Copyright Infringement Detection of Music Videos on YouTube by Mining Video and Uploader Meta-data

YouTube is one of the largest video sharing website on the Internet. Several music and record companies, artists and bands have official channels on YouTube (part of the music ecosystem of YouTube) to promote and monetize their music videos. YouTube consists of huge amount of copyright violated content including music videos (focus of the work presented in this paper) despite the fact that they have defined several policies and implemented measures to combat copyright violations of content. We present a method to automatically detect copyright violated videos by mining video as well as uploader meta-data. We propose a multi-step approach consisting of computing textual similarity between query video title and video search results, detecting useful linguistic markers (based on a pre-defined lexicon) in title and description, mining user profile data, analyzing the popularity of the uploader and the video to predict the category (original or copyright-violated) of the video. Our proposed solution approach is based on a rule-based classification framework. We validate our hypothesis by conducting a series of experiments on evaluation dataset acquired from YouTube. Empirical results indicate that the proposed approach is effective.