In recommendation network, the item vertexes have only context vectors ??

[GL123456]

When saw the embedded expression of the recommendation network, I was a little confused. As far as I know, LINE is an embedded representation for homogenous graphs,while the recommendation network is a Heterogeneous figure.So i would like to know how is a item embedding. Hope to be inspired by you, sincerely.