Authorship Attribution vs. Adversarial Authorship from a LIWC and Sentiment Analysis Perspective

Although Stylometry has been effectively used for Authorship Attribution, thereis a growing number of methods being developed that allow authors to mask theiridentity. In this paper, we investigate the usage of non-traditional featuresets for Authorship Attribution. By using non-traditional feature sets, one maybe able to reveal the identity of adversarial authors who are attempting toevade detection from Authorship Attribution systems that are based on moretraditional feature sets. In addition, we demonstrate how GEFeS (Genetic &Evolutionary Feature Selection) can be used to evolve high-performance hybridfeature sets composed of two non-traditional feature sets for AuthorshipAttribution: LIWC (Linguistic Inquiry & Word Count) and Sentiment Analysis.These hybrids were able to reduce the Adversarial Effectiveness on anadversarial test set by approximately 33.4%.