Privacy is one of the main societal concerns raised by critics of the uncontrolled growth and spread of information technology in developed societies. The purpose of this paper is to propose a privacy compliance engine that takes email messages as input and filters those that violate the privacy rules of the organization in which it is deployed. Our system includes two main parts: an information extraction module that extracts the names of the sender and recipients as well as sensitive information contained in the message; and an inference engine that matches the email information against a knowledge base owned by the organization. This engine then applies compliance rules to the information obtained from the extraction and database matching steps of the process. This prototype is currently being developed for a university setting. In this setting, it was shown to obtain a precision score of 77%. The next step of our research will be to adapt our system to the context of a health organization, where privacy rules are more complex and more sensitive.
展开▼
