Skip to main content

Table 1 Summary of the data manipulation and the desired effect on the graph data

From: The interplay between communities and homophily in semi-supervised classification using graph neural networks

Manipulation

Homophily

Mixing

Side-effects

\(Hom^+\)

Increased

Preserved

Binomial degree distribution

\(Hom^-\)

Decreased

Increased

Destroys community structure

\(Mix^-\)

Preserved(*)

Decreased

Binomial degree distribution

\(Mix^+\)

Preserved

Increased

Destroys sub-communities

  1. (*)\(Mix^-\) should result in a “homophilic” graph if the original was “homophilic”, and vice versa