In search of a gold standard in studies of deception

PDF

Abstract

In this study, we explore several popular techniques for obtaining corpora for deception research. Through a survey of traditional as well as non-gold standard creation approaches, we identify advantages and limitations of these techniques for web-based deception detection and offer crowd-sourcing as a novel avenue toward achieving a gold standard corpus. Through an in-depth case study of online hotel reviews, we demonstrate the implementation of this crowdsourcing technique and illustrate its applicability to a broad array of online reviews.