A Parallel Processing Architecture for Querying Distributed and Heterogeneous Data Sources

Collecting information from distributed data sources in the web of data is a topic of great interest since the number of data sources is eloquently increasing. The general objective of our work is to set up an aggregated search engine able to integrate an answer for a SPARQL query, and this, by collecting data from web accessible data sources alongside local databases via a single query interface. Runtime optimization is an essential task in such search engines, for this purpose, we present a general view on different processing steps in aggregated search engines to highlight the costliest step in terms of runtime, then, we propose an efficient parallel processing architecture to optimize runtime. The proposed parallel processing architecture allows collecting intermediate results from each data source independently, thus, this architecture supports answers integration from heterogeneous data sources, in this context, we present the ontology-based data access paradigm and its importance to query heterogeneous data sources in our aggregated search engine.