An experimental investigation of semi-automatic generation of concept maps from textbooks

Abstract

A concept map is a useful tool to organize and structure unstructured data, in this case, the information contained in a textbook. In education, a concept map can be used to set learning goals, monitor progress and visualize relationships between concepts. Constructing concept maps manually is a complex and time-consuming task which usually requires domain experts. Automatic creation of concept maps from documents is called concept map mining. In this study, we explore various technical approaches to semi automate the process of generating a concept map from an e-textbook. The approach selected is to use the topics listed in the appendix as concepts and mine the relation between pairs of topics using natural language processing and machine learning. Limitations of this approach and directions for future research are discussed.