Surprisingly popular voting

FAIR Lab, Penn State, 2023 to 2024

Advised by Dr. Hadi Hosseini; mentored by and in collaboration with Dr. Debmalya Mandal.

When a crowd votes, the majority can be wrong if most voters are non-experts. Surprisingly popular (SP) voting asks voters both for their own preference and for their prediction of how others will vote. That lets it find the answer that is more popular than people expected. This project designed mechanisms that implement SP voting with partial preference profiles, which makes it practical to recover the true ranking over many alternatives from mixed expert and non-expert populations. A follow-up paper analyzed SP voting under concentric rank-order models, and SP-Rank released a public dataset of ranked preferences with secondary information for benchmarking rank aggregation. This work formed my MS thesis, which received Penn State’s Best MS Thesis Award in Artificial Intelligence.