Chat Mining for Gender Prediction

Abstract

Abstract. The aim of this paper is to investigate the feasibility of predicting the gender of a text document's author using linguistic evidence. For this purpose, term-and style-based classification techniques are evaluated over a large collection of chat messages. Prediction accuracies up to 84.2% are achieved, illustrating the applicability of these techniques to gender prediction. Moreover, the reverse problem is exploited, and the effect of gender on the writing style is discussed.

Authorship Attribution

“…Other studies of this nature include [5,29,95] with largely similar results, although on different languages, genres, and methods. For example, [95] studied Internet chat logs (e.g., IRC, IM, ICQ, and such) in Turkish using a feature set most notable for its inclusion of smileys and a variety of classifiers including neural networks, k-nearest neighbor, and naive Bayesian analysis.…”

“…For example, [95] studied Internet chat logs (e.g., IRC, IM, ICQ, and such) in Turkish using a feature set most notable for its inclusion of smileys and a variety of classifiers including neural networks, k-nearest neighbor, and naive Bayesian analysis. The results from Baysian analysis were again in the 80% range, with other methods slightly or significantly less.…”

“…On the other hand, the 90%+ accuracy that most AAAC participants managed to achieve on the Paston letters are probably enough to cause investigators to start focusing on Margaret instead of John. Even a suggestion that there is an 80% likelihood that the culprit was female [89,95] is probably a useful finding.…”

Automatic Classification and Response of E-mails

“…A wide range of learning methods has been applied to this purpose, such as k-Nearest Neighbor [10], Naive Bayes [11], Support Vector Machines [12], Voting [3], centroid classifier [9], etc.…”

Language-independent Gender Prediction on Twitter

“…Most existing work on predicting gender focuses on exploiting the linguistic production of the users (Koppel et al, 2003;Schler et al, 2006;Kucukyilmaz et al, 2006;Burger et al, 2011;Miller et al, 2012;Rangel et al, 2016), just rarely using nonlinguistic information such as metadata (Plank and Hovy, 2015) or visual information (Alowibdi et al, 2013).…”