An interactive metaheuristic search framework for software service
identification from business process models

Aljawawdeh, Hamzeh

Abstract

In recent years, the Service-Oriented Architecture (SOA) model of computing has become widely used and has provided efficient and agile business solutions in response to inevitable and rapid changes in business requirements. Software service identification is a crucial component in the production of a service-oriented architecture and subsequent successful software development, yet current service identification methods have limitations. For example, service identification methods are either not sufficiently comprehensive to handle the totality of service identification activities, or they lack computational support, or they pay insufficient attention to quality checks of resulting services. To address these limitations, comprehensive computationally intelligent support for software engineers when deriving software services from an organisation’s business process models shows great potential, especially when the impact of human preference on the quality of the resulting solutions can be incorporated. Accordingly, this research attempts to apply interactive metaheuristic search to effectively bridge the gap between business and SOA technology and so increase business agility.
A novel, comprehensive framework is introduced that is driven by domain independent role-based business process models, and uses an interactive metaheuristic search-based service identification approach based on a genetic algorithm, while adhering to SOA principles. Termed BPMiSearch, the framework is composed of three main layers. The first layer is concerned with processing inputs from business process models into search space elements by modelling input data and presenting them at an appropriate level of granularity. The second layer focuses on identifying software services from the specified search space. The third layer refines the resulting services to map the business elements in the resulting candidate services to the corresponding service components. The proposed BPMiSearch framework has been evaluated by applying it to a healthcare domain case study, specifically, Cancer Care and Registration (CCR) business processes at the King Hussein Cancer Centre, Amman, Jordan.
Experiments show that the impact of software engineer interaction on the quality of the outcomes in terms of search effectiveness, efficiency, and level of user satisfaction, is assessed. Results show that BPMiSearch has rapid search performance to positively support software engineers in the identification of services from role-based business process models while adhering to SOA principles. High-quality services are identified that might not have been arrived at manually by software engineers. Furthermore, it is found that BPMiSearch is sensitive and responsive to software engineer interaction resulting in a positive level of user trust, acceptance, and satisfaction with the candidate services.

Citation

identification from business process models. (Thesis). University of the West of England. Retrieved from https://uwe-repository.worktribe.com/output/853842

• Thesis Type: Thesis
• Deposit Date: Oct 1, 2018
• Publicly Available Date: Jan 15, 2019
• Public URL: https://uwe-repository.worktribe.com/output/853842
• Award Date: Jan 15, 2019