link prediction in knowledge graph embedding

[afrooze]

i need to do link prediction in knowledge graph embedding. I must be doing the link prediction by Bayesian estimation.
can you guide me how can do it?
at least, how many relations must be in the train step?
I must compute a similarity matrix for head and tail too, can you introduce me any functions for it.