GUISE: a uniform sampler for constructing frequency histogram of graphlets

• Published in: Knowledge and information systems Vol. 38; no. 3; pp. 511 - 536
• Main Authors: Rahman, Mahmudur, Bhuiyan, Mansurul Alam, Rahman, Mahmuda, Hasan, Mohammad Al
• Format: Journal Article
• Language: English
• Published: London Springer London 01.03.2014; Springer; Springer Nature B.V
• Subjects: Analysis; Applied sciences; Approximation; Computation; Computer Science; Computer science; control theory; systems; Computer systems and distributed systems. User interface; Construction; Counting; Data Mining and Knowledge Discovery; Data processing. List processing. Character string processing; Database Management; Exact sciences and technology; Frequency distribution; Graph theory; Graphs; Information retrieval. Graph; Information Storage and Retrieval; Information systems; Information Systems and Communication Service; Information Systems Applications (incl.Internet); IT in Business; Memory organisation. Data processing; Networks; Regular Paper; Samplers; Social networks; Social sciences; Software; Studies; Tasks; Theoretical computing; MCMC sampling; Graph analysis; Graphlet sampling; Subgraph concentration; Uniform sampling; Graphlet degree distribution; Graphlet counting; Graph mining; Monte Carlo method; Histogram; Error rate; Graph theory; Distributed system; Data mining; Markov chain; Boarded computer; Sampler; Subgraph; Sampling; Counting
• ISSN: 0219-1377; 0219-3116
• DOI: 10.1007/s10115-013-0673-3

Graphlet frequency distribution (GFD) has recently become popular for characterizing large networks. However, the computation of GFD for a network requires the exact count of embedded graphlets in that network, which is a computationally expensive task. As a result, it is practically infeasible to compute the GFD for even a moderately large network. In this paper, we propose Guise , which uses a Markov Chain Monte Carlo sampling method for constructing the approximate GFD of a large network. Our experiments on networks with millions of nodes show that Guise obtains the GFD with very low rate of error within few minutes, whereas the exhaustive counting-based approach takes several days.