HIGHLIGHTS
SUMMARY
From timeSVD++, factorization machine, to the neural_network with attention-based novel RS models, recommendation methods keep including more forms of sensitive user data. While the users enjoying the convenience brought by a recommender system, it collects sensitive data such as private user profiles or interaction records from personal devices and stores them in a centralized server. In contrast, the k-anonymity based-RS, which keeps the central server, hides user identities and broadens certain sensitive information. The authors make use of clustering methods to replace users with centroids based on users` feedback on their . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.