IWBase : Provenance Metadata for Understanding and Trusting Answer Justifications

If users are expected to depend on answers from applications and services on the web, they need to have methods for asking for information that will increase their understanding and their trust in the answers. One way of increasing a user’s understanding and trust is to provide a summarized or detailed description of how the answer was obtained. This process would typically depend on some kind of proof structure defending the answer but this alone (no matter how understandable and/or thorough the proof is) is not enough. Any presentation of why a user should depend on an answer also critically depends on metadata concerning elements used in the proof such as underlying data sources, their authoritativeness, reasoning methods used, any uncertainties, etc. In this paper, we introduce IWBase, a comprehensive OWL-based distributed and scalable infrastructure for effectively creating, storing, evolving, and using proofrelated metadata for explanations. IWBase metadata is used to annotate answer justifications encoded in the Proof Markpup Language (PML). IWBase is currently in use by a number of DARPA and ARDA projects.