Abstract
Modeling tools are well established in software development. A model is the result of a series of modeling activities. The ability to recognize when a user is working on a certain modeling activity opens up a range of possibilities for context-sensitive support. One possible way to support the user is offering the auto-completion of the current task. The recognition of modeling activities is typically carried out by matching event patterns against events emitted by a user's editing operations. A user that intends to add or customize auto-completions must be able to easily understand and create activity definitions. However, defining the currently required complex event patterns is a challenging and error-prone task even for a person with an intensive knowledge of event-processing languages. In this paper, we propose the visual definition language VisPaRec accompanied by a method that allows creating activity definitions in a semi-automated and graphical way. We evaluate our visual definition language in a comparative user study against the generic event-processing language Rapide. We found that the proposed visual representation increases comprehensibility while reducing time for constructing and modifying activity definitions significantly.