Danmaku-Based Automatic Analysis of Real-Time Online Learning Engagement

Abstract

In recent years, there has been a rapid growth of online learning in higher education. Apart from professional online course platforms, many online video sharing websites have also provided online learning opportunities for college students. One of the most popular websites among college students in China is Bilibili, a Shanghai-based Chinese video sharing website known for its danmaku commenting system. This system enables users to post scrolling comments synchronized with the video timeline while the video is playing. Which attracts young students due to the lively user interaction. As a result, an increasing number of Chinese students are utilizing online courses on Bilibili as a supplementary learning resource alongside traditional classroom learning. Despite its popularity, online learning faces the challenge of students’ lack of participation more than traditional face-to-face learning does. To understand their learning involvement, we propose a novel danmaku-based automatic analysis model that extracts three dimensions of online learning engagement using the Text Mind software. This model enables us to understand the students’ learning patterns both as clusters and as individuals. Based on the model results, we present corresponding intervention strategies for different types of students based on their individual engagement characteristics.