A flexible sampling method for large-scale online social networks : Self-adjustable random walk (SARW).

Research output: Conference Papers (RGC: 31A, 31B, 32, 33)32_Refereed conference paper (no ISBN/ISSN)Not applicablepeer-review

View graph of relations

Author(s)

Related Research Unit(s)

Detail(s)

Original languageEnglish
Publication statusPublished - 21 May 2013

Conference

Title33th annual conference of the International Network for Social Network Analysis (Sunbelt 2013)
PlaceGermany
CityHamburg
Period21 - 26 May 2013

Abstract

Online social networks (OSNs) have become an increasingly attractive gold mine for academic and commercial researchers. However, research on OSNs faces a number of difficult challenges. One bottleneck lies in the massive quantity and often unavailability of OSN population data. Sampling becomes perhaps the only feasible solution to the problems. How to draws samples that can represent the underlying OSNs has remained a formidable task because of a number of conceptual and methodological reasons. On the one hand, the existing theories and methods of probability sampling are not practical for sampling of OSNs. On the other hand, most of empirically-driven studies on network sampling are confined to simulated data or sub-graph data, which are fundamentally different from real and complete-graph OSNs. Therefore, no single sampling method has been able to generate unbiased samples of OSNs with maximal precision and minimal cost. In the current study, we propose a flexible sampling method, called Self-Adjustable Random Walk (SARW), and test it against with the population data of a real, large-scale OSN. We evaluate the strengths of the sampling method in comparison with four prevailing methods, including uniform, breadth-first search (BFS), random walk (RW), and revised RW (i.e., MHRW) sampling.

Citation Format(s)

A flexible sampling method for large-scale online social networks : Self-adjustable random walk (SARW). / ZHU, Jonathan J. H.; XU, Xiaoke; ZHANG, Lun; PENG, Tai-Quan Winson.

2013. Paper presented at 33th annual conference of the International Network for Social Network Analysis (Sunbelt 2013), Hamburg, Germany.

Research output: Conference Papers (RGC: 31A, 31B, 32, 33)32_Refereed conference paper (no ISBN/ISSN)Not applicablepeer-review