Fixing Model Transformation Errors Using Heuristic Search
Department
Computer Science
Major
Computer Science
Research Advisor
Kessentini, Marouane
Advisor's Department
Computer Science
Funding Source
Missouri S& T Opportunities for Undergraduate Research Experiences (OURE) Program
Abstract
One of the efficient techniques used in model-driven engineering to ensure the quality of model-transformation mechanisms is testing. Two important steps should be addressed: the efficient generation/selection of test cases and the definition of oracle functions that assess the validity of the transformed models. This work is concerned with an additional step related to automatically fixing model transformation errors. In this work, we propose an evolutionary approach that uses good traceability links to fix detected transformation errors. This novel evolutionary approach is based on the dissimilarity between the new collected traceability links after fixing errors and the base of good traceability links. Thus, the best solution is the set of changes (on transformation rules or metamodels) that maximize the similarity between our base of good traceability links and the new ones collected after executing the changes to correct errors. The validation results on widely-used transformation mechanism confirm the effectiveness of our approach in terms of precision and recall on ten different scenarios.
Biography
Nathan is presently a junior at Missouri University of Science and Technology. Nathan is getting a BS currently in computer science and a minor in mathematics. Nathan interests include: graphical user interface, volunteer work, racquetball, soccer, and many more outdoor activities.
Research Category
Sciences
Presentation Type
Oral Presentation
Document Type
Presentation
Location
Upper Atrium/Hallway
Presentation Date
03 Apr 2013, 9:00 am - 11:45 am
Fixing Model Transformation Errors Using Heuristic Search
Upper Atrium/Hallway
One of the efficient techniques used in model-driven engineering to ensure the quality of model-transformation mechanisms is testing. Two important steps should be addressed: the efficient generation/selection of test cases and the definition of oracle functions that assess the validity of the transformed models. This work is concerned with an additional step related to automatically fixing model transformation errors. In this work, we propose an evolutionary approach that uses good traceability links to fix detected transformation errors. This novel evolutionary approach is based on the dissimilarity between the new collected traceability links after fixing errors and the base of good traceability links. Thus, the best solution is the set of changes (on transformation rules or metamodels) that maximize the similarity between our base of good traceability links and the new ones collected after executing the changes to correct errors. The validation results on widely-used transformation mechanism confirm the effectiveness of our approach in terms of precision and recall on ten different scenarios.
