Publishing Uncertainty on the Semantic Web: Blurring the LOD Bubbles

Abstract

The open nature of the Web exposes it to the many imperfections of our world. As a result, before we can use knowledge obtained from the Web, we need to represent that fuzzy, vague, ambiguous and uncertain information. Current standards of the Semantic Web and Linked Data do not support such a representation in a formal way and independently of any theory. We present a new vocabulary and a framework to capture and handle uncertainty in the Semantic Web. First, we define a vocabulary for uncertainty and explain how it allows the publishing of uncertainty information relying on different theories. In addition, we introduce an extension to represent and exchange calculations involved in the evaluation of uncertainty. Then we show how this model and its operational definitions support querying a data source containing different levels of uncertainty metadata. Finally, we discuss the perspectives with a view on supporting reasoning over uncertain linked data.