Using linkage information to approximate the distribution of relevant documents in DIR

Abstract

In this paper, a method for addressing the source selection problem in a distributed information retrieval environment is presented. The method proposes a solution to the source selection problem by making use of the outlink distribution extracted from a sampling collection in order to locate the most authoritative collections for a particular query. Experiments carried out with the algorithm show that its performance exceeds that of the uniform approach, with much more economical means. Specifically, the link-based method is much more efficient in terms of collection utilization than the uniform strategy, which can be applied under the same conditions. A key feature of the algorithm is that it can be combined and work in parallel with content-based source selection algorithms