Mining Minimal Contrast Subgraph Patterns

摘要: In this paper, we introduce a new type of contrast pattern, the minimal subgraph. It is able to capture structural differences between any two collections graphs and can be useful in chemical compound comparison building graph classification models. However, mining subgraphs challenging task, due exponentially large search space (sub)isomorphism problems. We present an algorithm which utilises backtracking tree first compute maximal common edge sets then uses hypergraph transversal algorithm, derive set subgraphs. An experimental evaluation demonstrates potential our technique for finding interesting data.