摘要：It is often desirable to be able to guarantee the integrity of historical data, ensuring that any subsequent modifications to the data can be detected. It would be especially convenient to extend such proofs of integrity to certain computations performed later using the historic data. However, current Recommender Systems Architecture are not suitable for use with extensive and sensitive user profile data. Thus, we propose an approach to agent-based Information Filtering resulting in an architecture preserving the privacy of all participants. The proposed solution covers trust relationships between participants and utilizes privacy-preserving implementations of existing filtering techniques
