Preserving the Original Query Semantics in Routing Processes

Abstract

In distributed data environments, peers (data sources) are connected with each other through a set of semantic correspondences in such a way that peers directly connected are called semantic neighbours. Queries are submitted considering partial information provided by a peer schema and may be answered by other neighbour peers. From the query submission peer, the original query is successively rewritten into queries over the peers, according to the correspondences between the original peer and the target ones. In this process, some of the original query terms may be lost while other ones may be added, leading to a semantic loss of the original query. In this work, we argue that it is essential to try preserving the original query semantics if we wish to hold what the users defined as important at query submission time. With this in mind, we propose an approach to preserve the original query semantics in query routing processes. Furthermore, we present a metric for assessing the relevance of neighbour peers according to an estimated query semantic value obtained at each query reformulation. In this paper, we present the developed approach and some experimental results we have accomplished.