Issue No.03 - May/June (2007 vol.22)
Neil J. Hurley , University College Dublin
Michael P. O'Mahony , University College Dublin
Guenole C.M. Silvestre , University College Dublin
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MIS.2007.44
Collaborative recommender systems are vulnerable to attacks that seek to manipulate recommendations made for target items. The authors examine such attacks from a cost perspective, focusing on the effect that attack size—in terms of the number of ratings inserted during an attack—has on attack success. They present a cost-benefit analysis that shows that attackers can realize profits, even when financial costs are imposed on the insertion of ratings.
recommender systems, cost-benefit analysis
Neil J. Hurley, Michael P. O'Mahony, Guenole C.M. Silvestre, "Attacking Recommender Systems: A Cost-Benefit Analysis", IEEE Intelligent Systems, vol.22, no. 3, pp. 64-68, May/June 2007, doi:10.1109/MIS.2007.44