TY - JOUR AU - Adomavicius, Gediminas AU - Zhang, Jingjing T1 - Impact of Data Characteristics on Recommender Systems Performance JO - ACM Trans. Manage. Inf. Syst. PY - 2012/04 VL - 3 IS - 1 SP - 3:1 EP - 3:17 UR - http://doi.acm.org/10.1145/2151163.2151166 M3 - 10.1145/2151163.2151166 KW - characteristics KW - dataset KW - dependence KW - evaluation KW - model KW - recommender L1 - SN - N1 - Impact of data characteristics on recommender systems performance N1 - AB - This article investigates the impact of rating data characteristics on the performance of several popular recommendation algorithms, including user-based and item-based collaborative filtering, as well as matrix factorization. We focus on three groups of data characteristics: rating space, rating frequency distribution, and rating value distribution. A sampling procedure was employed to obtain different rating data subsamples with varying characteristics; recommendation algorithms were used to estimate the predictive accuracy for each sample; and linear regression-based models were used to uncover the relationships between data characteristics and recommendation accuracy. Experimental results on multiple rating datasets show the consistent and significant effects of several data characteristics on recommendation accuracy. ER - TY - CONF AU - Narayanan, Arvind AU - Shmatikov, Vitaly A2 - T1 - Robust De-anonymization of Large Sparse Datasets T2 - Proc. of the 29th IEEE Symposium on Security and Privacy PB - IEEE Computer Society CY - PY - 2008/05 M2 - VL - IS - SP - 111 EP - 125 UR - http://www.cs.utexas.edu/~shmat/shmat_oak08netflix.pdf M3 - 10.1109/SP.2008.33 KW - anonymization KW - datamining KW - dataset KW - netflix KW - privacy KW - recommender KW - toread L1 - SN - N1 - N1 - AB - We present a new class of statistical de- anonymization attacks against high-dimensional micro-data, such as individual preferences, recommendations, transaction records and so on. Our techniques are robust to perturbation in the data and tolerate some mistakes in the adversary's background knowledge. We apply our de-anonymization methodology to the Netflix Prize dataset, which contains anonymous movie ratings of 500,000 subscribers of Netflix, the world's largest online movie rental service. We demonstrate that an adversary who knows only a little bit about an individual subscriber can easily identify this subscriber's record in the dataset. Using the Internet Movie Database as the source of background knowledge, we successfully identified the Netflix records of known users, uncovering their apparent political preferences and other potentially sensitive information. ER - TY - CONF AU - Song, Yang AU - Zhang, Lu AU - Giles, C. Lee A2 - T1 - A sparse gaussian processes classification framework for fast tag suggestions T2 - CIKM '08: Proceeding of the 17th ACM conference on Information and knowledge mining PB - ACM CY - New York, NY, USA PY - 2008/ M2 - VL - IS - SP - 93 EP - 102 UR - http://portal.acm.org/citation.cfm?id=1458098 M3 - http://doi.acm.org/10.1145/1458082.1458098 KW - bibsonomy KW - bookmarking KW - classification KW - dataset KW - ml KW - recommender KW - social KW - tag KW - tagging KW - taggingsurvey KW - toread L1 - SN - 978-1-59593-991-3 N1 - A sparse gaussian processes classification framework for fast tag suggestions N1 - AB - ER - TY - GEN AU - Narayanan, Arvind AU - Shmatikov, Vitaly A2 - T1 - How To Break Anonymity of the Netflix Prize Dataset JO - PB - AD - PY - 2006/ VL - IS - SP - EP - UR - http://www.citebase.org/abstract?id=oai:arXiv.org:cs/0610105 M3 - KW - Preis KW - anonymity KW - dataset KW - netflix KW - prize KW - recommender L1 - N1 - [cs/0610105] How To Break Anonymity of the Netflix Prize Dataset N1 - AB - We present a new class of statistical de-anonymization attacks against high-dimensional micro-data, such as individual preferences, recommendations, transaction records and so on. Our techniques are robust to perturbation in the data and tolerate some mistakes in the adversary's background knowledge. ER -