Topic popularity in a heterogeneous population of online social network users

Abstract

The popularity of a topic in a social network can be estimated from the number of posts about it. We investigate the effect of assuming homogeneity in a heterogenous population of users on the measured popularity. We found that having either equal users per tweets or equal topic popularity across the subpopulations would make the measured popularity accurate even if the population is still heterogenous. Guided by these special cases, we propose methods for accurately measuring topic popularity in a heterogeneous population of online social network users.