This paper focuses on P2P based data management and semantic mediation. We propose an approach based on a P2P for semantic interoperability of information sources that aims to combine the advantages of semantic mediation and peer-to-peer systems. It is based on a pure P2P with super peer architecture consisting of two types of peers. The super peer contains a reference ontology, which provides a common ontology (CO) of the domain. The peer contains export schema (ES), which represent local data. The approach based on semantic agreement between CO and ES, which called half agreement (HA). The halfagreements utilize for discovery sources and exchange information among peers.
Effective information and services sharing in distributed such as P2P based environments raises many challenges, including discovery and localization of resources, exchange over heterogeneous sources, and query processing. Traditional approach does not scale well when applied in dynamic distributed environments and has many drawbacks related to the large numbers of sources.
Several applications of P2P networks can be distinguished, ranging from
content sharing applications (e.g. Napster, Gnutella) to distributed computing
applications (e.g. SETI@home, Avaki) and development support platforms (e.g.
JXTA). Generally, two main categories of P2P systems can be distinguished.
Unstructured P2P systems organize peers in networked spaces. Each peer
controls and maintains its shared data. User queries are based on (
Survey on schema matching [1, 2] explained the general approach of schema matching based on terminology, structural and semantic. Kolfoglou [3] delivered state of the art ontology mapping, which consider some component in framework, methods, tools, translators, mediators, techniques and theoretical.
We propose an approach based on a P2P for information interoperability that aims to combine the advantages of semantic agreement and peer-to-peer systems. Main our contribution is how to create and implement peer semantic agreement for discovery process. The main difference of our approach to Remindin [4] and expertise-based [5] is in calculation of related peers based on similarity of halfagreement (semantic agreement) by calculating BindingValue.
The paper is organized as follows: section 2 presents the peer agreement based semantic approach. Section 3 presents an example. Finally, section 4 concludes the paper. 2 2.1
A P2P system Q =< P, A > is a set P = {P1, ..., Pn} of peers and a set A of agreements. Two types of peers can be distinguished in the approach. First, Super peers (SP) are used to maintain common ontology of a community. Peers (PP) represent another type of peers that provide or search information.
Figure 1 depicts the general processes of information exchange at P2P as
follow: (
Our approach utilizes available approaches that based on:
– Label Matching, a label has a part value of semantic, which presented at
taxonomy model such as WordNet [6]. There are two steps at label matching
[7]. First, a language preprocessing step is used to transform the labels into
words prior to linguistic analysis. For example, this step can be used to
expand abbreviations and reduce article such as the, a. Next, the labels are
matched by determining relations between them. This can be done based on
WordNet relations. The WordNet [8] is a brad coverage lexical network of
English words. Wu-Palmer (WUP) and Jiang-Conrath (JNC) methods are
utilized for the WordNet and combined with threshold value.
– Internal structure, a ’language’ attribute [9, 10] is property label of the
language such as owl:cardinality, rdfs:label. The similarity value between two
entities is derived by the ratio between numbers of similar properties over
the maximal number of both entities.
– The external structure takes into account the position of a concept in a
hierarchy. The method refers to upward cotopic distance [11] which compares
the similarity of the set of superclasses.
An agreement unit defines one-to-one or one-to-many mappings between (CCiO,
j
{CES }) where CCiO is an ontology and CEiS is an export schema concept. An
agreement unit encapsulates three main components that are described by RDF/
OWL schemas: (
Result of half agreement between P Pn to SP can be utilized to make direct mapping between peers. The approach of mapping composition based on inverse mapping as follow: ΩES1→ES2 = Ω1 ∗ Ω2−1. Result of mapping composition is called full agreement unit. 3
This example illustrates the steps for discovery sources using the general strategy of the agreement unit approach. Consider the peers PP1 and PP2 as providers and the fragments of ontology and export schemas shown in figure 2. Furthermore, assume that the two peers characterize roads differently. One peer PP1 classifies road according to speed limit while peer PP2 characterizes road according to size of road. Now consider a peer which characterizes roads by type (primary, secondary and so on) and which queries both peers PP1 and PP2 for a list of secondary street in an area. After developing of half agreement and look at meta information at super peer, the peer as request send his half agreement to PP1 and PP2. PP1 and PP2 calculate BindingValue. BindingV alue = ²/π, where ² = number of similar concept between request and provider peer, π = number of concept at request peer. PP2 will be selected as the interest parties because BindingValue of PP2 higher than threshold value (BindingValue of PP1=0/1, BindingValue of PP2=1/1). Result of discover can be continued to develop mapping composition and query process. 4
XMLS, RDFS and OWL and other ontology developments offer facility to enrich semantic description at P2P environment. We proposed a semantic agreement approach based on concept similarity values that take into account the place of a concept in a hierarchy and its structure consisting of directly linked properties and concepts. We utilize available approach of semantic mapping to develop half agreement. Result of half agreement can be utilized for discovery resources, mapping of concept between peers and handling of query to exchange information.
In the future work, we will focus on finalizing the architecture and prototype system to enhance negotiations between provider peer and a peer