METHOD FOR PRODUCING CO-OCCURRENT SUBGRAPH FOR GRAPH CLASSIFICATION

摘要

The present invention relates to a method to generate a frequent subgraph in a graph data base. More specifically, instead of generating all of the frequent subgraphs of a graph comprising a graph data base, group similar frequent subgraphs based on the structural similarity and generate a frequent subgraph by using a frequent subgraph having the highest grouping records from each similar group during graph mining. In addition, the repeated generation of similar frequent subgraphs can be prevented by generating representative frequent subgraphs among the groups of frequent subgraphs having similar structures, and a frequent subgraph which can accurately execute mining all the graphs included in the graph database by using a frequent subgraph having a limited number can be generated.