Browsing Linked Open Data with Auto Complete

Information in Linked Open Data is incomplete by nature and design. Nevertheless, the more complete information delivered on a subject is, the higher is its value to an end user. With our submission to the Semantic Web Challenge, we show how information, in particular types, can be completed automatically, based on heuristic rule learning. We introduce an approach which employs lazy learning instead of learning a global model, making the approach well scalable to large data, while at the same time providing results at an accuracy of 85.6%. The auto complete function is included in a modular semantic web browser.