Graph based molecular data mining - an overview

In the past years quite a lot of algorithms concerning frequent graph pattern mining have been published. In this paper an overview on the different methods for graph data mining is given, starting with the greedy searches proposed in the middle of the nineties. The ILP-based approaches are taken into account as well as ideas influenced by basket analyses proposed lately. A remaining question is how the different approaches can be tailored to meet the needs for mining molecules. In this area special problems occur as molecules are not just "normal arbitrary graphs". There are structures that are typical and frequent as rings and chains, some node types resp. atoms occur more often than others. It is an unsolved question how chemically isomorphic mining can be handled