PEEP- An Information Extraction base approach for Privacy Protection in Email

This paper presents a privacy compliance engine that monitors emails generated in an organization for violation of privacy policy of this organization. Our architecture includes four components: a domain knowledge defining the entities and private information we are dealing with, a pre-analysis component that extracts header information and segments the emails, an information extraction system that extracts private information, and an inference engine that matches the information extracted against a set of compliance rules. The compliance engine warns the sender about possible violations. A prototype has been developed for a university setting. Early empirical tests produced an Fscore of 69.3%.