Neighbor-base similarity matching for graphs

The rapid development of internet and data centers has made cloud data management a major issue in database management system. Various cloud data management related applications require the basic operation of graph pattern matching. Exact matching method for graph pattern matching is too restrictive and it incurs very high computational cost as an NP-complete problem. Thus, it cannot be applied to most cloud applications. So several approximate notions are proposed. However, traditional approximate matching methods are still too restrictive in some situations, and some of them may neglect important nodes in the pattern. To address these problems, we propose a novel notion for graph pattern matching, and show that it can be processed in polynomial time. In addition, our method is flexible, free of thresholds and does not leave out any node in the pattern.