ESEML: Empirical Softare Engineering Modeling Language

Transcript

1. ESEML Empirical Softare Engineering Modeling Language Bruno Cartaxo [bfsc@cin.ufpe.br] Ítalo Costa [imac@cin.ufpe.br] Dhiego Martns [daom@cin.ufpe.br] André Santos [alms@cin.ufpe.br] Sérgio Soares [scbs@cin.ufpe.br] Vinícius Garcia [vcg@cin.ufpe.br]
2. MOTIVATION Researches in Softare Engineering normally proposes net practces to increase productvity and quality. A great part of these researches fail to present empirical evidence.
3. EMPIRICAL SOFTWARE ENGINEERING There are several types of empirical studies. Such as, surveys, case studies, secondary studies, acton research and controlled experiments.
4. CONTROLLED EXPERIMENTS According to Sjoberg only 1.9% of artcles has a controlled experiment and the quality is not very high. With Experiment s Without Experiment
5. CONTROLLED EXPERIMENTS Wide range of skills is necessary to conduct experiments that ofen create a barrier for adoptng it. Skills in terminology, statstcs knot hot and expertse in experimental design.
6. OBJECTIVE Facilitate the modeling process and defniton of an experimental plan. By mitgatng social barriers betteen stakeholders. Such as statstcians, experiments designers, and domain experts.
7. PROPOSAL DSLs are efcient to model specifc domains + Controlled experiments have their specifc domain elements = ESEML guides controlled experiments modeling in softare engineering and reduces social barriers
8. ESEML A visual DSL for modeling controlled experiments in softare engineering. That Automatcally generate the experimental plan from an instantaton of a domain model.
9. METHODOLOGY Informal reviet of models, ontologies and formal representatons for controlled experiments. Meta-model based on the reviet. Microsof DSL Tools to create the DSL and its related torkbench.
10. META-MODEL
11. LANGUAGE WORKBENCH ELEMENTS PALLETE EXPERIMENT MODEL
12. LANGUAGE WORKBENCH Parameter Hypothesis Dependent Variable Tratment Factor Experiment Validity Goal Queston Metric
13. GENERATED DOCUMENT
14. DOCUMENT PARTS
15. CONCLUSION ESEML is the kickof to a major initatve for defne a platorm of empirical studies in softare engineering.
16. FUTURE WORK Automatcally generaton of artfacts to collect data and execute experiments. Systematc reviet to more accurate meta-model . Empirical evaluaton of ESEML.
17. ?