Using Background Knowledge for Graph Based Learning: A Case Study in Chemoinformatics

The benefit of incorporating background knowledge in the learning process has been successfully demonstrated in numerous applications of ILP methods. Nevertheless the effect of incorporating background knowledge in graph learning has not yet been systematically explored. A first step in this direction is taken in this work, where a case study in chemoinformatics is presented, in which various types of background knowledge are encoded in graphs that are given as input to a graph learner. It is shown that the type of background knowledge encoded indeed has an effect on the predictive performance, and it is concluded that encoding appropriate background knowledge may even be more important than selecting which graph learning algorithm to use.

• Type: Conference
• Year: 2007
• Where: IMECS
• Authors: Thashmee Karunaratne, Henrik Boström
• Keywords: Background Knowledge, Graph Learning, IMECS 2007, Knowledge In Graph, Theoretical Computer Science