Design and Implementation of a CBR-based Privacy Agent

This paper presents Privacy Advisor; a software which uses machine-learning techniques to help the users make online privacy decisions. Privacy Advisor is based on Case Based Reasoning (CBR), which relies on the ability to identify similar situations from the past and use these to provide recommendations in new situations. This paper focuses on the algorithms necessary to calculate the similarity of privacy policies. In addition, we provide results from a focus group study on the perceived similarity of data items and data handling purposes from a privacy point of view.