Humboldt-Universität zu Berlin - Faculty of Mathematics and Natural Sciences - Model Driven Software Development


Model Driven Software Development

Current theses


  • Application of Multidimensional Search Structures to Analyze Similar Model Elements in Model-based Software Product Families
    David Császár, Diploma thesis
    Ad-hoc re-use mechanisms like the “clone and own” paradigm lead over time to a set of similar, but not identical variants of a software product. Such software product families are hard to understand and cause high maintenance expenses. Different methods and techniques have already been suggested in the research literature, both for analyzing those ad-hoc developed software product families and for restructuring them via systematic re-use mechanisms.  A first, fundamental step in analyzing those software families is the identification of joint and different parts in the individual variants. In case of model-based product families, the key problem is, regarding n variants, to identify an optimal (or sufficiently good) set of n-tuples consisting of equal (or similar) model elements. Actually, this already constitutes a difficult problem with two models (n = 2), in other words, determining corresponding elements in two models to be compared. This especially applies when the similarity of model elements should be considered for comparison. Multidimensional search trees have already been used successfully as a scalable process for identifying similar elements in pairs of models. This work should evaluate the possibility of generalizing this approach for N>2.
  • On the Handling of Ordered Collections of Model Elements within the Model Comparison Tool SiLift
    Tilo Prütz
    UML models may contain various ordered collections of model elements, e.g. the parameters of a class in a UML class diagram. For some collections, including the quoted example of parameter lists, the order is also relevant for model comparison. Local rearrangements of elements of an ordered collection in a model difference, it means the description of differences between two models, as an example, should be reported correctly. The model comparison tool SiLift does not ensure this at the moment. Even though there are initial outlines for a solution, a detailed elaboration and systematic evaluation are still missing and subject of the work.

  • Investigations on the Suitability of Freely Available UML Models for Evaluation of Tools in Model-based Software Engineering
    Michael Amboß
    Tools and methods of model-based software engineering (MBSE) have to be evaluated and tested. Suitable data sets for evaluation and testing, ideally models from the industrial practice, are often not freely available. The extraction of models used in OpenSource projects is one option in dealing with this shortage. Actually, on large number of GitHub projects have been examined lately, and many UML models, stored in image, .xmi and .uml files have been extracted. The key research question of the recent study was how UML models are used in OpenSource projects. It is still indetermined, how far those models can be used for evaluation of MBSE research tools and methods. Open questions concern for example the representation of extracted models, their consistency level as well as available development and evolution scenarios. This work aims on analyzing the available models and answer those questions. In best case, based on this, subsets of models can be identified which could serve as case studies for evaluation of certain MBSE tool classes.