Efficient respondents selection for biased survey using homophily-high social relation graph

Abstract

Online social relationships which can be extracted from various online resources such as online social networks are getting much attention from the research communities since they are rich resources to learn about the members of our society as well as the relationships among them. With the advances of Internet related technologies, online surveys are established as an essential tool for a wide range of applications. One significant issue of online survey is how to select a quality respondent group so that the survey result is reliable. This paper studies the use of pairwise online social relationships among the members of a society to form a biased survey respondent group, which might be useful for various applications. We first introduce a way to construct a homophily-high social relation graph. Then, we introduce the minimum inverse k-core dominating set problem (MIkCDSP), which aims to compute a biased respondent group using the homophily-high social relation graph. We show the problem is NP-hard and most importantly propose a greedy approximation for it. Our simulation based on a real social network shows the proposed algorithm is very effective.