Method and system of privacy protection recommendation based on matrix factorization and ridge regression

摘要

Method and system for generating a privacy protection recommendations from the first set of users, begin by receiving the first of the record set that contains the token and items as input; matrix factorization with respect to the first set of records the first of garbled circuit design and evaluation on the basis of, the second garbled circuit based on the ridge regression designed and evaluated with respect to the second record from the requesting user, and privacy protection, the recommendation related to at least one of the items produced, the record and the party information is in addition to the source to be extracted from it is placed in the secret, the recommendation is known only by the requesting user. The system includes a cryptographic service provider to design a garbled Circuit, also recommender for evaluating the first circuit. Requesting user evaluates the second circuit, not included in the first set of users.