Table 2 Statistics hyper-parameter settings for the datasets.
From: Knowledge graph convolutional networks with user preferences for course recommendation
Hyper-parameter settings | MOOCCube | Book-Crossing | MovieLens-1 M |
|---|---|---|---|
h (hop numbers) | 2 | 3 | 2 |
n (neighbor sampling size) | 8 | 8 | 8 |
s (the size of each triplet) | 32 | 64 | 32 |
\(\:{\lambda\:}_{1}\)(kge weight) | 10−2 | 10−2 | 10−2 |
\(\:{\lambda\:}_{2}\:\)(regularizer weight) | 10−4 | 2 × 10−5 | 10−4 |
\(\:\eta\:\:\)(learning rate) | 10−4 | 10−4 | 10−4 |
