Unsupervised and scalable algorithm for learning node representations

authors : Tiago Pimentel1, Adriano Veloso1 and Nivio Ziviani1,21) using each node as root once

2) skipgram learn a model by maximizing the log probability

    (1)

 

the probability of the model in this paper

Vt : a vertex in the graph

Vt+1 : another vertex in the graph             : the representation of Vi

P(Vt+1|Vt , r)  can be written as 

   (2)  

 : the transposed output feature vector of vertex j     used to make prediction

  are learned simultaneously by optimizing (1)