70 likes | 186 Vues
This project explores the benefits of linking MOOC contents through human language technologies, improving search, summarization, and problem-solving. It delves into user studies and crowdsourcing strategies to establish ground truth and enhance learning outcomes. The research compares methods for automatic linking and highlights the potential of graphical models in facilitating better results. Future work involves refining algorithms and expanding the approach to more MOOCs to enhance scalability.
E N D
Linking MOOC Contents Using Human Language Technologies Shang-Wen Daniel Li,Victor Zue, with Hung-yi Lee and Chengjie Sun
Linking MOOC Contents Using Human Language Technologies Shang-Wen Daniel Li,Victor Zue, with Hung-yi Lee and Chengjie Sun
Linking MOOC Contents Using Human Language Technologies Shang-Wen Daniel Li,Victor Zue, with Hung-yi Lee and Chengjie Sun
Linking MOOC Contents Using Human Language Technologies Shang-Wen Daniel Li,Victor Zue, with Hung-yi Lee and Chengjie Sun
Linking MOOC Contents Using Human Language Technologies Shang-Wen Daniel Li,Victor Zue, with Hung-yi Lee and Chengjie Sun
Establish Ground Truth Can linking help? • User study on problem solving • Linking seems to help • Search faster • Summarize better • Obtain annotated data w/ a crowdsourcing strategy • Crowdsourcing can achieve quality similar to that from experts (Kappa Score) Search (Seconds) Summary (Score)
Conclusion and Future Work Linking Contents Automatically • Compare graphical model-based method to cosine similarity • Graphicalmodelyieldsbetterresult • Linking does help learning • HLT is promising for a scalable solution • Refine algorithms and expand to more MOOCs Linking between textbook and other contents (Sentence accuracy)