Taxonomy-Driven Lumping for Sequence Mining
In many application domains, events are naturally organized in a hierarchy. Whether events describe human activities, system failures, coordinates in a trajectory, or biomedical phenomena, there is often a taxonomy that should be taken into consideration. A taxonomy allow us to represent the information at a more general description level, if we choose carefully the most suitable level of granularity.
Given a taxonomy of events and a dataset of sequences of these events, we study the problem of finding efficient and effective ways to produce a compact representation of the sequences. This can be valuable by itself, or can be used to help solving other problems, such as clustering.
KeywordsData Mining Markov Model Knowledge Discovery Application Domain System Failure
- 1.Bonchi, F., Castillo, C., Donato, D., Gionia, A.: Taxonomy-driven lumping for sequence mining. Data Mining and Knowledge Discovery (2009) doi: 10.1007/s10618-009-0141-6 Google Scholar