CECR: Collaborative Semantic Reasoning on the Cloud and Edge

Semantic reasoning could exploit the implicit information hidden in the graph and enrich the incomplete knowledge graph. Most existing research in semantic reasoning mainly focuses on the completeness of reasoning results and the efficiency of the reasoning algorithm, which neglects the practicality and scalability of reasoning systems. Especially in the Internet era of data explosion, traditional reasoning systems are gradually struggling to meet the demands of large-scale data reasoning. Therefore, scalability has become a focus for reasoning systems. In this paper, we combine cloud and edge computing for scalability with distributed storage based on data correlation and task scheduling. We specifically propose the query-driven backward reasoning optimization algorithm to improve efficiency and overcome the resource limitation of edge nodes. A reasoning system named CECR (Cloud and Edge Collaborative Reasoning System) is implemented to validate our approach. Experiments on three benchmarks demonstrate the soundness and completeness of reasoning results and scalability of CECR.