Analyzing Logs of a Media Server for Diagnosing Learners' Behaviors in a Video- Based Learning Environment

摘要

As Web-based courses using videos become popular in recent years, the issue of managing a media server becomes noteworthy. In general, the video signals are transmitted over Internet in the streaming format that can make the quality of Internet-based videos acceptable to learners using a limited bandwidth. Furthermore, a distance learning instructor can observe a learner's behaviors of using Internet-based videos for learning something. Because a Web-based course uses a media server to provide streaming videos for learning, logs of media server usage will be generated when a learner plays, pauses, stops, forwards, or rewinds a clip of a teaching video. An instructor needs analysis tools to manage the logs and discover unusual patterns within logs of a media server to help improve instruction. Although statistical tools for managing logs of a media server exist, none specifically address the needs of a distance learning instructor. The major reason is that logs of a media server can not server as records of a learner's behaviors to satisfy the requirements of behavior diagnosis properly. To resolve this problem, a data cube model is proposed to store learners' records of using Internet-based videos for analysis. Then, a data mining technique, called associations discovery, is used to analyze learners' online behaviors for mining and verifying learner's behavior patterns. For instance, a behavior pattern has the form "when learners use a specific clip, 80% of learners will repeat the clip of a teaching video more than three times". Finally, an instructor can reevaluate the clips of teaching videos with unusual patterns discovered.