Graph mining is an important research area in data mining , while graph mining mainly focuses on frequent subgraph mining in graph datasets. The key step in the research of frequent subgraph mining techniques is to establish an effective mechanism to reduce the generation of redundant candidate subgraphs in order to efficiently calculate and process the required frequent subgraphs. A path-based frequent subgraph mining algorithm is proposed. The algorithm first finds all frequent edges to mine frequent single paths, then expands large frequent paths through combination operation and bijection sum operation, and then uses the connection operator to generate all frequent subgraph candidate sets. The correctness and completeness of the algorithm are proved by theorem. Theoretical analysis shows that the algorithm has lower time complexity than the existing algorithm. Finally, experiments are carried out on two graph datasets, and the results verify that the algorithm is superior in terms of quality and time performance when generating candidate sets.

 

 ,