There is general agreement that the effectiveness of Machine Learning and Knowledge Discovery output strongly depends not only on the quality of source data and the sophistication of learning algorithms, but also on additional input provided by domain experts. There is less agreement on whether, when and how such input can and should be formalized as explicit prior knowledge.

The six chapters in the first part of the book aim to investigate this aspect by addressing four different topics: inductive logic programming; the role of human users; investigations of fully automated methods for integrating background knowledge; the use of background knowledge for Web mining. The two chapters in the second part are motivated by the Web 2.0 (r)evolution and the increasingly strong role of user-generated content. The contributions emphasize the vision of the Web as a social medium for content and knowledge sharing.