Ontology based Approach for Semantic Service Selection in Business Process Re-Engineering

TL;DR

This paper presents BPMNSemAuto, an ontology-based model that enables business analysts to verify the feasibility of business process designs and automatically select suitable web services for execution.

Contribution

It introduces a novel ontology-driven framework integrating service registry data, QoS properties, and synonym handling for automated business process service selection.

Findings

Automates business process feasibility checking.

Integrates WordNet for synonym resolution.

Provides an effective service selection algorithm.

Abstract

This research aims to provide the possibility to the business analysts to be able to know whether their design business processes are feasible or not. In order to solve this problem, we proposed a model called BPMNSemAuto that makes use of the existing services stored in the service registry UDDI (Universal Description Discovery and Integration). From the data extracted from the UDDI, the WSDL files and the tracking data of service execution on the server, a Web Service Ontology (WSOnto) is generated to store all the existing services. The BPMNSemAuto model takes an input of business process design specifications, and it generates an executable business process as an output. It provides an interface for business analysts to specify the description of each service task of the design business process. For each service task, the business analysts specify the task objective (keywords), inputs, outputs and weights of the Quality of Service (QoS) properties. From the design business process with the service task specifications, a Business Process Ontology (BPOnto) is generated. A service selection algorithm performs the mapping between the instances of the WSOnto and the BPOnto to obtain possible mappings between these two ontologies. The obtained mappings help the model to acquire web services to execute the desired service tasks. Moreover, the consistency checking of the inputs of the proposed model is performed before executing the service selection algorithm. WordNet is used to solve the synonym problems and at the same time a keyword extraction method is presented in this paper.