JMaPSS: Spreading Activation Search for the Semantic Web

The semantic Web augments search by providing meta-information to structure knowledge. Challenges associated with search technology, such as accessing a large knowledge base with limited processing capability, may be addressed by AI techniques that provide greater flexibility albeit with less precision. In this paper we present JMaPSS, which applies a parallel search algorithm known as marker-passing to improve search relevancy results. We describe an instantiation of JMaPSS implemented specifically for semantic Web search. Our investigations suggest that such techniques, using an expanded notion of recall emphasizing relevance, deserve additional exploration.