“…It generalizes several exploration-based sampling schemes (e.g., Snowball sampling, Forest Fire sampling, graph-expander sampling etc. ), and can be applied to any large graphs due to its flexibility [58]. Specifically, for citation/co-occurrence graphs (e.g., Cora, Citeseer, and Actor), we use the publicly available citation graphs -Pubmed and DBLP -to estimate the average node degree.…”