A Middleware Architecture to build Scalable and Efficient Reasoning Infrastructures for the Semantic Web
A poster presentation written by Alissa Kaplunova, Atila Kaya and Ralf Möller. It was presented at the ESWC2007. It is about Semantic Middleware, Querying over Ontologies and Reasoning on the Semantic Web
A crucial requirement of the Semantic Web vision to come true is the efficiency of reasoning over ontologies distributed Web-wide. The typical Semantic Web scenario where software agents accomplish complex tasks requires a scalable inference infrastructure that provides for efficient reasoning on ontologies with respect to implicit knowledge. Nowadays, state-of-the-art reasoning engines are highly-optimized for standard reasoning tasks and can even deal with large but not very expressive ontologies e.g., by supporting incremental query answering. However, they are not well-equipped to offer the same quality of service in case multiple clients pose queries concurrently. In order to close this gap we propose a semantic middleware architecture to incorporate multiple reasoners. The primary goal of the middleware is to minimize the query answering time for each query by employing sophisticated dispatching algorithms. We develop different strategies to investigate the resulting problems of load balancing and caching in the presence of incrementally answered queries. Based on the design of the proposed middleware architecture and practical experiments with an implementation, we study the effects of concurrent query executions and iterative transmission of corresponding results for queries. Our case study shows that scalability in server-based Semantic Web applications can be achieved in the near future.
This data has been imported from the ESWC2007 RDF