Monitoring Online Topic Evolvements with Near-duplicate Videos (2011–2013)
Online video data are growing exponentially in Web social environments, where a large number of new topics keep emerging and evolving. The open nature and huge popularity of online social websites give rise to the existence of a large portion of near-duplicate content. This project aims at investigating effective methods to discover online topics and monitor their temporal evolvements by analyzing the dynamic textual data and near-duplicate videos. Detecting topic changes will help leverage collective intelligence for many business and services. The expected innovative outcomes include new topic models and novel methods for online near-duplicate video detection, topic discovery and mining.