Efficient Graph Similarity Join with Scalable Prefix-Filtering Using MapReduce

The graph similarity join retrieves all pairs of similar graphs on graph datasets. In this paper, we propose an efficient MapReduce-friendly algorithm tackling with the graph similarity join problem on large-scale graph datasets. In particular, the efficiency of our algorithm is guaranteed by: 1) scalable prefix-filtering suitable for q-gram alphabet that is beyond the memory; 2) an effective candidate reduction strategy that greatly cuts down the data communication cost; 3) a two-round data access proposal that reduces the data access overhead. Extensive experiments on large-scale real and synthetic datasets demonstrate that our proposal outperforms the state-of-the-art method with higher system scalability and faster speed.