Unbiased offline recommender evaluation for missing-not-at-random implicit feedback
Research output: Contribution to journal › Conference article › Research › peer-review
Standard
Unbiased offline recommender evaluation for missing-not-at-random implicit feedback. / Yang, Longqi; Wang, Chenyang; Cui, Yin; Belongie, Serge; Xuan, Yuan; Estrin, Deborah.
In: RecSys 2018 - 12th ACM Conference on Recommender Systems, 27.09.2018, p. 279-287.Research output: Contribution to journal › Conference article › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - Unbiased offline recommender evaluation for missing-not-at-random implicit feedback
AU - Yang, Longqi
AU - Wang, Chenyang
AU - Cui, Yin
AU - Belongie, Serge
AU - Xuan, Yuan
AU - Estrin, Deborah
N1 - Publisher Copyright: © 2018 Association for Computing Machinery.
PY - 2018/9/27
Y1 - 2018/9/27
N2 - Implicit-feedback Recommenders (ImplicitRec) leverage positive only user-item interactions, such as clicks, to learn personalized user preferences. Recommenders are often evaluated and compared offline using datasets collected from online platforms. These platforms are subject to popularity bias (i.e., popular items are more likely to be presented and interacted with), and therefore logged ground truth data are Missing-Not-At-Random (MNAR). As a result, the widely used Average-Over-All (AOA) evaluator is biased toward accurately recommending trendy items. In this paper, we (a) investigate evaluation bias of AOA and (b) develop an unbiased and practical offline evaluator for implicit MNAR datasets using the Inverse-Propensity-Scoring (IPS) technique. Through extensive experiments using four real-world datasets and four widely used algorithms, we show that (a) popularity bias is widely manifested in item presentation and interaction; (b) evaluation bias due to MNAR data pervasively exists in most cases where AOA is used to evaluate ImplicitRec; and (c) the unbiased estimator significantly reduces the AOA evaluation bias by more than 30% in the Yahoo! music dataset in terms of the Mean Absolute Error (MAE).
AB - Implicit-feedback Recommenders (ImplicitRec) leverage positive only user-item interactions, such as clicks, to learn personalized user preferences. Recommenders are often evaluated and compared offline using datasets collected from online platforms. These platforms are subject to popularity bias (i.e., popular items are more likely to be presented and interacted with), and therefore logged ground truth data are Missing-Not-At-Random (MNAR). As a result, the widely used Average-Over-All (AOA) evaluator is biased toward accurately recommending trendy items. In this paper, we (a) investigate evaluation bias of AOA and (b) develop an unbiased and practical offline evaluator for implicit MNAR datasets using the Inverse-Propensity-Scoring (IPS) technique. Through extensive experiments using four real-world datasets and four widely used algorithms, we show that (a) popularity bias is widely manifested in item presentation and interaction; (b) evaluation bias due to MNAR data pervasively exists in most cases where AOA is used to evaluate ImplicitRec; and (c) the unbiased estimator significantly reduces the AOA evaluation bias by more than 30% in the Yahoo! music dataset in terms of the Mean Absolute Error (MAE).
KW - Bias
KW - Evaluation
KW - Implicit feedback
KW - Propensity
KW - Recommendation
UR - http://www.scopus.com/inward/record.url?scp=85056795110&partnerID=8YFLogxK
U2 - 10.1145/3240323.3240355
DO - 10.1145/3240323.3240355
M3 - Conference article
AN - SCOPUS:85056795110
SP - 279
EP - 287
JO - RecSys 2018 - 12th ACM Conference on Recommender Systems
JF - RecSys 2018 - 12th ACM Conference on Recommender Systems
T2 - 12th ACM Conference on Recommender Systems, RecSys 2018
Y2 - 2 October 2018 through 7 October 2018
ER -
ID: 301825505