Table 2 Symbol description.
From: Music recommendation algorithms based on knowledge graph and multi-task feature learning
Symbol | Description |
|---|---|
u | User |
v | Item |
M | Number of users |
N | Number of items |
G | Knowledge graph |
(h,r,t) | Three-tuple |
h, r, t | Head entity, tail entity, relationship |
θ | A parameter of the function f |
Y | A user-item interaction matrix |
W,\({{\text{W}}}_{{\text{l}}}^{{\text{VV}}}\),\({{\text{W}}}_{{\text{l}}}^{{\text{EV}}}\),\({{\text{W}}}_{{\text{l}}}^{{\text{VE}}}\),\({{\text{W}}}_{{\text{l}}}^{{\text{EE}}}\) | Weight |
b,\({{\text{b}}}_{{\text{l}}}^{{\text{V}}}\),\({{\text{b}}}_{{\text{l}}}^{{\text{E}}}\) | Bias |
S(v),S(u) | A set of entities associated with music item item v (user u ) |
\(f_{{R{\text{S}}}}\) | The prediction probability of the recommendation module |
\(\alpha ()\) | The tanh function |
\(\sigma ()\) | The Sigmoid function |
\(\beta ()\) | The softsign function |
\(\delta ()\) | The softplus function |
\(f_{{{\text{KG}}}}\) | The probability prediction of the KGE module |
d | The dimensions of the hidden layer |
\({{\text{C}}}_{{\text{l}}}\) | The cross feature matrix of layer l |
\({{\text{v}}}_{{\text{l}}+1}\),\({{\text{e}}}_{{\text{l}}+1}\) | Item and entity feature vectors for layer l + 1 after cross-compression unit processing |
λ1,λ2 | The equilibrium parameters |
J | The cross-entropy function |
