An Embedded Domain Specific Language to Model, Transform and Quality Assure Business Processes in Business-Driven Development

Micallef, Luana (2008) An Embedded Domain Specific Language to Model, Transform and Quality Assure Business Processes in Business-Driven Development. Other thesis, University of Malta. (Full text available)

PDF - Publisher pdf
Download (3MB) Preview
[img]
Preview

Abstract

Business process models are produced by business analysts to graphically communicate the business requirements to IT specialists. As business processes are updated to meet the new demands in the competitive market, the underlying IT solution is adapted, to reflect precisely the current goals of the organisation. The models should then act as an abstract representation of the solution. It is essential to adapt to Business-Driven Development (BDD), whereby models are refined into the IT solution and implemented in a Service-Oriented Architecture. This means that models must be free from data and control-flow errors, such as deadlocks. If models are not quality assured at the modelling phase, errors would be discovered later and the entire BDD lifecycle would have to be repeated. Combining model transformations with quality assurance would help modellers to preserve the correctness of models and rapidly carry out modifications. Although various modelling languages have been developed to assist modellers in the production of high quality business process models, none of them adopted a functional approach, based on higher-order logic. As BDD is being adopted by most organisations, the need for such a language is becoming more evident. Since specialized functionality is required, a general-purpose language is not really necessary. Instead, a domain-specific language which provides the right abstraction and captures precisely the semantics of the business process modelling domain, should be developed. The definitions of the models would be easy to comprehend and reason about, by anyone who is not necessarily an IT specialist. However, since languages are made up of domain independent and dependent linguistic components, it is more cost effective and feasible to embed the new language in a general-purpose language. In this project we present a domain specific language embedded in the functional language, Haskell, to model, transform and quality assure business processes in Business-Driven Development. By adopting a functional approach, we developed a language: 1) with which various models can rapidly be produced in a concise and abstract manner, 2) allows users to focus on the required behaviour rather than its implementation, 3) ensures that all the required details, to generate the executable code, are specified, 4) the abstract representation can be interpreted, analysed and transformed in various ways, 5) quality assures models by carrying out three types of checks; by Haskell’s type checker, at construction-time through our embedded type system, and by specialised functions that analyse the components in the model. By embedding our language in Haskell, the models, quality assurance checks and transformations are essentially functions which can easily be composed and defined. Connection patterns, defined in the language, play an important role to ensure that definitions are concise, readable and easy to comprehend. Different from other previous modelling tools, users are able to define their own parameterized models and transformations. By generating a directed graph for the models, various types of analysis can be carried out with greater ease. Moreover, quality assurance can be combined to model transformations by declaratively defining pre and post conditions for each transformation. These conditions as well as transformations can easily be composed of other previously defined checks or transformations. With this language, we aim to capture the domain semantics of IBM’s WebSphere Business Modeler Advanced v6.0.2.

Item Type: Thesis (Other)
Additional information: BSc IT (Hons) Thesis
Subjects: Q Science > QA Mathematics (inc Computing science)
Divisions: Faculties > Sciences > School of Computing
Depositing User: L. Micallef
Date Deposited: 17 Apr 2015 16:14 UTC
Last Modified: 26 Apr 2018 22:09 UTC
Resource URI: https://kar.kent.ac.uk/id/eprint/47997 (The current URI for this page, for reference purposes)
  • Depositors only (login required):

Downloads

Downloads per month over past year