Classificação de traços de personalidade através de Linguagem Natural aplicada a ambientes virtuais de aprendizagem

Abstract

In a context of collaborative computer-mediated learning, it is important to include groups formation methodologies (grouping) of students so that interactions between them can lead to the improvement of learning. The research project, in which this work is inserted, consists of identifying the personality traits of the learner through Natural Language. From these, it will be possible to predict the best group formations. To do so, a collection of texts written by students was made, to then classify their personality traits, according to the Big Five model. Thus, after processing the text using the NLTK library available in Python, the categorization of the useful words mined in the text was performed using the LIWC dictionary. The percentages of each category of word present in the text allow to identify the traits of personality of the student, using some classification algorithm. It should be emphasized that the Big Five model has already been associated with the formation of groups and that obtaining texts written in virtual learning environments is feasible, since they contain tools such as chats and forums. Thus, this work can be inserted in these environments in order to provide information for group composition.