ENHANCING SEARCH PERFORMANCE IN PEER-TO-PEER NETWORKS
Abstract (Summary)In the past years, peer-to-peer (P2P) computing has attracted tremendous attention from both user and research communities. A large body of research work has been focusing on P2P search. There are two important factors driving research interests into P2P search. First, the trend of information proliferation eagerly calls for a scalable information infrastructure capable of indexing and searching rich content such as HTML, music and image files. A recent study has shown that 93% of information produced worldwide is in digital form. The volume of data added each year is over one terabytes and is expected to grow exponentially. Such huge amount of information poses many challenges for existing search systems. Second, compared to centralized search systems, P2P search systems are particularly attractive and promising due to their scalability, availability, low cost, easy of deployment, and data freshness. In this thesis, we combine techniques from information retrieval (IR) and P2P computing to tackle the issue of how to improve P2P search performance in terms of search efficiency and quality of search results. We have designed and experimentally evaluated two P2P search frameworks on distributed hash tables (DHTs) and unstructured P2P networks such as Gnutella. The results have shown that our proposed solutions can make search very efficient and improve quality of search results.
School:University of Cincinnati
School Location:USA - Ohio
Source Type:Master's Thesis
Date of Publication:01/01/2005