Incremental Reasoning on Streams and Rich Background Knowledge

Presented at: 7th Extended Semantic Web Conference (ESWC2010)

by Davide Francesco Barbieri, Daniele Braga, Stefano Ceri, Emanuele Della Valle, Michael Grossniklaus

Innovative context aware mobile applications require performing complex reasoning tasks that combine streaming information with rich background knowledge. For instance, from a stream of quantitative information about latitude and longitude such applications need to derive qualitative information about places like home, office, or gym. Then, rising up the level of abstraction, they need to reason about concrete problems, such as deciding how to reach such places, which means of transportation to choose, how to attend ongoing events nearby, or how to skip traffic jams. This article presents a technique for Stream Reasoning consisting in incremental maintenance of materializations of ontological entailments in presence of streaming information. In this article we elaborate on previous papers that extend to logic programming results from incremental maintenance of materialized views in deductive databases. Our contribution is a new technique that takes in explicit consideration the order in which streaming information arrives to the Stream Reasoner. By adding time validity information to each RDF statement, we show that is possible to compute a new complete and correct materialization by (a) simply dropping explicit statements and entailments that are no longer valid, and (b) evaluating a maintenance program that propagates changes in explicit RDF statements to the stored implicit entailments. In this paper, we also provide experimental evidence that our approach significantly reduces the time required to compute the new materialization, and that our approach opens up for several further optimizations.

Keywords: Context-aware, Incremental Maintenance, Materialization, Reasoning, Stream


Resource URI on the dog food server: http://data.semanticweb.org/conference/eswc/2010/paper/mobility/10


Explore this resource elsewhere: