Graph sampling techniques are used to get a
smaller but representative snapshot of social
graphs, which preserves properties such as
degree distribution. As shown in [4], the sampling
result of Breadth-First Sampling (BFS) and
Random Walk (RW) are biased toward highdegree
vertices, although they have been widely
used in social graph analysis. The Metropolis-
Hasting RW (MHRW) and a Re-Weighted RW
(RWRW) are proposed and proved to perform
uniformly in sampling Facebook. The article also
introduces online convergence diagnostics to
assess sample quality during the sampling process.
Frontier Sampling (FS) [5], which leverages
multidimensional RW, is proposed to achieve
lower estimation errors than RW, especially in
the presence of disconnected or loosely connected
graphs. Ribeiro et al. [5] show that FS is more
suitable for estimating the tail of degree distribution
than random vertex sampling. Moreover,
FS can be made fully distributed without any
coordination costs.