A Scalable Parallel Semantic Reasoning Algorithm-Based on RDFS Rules on Hadoop

The rapid growth of semantic web utilization in the cloud has resulted in massive amounts of RDF data, which is challenging large-scale RDF semantic reasoning. The traditional semantic reasoning process is very time-consuming and lacks scalability. In this paper, we present a scalable method for RDF Srule-based semantic reasoning using a distributed framework of Hadoop MapReduce, and propose an optimized semantic reasoning algorithm based on RDFS rules. The reasoning algorithm first classifies RDFS entailment rules to build different reasoning rule models, and then orders the rule execution sequences according to the relation of RDFS entailment rules to reduce reasoning time. During algorithm execution in MapReduce, the reasoning work handles RDFS rules in the Map process phase, and data duplication elimination is handled in the Reduce process phase. The experiment results on the LUBM benchmark show that our optimized reasoning algorithm outperforms Urbani's reasoning method in efficiency, stability, and scalability. The average reasoning time of our algorithm is only 1/3 that of Urbani's algorithm with different RDF dataset scales.