You are on page 1of 6

2009 Fourth International Conference on Internet and Web Applications and Services

A Generic P2P Collaborative Strategy for Discovering and Composing Semantic Web services

Mohamed Gharzouli
LIRE Laboratory, Department of Computer Sciences, Mentouri University of Constantine Constantine, 25000, Algeria gharzouli@gmail.com
AbstractThe Web services paradigm is considered as a revolution for the web in which a network of heterogeneous software components interoperates and exchanges a dynamic information. This computing technology can be used potentially to compose new distributed collaborative applications and make easy the development of more complex Web process. In this context, Web services based on Peer-toPeer computing require special attention from collaboration and interoperability in a distributed computing environment. The P2P systems provide a decentralized model, which offers a better scalability, reliability, and performance for Web services composition. We present an unstructured P2P strategy for composing semantic Web services. We propose a collaborative workspace between different participants to achieve a specific goal through the composition of Web service. In this strategy, we define two principal concepts: the epidemic discovery algorithms and the composition table. The first one describes a distributed solution to discover semantic Web services in unstructured Peer-to-Peer networks. The second one is a distributed method to publish and preserve the description and the composition way of the peer-to-peer composed Web services. Keywords: Semantic Web services; P2P computing; Web services composition; Web services discovery;

Mahmoud Boufaida
LIRE Laboratory, Department of Computer Sciences, Mentouri University of Constantine Constantine, 25000, Algeria mboufaida@umc.edu.dz propose different solutions based on centralized discovery methods (such as UDDI), where Web services are described by service interface functions and they publish their capabilities and functionalities with a registry. Moreover, the centralized control of published services suffers from problems such as high operational and maintenance cost. In addition, these solutions do not ensure the required scalability to support a flexible and dynamic environment [1][2]. Recently, several frameworks, tools, algorithms and applications have been developed to increase the use of semantic Web services. An important improvement has been made for making Web services based P2P computing [10][13][14]. P2P systems provide a scalable alternative to centralized systems by distributing the Web services among all peers. The P2P based approaches offer a decentralized and self-organizing context, where Web services interact with each other dynamically. Many of recent articles related to automated composition of semantic Web services in the P2P networks have been published [18]. Although, a considerable number of them discuss the dynamic discovery and the distributed composition of semantic Web services in the structured Peer-to-Peer networks like Chord [12], Pastry [16], and CAN [17]. In this paper, we describe an unstructured P2P based strategy of Web services discovery and composition. The principal idea of this paper is to present research epidemic algorithms based on logic planning. These algorithms enable us to find basic services distributed among all the peers to compose a personalized service. Moreover, in this solution we use a table of composition to preserve the trace of the composition, for a possible future re-uses. This table is considered as a cache memory, which preserves the P2P composition ways. Its objective is to accelerate the research of the services already composed. Furthermore, peers which have similar caches (the same interests) can be gathered in communities to achieve common goals. The rest of this paper is structured as follows. In Section 2, we briefly introduce the distributed composition of

I.

INTRODUCTION

The Web service technologies continue to happen in many different domains from business environments to scientific applications. This technology promises to provide a dynamic integration and an interaction of heterogeneous systems, and thereby, to facilitate a fast and efficient cooperation among the entities in cooperative environments [4][5][6]. To achieve these objectives, Web services should act autonomously with a minimal human intervention as possible [6]; they should be able to discover other services, which have particular capabilities and realize precise tasks. The Web service discovery is a critical problem since the beginning of the service oriented paradigm. The first generation of works done on Web services architectures
978-0-7695-3613-2/09 $25.00 2009 IEEE DOI 10.1109/ICIW.2009.72 449

Authorized licensed use limited to: B S ABDUR RAHMAN CRESCENT ENGINEERING COLLEGE. Downloaded on July 30,2010 at 06:41:54 UTC from IEEE Xplore. Restrictions apply.

Semantic Web Services (SWS) in the Peer-to-Peer networks. Section 3 describes our solution for composition of WS and presents how we use epidemic algorithms to discover distributed WS in the unstructured P2P networks. Section 4 reviews related work and Section 5 concludes the paper and discusses some further research works. II. OVERVIEW AND MOTIVATION OF THE WORK P2P computing is the next evolutionary step in computing and a new generation of the networks. This new direction in distributed computing focuses on networking and resource sharing with better reliability and scalability [7][8]. The increasing popularity of P2P systems (such as Overnet, Kazaa and Gnutella) for file sharing indicates general interest in resource sharing [3], especially, the semantic Web service. This paper is interested to decentralized P2P networks. In these systems, there are two methods to send queries: Distributed Hash table (DHT) and flooding-based algorithms. DHT-based networks, known as structured P2P systems are proposed to solve a number of problems appeared in the first generation of the P2P architectures (hybrid and pure P2P systems). Structured P2P systems have several advantages like: scalability, availability, and management of the fault tolerance. However, DHT-based protocols are difficult to maintain because of their static topologies (for example, a ring for Chord). Moreover, the DHT cannot allow complex research. On the other hand, unstructured P2P networks (such as Gnutella) require no centralized directories and no precise control over network topology or data placement. Though, the ooding-based query algorithm used in these systems does not scale; a query generates a large amount of traffic hence large systems become quickly overwhelmed by the query-induced load [11]. The decision to employ the unstructured P2P networks in this work is justified by several reasons. Generally, in the context of semantic Web services discovery and composition, the requester (service customer) searches a capacity, a goal or a property of a resource (service). So, when it sends the request, the invoker requires no placement or identifier of the resource. This characteristic does not exist in the structured P2P systems, to find a resource in these networks, it is necessary that the requester must be known beforehand the identifier of the resource. So, the unstructured P2P systems are more adapted to discover distributed Web services. Furthermore, the unstructured P2P systems are used by very large communities of net surfer. In addition, our approach explores various solutions to unstructured P2P discovery algorithms. We propose a distributed replication strategy that resolves the problem of ooding-based methods, while reducing the network traffic, and accelerates the discovery of semantic web services. To achieve this objective, we propose the use of a replication table (the composition table) that preserves the composition way of precedent composed Web services. This table has

many advantages that will be exposed in the following paragraph. III. AN UNSTRUCTURED P2P BASED STRATEGY TO DISCOVER AND COMPOSE SWS

The most important objective of this paper is to propose a decentralized scenario that supports the automatic composition of the semantic Web services distributed among an unstructured P2P network. Starting from a pure unstructured network, we need to compose a Web service which answers a particular request. This service can be composed from a set of Web services distributed through several peers of the network. To discover the appropriate Web services, we implement two epidemic algorithms based on Input/output matching and oriented by the achieved goal. Furthermore, all participant peers store in his composition table all traces of preceding composed Web services. This historic can be used to give a rapid response from future similar requests. In the following paragraphs we expose the composition table and the epidemic discovery algorithms. A. The composition table The main idea is to create a distributed table, namely table of composition (Table 1). Each peer creates this table to store all compositions in which it has already participated to realize them. It is necessary to give the various definitions of the different attributes of the composition table. These different concepts are well presented in the following example (Figure 1, Figure 2): Initiator Peer: the peer that begins the composition. compoID: the composition identifier (each composition is defined in the network by its initiator peer and its identifier). Init-input: initial input of the composite Web service. Init-output: initial output of the composite Web service. Goal: the goal of the composite Web service Executed services: Web services executed locally to compose the achieved Web service. Reserve services: local web services which can be executed in the same composition. Precedent peers: peer, which executes the precedent Web services for the composition. Next peers: peer, which execute the next Web services for the composition. Reserve peers: peers, which can replace the next peer. State of the composite Web services: the composite Web service is active, if all the participants peers are joining the network. We have added these five last attributes, to repair the composition ways, in case where a participant peer has left the network. This solution gives a possibility to manage the scalability and the fault tolerance of the participant peers (In this paper, this aspect is not tackled). The Table 1 shows the content of the composition table of the peer P5 (Figure 1, Figure 2). The framed line of this

450

Authorized licensed use limited to: B S ABDUR RAHMAN CRESCENT ENGINEERING COLLEGE. Downloaded on July 30,2010 at 06:41:54 UTC from IEEE Xplore. Restrictions apply.

table, describe a composite web service which computes the price of books in Algerian Dinar (AD). This Web service is composed in the Network from four services (Figure2), each one is deployed by a peer (P2, P4, P5 and P7 respectively). In Figure 1, the invoker Peer (P1) receives a request (a Web service that reads the title and the author of a book and displays its price in Algerian Dinar). If the invoker peer (P1) cannot answer the request directly through the execution of a local web service or a local composition, it searches in his composition table if there is a P2P-composition, which is already realized in the network, and answers this request. Else, P1 sends a message to different Peers to realize a new P2P-composition. After the reception of the favorable answers, the invoker peer (P1) filters the list of the peers, creates a peers-list and it sends the parameters of the composition to the peer, which is at the head of the list (initiator Peer: P2 in the Figure 1 and Figure 2).
Request P3 P1

P6

P4
P2

P5

P7
P1: invoker peer, P2: initiator peer P4, P5, P7: participants peers P6: reserve peer (for P5) : The composition Way : ask : The composition table : Local Web services Figure 1. Example of an unstructured P2P composition

In this example (Figure 2), the initiator Peer (P2) has a Web service which is the same input from the requested service (book title & author). But the output of the P2 service is the ISBN of the book. So, P2 sends a request to search a service which reads the ISBN of a book and displays his price in AD. So, each peer propagates the request until the initial output has been obtained (Web service deployed by P7). This discovery method is implemented in epidemic algorithms, clearly exposed in the section B. Instead of using a centralized repository of goals like in [1], the table of composition offers a distributed solution which has several advantages: Each Peer stores a historic of compositions, already realized in the network. In addition, each peer can exploit the experiment of other peers. The Peer which participates in several compositions has a very rich table, which gives it more probability to discover composite Web services for future requests. This property encourages the peers to collaborate with other peers, in order to achieve common goals (this property minimizes the number of the egoistic peers in the network). The composition table permits to create a collaborative workspace. Peers have similar tables that can be grouped in collaborate communities. Automatically, the peers which have the similar tables, they have the similar interests. Epidemic discovery algorithms based on input/output matching Initially, its important to give the various definitions of the basic concepts used in the following algorithms. Definition 1: Web service is 3-uplet defined as: WS= (WSinput, WS-output, WS-goal). Definition 2: A Goal is the conceptualization of a domain of services whose ultimate aims are identical or similar [1]. A goal is specified in the semantic description of the web service. Definition 3: A composition is a process constituted from a whole of basic services. In our context, there are two types of compositions: local composition and P2P-composition. Definition 4: A Local Composite Web Service (LocCWS) is built from a whole of Web services belonging to the same peer. Definition 5: A P2P-composition is a composition where the P2P composite Web service (P2P-CWS) is build from a whole of Web services belonging to different peers of the network. A P2P-composition (P2P-Comp) is tuplet (Initiator-Peer, CompID, Init-Input, Init-Output, Goal). Definition 6: Participant Peers: all the Peers participate (collaborate) in a P2P-composition (the initiator peer is a participant peer). B.

P2
Input Book title + Author

P4
Output ISBN Input ISBN

WS

WS

Output Price in Dollar

P7
Output Price in AD

P5
Input Price in Euro Output Price in Euro

WS

WS

Input Price in Dollar

Figure 2. P2P composed Web service, computing the price of books in AD.

451

Authorized licensed use limited to: B S ABDUR RAHMAN CRESCENT ENGINEERING COLLEGE. Downloaded on July 30,2010 at 06:41:54 UTC from IEEE Xplore. Restrictions apply.

1) Main Algorithm: Each peer in the network executes the following main algorithm when receiving a request:
Begin 1. Evaluate the request (define the init-Input, the initOutput and the goal). 2. Search a local Web service where [(WS-input=InitInput) and (WS-output=Iinit-Output) and (WSgoal=Goal)] 3. If (there is a local WS) then execute WS; send the response; go to END; 4. If (there is not a local WS) then Compose a local Web service where [(LocCWSinput=Init-Input) and (LocCWS-Output=Init-Output) and (LocCWS-goal=Goal)]; 5. If (there is a LocCWS) then execute LocCWS; send the response; go to END; 6. If (there is not a LocCWS) then Search-in-Composition-table (Init-Input, Init-Output, Goal); 7. If (there is not P2PCWS in the composition table) then Launch-a-New-P2P-composition( ); 8. End.

composition process. To start a P2P discovery based on the before chaining algorithm, we use the List-WS-Init-Input.
Begin 1. Create List-WS-init-Input (the list of Web services with the same input). 2. While (List-WS-Init-Input not empty) do a) Select the Heading of List-WS-Init-Input; b) Init-Input :=Output_of_Heading-List-WS-Init-Input; calculate the TTL; c) Send message search (initiator- Peer, Iinit-Input, InitOutput, goal, TTL); d) If (time TTL and response success research) then Stop research; send a success response; go to End; Else If (time TTL and response not success research) then Select a reserve service from the list-WS-Init-Input; End While; END.

The procedure search-in-composition-table is defined as follow:


Search-in-Composition-table () Begin Select [Initiator-Peer, CompID] from the composition table where [(P2PCWS-Input=Init-input) and (P2PCWSoutput=Init-Output) and (P2PCWS-goal=goal) and (state-ofP2PCWS=Active)]; /*create a selection-list*/ 1. If (the selection-list empty) then While (selection-list empty) and (P2P-composition is not possible) do a) Select an Initiator-Peer from the selection-list; b) Send a message: search (CompID, Init-Input, init-Output, Goal) from the initiator peer; c) If (success P2P-comp) then send the response; go to END; /* the end of the main program*/ End While; END.

3) Back chaining algorith:. In this discovery method, each peer searches the Output of the service which can grow the composition process. This algorithm uses the List-WSInit-Output.
Begin 1. Create List-WS-init-Output (the list of Web services with the same Output). 2. While (List-WS-Init-Output not empty) do a) Select the Heading of List-WS-Init-Output; b) Init-Output :=Input_of_Heading-List-WS-Init-Output; calculate the TTL; c) Send message search (initiator- Peer, Iinit-Input, InitOutput, goal, TTL); d) If (time TTL and response success research) then Stop research; send a success response; go to End; Else If (time TTL and response not success research) then Select a reserve service from the list-WS-Init-Output; End While; END.

To start a new P2P composition (Step 7 of the main algorithm), we propose two epidemic discovery algorithms based on logic planning methods. We present in what follows two types of algorithms: before chaining and back chaining. When it receives a request for P2P-composition, each peer executes initially the main algorithm (step 1 to step 6). If there is not a possibility to answer the request locally, it prepares initially two lists: the list of the services which have an input equal to the init-input (List-WS-Init-Input) and the list of the services which have an output equal to the init-output (List-WS-Init-Output). These local services can be basic or partially composed locally. In addition, each peer sends to the next peer a TTL (Time-To-Live) which is computed starting from the TTL of the precedent participant peer. 2) Before chaining algorithm: In this algorithm, each peer searches an input of a service, which can progress the
452

The following figure describes the before chaining and Back chaining discovery (the peer P1 is the initiator peer):
P1 Init Input
WS WS

P2

P3

Init Output

Init Output

Init Input

WS

: Before chaining discovery : Back chaining discovery Figure 3. Before chaining and back chaining Algorithms.

Authorized licensed use limited to: B S ABDUR RAHMAN CRESCENT ENGINEERING COLLEGE. Downloaded on July 30,2010 at 06:41:54 UTC from IEEE Xplore. Restrictions apply.

In addition, it is important to note if the composition is successful starting from a back_chainnig discovery, this composition will be saved in the composition table as follows: the initiator peer is the last participant peer and each participant peer permutes its precedent and its next went. Finally, its important to mention that in our context, each Web service is described with a semantic description (in our work we use OWL-S) and domain ontology which represents the metadata of the web service domain. For example, the input and the output are described in the service profiles of the OWL-S description. Now, we are developing a framework, which contains a WSDL to OWLS generator. IV. RELATED WORK In general, the research works suggested in the domain of the semantic Web services composition are classified in the following principal categories: Importation of the Web semantic technologies to the Web services discovery and composition [14] [9]. Use of the Web semantic (especially, the ontology) to search and classify the Peers contents in P2P networks [19][22]. Convergence between the systems Multi agents (MAS) and the Peer-to-Peer networks for the composition of the semantic Web services [3]. Use of the formal methods to express the semantic of the specification languages of composite Web services [21]. Evaluation of the performance and the quality of the composite Web services (QoS) [15][20]. Recently, many of papers have been published to propose solutions of the peer-to-peer composition of semantic Web services. Several research works improve the decentralized discovering methods of Web services in the Peer-to-Peer networks. Various of centralized and peer-topeer Web service discovery methods have been proposed in the context of the Web services composition and Web services based business process management. Among these, [1][4][5] have similar concepts to those which are used in our method. F. Mandreoli et al [1] present the architecture of FLO2WER, which is a framework that supports large scale interoperation of semantic Web services in a dynamic and heterogeneous P2P context. They have adopted a hybrid approach that exploits the advantages of centralized registries for service discovery and composition, as well as the dynamism and the scalability of non-structured P2P networks. The main idea of FLO2WER framework is that, while decentralizing the knowledge of what specific services are available in the system, they keep centralized knowledge of what objectives may be satisfied within the network, namely Goals. Each Goal specifies therefore a sub

network of specific services, and it is stored in an appropriate repository, called Goal Repository. However, it is not described and detailed how the goals have been discovered. Moreover, the use of a central repository of goals is similar to central discovery methods based on Web services functionalities. In addition, a central repository creates single point of failure and a centralized control of published services. Then, this solution suffers from problems such as high operational and maintenance cost. In contrast, our discovery method is a pure decentralized solution with any central repository. The composite Web services already realized in the network are published with a distributed description among all participant peers. [2][4][5] define a composite web service as finite automata. J. Hu et al [2] and F. Emekci et al [4] propose a structured peer-to-peer framework for Web service discovery in which Web services are located based on both service functionality and process behavior. They represent the process behavior of the Web services with finite automata and use these automata for publishing and querying the Web services within the system. This framework is scalable due to the underlying peer-to-peer architecture because the Web services can join and leave the system dynamically. In our work, we represent the process behavior of the Web services by finite automata (a finite sequence of basic Web services belong to several Peers). In addition, in our context we proposed an unstructured Peerto-Peer solution where the participant peers and the Web services (which compose the resulting service) are not beforehand known. In addition, we publish the P2P composed Web services with a distributed table (composition table). If a Peer wants to re-use a Web service which has been already composed, it transfers the request from the initiator peer of this composite Web service. V. CONCLUSION AND FUTURE WORKS

In this paper, we proposed an unstructured P2P strategy for semantic Web service discovery in which Web services are distributed among all peers of the network. We provided the different algorithms used in this strategy to discover the adequate Web services during the composition of a new Web service, which answers a received request. At this purpose, we adopted an epidemic research method that exploits the advantages of the decentralized architecture of non-structured P2P networks. Moreover, the use of composition table has several advantages. Firstly, each peer has its private table that provides a purely distributed method of discovery, composition and publication. Secondly, the distribution of this table creates a collaborative workspace where each peer can exploit the experiment of the other peers. Furthermore, this table permits to preserve the trace of the various successes compositions for a possible future re-use. Now, we are developing a JXTA-based distributed framework, which implements this solution. Furthermore, as

453

Authorized licensed use limited to: B S ABDUR RAHMAN CRESCENT ENGINEERING COLLEGE. Downloaded on July 30,2010 at 06:41:54 UTC from IEEE Xplore. Restrictions apply.

a future work, we plan to optimize the proposed discovering algorithms through the evaluation of the Quality of the discovered Services (QoS) and that of services, which will be composed. REFERENCES
F. Mandreoli, A. M. Perdichizzi, and W. Penzo, A P2P-based Architecture for Semantic Web Service Automatic Composition, IEEE computer Society DOI 10.1109/DEXA2007, Regensburg Germany, 2007, pp. 429-433. [2] J. Hu, C. Guo, H. Wang, and P. Zou, Web Services Peer-to-Peer Discovery Service for Automated Web Service Composition, Springer-Verlag Berlin Heidelberg (LNCS 3619), ICCNMC 2005, Zhangjiajie China, 2005, pp. 509-518. [3] P. Kungas and M. Matskin, Semantic Web Service Composition Through a P2P-Based Multi-agent Environment, Springer-Verlag Berlin Heidelberg (LNAI 4118), AP2PC 2005, Utrecht The Netherlands, 2006, pp. 106-119. [4] F. Emekci, O.D. Sahin, D. Agrawal, and A. El Abbadi, A Peer-toPeer Framework for Web Service Discovery with Ranking, In the Proc of the IEEE International Conference on Web Services (ICWS04), California USA, 2004, pp. 192-199. [5] O. D. Sahin, C. E. Gerede, D. Agrawal, A. El Abbadi, O. Ibarra, and J. Su, SPiDeR: P2P-Based Web Service Discovery, In the Proc of Third International Conference on Service-Oriented Computing (ICSOC), Amsterdam, The Netherlands, 2005, pp. 157-169. [6] T. ESSAFI, N. DORTA and D. SERET, A Scalable Peer-to-Peer Approach To Service Discovery Using Ontology, In the Proc of 9th World Multiconference on Systemics, Cybernetics and Informatics. Orlando, 2005. [7] R. Schollmeier, A Definition of Peer-to-Peer Networking for the Classification of Peer-to-Peer Architectures and Applications, In the Proc of the First International Conference on Peer-to- Peer Computing, Linkoping, Sweden, 2001, pp. 101-102. [8] K. Verma, K. Sivashanmugam, A. Sheth, A. Patil, S. Oundhakar, and J. Miller, METEORS WSDI: A Scalable P2P Infrastructure of Registries for Semantic Publication and Discovery of WS, Journal of Information Technology and Management, Special Issue on Universal Global Integration, Vol. 6, No. 1, 2005, pp. 17-39. [9] D. Berardi, D. Calvanese, G. De Giacomo, R. Hull, and M. Mecella, Automatic Composition of Transition-based Semantic Web Services with Messaging, In Proc. of VLDB, Trondheim, Norway, 2005, pp. 613-624. [10] M. Barhamgi, P-A Champin, D. Benslimane, and A.M. Ouksel, Composing Data-Providing Web Services in P2P-Based Collaboration Environments, Springer-Verlag Berlin Heidelberg, [1]

Proc. of CAiSE 2007. LNCS 4495, Trondheim, Norway, 2007, pp. 531-545. [11] Q. Lv, P. Cao, E. Cohen, K. Li, S. Shenker Search and Replication in Unstructured Peer-to-Peer Networks, Proc of the 16th

international conference on Supercomputing, New York, USA, 2002, pp. 84-95.

[12] I. Stoica, R. Morris, D. Karger, M. F. Kaashoek, and H. Balakrishnan, Chord: A scalable peer-to-peer lookup service for internet applications, In Proceedings of the ACM SIGCOM, California, USA, 2001, pp.149-160. [13] I. Toma, B. Sapkota, J. Scicluna, J. M. Gomez, D. Roman, and D. Fensel, A P2P Discovery mechanism for Web Service Execution Environment, In the Proc of the 2nd WSMO Implementation Workshop (WIW05). Innsbruck, Austria, 2005. [14] H. Jin, H. Wu, Y. Li, and H. Chen, An Approach for Service Discovery based on Semantic Peer-to-Peer, In the Proc of ASIAN, Kunming, China, 2005, pp. 259-260. [15] L-H. Vu, M. Hauswirth and K. Aberer, Towards P2P-based Semantic Web Service Discovery with QoS Support, In the Proc of Business Process Management Workshops, Nancy France, 2005, pp.18-31. [16] A. Rowstron and P. Druschel, Pastry: Scalable, distributed object location and routing for large-scale peer-to-peer systems, In the Proc of the 18th IFIP/ACM International Conference on Distributed Systems Platforms (Middleware 2001), Heidelberg, Germany, 2001. [17] S. Ratnasamy, P. Francis, M. Handley, R. Karp, and S. Shenker, A scalable content addressable network, In Proc of the 2001 conference on Applications, technologies, architectures, and protocols for computer communications, ACM SIGCOMM, California, USA, 2001, pp. 161-172. [18] D.G. Galatopoullos, D.N. Kalofonos, E.S. Manolakos A P2P SOA Enabling Group Collaboration through Service Composition In proc of ACM ICPS08, Sorrento, Italy, 2008, pp. 111-120. [19] P. Haase, R. Siebes, F. v.Harmelen, Peer Selection in Peer-to-Peer Networks with Semantic Topologies, In the Proc of ICSNW, Paris, France, 2004, pp. 108-125. [20] J. Liu, N. Gu, Y. Zong, Z. Ding, S. Zhang and Q. Zhang, Web Services Automatic Composition Based on QoS, IEEE Computer Society, In the Proc of the 2005 IEEE International Conference on eBusiness Engineering (ICEBE05), Beijing China, 2005, pp. 607-610. [21] X. Tang, C. Jiang, Z. Ding, Automatic Web Service Composition Based on Logical Inference of Horn Clauses in Petri Net Models, In Proc of IEEE International Conference on Web Services (ICWS), Utah, USA, 2007, pp. 1162-1163. [22] C. Wang, J. Lu, and G. Zhang, Generation and Matching of Ontology Data for the Semantic Web in a Peer-to-Peer Framework, Springer-Verlag Berlin Heidelberg, Proc of APWeb/WAIM. LNCS 4505, Huang Shan, china, 2007, pp. 136143.

TABLE I
THE CONTENTS OF THE COMPOSITION TABLE (EXAMPLE OF P5 TABLE, APPEARED IN FIGURE 1)

Initiator Peer P11 P2 . P23

CompI D 0001 0003 0010

Initinput . String String . .

Initoutput .. String ..

Goal Compute the price of a book in AD .

Executed services .. WS2 ... ..

Reserve services WS3 .. .

Precedent Peers . P4 .. ..

Next Peers P7 . .

Reserve Peers .. P6 . ..

State of Comp services Active . .

454

Authorized licensed use limited to: B S ABDUR RAHMAN CRESCENT ENGINEERING COLLEGE. Downloaded on July 30,2010 at 06:41:54 UTC from IEEE Xplore. Restrictions apply.