Various kinds of inquiry, scenario, and project based teaching and learning have been promoted by educational thinkers over the last 100 years, but they are by no means typical in mainstream classrooms: "[...] projects offer many attractive promises, but they are often difficult to implement" (Barron et al., 1998, p. 306; cf. Bundsgaard, 2005, p. 315ff.). At least five challenges can be identified (Barron et al., 1998; Bundsgaard, 2009, 2010; Gregersen & Mikkelsen, 2007; Krajcik et al., 1998): Organizing collaboration, structuring workflows, integrating academic content, sharing products, and differentiating teaching. Technology can help respond to these challenges (Brush & Saye, 2008; Bundsgaard, 2009, 2010; Ge, Planas, & Er, 2010; Helic, Krottmaier, Maurer, & Scerbakov, 2005; Daniel Schneider & Synteta, 2005; D. Schneider, Synteta, & Frete, 2002), but platforms are very expensive to build from the ground up. If these platforms are to find their way into everyday teaching and learning, they have to be easy and cheap to develop. Thus there is a need for easy to use application programming platforms. This paper argues that a visual modeling programming language would be an important part of such a platform. The Collaborative Learning Modeling Language (ColeML) makes it possible to articulate complex designs for learning visually and to activate these design models as interactive learning materials. ColeML is based on research in workflow and business process modeling. The traditional approach in this area, represented by, for example, the Workflow Management Coalition (Hollingsworth, 1995) and the very widespread standard Business Process Modeling and Notation (BPMN), has been criticized on the basis of research in knowledge work processes. Inspiration for ColeML is found in this research area, which calls for interactive process modeling: " An interactive process system should / - Enable modelling by end users,/ - Integrate support for ad-hoc and routine work, / - Dynamically customise functionality and interfaces, and / - Integrate learning and knowledge management in everyday work" (Jorgensen, 2004, p. iii) It should not only be expert process modelers who are able to model processes; knowledge workers should also be able to do so themselves while the model is in use. ColeML is developed according to Jorgensen's (2004) design principles: 1) The modeling language should be simple with a few basic concepts, 2) the language should make possible a visual graphic representation of the model, 3) elements of the model should be able to change status during the articulation, 4) the system should accept unfinished models, 5) models should be able to be built by integrating other models, and 6) it should be possible to build a systematically designed and searchable repository of models and templates (Jorgensen, 2004, pp. 27f., 32, 36). In addition to these design demands it should be possible to integrate software components that solve subtasks in the process, for example, interactive assistants (cf. Bundsgaard, 2005, 2009). In this paper the basics of ColeML are presented together with the logical architecture of a system based on ColeML.
