Approaches Of Semantic Web Service Composition

Published Date: 02 Nov 2017

Abstract— the web has been changed a lot after the introduction of a Service Oriented Architecture (SOA). The business processes are more efficient to perform a task by composing the number of web services. But, all these things are done statically. They are not flexible in nature. Semantic web comes here to rescue. Semantic Web Service (SWS) is the form of Semantic Web. Semantic Web Service (SWS) allows services to be discovered and execute them at runtime. The purpose of this paper is to compare SWS composition approaches.

Keywords— Semantic Web, Web Services, Service Oriented Architecture

Introduction

Service Oriented Architecture (SOA) has brought revolution, the way resources or knowledge is shared on to the web. It is far better than the traditional client Server Architecture Web Service is the best suited example of SOA. Most of the Web Applications today, are developed as a Web Service. For almost all things, Web Service is available. Even for the same purpose multiple implementation are available. And to deal with such situation with traditional techniques like WSDL, UDDI etc. is not enough. As the growing number of implementation, it is very difficult for manual discovery and sometime composition. To tackle this issue, Web services are proposed with semantic enhancement. Three major efforts (i.e. OWL-S, WSMO and WSDL-S) are going on addressed this enhancement. The SWS community has presented some approach or solution, so that automatic composition and integration of web services is possible. But the fact is all of them failed addressed dynamic composition issues. Major issues in composition and integration of semantic based web services are:

•With the large number of services, manual discovery and composition gives very poor performance and its non-flexible approach.

•Static composition is not able to handle services, which change on the fly.

•When composition of web service is hard coded then a single service within the composition may create failure due accessibility of network.

Traditional web services have Syntactical description (WSDL document) and index based (UDDI Registry) search mechanism. Web services and SOA need to be semantically enhanced to support semantic based web services machine understandable way.

SWS Effort

Semantic web offers various machine understandable semantic descriptions to facilitate automatic and efficient selection of service. Hence services on the web able to find each other by matching their requirements and needs. Semantic web Service will improve the discovery, composition and interaction among them with Services. It also helps to build up more complex workflows. As shown in figure the semantic web is attach to web service (or SOA) to come up with the new Semantic Web Service(SWS).Semantic Web is the approach to facilitate machine understandable description and to consume it. By combining this capability to Web service, Web services can now search automatically best suited and optimal implementation. As we no longer tied to a specific implementation. By doing this, we can also protect our System from crash, when a given service is not available. Semantic web uses Resource Description File (RDF) to describe the resource or we can say web service. This RDF files are nothing but the pointer to other RDF file. These files are being used because these files have data merging feature even if the underlying schema does not matched.

<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:si="http://www.example.com/rdf/">

<rdf:Description rdf:about="http://www.publisher.com">

<si:title>Book Name</si:title>

<si:author>Author Name</si:author>

</rdf:Description>

</rdf:RDF>

Fig. 1 Data Model for Sample RDF Code

The Web Ontology Language for Services (OWL-S) is an ontology built on the top of OWL to describe web Service. It’s a set of markup language constructs that can be used to define properties and capabilities of Web services in a way that computer understands. It main goal is to provide a semantic description of Web service to provide facilities like dynamic and automated discovery, invocation and composition of Web Service. OWL-S provides Web service semantics by ontologically annotating: (1) the input to a Web service (2) the output produced by a Web service, (3) pre-conditions, that is required to perform a service and (4) Final outcome that the service will produce after its execution. OWL-S is a stack of OWL ontologies (Profile, Process Model and Grounding ontologies). The Process Model ontology can be used to model the composition of SWSs by defining the control flow and data flow on the basis of matching semantics of sub processes. One more option for SWS is WSDL-S. WSDL-S approach extends tags of the existing Web Service Description Language (WSDL), instead of defining separate ontologies to provide service semantics. It not only supports input/output annotation, but WSDL-S extension also enables the description of precondition and effect of a Web service operation. Sample snippet of WSDL-S is shown below:

<wsdl:message name="CarRequest">

<wsdl:part name="inO" type="tnsl:TypesOfAvailableCars"

LSDISExt:onto-concept="LSDISOnt:Roadster"/>

</wsdl:message>

WSMO is another candidate to develop specifications for SWSs. It has three approaches to model Web services composition (i.e. state machine, structured and data flow models). State-based model is somehow related to WSFL in which each state defines control flow to control activities. Structured model is based on structured design methodology. It is used in workflow languages (e.g. BPEL). Data flow model is based on parallel programming languages. And it is based on concurrent control components of structured model.

Existing Composition Approaches

A Bottom-Up Approach

The work discussed in [3] presents a bottom-up approach by integrating the semantic Web technology into Web service technology. It considers BPEL as a composition technique of Web services. The main goal is to add semantic in BPEL that provide machine understandable or semantic description of required web services within process and extending workflow execution engine (BPWS4J) to realize these semantic description. With the help of these semantic description of service the bottom-up approach uses Semantic Discovery Services(SDS) to Dynamically discover a desired service on the basis of matching semantics and bind it within composition. If in case a single service does not meet a service requirements, SDS uses a recursive back-tracking algorithm to determine a sequence of service invocation or a service chain, which takes input generated by BPWS4J and returns the output which is expected to be generated by BPWS4J. The major limitation of this approach is that the performance or efficiency decreases as the number of service profile increases in service chain. The disadvantage of this approach is that it does not consider any condition for discovery and composition purposes.

METEOR-S Approach.

In the METEOR-S project[12], a tool which is developed for dynamic composition of Web services. The METEOR-S tool or METEOR-S process designer allows process designers to design processes on the basis of constraints which are required to form a process. Constraints can be on the basis of business and processes. The main idea behind this Web Service Composition Tool is to write required service specification as an abstract process in BPEL process and to discover service whose semantic description matches to defined abstract process. Once desired services are discovered, candidate services are selected on the basis of process constraints given. Here, BPEL is used for process modelling. The major drawback of the approach is that user has to select services which were discovered dynamically manually from the bundle of services returned.

Template Based Composition.

This is an AI approach. In [5], author has used workflow template to write abstract activities. These abstract activities can be used to describe required services. On the basis of these specifications required services can be discovered to have executable workflows. This approach focuses on importance of adding preferences in templates so that services can be ranked to find most suitable one from the bunch of discovered services. Author proposes the use of semantic Web technology (OWL) for writing template so that it allows reasoning for flexible and more consistent selection of required services. This approach focuses on extending existing OWL-S process ontology to include abstract processes. This approach proposes that process ontology should have abstract process that can be used to refer to the Profile ontology of an OWL-S service with other specifications that can be used to rank and find best matching service.

WSMO Composition Approach.

WSMO team has developed a tool [4] for dynamic composition tool of Web Services. The composition tool allows designers to select targets, mediators and control flow operators to define control flow between components. The composition process starts when composition goal is selected from the list of available goals defined in IRS-I11[9] server. Data flow between these goals can be defined by specifying the data source as input of goal and the data destination as an output of the goal. By using mediators, type mismatch between input and output of goal can be managed. Mediator map and perform transformation between goals. By defining XSL transformation, data mapping between messages of different types can be supported.

Automated Composition by Using SHOP2.

The work discussed in [6] describes how an Al planning system (SHOP2) can be used with the DAML-S (OWL-S) Web service description to automatically compose Web services. This approach gives partial support for composing services on the basis of their matching functional and non-functional semantics. [6] Does not support the creation of a composite process with all OWL-S supported control constructs (e.g. this approach does not support synchronization between processes components by implementing support for OWL-S Split-Join control construct).

SWORD

The method reported in [14] provides a set of tools for composition of a class of Web services. The SWORD implements use of rule-based expert system that determines possibility of automatic creation of composite service from existing services. In case of such possibility a plan is created. Execution of such a plan generates composite service. This approach is limited with respect to selecting Web services for composition just on the basis of input and output and does not handle services that have certain pre-conditions or effects.

Plengine.

The approach in [8] is a software system that supports planning for service composition and service enactment. The Plengine uses integrated meta-model approach to plan for Web services composition. The Plengine consists of two components: a composer and an enactor. The composer is responsible to generate composition with the help of its sub-component Composer Thread that uses search-planning algorithm to perform composition. The enactor is responsible for scheduling and execution of individual services within a composition. This work focuses on overcoming limitations (e.g. handling exceptions, sophisticated support for control flows and extending architecture of meta-models). paragraphs must be indented. All paragraphs must be justified, i.e. both left-justified and right-justified.

Limitations of Existing Approaches

We have seen various approaches for semantic web service composition. Now, we will compare above approaches on the basis various:

Service Discovery and Selection based on matching Functional and Non-Functional Semantics: This issue addresses the discovery of a service on the basis of matching Functional semantics like input, output and constraints and also non-functional semantics like response time, geographical location, etc. It also concerned with finally selecting a service from bunch of discovered services.

Service Binding & Referencing: In case of a workflow language as Web services composition, Service Binding & Referencing describes that how a selected service is bound in final composition. In case of an Al planning approach, it describes how a service is referred in final composition generated by an Al plan.

Composition Strategy: This employs the composition approach used for SWS composition. For example in case of a workflow language as Web services composition, composition strategy describes that either composition is dynamic or not. Or, in case of an Al planning approach composition strategy describes that either the final composition is generated automatically (automatic) or semi-automatically (semi-automatic).

Execution: This issue focuses on execution support for the execution of final composition.

Semantic Web Technology: It concerns with approach used to add semantics to Web service technology (e.g. OWL-S, WSDL-S or WSMO etc.).

Service Discovery

Service Selection

Service Binding & Referencing

Composition Strategy

Execution

Semantic

Web Technology

Bottom-up Approach

Partial

Not Available

Run-Time

Dynamic

True

OWL-S

METEOR-S

Available

Not Available

Deployment/Design Time

Dynamic

True

WSDL-S

Template Based Composition

Partial

Available

Dynamic

Automatic

True

OWL-S

WSMO Approach

Partial

Available

Dynamic

Semi- Automatic

True

WSMO

SHOP2

Partial

Partial

Dynamic

Automatic

True

OWL-S

SWORD

Available

Available

Composition Time

Semi-Automatic

True

Independent of Standards

Plaengine

Available

Available

Dynamic

Automatic

True

Integrated Meta-Model