User profiling with privacy: A framework for adaptive information agents

This paper presents a framework for personal agents that respect the privacy of the individual. We present some motivations and outline a framework for the use of personal agents and user profiling for information systems designed around web services. A key element of our approach in general is to consider the impact of user-profiling and autonomous agents on the user. One particular aspect, which we explore in this paper, is the need to respect user's privacy. One often-cited benefit of using personal agents is for personalising interaction. However, personalisation and privacy have contradictory goals in disclosing personal information. We explore some elements of our framework that allow the user to control the trade-offs around disclosure of personal information. We conclude with some motivating examples of the use of our framework in information-based tasks.