OntoCA: Ontology-Aware Caching for Distributed Subgraph Matching

Abstract

With the growing applications of knowledge graphs in diverse domains, the scale of knowledge graphs is dramatically increasing. Based on the fact that a high percentage of queries in practice is similar to previous queries, extensive caching methods have been proposed to accelerate subgraph matching queries by reusing the results of previous queries. However, most existing methods show poor performance when dealing with distributed subgraph matching queries, as numerous intermediate results from the caching should be transmitted to the worker nodes for further validation, leading to extra communication and computation overhead. In this paper, we propose a novel ontology-aware caching method, called OntoCA, which leverages ontology information for efficient distributed queries. Unlike the existing caching methods, our approach fully employs semantic reasoning to filter intermediate results at an early stage, thus improving the query performance. Furthermore, a workload-adaptive prefetching strategy is proposed to increase the hit ratio of OntoCA. The experimental results show that our proposed OntoCA and prefetching strategy outperforms the existing state-of-the-art distributed query method, reducing the query times by 56.16%.