Classificação automática de questões conforme a Taxonomia de Bloom.

Abstract

The diversification of the difficulty in the questions proposed to students has a direct impact on their learning process. Thus, the fundamental role of the teacher is to provide questions that encourage the critical sense of students to improve their cognitive skills. Therefore, a taxonomy that aims to assist this process, classifying the cognitive level required by the questions, is Bloom's taxonomy. In this context, computing can offer tools for the automatic classification of questions according to Bloom's taxonomy, benefiting teachers and students. Although there are works with the objective of creating automatic classifiers for Bloom's taxonomy, some algorithms that use Gradient Boosting techniques are not commonly used for this process. Therefore, this work proposes the use of the XGBoost and CatBoost algorithms to be compared with the SVM and Random Forest algorithms in the question classification process. In addition, we propose the use of automatic techniques to increase the number of questions, classified according to Bloom's taxonomy, available in the database. We believe that the result of this work contributes to the improvement of the question classification models.