Enabling Semantic Search in Structured P2p Networks via Distributed Databases and Web Services

First generation of overlay P2P networks had scalability problems although offering useful keyword-base searching functionality. The new generation of structured networks, based on Distributed Hash Tables (DHT), solved the scalability problem but at the same time eliminated the possibility of performing searches by proximity. This can be considered as an effective limitation as users of P2P networks will most often not have the complete information about the data they are looking for. Accordingly they tend to submit general queries to the network, hoping to receive back a list of results that match their needs. In this paper we describe a solution that is able to profit from the best of the two approaches. It relies on a two-layer architecture, offering on the top layer a set of Web Services to upload, search and retrieve content and on the bottom layer a fully structured P2P network (DHT-based) for distribution and downloading efficiency. Central to our solution, is the use of distributed databases, implemented at the top layer, holding descriptive metadata of the uploaded digital objects. The developed system also makes use of the JXTA project at the top layer and the Bit Torrent suite of protocols at the bottom layer.