iMap : um mecanismo de inferência para mapas conceituais

Abstract

Concept maps are graphical representations of a person’s knowledge at a given time and area of expertise. For his investigative nature, they are used as tools to support pedagogical approaches that aim to promote meaningful learning. However, the maps evaluation process tends to be costly because it causes a heavy load of cognitive processing on the part of the evaluator, since it needs to map out the concepts and relations in order to find nuances of knowledge present there. This research aims to increase the level of abstraction in the interactions between the evaluator and the concept maps by providing an intermediate layer of computational intelligence that favors the communication through questions and answers in natural language, producing tools that allow the evaluator to examine the content of the concept map without requiring the visual mapping of concepts and relations present in the evaluated maps. A tool is prototyped and a proof of concept is presented. The analysis of the proposed architecture has led to the definition of a final architecture with features that allow enhancing the use of concept maps and facilitating various pedagogical applications. This research is dedicated to the investigation of Question Answering Systems, by applying natural language processing techniques to analyze the question and interprete the concept map, as well as applying techniques of artificial intelligence to infer answers to the questions.