Constructing Progressive Learning Routes through Qualitative Simulation Models in Ecology


Paulo Salles

Universidade de Brasilia Instituto de Ciências Biológicas Campus da Asa Norte Brasilia - DF 70.710-900, Brasil Phone:+55 61 348-2286 E-mail:

Bert Bredeweg

University of Amsterdam Department of Social Science Informatics Roetersstraat 15 1018 WB Amsterdam The Netherlands Phone: +31-20-525 6788 E-mail:

Submitted to the 10th International Conference on Artificial Intelligence in Education AI-ED 2001, May 19-23, San Antonio, Texas, USA.


Keywords: Qualitative Simulation, Model Progression, Model Dimensions, Model-based Reasoning, Sequencing Subject Matter

Abstract: Qualitative models support interactive simulations that are well suited to help learners in acquiring causal interpretations of physical systems and their behaviour. Such simulation models can be large, particularly if they include many subsystems. When simulations are too big they hardly can be used effectively for teaching purposes. They have to be reorganised into smaller sets of simulation models and ordered in a sequence for the learner to progress through. Model-dimensions and techniques such as Causal Model Progression have been presented as means to address this problem. In this paper we investigate how to decompose a large qualitative simulation into a progressive sequence of smaller simulations, useful for teaching purposes, in the domain of ecology. Based on notions introduced by Causal Model Progression, the Genetic Graph, and the Didactic Goal Generator, we have constructed a set of dimensions that can be used in this respect. Following these dimensions we show how a large qualitative simulation model of the Brazilian Cerrado vegetation dynamics can be rearranged into a sequence of clusters, each representing and simulating distinct features of such ecological systems. These clusters are ordered in evolutionary model progression lines according to movements from static to dynamic models and, by incorporating structural changes, from less complex to more complex models. The approach presented in this paper thus provides means, in terms of knowledge characteristics, to effectively reorganise qualitative simulation models for teaching purposes. In the discussion we briefly argue that this approach may also be applicable to qualitative simulation in other domains.


Full Paper (PDF 50.7 KB)