Multi-hop temporal knowledge graph reasoning with temporal path rules guidance

Abstract

Temporal Knowledge graph (TKG) is a key component of artificial intelligence. In recent years, many large-scale temporal knowledge graphs have been produced and put into practical applications. At present, researchers have proposed many methods to reason facts that do not exist in temporal knowledge graph using the existing information. However, most traditional embedding-based reasoning methods of temporal knowledge graph lack interpretability, cannot get the reasoning paths. Therefore, the multi-hop path reasoning method of knowledge graph has gradually become a hot-spot. However, technologies related to multi-hop path reasoning in temporal knowledge graph are still in their infancy, and existing methods ignore the importance of logical rules in temporal knowledge graph and do not consider the rules between multi-hop paths. To solve this problem, we propose a multi-hop path reasoning method of temporal knowledge graph based on temporal path rules, named TPRG. TPRG further analyzes the multi-hop paths that may be formed in the multi-hop path reasoning task of temporal knowledge graph, deeply studies the logical connection between the multi-hop paths of temporal knowledge graph and the potential rules between these paths. We define a total of fourteen temporal path rules in TPRG, and these rules can be divided into three different lengths and five different types. The qualified rules are extracted from the datasets according to the proposed temporal path rules. Then, the temporal path rule confidence calculation strategy is proposed. Finally, we conduct temporal knowledge graph reasoning through temporal path rules confidence. Experimentally, our model outperforms other state-of-the-art reasoning methods in several aspects over four public temporal knowledge graph datasets.