|
|
|
| Integrating Local Relationships and Entities in Knowledge Graph Completion Model |
| GAO Rui, SUN Gengxin, BIN Sheng
|
| College of Computer Science & Technology, Qingdao University, Qingdao 266071, China |
|
|
|
|
Abstract Given that the majority of existing knowledge graph completion methods adopt an independent processing approach for triplets, overlooking the varying contributions of neighborhood relations and entities to the central entity,this paper introduces a graph neural network model called REGNN that integrates neighborhood relations and entities. In this model, feature information from relations and entities within the neighborhood is incorporated into the central entity′s update, enriching the representation of the central entity through the aggregation of entity and relation features. Experimental results demonstrate that, in comparison to traditional graph neural network models, REGNN model achieves improvements of 3.3% and 1.5% in terms of the MMR and Hits@10 metrics on the FB15K-237 dataset, and improvements of 1.4% and 3.6% on the WN18RR dataset, thus validating the effectiveness of REGNN model.
|
|
Received: 20 October 2023
Published: 13 February 2026
|
|
|
|
|
|
|
|
|
