A case study of collaborative modelling: building qualitative models in ecology

Modelling is seen as a learning activity in itself and qualitative modelling environments start to play a role in this respect. However, building (qualitative) models is not an easy task. It is therefore necessary to develop support for teachers and students. This paper describes an experience in which Artificial Intelligence (AI) undergraduate students from the University of Amsterdam and graduate ecology students from the University of Brasilia were engaged in a collaborative model building activity. The objective was to build qualitative models about the carbon cycle and the greenhouse effect in GARP. A questionnaire was used to obtain the students opinion about different aspects of the modelling effort. Almost all the students (94%) reported an increase in their understanding of the ecological problems after the modelling activity, (an observation that supports the idea of modelling as a learning activity in itself). In certain aspects, being an ecologist (and therefore possessing relevant domain knowledge) made some parts of the model building activity easier. For example, global identification of the processes involved. Contrary, the AI students found it easier to construct typical AI representations, such as subtype hierarchies. The most difficult task for both groups was to build a library of model fragments. Identifying quantities and their quantity spaces were also mentioned as difficult. In order to improve the performance of the qualitative modelling environments the QR community has to put effort in developing authoring tools with explanatory facilities. The study reported here provides some insights on how to scaffold such model building tools.