Can we infer the preference of a networked population by sampling a small subset ?

Arriving at an aggregate preference that summarises accurately the preferences of a large population of agents, by gathering individual preferences, is a challenging task. We explore the possibility of harnessing information available on the underlying social network and collecting  preferences from a small sample of representative nodes. Using a Facebook App that we designed and delpoyed, we modeled how preferences are distributed among nodes in a social network. We formulated an an objective function and designed algorithms for selecting the best representative nodes. We discover that selecting representatives using social network information provides an excellent way for aggregating preferences related to personal topics such as lifestyle.  However, for social topics like government policies, simple random polling with a reasonable sample size suffices.

Swapnil Dhamal, Rohith D. Vallam, and Y. Narahari. Modeling spread of preferences in social networks for sampling-based preference aggregation. IEEE Transactions on Network Science and Engineering (TNSE), vol. 6, no. 1, pages 46-59. IEEE, 2019.

Link to the paper: