Improving Judgment Reliability in Social Networks via Jury Theorems

Paolo Galeazzi, Rasmus K. Rendsvig, Marija Slavkovik

Abstract

Opinion aggregators—such as ‘like’ or ‘retweet’ counters—are ubiquitous on social media platforms and often treated as implicit quality evaluations of the entry liked or retweeted, with higher counts indicating higher quality. Many such aggregators are poor quality evaluators as they allow disruptions of the conditions for positive wisdom-of-the-crowds effects. This paper proposes a design of theoretically justified aggregators that improve judgment reliability. Interpreting states of diffusion processes on social networks as implicit voting scenarios, we specify procedures for isolating sets of independent voters in order to use jury theorems to quantify the reliability of network states as quality evaluators. As real-world networks tend to grow very large and independence tests are computationally expensive, a primary goal is to limit the number of such tests. We consider five procedures, each trading a degree of reliability for efficiency, the most efficient requiring a low-degree polynomial number of tests.

OriginalsprogEngelsk
TitelLogic, Rationality, and Interaction. LORI 2019.
Antal sider14
Vol/bind11813
ForlagSpringer
Publikationsdato2019
Sider230-243
ISBN (Trykt)9783662602911
ISBN (Elektronisk)9783662602928
DOI
StatusUdgivet - 2019
NavnLecture Notes in Computer Science
ISSN0302-9743

Fingeraftryk

Dyk ned i forskningsemnerne om 'Improving Judgment Reliability in Social Networks via Jury Theorems'. Sammen danner de et unikt fingeraftryk.

Citationsformater