Trust and Provenance You Can’t Have One Without The Other

On the social web content is no longer generated by a small number of established authorities but by a huge number of mostly anonymous users. On the semantic web new content is created by combining existing information from different sources on the fly. In both cases trust is of fundamental importance. Is the user providing new content trustworthy? How trustworthy is information inferred from such content? Trust ratings are one approach to define a measure for user reputation. Provenance in turn, provides meta information about the creation and processing of content. It answers questions such as who created the original source, under which conditions, and how it was further manipulated before being published. Provenance makes trust ratings transparent, i.e., reproducible. On the semantic web trust and provenance are on the top of the so-called semantic web layer cake and hence necessary parts of its infrastructure. In contrast, while trust ratings are popular on the social web, provenance plays a minor role so far. In this paper we use a simple example to demonstrate why this is questionable and how the absence of provenance may render trust ratings useless.