RQ1
How should the degree of negativity be inferred in DIVINE?
(1) DIVINE-I
(1) DIVINE-I
Comparisons of several methods for inferring the degree of negativity. When it is equipped with WRMF, DIVINE-I consistently achieves high AUC in all datasets.
(a) U-LP task
(a) U-LP task
(b) M-LP task
(b) M-LP task
(c) B-LP task
(c) B-LP task
(2) DIVINE-T
(2) DIVINE-T
(a) U-LP task
(a) U-LP task
(b) M-LP task
(b) M-LP task
(c) B-LP task
(c) B-LP task