Fig. 1. The learning model of GCE.
Department of Information Management,
Ming-Chuan University, Taipei, Taiwan, R.O.C.
fhwang@mcu.edu.tw
In this paper, we design a learning environment called GCE (Geographic Climate Exploration) equipped with intelligent learning support agents for geographic climate exploration on the Web. In this environment, students have full control over the entire learning process. The main learning strategies supported by GCE is the rule induction learning strategy [1,3] and the explanation-based learning strategy [2]. The GCE environment scaffolds the students with real world cases presented as multimedia medium to help them induce the characteristics of geographic climate patterns. Besides, an intelligent learning support agent could be asked to give guidance in a gradual manner to improve students' induction effectiveness. Students are also expected to construct a scientific causal model of the earth geographic climates through the explanation-based generalization strategy. They should be able to explain the reasons that constitute a region's climate by considering the interactions among a variety of climatic factors. Again, an intelligent agent could be asked to intervene if students failed to give the correct explanations. The domain knowledge structure is represented as a two-level knowledge abstraction. In the first knowledge level, each type of climate can be recognized as a specific configuration of the constituting factors. This knowledge level is a type of declarative knowledge, and traditionally students have to learn it by memorizing. The second knowledge level is a deeper, causal knowledge that involves the influence factors and their interactions, which comprise a scientific causal model that explains a region's climate.
The learning model is shown in Fig. 1. Students could enter the knowledge construction mode that consists of rule induction and explanation-based generalization, or enter the two test zones in any orders. In this exploration environment, we provide the following learning resources and activities. They are (1) the geographic climate multimedia material base, (2) the knowledge construction activity, and (3) the test zones for testing knowledge transfer capability. The geographic climate database contains hypermedia materials about the geographic climates. It also serves as an exploration database for the students to engage themselves in the knowledge construction activities. Students are expected to identify the characteristics of geographic climate patterns through the rule induction strategy. Besides, to get a better understanding of the climate patterns, students are required to construct a scientific causal model of the earth geographic climates through the explanation-based generalization strategy. The GCE environment provides two test zones to enhance students' learning. The first test zone provides a statistical graphical diagram that represents the changes of rain and temperature of a region over the twelve months of a year. Students are asked to figure out what climate of the region is from this diagram. This test zone aims to enhance students' memorized knowledge about climate patterns. On the other hand, the second test zone is a set of games whose objectives are to provide real life situations and problems that students have to solve by applying various kinds of knowledge, including the climatic knowledge.
A prototype of the GCE is under implementation based on the ILSE architecture [4]. A preliminary evaluation result shows that the system is effective to some extent. However, the result also shows that more exploratory tools are needed to help students investigate the climatic data in more than one perspectives. Further evaluation tests will be conducted to provide evidence of learning and transfer. One direction for future work would involve manipulating students' learning strategies and examining their transfer performance in the consequence of problem solving.
[1] Dejong, G.F., An approach to learning from observations, in: R. . Michalski, J.G. Carbonell and T.M. Mitchell (Eds.), Machine Learning: An Artificial Intelligence Approach, Vol. 2. Morgan Kaufmann, Los Altos, CA, 1986.
[2] Mitchell, T.M., Kelle, T., and Kedar-Cabelli, S., Explanation-based generalization: A unifying view, Machine Learning, 1: 47–80, 1986.
[3] Simon, H.A., and Lea, G., Problem solving and rule induction: A unified view, in: L.W. Gregg (Ed.), Knowledge and Cognition. Erlbaum, Hillsdale, NJ, 1974.
[4] Wang, F. H. (1997). ILSE: A three-tier client/server architecture for intelligent learning-support environments on World Wide Web, in: Proc. of the 3d Conference on MIS Research and Practice, Nov., 1997, pp. 199–206.
Fig. 1. The learning model of GCE.