Abstract
With the explosion of learning data in online educational platforms, many educators and researchers have developed a keen interest in learning analytics, in which social network analysis (SNA) and sentiment analysis are the two critical methods for exploring collective learning processes. In this study, we extracted textual data from the discussion forum of a "contract law" course in a university learning platform. We examined the differences in distributions of males and females in the overall sociogram. As well, through calculating learners' social characteristics and sentiment densities across a semester, we explored the relationship among social characteristics, sentiments and learning outcomes. The experimental results showed that females tended to be more active than males in forum interactions, and high-performing learners participated more actively in interactions than did low-performing learners. Moreover, there was a strong positive correlation between confusion and learning outcomes. As well, the learners who were involved actively in discussions were more likely to express the confused sentiment.