A coverage-based approach for filtering and prioritizing regions in a semantic cache

摘要

Ensuring quick responses as well as a high degree of reliability remains one of the biggest challenges in mediation systems. Generally, in order to reduce their response time and to be able to provide answers in case the sources are inaccessible, mediators can resort to semantic caching. However, this may lead to additional costs, especially if there is a large number of semantic regions in the cache. In this work, we propose a coverage based approach in order to decide whether it is optimal to use the cache if a new user query is submitted and to identify the semantic regions that can contribute in providing answer if it is. Our approach is based essentially on calculating attribute and predicate coverage rates of the new user query in each semantic region. To validate our proposition, we have performed an experimental evaluation using a prototype that generates random semantic regions.