Classifying Research on UML model Inconsistencies with Systematic Mapping

Sammanfattning: Context: Unified Modeling Language (UML) is a universal and standard modeling language that has been extensively used in software development process. Due to overlap and synchronous nature among different modeling artefacts in UML, several consistency issues have been identified in many software development projects that may lead to project failure. To reduce the level of such threat, over the past decade, a substantial research addressing those problems has been done both in academic and industry. This study is intended to investigate the reported research and to provide a systematic picture on different researched aspects of UML model inconsistencies, using the systematic mapping method. Objectives: The overall goal was to be achieved by fulfilling the following two main objectives: elaborating a proper and justified tool for performing the mapping and later used the tool in order to obtain a systematic and multidimensional picture of the approaches and the performed research in the area relating to different issues considering inconsistencies when using UML in software development. Research Methods: In order to ensure quality of the final foreseen systematic picture of the conducted research, a considerable effort was put first on a preparation of the tool that was used to obtain the mapping. The tool was a rigorous process based on classification methods and mapping guidelines obtained from a systematic literature review on the systematic mapping in software engineering. Then the tool was applied in a systematic way to obtain a number of mappings, followed by the analysis of the obtained results. Results: The systematic literature review resulted in identifying 5 mapping guidelines, 21 classifications, and 2 categorization methods. After analysis of them, a justified mapping process was developed by selecting standard guidelines, appropriate classifications and categorization methods. The mapping process applied for the period of 1999-2012 revealed 198 relevant studies developed by 321 researchers. On the basis this evidences, a number of mappings illustrating the conducted research on UML model inconsistencies ware obtained. The mapping reviled that the published research is mostly focused on rather formal issues such as semantic, syntactic, intramodel, inter-model and evolution problems, while a less attention is placed on more practical on time, and security problems. When the quality of research is concerned, 38% of papers proposed solutions as well as validated them through academic, industry or both, 35% of papers proposed only solutions. When the usage of empirical methods is considered, case studies are most frequently used (in almost half of the relevant papers) and followed by experiments (reported in 15% of papers), while 25% carried works do on report a systematic method used. Conclusions: The findings of systematic mapping study revealed that there are some aspect related to consistency such as time and security that are not given big attention. Identification and in-depth studying of inconsistencies in UML designs along with their dependencies are also missing. Most of the investigations are also academic with no evidence whether these reports produce interest for industry or not. State-of-the-art followed by state-of-the-practice studies related to consistency checking techniques and validating them in real industrial setting could be recommended.