Avaliação automática de mapas conceituais para identificar indícios de aprendizagem significativa.

Abstract

The evaluation of concept maps is a complex process and many challenges currently exist,especially when it comes to automating the evaluation of maps, which should allow the creation of mechanisms to verify possible changes in the learner's cognitive structure, as well as providing information about the process of acquisition of new knowledge. Map evaluation requires structural and semantic analysis to evaluate the correctness of propositions, as proposed in several works in the literature, however, these evaluations do not consider certain cognitive aspects during the construction of learner's knowledge. In this sense, it is important to observe the map construction process, as through this it is possible to identify aspects related to meaningful learning, such as the cognitive processes of progressive differentiation and integrative reconciliation of Ausubel's theory and other cognitive aspects, such as difficulties in conceptual constructions and conceptual errors made by learners during the construction of their knowledge. The combination of evaluating the correctness of the propositions and the cognitive aspects identified during the map construction process allows us to identify signs related to the quality of learning. This thesis proposes a conceptual framework for the construction of computational systems, which provides the organization and structuring of the evaluation of concept maps to automate the evaluation. For conception of the framework, we developed a model and a method for qualitative and quantitative evaluation of concept maps that considers structural and semantic analysis as well as the map construction process, proposing criteria to identify signs of rote, meaningful and non-learning learning. In this research, we explored signs of learning through case studies, which were also used as proofs of concept for the proposed map evaluation method.