Link Mining for the Semantic Web, Position Statement

A key challenge for data mining is tackling the problem of mining richly structured datasets, where the objectsare linked in some way and we have some ontological information describing the domain. Links among the objects
demonstrate certain patterns, which can be helpful for many data mining tasks and are usually hard to capture with
traditional statistical models. Our goals are to develop machine learning algorithms which can a) make use of the
ontological theory during learning b) learn how to refine, or update the ontological theory, based on learned model
and c) do a mix of both.