Trust-Based Probabilistic Query Answering

On the semantic web, information from several sources (with disparate trust levels) may be fused with the goal of answering complex queries from end users. In such context, it is critical to aggregate information from heterogeneous sources and support trust-based query answering over the integrated knowledge base. In this paper, we describe a scalable probabilistic query answering over complex, uncertain and partially consistent knowledge bases. The key contributions in this paper are fourfold. First, our approach provides an intuitive query answering semantics over a probabilistic knowledge base. Second, our approach tolerates inconsistencies in knowledge bases. Third, we propose and evaluate a scalable error-bounded approximation query answering over a large knowledge base. Finally, we empirically show the value of taking into account trust information in the query answering process.