Explainable Knowledge Reasoning on Power Grid Knowledge Graph

The smooth operation of the power grid is closely related to the national economy and people's livelihood. The knowledge graph, as a widely-used technology, has made considerable contributions to power grid dispatching and query answering. However, explainable reasoning on grid defects datasets is still of great challenge, most models cannot balance effectiveness and explainablity. Therefore, their assistance in grid defects diagnosis is minimal. To address this issue, we propose the rule-enhanced cognitive graph for power grid knowledge reasoning. Our model consists of two modules: expansion and reasoning. For the expansion module, we take into consideration that path-based methods often ignore graph structure and global information and combine the local cognitive graph and global degree distribution. For the reasoning module, weprovide reasoning evidence from two aspects: logical rule learning for strong evidence and cognitive reasoning for possible paths. Experiment results on our grid defects dataset make known that our model achieves better performance with explainablity.