Learning from What Others Know: Privacy Preserving Cross System Personalization

摘要: Recommender systems have been steadily gaining popularity and deployed by several service providers. Large scalable deployment has however highlighted one of the design problems recommender systems: lack interoperability. Users today often use multiple electronic offering recommendations, which cannot learn from another. The result is that end user to provide similar information in some cases disjoint information. Intuitively, it seems much can be improved with this situation: learnt system could potentially reused another, offer an overall personalization experience. In paper, we effective solution problem using Latent Semantic Models learning a model across systems. A privacy preserving distributed framework added around traditional Probabilistic Analysis framework, practical aspects such as addition new items are also dealt work.