iSeeker: Towards an Engine for Processing Aggregated Search on Linked Data

Aggregated search consists in retrieving, synthesizing and assembling fragments of information from a set of distributed and autonomous sources. Search engines, such as Google and Bing, have adopted aggregated search. The effectiveness and extensive potentiality of aggregated search have also attracted researchers from the Semantic Web community. In Semantic Web, aggregated search mainly targets aggregated SPARQL query processing. The goal of this approach is to apply the principle of aggregated search in querying distributed RDF graphs stored in a wide number of verticals (a.k.a sources or endpoints). The critical characteristics of an efficient aggregated SPARQL processing engine include completeness of meaningful and useful information and performance. The straightforward queries that are submitted by the users do not always cover everything what users had in their mind. The key reason is that the users are unaware of all the sources available on the Web. The query processing engine should take some responsibility in this occasion, to provide additional information which the users expect. In fact, this is one of the core principles of aggregated search. We propose a preliminary framework for aggregated SPARQL query processing engine. We describe an architecture with its components designed to increase the performance of the underlying search engine.