ExeVis: concept-based visualization of exercises in online learning

In recent years, online learning has gained popularity and proven to be an effective way of education. Numerous studies have analyzed teaching materials and learning behaviors. However, most of the existing studies ignore the relationships between learning concepts and exercises, which can convey teaching performance and student behaviors. Presenting the relationships between concepts and exercises in online learning not only can help educators explore the distribution of exercises and concepts to consolidate knowledge but also can provide intuitive feedback on student behavior in online courses, which can enhance the teaching strategy. In this work, we extract learning concepts from exercises, establish logical relationships between concepts and exercises, and construct the hierarchical structures of concepts via both automatic models and semi-automatic models. To help users analyze and evaluate concepts and exercises effectively and intuitively, we design and implement a visual analysis prototype system, named ExeVis, integrating multiple interactive visualization graphs. ExeVis is equipped with multiple interactive and intuitive visualization charts including a control view to select and display basic information, an overview with hierarchical structures to present the distribution and mastery of concepts and exercises, a correlation view to reveal relationships between exercises, and a performance view to show individual capability. Case studies with real data and expert interviews demonstrate the usefulness and effectiveness of ExeVis in providing educators with valuable insights into the appropriateness of exercises and enabling them to adjust their teaching methods.