About the adjacency matrix

[MTandHJ]

Hi,

I want to know why the adjacency matrix is normalized by

        self.adjacency = adj.multiply(1.0/adj.sum(axis=0).reshape(1, -1))

rather than

        self.adjacency = adj.multiply(1.0/adj.sum(axis=1).reshape(-1, 1))

following the Equation (1).

Thanks !