Parallel Clustered Low-Rank Approximation of Graphs and Its Application to Link Prediction

Abstract

Abstract. Social network analysis has become a major research area that has impact in diverse applications ranging from search engines to product recommendation systems. A major problem in implementing social network analysis algorithms is the sheer size of many social networks, for example, the Facebook graph has more than 900 million vertices and even small networks may have tens of millions of vertices. One solution to dealing with these large graphs is dimensionality reduction using spectral or SVD analysis of the adjacency matrix of the network, but these global techniques do not necessarily take into account local structures or clusters of the network that are critical in network analysis. A more promising approach is clustered low-rank approximation: instead of computing a global low-rank approximation, the adjacency matrix is first clustered, and then a low-rank approximation of each cluster (i.e., diagonal block) is computed. The resulting algorithm is challenging to parallelize not only because of the large size of the data sets in social network analysis, but also because it requires computing with very diverse data structures ranging from extremely sparse matrices to dense matrices. In this paper, we describe the first parallel implementation of a clustered low-rank approximation algorithm for large social network graphs, and use it to perform link prediction in parallel. Experimental results show that this implementation scales well on large distributed-memory machines; for example, on a Twitter graph with roughly 11 million vertices and 63 million edges, our implementation scales by a factor of 86 on 128 processes and takes less than 2300 seconds, while on a much larger Twitter graph with 41 million vertices and 1.2 billion edges, our implementation scales by a factor of 203 on 256 processes with a running time about 4800 seconds.

Heat kernel based community detection

“…Many of these studies involve fast ways to approximate the entire matrix exponential, instead of a single column as we study here. For instance, Sui et al [51] describe a low-parameter decomposition of a network that is useful both for estimating Katz scores [25] and the matrix exponential. Orecchia and Mahoney [46] show that the heat kernel diffusion implicitly approximates a diffusion operator using a particular type of generalized entropy, which provides a principled rationale for its use.…”

Scalable and Memory-Efficient Clustering of Large-Scale Social Networks

“…Sui et al [16] presented the first implementation of a clustered low-rank approximation algorithm for large social network graphs, and its application to link prediction. As one part of their parallel clustered low-rank approximation algorithm, they developed a parallel clustering algorithm, called PEK.…”

A Study of Clustering Techniques and Hierarchical Matrix Formats for Kernel Ridge Regression

“…Some of the previous algorithms using this idea are the following. Clustered Low-Rank Approximation (CLRA, [16] and its parallel version [21]) was created to process large graphs. It starts with the clustering of the adjacency matrix, and then computes a low-rank approximation of each cluster (i.e., diagonal block), e.g.…”