An Evaluation of a Cluster-based Testbed for Peer-to-Peer Information Retrieval
Saloua Zammali, Khedija Arour · 2011
Abstract—Peer-to-Peer (P2P) systems present an advanta-geous way to provide and share services [1]. Hence, P2P are the major technology of access upon various resources on Internet. Hence, P2P are the major technology of access upon various resources on Internet. A particularly intriguing class of distributed applications consists in Information Retrieval (IR) systems. The issue of Peer-to-Peer Information Retrieval (P2PIR) is being tackled by researchers attempting to provide valuable insights and to propose solutions to use it successfully. Nearly, all published studies have been evaluated by simulation means, using well known document collections (usually ac-quired from TREC). This practice leads to two problems: First, there is little justification in favor of the document distributions used by relevant studies and second, since different studies use different experimental testbeds, there is no common ground for comparing the solutions proposed. In this paper, we propose, CB a testbed for P2PIR based on P-Kmeans. CB, a cluster-based testbed, allows to distribute documents. This work marks the start of an effort to provide more realistic evaluation environments for P2PIR systems as well as to create a common ground to compare the current and future architectures. Keywords-Testbed; P2P systems; Information retrieval. I.