Querying communities of interest in peer database networks

Abstract

Peer-to-Peer (P2P) networks are used primarily for file sharing. P2P user communities are continuing to grow rapidly, but there is no specific query mechanism for communities of interest in peer database networks. Peer databases are linked to each other through acquaintances. These are individual, independently developed databases that contain local data. Many researchers have tackled the problem of query processing in P2P networks. Moreover, researchers have started to investigate community-based querying in peer database networks. The focus of this thesis is to study a query translation mechanism, to develop an algorithm for querying communities of interest in peer database networks, and to implement a prototype of this algorithm. P2P communities are formed using common claimed items. First, we investigate an existing community formation and discovery algorithm for a text file, as well as an existing query translation mechanism for peer databases. After that, we develop a community-based search algorithm for peer databases. The developed algorithm combines the aforementioned community formation and discovery with the query translation mechanisms. A prototype has been developed and experimental results are shown. An approach for discovering communities 'on the fly' is introduced and a method for optimizing the community discovery technique is shown.