We present exact and approximate MapReduce estimators for the number of cliques of size k in an undirected graph, for any small constant k >= 3. Besides theoretically analyzing our algorithms in the computational model for MapReduce introduced by Karloff, Suri, and Vassilvitskii, we present the results of extensive computational experiments on the Amazon EC2 platform. Our experiments show the practical effectiveness of our algorithms even on clusters of small/medium size, and suggest their scalability to larger clusters.
View on arXiv