Since the edge prior is based only on degree, it is unsurprising that it exhibits weak performance in predicting a network with a different degree distribution.
We have considered the edge prior as a baseline edge predictor, whose performance indicates the utility of degree for a specific prediction task.
The edge prior's low performance in the third task indicates that degree is less helpful for edge prediction tasks in which training and testing networks do not share their degree distributions.
Moreover, we believe such between-distribution prediction may be a relatively common task, with examples given by the networks in Figure {@fig:degree-bias}.
I feel like this paragraph could use some tuning. It would help to give the take-home message more clearly to aid readers. I think that it's there, but it's hidden a bit between the lines.
I feel like this paragraph could use some tuning. It would help to give the take-home message more clearly to aid readers. I think that it's there, but it's hidden a bit between the lines.