| dc.description.abstract | Currently, many organizations employ business logic, and therefore business rules, as part of their products and/or services to deliver added value to their customers. However, as the amount of business rules utilized increases, the cohesion between those business rules is not taken into account. This makes it practically impossible to perform proper impact analysis before changes need to be implemented. Furthermore, the configuration of the business logic in products and/or services containing business rules are generally not designed to cope with modifiability. This seems very counterintuitive as previous research already proved that business logic is characterized by the highest change frequency in combination with the highest amount of effort required to perform the changes when compared to other aspects of IS development and maintenance.
First, a literature review is conducted to provide an overview of the context of business logic, business rules, the curent state-of-the-art on business rule architectures, neighbouring fields, and modifiability as a measurable quality attribute of a architecture. The literature review was followed by the exploration of three rule-oriented architectures for application in the BRM domain. The three rule-oriented architectures are utilized to construct logical architectures for the cases Chronic Obstructive Pulmonary Disease (COPD) and Diabetes Mellitus (DM) of the NHS, which concern the remuneration of healtcare services. Both the exploration and constuction of logical architectures regarding the three selected architectural candidates is performed according to the design science research approach, including a multitude of validation cycles with subject-matter experts. This was followed by selecting an architecture evaluation method from the body of knowledge to evaluate the architectural candidates on effort required to apply modifications, and therefore a means to analyse modifiability.
Based on four criteria the ALMA-method from the software architecture domain was selected. As ALMA is dependent on productivity measures of individual cases, we conducted two interviews with NHS personnel who transform requirements from multiple stakeholders into business rules. The interviews resulted in two levels of experience, which we interpreted carefully as worst and best case productivity figures for the NHS as input for ALMA. Lastly, we gathered additional historic data regarding change history of the NHS business rule sets as applied in the clinical conditions COPD and DM concerning the previous eight years.
ALMA’s results reveal that from all three included architectural candidates, the rule family-oriented architecture performs best on effort prediction. Analysis of the results shows that the case, the clinical condition COPD, containing less business rule sets, shows relatively equal modifiability performance concerning the included architectural candidates. This changes considerably when a larger case is analysed, which increases difference in modifiability performance between the included architectural candidates. We believe this difference is caused by the strutctural design decision of an architectural candidate which enforces the logical architectures either to utilize redundant elements or reuse existing elements. | |
