Article Preview
TopIntroduction
The Semantic Web idea (Berners-Lee, Hendler, & Lassila, 2001) in unstructured adaptive P2P networks is an approach to represent, manage, and retrieve distributed knowledge in an efficient manner. To pursue the Semantic Web vision, metadata is an essential add-on for information resources. This rich representation of data aims to improve knowledge discovery and data management. Traditionally, adding metadata to resources is a manual and expensive process and probably the main cause of the slow growth and difficult implementation of the Semantic Web.
Various efforts have been made from the P2P research community to create P2P systems that can discover resources efficiently and accurately. Standard P2P searching technologies like Gnutella 2 (Stokes, 2002) produce irrelevant search results with low precision rate. Kazaa (Liang, Kumar, & Ross, 2004) is a very popular super-peer based application, which appears to offer better scaling; however it inherits the limitations that come with flooding algorithms. Structured P2P architectures on the other hand, like Chord (Stoica, Morris, Karger, Kaashoek, & Balakrishnan, 2001), CAN (Ratnasamy, Francis, Handley, Karp, & Schenker, 2001) and Pastry (Rowstron & Druschel, 2001), guarantee retrieval of existing network resources with the use of Distributed Hash Tables (DHTs). The maintenance of a distributed index however comes with high cost, as additional traffic is generated for maintaining the routing information in a highly dynamic environment.
Trying to overcome the problems mentioned above, there are a number of P2P research systems (Cudré-Mauroux, Agarwal, & Aberer, 2007; Ehrig et al., 2003; Nakauchi, Morikawa, & Aoyama, 2004; Nejdl et al., 2002; Tatarinov & Halevy, 2004) that try to encompass the semantic technology notion, to represent meaning and knowledge, as well as to use reasoning for retrieving the knowledge. Knowledge needs to be encoded in a structured form to become widely accessible. An ontology structure (Jepsen, 2009) is an important part of the semantic web as it represents knowledge at the level of concepts; ontology provides a shared understanding through conceptualisation. In P2P systems, a domain specific ontology can be used for classifying network resources to ontology related concepts. Related research methods, are trying to exploit the benefits provided by semantics through an ontology structure in order to enhance data management and to improve resources discovery. As a result of that, the query can also be categorised to a specific concept and be routed to the peer that supports the specific concept.
An ontology-based P2P topology for service discovery has been described by (Schlosser, Sintek, Decker, & Nejdl, 2003), where the network is organised in a HyperCup topology. Information or services that peers provide are categorised to general concepts; these concepts and their relationships form the network ontology. Peers with similar interests are organised in a concept cluster. Each cluster is assigned a combination of concepts which best describe the peers that belong to the cluster. This network organisation aims to efficiently route the query to peers that can satisfy it. This technique uses broadcasting for query forwarding; when the query reaches the most appropriate cluster is then broadcasted to all peers. The network uses globally known ontologies for clustering its nodes. This approach however is an example of structured P2P networks. Yongxiang Dou and Xiaoxian Bei (2008) also presented a semantic information retrieval system based on a hybrid ontology integration approach. In hybrid approaches, each network source employs its own local ontology but all ontologies are built upon a global shared vocabulary. The authors are focusing on the problems involved with information representation and integration in P2P semantic retrieval networks. This research and the system proposed by the authors are in preliminary stages and no experiments have proven the validity of the approach yet.