Fine-Tuning Model Transformation: Change Propagation in Context of Consistency, Completeness, and Human Guidance
An important role of model transformation is in exchanging modeling information among diverse modeling languages. However, while a model is typically constrained by other models, additional information is often necessary to transform said models entirely. This dilemma poses unique challenges for the model transformation community. To counter this problem we require a smart transformation assistant. Such an assistant should be able to combine information from diverse models, react incrementally to enable transformation as information becomes available, and accept human guidance – from direct queries to understanding the designer(s) intentions. Such an assistant should embrace variability to explicitly express and constrain uncertainties during transformation – for example, by transforming alternatives (if no unique transformation result is computable) and constraining these alternatives during subsequent modeling. We would want this smart assistant to optimize how it seeks guidance, perhaps by asking the most beneficial questions first while avoiding asking questions at inappropriate times. Finally, we would want to ensure that such an assistant produces correct transformation results despite the presence of inconsistencies. Inconsistencies are often tolerated yet we have to understand that their presence may inadvertently trigger erroneous transformations, thus requiring backtracking and/or sandboxing of transformation results. This talk explores these and other issues concerning model transformation and sketches challenges and opportunities.
Prof. Dr. Alexander Egyed is a full professor at the Johannes Kepler University (JKU), Austria and head of the Institute for Systems Engineering and Automation (SEA). He received his Doctorate degree from the University of Southern California, USA under the mentorship of Dr. Barry Boehm in 2000 and, before joining the JKU in 2008, he worked as a Research Scientist for Teknowledge Corporation, USA (2000-2007) and then as a Research Fellow at the University College London, UK (2007-2008). He is most recognized for his work on software and systems modeling – particularly on consistency (=i.e., correctness and completeness) and traceability of models (i.e., where is information coming from, where is it being used?) and developed the fastest-known model consistency checker – a technology which is currently being transitioned to industrial use. Dr. Egyed’s work has been supported by research grants from Austria, Canada, UK, and USA and his work has been published in over 80 refereed scientific books, journals, conferences, and workshops. He was recognized as the 10th best scholar in software engineering in a study by Ren-Taylor in the Communications of the ACM, 2007, was named an IBM Research Faculty Fellow in 2010 in recognition to his contributions to consistency checking, and received a Recognition of Service Award from the ACM. His research interests and expertise cover model-driven engineering and includes variability, consistency checking and resolution, transformation, and traceability. He is a member of the IEEE, IEEE Computer Society, ACM, and ACM SIGSOFT. Contact him at firstname.lastname@example.org.