PrivaSeer: A Privacy Policy Search Engine

Web privacy policies are used by organisations to disclose their privacy practices to users on the web. However, users often do not read privacy policies because they are too long, time consuming, or too complicated. Attempts to simplify privacy policies using natural language processing have achieved some success, but they face limitations of scalability and generalization. While this puts an onus on researchers and policy regulators to protect users against unfair privacy practices, they often lack a large-scale collection of policies to study the state of internet privacy. To remedy this bottleneck, we present PrivaSeer, the first privacy policy search engine. PrivaSeer has been indexed on 1,400,318 English language website privacy policies and can be used to search privacy policies based on text queries and several search facets. Results can be ranked by PageRank, query-based document relevance, and the probability that a document is a privacy policy. Results also can be filtered by readability, vagueness, industry, and mentions of tracking technology, self-regulatory bodies, or regulations and cross-border agreements in the policy text. PrivaSeer allows legal experts, researchers, and policy regulators to discover privacy trends and policy anomalies in privacy policies at scale. In this paper we present the search interface, ranking technique, and filtering techniques for PrivaSeer. We create two indexes of privacy policies: one including supplementary non-policy content present in privacy policy web pages and one without. We evaluate the functionality of PrivaSeer by comparing ranking techniques on these two indexes.