VidBot: Intelligent Video Learning Tool for Content Mining and Playback Traffic Statistics

With the rapid development of online education, video-based learning has become a significant mode of learning. However, existing video learning tools often fail to fully meet learners' needs for in-depth learning and interactive communication, thereby hindering learning outcomes. To address this challenge, this study proposes and implements an innovative intel-ligent video learning tool named VidBot, which is based on large language models for deep content mining and playback traffic analysis of videos. VidBot integrates Whisper and ChatGLM3-6B language models as well as the Pyltp library to build seven major functional modules, including Video Playback Visualization, Mind Mapping, Knowledge Graph, Cross-referencing of Knowledge Points, Automatic Video Segmentation, Intelligent Tutor, and Shared Notes. These modules aim to provide learners with deeper and more interactive learning experiences. The client of VidBot is released in the form of a plugin, and the server is integrated into the online education platform developed independently by our university.