Paper 59 (Research track)

DRAGON: Decision Tree Learning for Link Discovery

Author(s): Daniel Obraczka, Axel-Cyrille Ngonga Ngomo

Abstract: The provision of links across RDF knowledge bases is regarded as fundamental to ensure that they can be used in real-world applications. The growth of knowledge bases both with respect to their number and size demands the development of time-efficient and accurate approaches for the computation of such links. In this work we present DRAGON, a fast decision-tree-based approach that is both efficient and accurate. Our approach is based on the insight that the semantics of decision trees is equivalent to that of link specifications, which are commonly used to compute links across knowledge bases. We use this insight and the characteristics of link specifications to derive a decision-tree learning algorithm dedicated to discovering link specifications. We evaluate DRAGON by comparing it with state-of-the-art link discovery approaches as well as the common decision-tree-learning approach J48. Our results suggest that our approach achieves state-of-the-art performance in regards to F-measure while being up to 6 times faster than existing algorithms for link discovery on RDF knowledge bases.

Keywords: Decision Trees; Machine Learning; Linked Data; Semantic Web; Link Discovery

Share on

Leave a Reply

Your email address will not be published. Required fields are marked *