Privacy-aware smart city: A case study in collaborative filtering recommender systems

摘要: Ensuring privacy in recommender systems for smart cities remains a research challenge, and this paper we study collaborative filtering privacy-aware cities. Specifically, use the rating matrix to establish connections between city κ-coRating, novel privacy-preserving data publishing model. First, model concerns as problem of recommendation. Then, introduce κ-coRating address published matrices, by filling null ratings with predicted scores. This allows us mask original preserve κ-anonymity-like privacy, enhance utility (quantified using prediction accuracy paper). We show that optimal κ-coRated mapping is an NP-hard design efficient greedy algorithm achieve κ-coRating. then demonstrate our approach empirically.