I'm interested in machine learning, data mining, graphs, social networks, ontologies, and information retrieval. These topics may sound diverse but they are closer than they appear.
Mainly, I think statistical machine learning is a skeleton key that unlocks a great deal of information about data, and since most data can be represented as graphs, the combination of the two just seems common-sensical. Add to that the fact that lots of relational and semantic data is embedded in natural language documents scattered all over the place, and you see why the confluence of these topics is so interesting from a research perspective. Well, ok, I may be a bit biased.