A supervised learning approach to detect copyright infrigments

In the age of communication, piracy and copyright violations are emerging as a primary problem in our society. These issues produce an increasing loss of money and non-desirable situations in business. In order to provide some support, governments have approved multiple laws projects to toughen the legality. However, more efforts must be made in order to detect this illegal content and enforce the law. For this purpose, it is desirable to develop automatic tools which achieve this job. In this paper it is presented a framework based on a binary classifier which has been trained through supervised learning techniques. This classifier is able to label a website (as legal or not) focusing only on the URL domain. The work aim is experimental, hence we have used text processing techniques, several similarity metrics and different classification models, in order to evaluate the resulting outcomes. These latter have been promising, given a firm line of work in future.