Construção, comparação e agrupamento automáticos de mapas conceituais com apoio a atividades pedagógicas

Abstract

Concept Maps (CMs) are extremely useful tools in the construction and representation of knowledge. Formed by concepts and their relationships, an CM is a powerful tool for summarizing, simplifying, organizing and representing knowledge. Researches have investigated and proposed software resources with the purpose of favoring activities with CMs, promoting their creation in digital media, and proposing features such as automatic generation of maps from text, construction of collaborative maps, map comparison, junction of maps, among others. However, these resources are mostly isolated and difficult to integrate, which does working with CMs often still a challenge for the teacher, especially in large classes, requiring a lot of time and generating cognitive overload, especially in evaluation and comparison of maps. Therefore, this dissertation proposes an architecture that aims to support the teacher in the elaboration and application of activities with CMs more frequently, reducing the time necessary for this, as well as reducing their cognitive overload. The results obtained are promising, demonstrating that Natural Language Processing tools can greatly contribute to the automation of processes involving CM activities.