Utilização de Técnicas de Aprendizado de Máquina para Acompanhamento de Fóruns Educacionais

Abstract

With the growth of distance education, and the adoption of computational systems as an educational support tool, the use of virtual learning environments has also grown. These environments offer various tools that could be used to assist in the teaching-learning process, one of these tools is the forum. Discussion forums in online educational platforms often require monitoring of thousands of users. Due to the large number of posts, it is difficult to supervise all students. In order to perform an effective monitoring, it is necessary computational tools to assist the teacher. This study aims to develop an approach to manage educational forums posts. To achieve this goal, two steps were performed: the identification of posts classified as doubt, and the extraction of the subject from post. In the Identification stage, the neural network algorithm achieved 97% of accuracy. For subject extraction step was proposed an algorithm based on unsupervised learning and Wikipedia. This algorithm achieved a 76.1% of accuracy. Moreover, the system recommended videos to the student according to the subject extracted from posts. This approach assists the teacher by reducing the time used to answer the students’ questions, and the student by providing supplementary materials to help solve his doubt.