Automated abuse detection of privacy policy

Abstract

With the wide adoption of smart devices and mobile apps, users are able to perform daily activities such as internet banking, shopping and even instant messaging. These mobile apps collect a variety of information from their users which poses significant risks to data privacy. Therefore, privacy policies are intended to describe their data privacy practices and in recent years, there have been regulatory restrictions such as the General Data Protection Regulation (GDPR) that serves as a guideline for such practices. However, due to a lack of understanding of GDPR, privacy policies might be vague and incomplete which fails to inform users how data is being stored, used or shared. Furthermore, due to the complexity and length of privacy policies, users tend to ignore them. As such, this report proposes an automated privacy policy classification tool to determine if a privacy policy is complete through the use of machine learning and deep learning techniques. These techniques will be used to learn the input features and patterns of various sentences that constitute a complete privacy policy. At the same time, a comparison was made to determine which algorithm performs the best in the classification.