CLEM - Combining Live and Exploratory Modeling

Behavioral models (e.g., programs or domain‐specific models) typically have to be first designed, implemented, and executed, before results can be observed. Every model change requires re‐triggering this pipeline to see the effects. Live modeling pushes towards changing this paradigm by allowing users to (almost) freely switch between model design and execution to receive immediate feedback (IF), i.e., rendering a model change’s effects visible directly on an already running execution, and to allow the direct manipulation (DM) of a model’s runtime state and based on this, derive the corresponding model design adaptation. DM is particularly challenging as propagating changes from the runtime state back to the model’s design is associated with an intractably huge search space of potential solutions. Exploratory modeling (EM) is a technique for enabling the systematic comparison of design alternatives, and thereby, encouraging design space exploration. Thus, EM seems a good candidate for tackling DM. Both live and exploratory modeling have seen preliminary studies before yet there is still no systematic approach to engineering new modeling languages with the native support of such facilities. Thus, these features have to be redeveloped for each new, possibly domain‐specific language.