UC San Diego

The adoption of digital computers and the digitization of record keeping marked the onset of the Digital Revolution bringing us into the Information Age. Among other things, the Internet is often regarded as a central force behind this revolution. As the name suggests, it means a web of interconnected computer networks. In recent years it has grown exponentially, allowing users to share and access information at an unprecedented scale. This freedom has its own set of challenges; the Internet unfortunately is often used for illegal sharing of copyrighted content and the traditional copyright laws were not well equipped to handle such scenarios. Hence, the Digital Millennium Copyright Act was signed into law as an attempt to tackle these challenging issues. It provides an extra-judicial process, Section 512, by which copyright holders can issue takedowns notices of allegedly infringing material. In this work we attempt to look at the takedown notices, available from online repository like Chilling Effects, and try to analyze them in a systematic fashion. We mainly focus on using machine learning techniques such as latent Dirichlet allocation, k-means, support vector machines and random forests to find interesting patterns in the dataset and try to reason about different challenges faced while working with this dataset