QUALITY EVALUATION TECHNIQUES FOR ENTERPRISE ARCHITECTURE MODELS

• Main Author: Agustina Rumapea, Sri
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/69768
• Institution: Institut Teknologi Bandung

The enterprise architecture (EA) model is important for managing the complexity of organizational structures, information technology &business environments, and facilitating the integration of strategies, personnel, business &IT to achieve common goals. The EA model is a structural model (design) that provides a comprehensive view of an organization and can be used as documentation, communication, diagnosis, analysis, discussion, and guidelines for a dynamic company. Therefore many organizations allocate resources and invest in EA projects to understand and realize their business. Unfortunately, enterprises often fail to benefit from the AE model because they do not meet the expectations of stakeholders. In the context of EA products and services, how well EA products or services meet the needs of stakeholders is referred to as quality. Quality is the totality of features and characteristics related to the ability to meet both express and implied needs. Therefore, an evaluation of quality is needed because it will be useful and bring value to the enterprise. This is because a positive perception of the value of EAs is essential to ensure ongoing commitment from stakeholders. Studies related to EA evaluation itself are quite widely carried out and mostly focus on the planning and modeling process of EAs. On the other hand, research related to the evaluation of the AE model itself is still limited and in general is still carried out manually through questionnaires and interviews. But with the characteristics of a complex and dynamic EA model, evaluation takes a long time with resources that must have certain expertise. The result will also depend heavily on human interpretation. For this reason, an automation process is needed in conducting evaluations. A formal verification approach will be used for constructed evaluation techniques. Evaluation techniques that are carried out automatically allow management to carry out cost- and time-effective evaluations and get objective results. Addressing this issue, this study proposes the right quality factors for AE models and quality evaluation techniques that can be carried out automatically. The quality factors of an evaluation enterprise architecture models have been identified through systematic literature review and also classified through framework Sequal. There are 17 quality factors proposed and mapped into 6 types of the Sequal framework, namely semantic, syntactic, pragmatic, physical, deontic and empirical. To find out the degree of relevance of the factors that have been identified to the quality of the EA model is carried out by quantitative studies. Quantitative studies are analyzed statistically using validity and reliability tests. The validity test is carried out by using the Pearson product moment correlation method between each statement item and its total score, where a statement item is declared valid if the validity coefficient is greater than or equal to 0.300. Meanwhile, the data reliability test aims to test the consistency of a set of measuring instruments in each type of quality. The method used to test the reliability of the questionnaire was cronbach alpha, where a quality type was declared reliable if it produced croncbach alpha greater than or equal to 0.700. The result is that each indicator is declared valid because it has a validity coefficient above 0.300 so that it is able to explain each type of quality well. Similarly, the resulting cronbach alpha is all greater than the permissible critical point of 0.700 so that each quality type instrument is declared reliable or reliable in measuring each of the quality types that have been proposed. The quality factor is then evaluated to assess the level of importance of each factor in real practice through a survey to tenaga ahlis and a group of practitioners in the field of EA. The survey was conducted using the interview and questionnaire method. As a result of 17 factors, there are 16 quality factors that are relevant and important to use in real practice while the attractiveness factor is not very convincing. Taking into account the need for simplification and priority some factors have a higher level of importance than other factors. The survey results determine the 6 most important quality factors are completeness, correctness, interpretability, alignment, integration and readability&understanbility. The quality factor evaluation technique was developed using a formal method approach with a checking model technique. The case for the checking model chosen is Alloy. The development of evaluation techniques will follow 3 phases in the model checking technique, namely modelling, formal specification and verification. In the modelling phase, the Archimate2Alloy plugin is customized and subsequently used to transform EA models developed using the Archimate language into an alloy format. The metamodel for the Archimate language is also built according to archimate specification 3.0 so verification can be done. As for the specification phase, because the alignment quality factor will be verified, the specification for the alignment quality factor is developed. Specifications will be developed based on alignment rules identified using a heuristic alignment approach. Of the 52 alignment rules that were successfully identified, it was further analyzed and produced 21 formal specifications. The verification technique was further tested on 4 EA models developed by the Open Group. As a result, the evaluation technique developed can find existing violations in the EA model that can be used as a recommendation to improve the EA model. Automation of the use of quality model evaluation techniques and the speed of obtaining evaluation results is expected to increase the participation of architects and stakeholders in carrying out the evaluation process. With this, it is hoped that the quality of the EA model will be easier to achieve so that it will increase stakeholders' positive perception of the value of the EA model.