In recent years, data mining in graphs or graphmining have attracted much attention due to explosivegrowth in generating graph databases. The graph databaseis one type of database that consists of either a single largegraph or a number of relatively small graphs. Someapplications that produce graph database are as follows:Biological networks, semantic web and behavioral modeling.Among all patterns occurring in graph database, miningfrequent subgraphs is of great importance. The frequentsubgraph is the one that occurs frequently in the graphdatabase. Frequent subgraphs not only are importantthemselves but also are applicable in other aspects of dataanalysis and data mining tasks, such as similarity search ingraph database, graph clustering, classification, indexing,etc. So far, numerous algorithms have been proposed formining frequent subgraphs. This study aims to createoverall view of the algorithms through the analysis andcomparison of their characterizations. To achieve the aim,the existing algorithms are classified based on their graphdatabase and their subgraph generation way. The proposedclassification can be effective in choosing applicationsappropriate algorithms and determination of graph miningnew methods in this regard.
展开▼