Learning Track Representation and Trends for Conference Analytics

—We propose a new system for facilitating the co-creation of conference tracks through data analytics and human knowledge. The system attempts to learn track representations based on a topic-track matching framework, infer historical track representations by topic evolution paths and investigate the evolution of track representations to figure out track trends. We thus aim to develop a data-driven approach for improving expert-designed track descriptions in future conference organization. One challenge in our work is how to learn track representations from limited publication papers of each year, and another challenge is to how to figure out track trends when track descriptions are not readily available in some years. We present two novel approaches on learning track representation by topic-track matching and analyzing track trends by constructing topic evolution paths, respectively. We also show interesting results on topical leaps and branches from year to year, obtained from pap...

• Type: Journal
• Year: 2016
• Where: HICSS
• Authors: Pengfei Liu, Shoaib Jameel, King Keung Wu, Helen M. Meng
• Keywords: Biometrics, HICSS 2016