Negative-Aware Collaborative Filtering Sheng-Chieh Lin Academia Sinica 0 1 2 Taiwan jacklin_ 0 1 2 @citi.sinica.edu.tw 0 1 2 Chuan-Ju Wang Academia Sinica 0 1 2 Taiwan cjwang@citi.sinica.edu.tw 0 1 2 Ming-Feng Tsai National Chengchi University , Taiwan Sheng-Fang Yang National Chengchi University , Taiwan Yu-Neng Chuang National Chengchi University , Taiwan 2019

-

observed association unseen association INTRODUCTION

N METHODOLOGY 1htps://grouplens.org/datasets/movielens/ 2htp://www.wanghao.in/data/ctrsr_datasets. rar

Dataset Movielens MF 0.237 0.169 0.199

N LBPR = − Õ log u,(i,i′)

1 nui′

⊺ ⊺ log σ (θu θi − θu θi′) + λ ∥Θ∥22 .

(2)

Note that as the proposed negative-aware matrix is independent of the recommendation models, it can be seen as a generic device applicable to other recommendation algorithms with the use of negative sampling.