Answering Queries by Semantic Caches

Abstract

There has been growing interest in semantic query caches to aid in query evaluation. Semantic caches are simply the results of previously asked queries, or selected relational information chosen by an evaluation strategy, that have been cached locally. For complex environments such as distributed, heterogeneous databases and data warehousing, the use of semantic caches promises to help optimize query evaluation, increase turnaround for users, and reduce network load and other resource usage. We present a general logical framework for semantic caches. We consider the use of all relational operations across the caches for answering queries, and we consider the various ways to answer, and to partially answer, a query by cache. We address when answers are in cache, when answers in cache can be recovered, and the notions of semantic overlaps, semantic independence, and semantic query remainder.

While there has been much work relevant to the use of semantic caches, no one has addressed in conjunction the issues pertinent to the effective use of semantic caches to evaluate queries. In some cases, this is due to overly simplified assumptions, and in other cases to the lack of a formal framework.We attempt to establish some of that framework here. Within that framework, we illustrate the issues involved in using semantic caches for query evaluation. We show various applications for semantic caches, and relate the work to relevant areas.