FSP: Frequent Substructure Pattern mining

Graphs have become increasingly important in modeling the complicated structures. Mining frequent subgraph patterns is an important research topic in graph mining that helps to analyze the structured database. It has been applied in many applications, such as chemistry, biology, computer networks, and world-wide web. In this paper, we propose a new algorithm called FSP (frequent substructure pattern mining), which improves the state-of-the-art algorithm - gSpan. Our algorithm has reduced the number of graph and subgraph isomorphism tests and the number of accessing the graph database. The performance of FSP was evaluated base on a chemical compound dataset, which is widely used by subgraph mining algorithms. The experimental results show that FSP overcomes with the state-of- the-art gSpan algorithm.