2009 · 318 citations · 19 references
ComputingInformation SystemsRecommender SystemsStatistical ModelingStatisticsCollaborative FilteringMissing Data ImputationMissing Data MethodsRecommendation SystemsData SparsityMissing At RandomMissing Not At Random AssumptionMissing-not-at-random ModelingRanking With Incomplete InformationRating PredictionStatistical Decision TheoryStatistical Learning Theory
Collaborative filtering assumes missing ratings are random, but the observation process and missing‑data theory show that incorrect assumptions can bias predictions. The study investigates how non‑random missing data affect collaborative ranking and builds on prior findings about prediction. The authors analyze collaborative ranking under non‑random missing data, extending earlier prediction studies. Evidence shows that missing‑at‑random assumptions are violated in real recommender systems.
A fundamental aspect of rating-based recommender systems is the observation process, the process by which users choose the items they rate. Nearly all research on collaborative filtering and recommender systems is founded on the assumption that missing ratings are missing at random. The statistical theory of missing data shows that incorrect assumptions about missing data can lead to biased parameter estimation and prediction. In a recent study, we demonstrated strong evidence for violations of the missing at random condition in a real recommender system. In this paper we present the first study of the effect of non-random missing data on collaborative ranking, and extend our previous results regarding the impact of non-random missing data on collaborative prediction.
19
Item-based collaborative filtering recommendation algorithms
Badrul Sarwar, George Karypis, Joseph A. Konstan et al. · 2001 · 9K citations
Probabilistic Matrix Factorization
Andriy Mnih, Ruslan Salakhutdinov · Neural Information Processing Systems · 2007 · 3.6K citations