An approach for thematic relevance analysis applied to textual contributions in discussion forums

Abstract

Discussion forums in learning management systems (LMS) have been shown to promote student interaction and contribute to the collaborative practice in the teaching-learning process. By evaluating the postings, teachers can identify students with learning difficulties. However, due to the large volume of posts that are generated on a daily basis in these environments, manual analysis becomes impractical. This article proposes a mechanism to support teaching through the thematic relevance analysis of the posts made by students in discussion forums. For this, text mining and metrics from network science were used to process and extract characteristics of the texts. Then, the processed texts were classified through supervised learning algorithms. The results show that the use of these techniques may generate potentially useful indicators for teachers to help them improve their pedagogical practices.