Web Services Composition based on Domain Ontology and Discrete Particle Swarm Optimization

This paper proposes an approach for web services composition based on domain ontology and discrete particle swarm optimization (DPSO) algorithm. This method builds an optimized graph for service composition based on domain ontology and its reasoning capability, and then a discrete particle swarm optimization algorithm based on the graph is proposed to accomplish service composition. The simulation results show that it can produce good results, especially when the amount of web services is large. The rapid development of web services shows that Internet software will be integrated following the service-oriented model (big granularity, loose coupling and dynamic banding) in the future. Single web service only provides the limited function, so web services composition is very important. Now most web services are composed by hand, so it is necessary to study web services composition based on semantic. Literature [1] adopted the DOSCM method to study web services dynamic composition based on domain ontology, but it needs sort all the web services following the connection degrees among web services before composing them. Obviously it is very difficult to do so in the case of too many web services. In this paper, we adopt the discrete particle swarm optimization algorithm and domain ontology to compose web services dynamically based on literature [1],we can see that it has obvious effect when the service amount is large. The remainder of this paper is organized as follows. Section 1 gives the description of web services composition based on domain ontology; Section 2 is the overview of the discrete particle swarm optimization (DPSO) algorithm; Section 3 details the proposed web services composition method which bases on DPSO and domain ontology (we called this method as DOPSO); Section 4 is the experimental result, and finally section 5 concludes this paper. Web Services Composition based on Domain Ontology and Discrete 1021 1 The problem of web services composition based on domain ontology The basic conceptions of domain ontology can refer to literature [1], on account of the paper’s length,this paper will not discuss them. The problem of web services composition based on domain ontology can be described as follows. Fig. 1. WSC-Graph The basic idea of DOPSO is that we can compute the connection degrees among web services based on conception similarity degrees and domain ontology to form an optimized web services composition graph (WSC-Graph), and then we can switch the web services composition problem to the problem that we find a path on the WSC-Graph which satisfies the users’ requirement best. As described in figure 1,supposing composed service W has m basic web services, that is 1 2 ( , ,..., ) m W w w w 1 For the web service requirement ( , ) k k k WSR I O , there are n1 web services which have the semantic connection relationship(that is to say, w1 has n1 candidate web services), and these n1 web service also have n2 subsequent web services, the rest may be deduced by analogy, and m W has s n candidate web services, that is 1 1,1 1,2 1, ... k n I I I I , 2 1,1 2,1 2,n O I I ,......, ,1 ,2 , ... S m m mn k O O O O . k I in figure 1 presents the input of ( , ) k k k WSR I O and k O is the output. The circles in figure 1 denote web services, the two web services which are connected by the directed edge have the signification of semantic connection, the broken lines in figure 1 show that the two web services connected by them have no signification of semantic connection, but for the simple aim, we still connect them and their semantic connection degrees are zero. Each path from k I to k O in WSC-Graph is a service composition result and the best one is the one which has the biggest web services composed satisfaction degree. The details of conceptions, such as the similarity degree between two domain ontology concepts, the semantic similarity degree between two conception sets and the formula about connection degree of two services, can refer to literature [1]. 2 The discrete particle swarm optimization Particle swarm optimization (PSO) is one of the evolutionary computational techniques. Suppose that the search space is D-dimensional, and the position of the ith particle of the swarm can be represented by a D-dimensional vector, ,1 ,2 , { , ,..., } i i i i D x x x x .The velocity (position change per generation) of the particle i x can be represented by another D-dimensional vector, ,1 ,2 , { , ,..., } i i i i D v v v v .The best position previously visited by the ith particle is denoted as , ,1 ,2 , { , ,..., } i best i i i D PC p p p and its corresponding fitness value is denoted as PLi,best.If the topology is defined such that all particles are assumed to be neighbors and g as 1,1 w