Content area
Full text
Asbtract - P2P systems allow users to share and access different resources over Internet. Actually, they are an excellent alternative to client/server ones, since they are fault tolerant and more scalable. Nonetheless, searching pertinent peers (i.e., peers which share pertinent resources) to user queries in large scale P2P systems is a major problem. In the literature, this task is commonly named query routing. In this respect, the author introduces in this paper a new query-based routing method for unstructured P2P systems. The main originality of our method is the exploitation of the strong connections between query terms and relevant peers, which provided pertinent results for those queries, to define a user profile model. Thus, the computation of the user profile comes down to extracting special types of triadic formal concepts. The user profile is used latter to efficiently route the forthcoming queries. Simulation results show that our method is more effective than the baseline method.
Keywords: P2P, User Profile, Learning, Query Routing
(ProQuest: ... denotes formulae omitted.)
1Introduction
P2P systems allow users to share and access different resources over Internet. Indeed, in P2P systems, peers play equal roles (i.e., client and server at the same time). Hence, each peer is responsible for indexing its shared documents and for searching pertinent documents to queries received from other peers (i.e., server role). Furthermore, it can also send queries to retrieve pertinent documents shared by other peers (i.e., client role). Actually, P2P systems are an excellent alternative to client/server systems, since they are fault tolerant, autonomic, scalable and cost effective. Nonetheless, locating relevant peers (i.e., peers that share pertinent resources) in large scale P2P systems remains a challenging issue. In the literature, this task is commonly named query routing [1, 2]. Existing query routing approaches in fully distributed P2P systems (i.e., unstructured P2P systems) rely on pure flooding or controlled flooding mechanisms. Within the former mechanism the requesting peer broadcasts the search query to all its neighbors. Whenever a neighboring peer receives the query it returns a response to the requesting peer if it shares pertinent documents. Thereafter, it continues to forward the query to all its neighbors and so on until a Time To Live (TTL) threshold is decremented to 0. Indeed, this method generates an...