FindXpRT
Find an eXpert via Rules and Taxonomies

The Semantic Web contributes to making AI mainstream. AI knowledge representations such as ontologies and rules can be used to formalize Web metadata vocabularies, where constants, classes, and relations carry URIs. This permits AI reasoning over metadata both for direct question answering and high-precision information (URI) retrieval. In parallel, the Web 2.0 has been developed with social networking portals as a main component. The Friend of a Friend (FOAF) project uses Semantic Web metadata for social networking: The FOAF vocabulary defines classes and properties (binary relations) for describing profiles of persons and organizations, where a 'knows' property establishes the social network. While current FOAF profiles group ontology-defined class and property facts around persons, this talk proposes to also allow person-centric rules. In particular, a Horn rule can derive a new property (e.g., how to contact John) from a conjunction of given properties (e.g., the relationship between John and the caller, their urgency, John's location, and the time of day). We propose rules also for creating new connections between persons. Expert finding is considered as a special case of social networking where the ontology and rules define areas and degrees of expertise as well as the availability of an expert for a user. When an expert is not available, their 'knows' property can be used for (possibly recursive) referrals to other experts. Our FindXpRT system implements these methods in a RuleML-extended FOAF. We propose a benchmark suite for expertise exchange more generally, testing expert-finding systems against expert profiles. This is exemplified with our implemented system, tested against expertise and co-expertise areas in computer science and music, respectively.

The FindXpRT system -- its theory and implementation -- is described in detail in the documents below. Meanwhile, improvements of the underlying OO jDREW engine led to a 160-fold speed-up of the numbers published in the benchmarks there.