Predicting problem-solving performance with concept maps: An information-theoretic approach

a r t i c l e i n f o An increasing number of researchers and educational practitioners have shown their interest in the use of concept mapping techniques to elicit and represent individuals' knowledge structures. By extending previous research on concept map assessment, this study aims to develop a new evaluation metric enabling the " information uncertainty " embedded in concept maps to be assessed so as to predict individuals' problem-solving performance. In particular, our novel metric EntropyAvg is underpinned by the notion of entropy in information theory. A controlled experiment is carried out to evaluate the effectiveness of our proposed metric. Our experimental results demonstrate that the entropy values computed according to EntropyAvg strongly correlate to individuals' problem-solving performance and that the predictive power of EntropyAvg is significantly higher than that of existing widely used concept map evaluation metrics. Our research work opens the door to the application of concept mapping techniques in enterprise-wide knowledge management in general and enterprise-wide knowledge assessment in particular. Knowledge structure (or cognitive structure) refers to a set of concepts and the relationships among such concepts [12,28]. The presence of certain knowledge structures in individuals' memories largely determines their ability to characterize a particular problem domain and their problem-solving performance [2]. Several mapping techniques such as concept maps [24], causal maps [4,20], and mind maps [5] have been developed to elicit implicit knowledge structure from an individual's memory. Among these mapping techniques, concept mapping is unique in the sense that concepts and propositions play the central role in the representation of knowledge structures and the construction of meanings [26]. Assimilation theory is generally considered the theoretical foundation of the concept mapping technique [2,25]. According to assimilation theory, new knowledge is acquired by linking novel concepts to the existing concepts of a knowledge structure. Both the newly acquired and the existing knowledge structures are modified during a learning process. The construction of a concept map through hierarchical organization, progressive differentiation, and integrative reconciliation shapes an individual's ability to assimilate and integrate knowledge and solve problems creatively. Accordingly, certain properties such as the structure and content of the concept map constructed can be used to predict an individual's problem-solving performance [2,18,31,37]. As the structural elements are the main building blocks of a concept map, it is technically feasible to develop a formal computational model to estimate the quality of the concept map based on …