Extracting discriminative patterns from graph structured data using constrained search

A graph mining method, Chunkingless Graph-Based Induction (Cl-GBI), finds typical patterns appearing in graph-structured data by the operation called chunkingless pairwise expansion, or pseudo-chunking which generates pseudo-nodes from selected pairs of nodes in the data. Cl-GBI enables to extract overlapping subgraphs, but it requires more time and space complexities than the older version GBI that employs real chunking. Thus, it happens that Cl-GBI cannot extract patterns that need be large enough to describe characteristics of data within a limited time and given computational resources. In such a case, extracted patterns maynot be so interesting for domain experts. To mine more discriminative patterns which cannot be extracted by the current Cl-GBI, we introduce a search algorithm in which patterns to be searched are guided by domain knowledge or interests of domain experts. We further experimentally show that the proposed method can efficiently extract more discriminative patterns using a real world dataset.