@Article{Elmisery/Rho/Sertovic/Boudaoud/Seo/2017/Springer,
AUTHOR = {Elmisery, A.M. and Rho, S. and Sertovic, Mirela and Boudaoud, K. and Seo, Sanghyun},
TITLE = {Privacy-Aware Group Based Recommender System in Multimedia Services.},
YEAR = {2017},
MONTH = {July},
JOURNAL = {Springer International Journal of Multimedia Tools and Applications},
PAGES = {1-25},
URL = {https://link.springer.com/article/10.1007/s11042-017-4950-0},
ABSTRACT = {Recommending similar-interest users{\^a}€™ groups in multimedia services is the problem of detecting for each registered user his/her membership to one interest-group of relevant consumers. The consumers in each interest-group share some relevant preferences which guarantee that the interest-group as a whole satisfies some desired properties of similarity. As a result, forming these interest-groups requires the availability of personal data of different consumers. This is a crucial requirement for different recommender systems. With the increasing trend of service providers to collect a large volume of personal data regarding their end-users, presumably to better serve them. However, a significant part of the data that is typically collected is not essential to the service being offered, or to the completion of the services it was presumably released for. Gathering such unnecessary data can be seen as a privacy threat, and storing it exposes the end-users to further unavoidable risks. In this paper, a privacy aware group based recommender system is proposed for the automated discovery of appropriate interest groups in multimedia services. A fog based middleware (FMCP) was introduced that runs at end-users{\^a}€™ sides and allows exchanging of their information to facilities recommending and creating interest-groups without disclosing their real preferences to other consumers. The membership of consumers in various interest groups allows receiving highly appropriate and reliable multimedia content-related advices. The system utilizes two protocols to attain this goal. Experiments were performed on real dataset.},
NOTE = {ISI}
}
|
