Rule-Based Trust Assessment on the Semantic Web

Abstract

The Semantic Web is a decentralized forum on which anyone can publish structured data or extend and reuse existing data. This inherent openness of the Semantic Web raises questions about the trust-worthiness of the data. Data is usually deemed trustworthy based on several factors including its source, users' prior knowledge, the reputation of the source, and the previous experience of users. However, as rules are important on the Semantic Web for checking data integrity, representing implicit knowledge, or even defining policies, additional factors need to be considered for data that is inferred. Given an existing trust measure, we identify two trust axes namely data and rules and two trust categories namely content-based and metadata-based that are useful for trust assignments associated with Semantic Web data. We propose a meta-modeling framework that uses trust ontologies to assign trust values to data, sources, rules, etc. on the Web, provenance ontologies to capture data generation, and declarative rules to combine these values to form different trust assessment models. These trust assessment models can be used to transfer trust from known to unknown data. We discuss how AIR, a Web rule language, can be used to implement our frame-work and declaratively describe assessment models using different kinds of trust and provenance ontologies.