation and context association, so that omission of breeding workload is avoided. And then converting the mapped drawing event data into drawing event vectors based on the graph neural network model, thereby quantifying the complex drawing work of teachers, being suitable for business logic of different business scenes and improving the flexibility of the performance management method. Finally, matching the guiding event vector with a preset guiding target rule base, thereby realizing intelligent accounting of the guiding target completion degree, improving the accounting precision and efficiency of performance, supporting flexible adjustment of guiding rules, adjusting and expanding, and improving the suitability of the performance management method. Drawings Fig. 1 is a schematic flow chart of a performance management method based on a knowledge graph provided by the application; Fig. 2 is a flow chart of another performance management method based on a knowledge graph provided by the present application. The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments. Detailed Description The following description of the embodiments of the present application will be made clearly and fully with reference to the