Automatic Misogyny Detection in Social Media: A Survey

Abstract

This article presents a survey of automated misogyny identification techniques in social media, especially in Twitter. This problem is urgent because of the high speed at which messages on social platforms grow and the widespread use of offensive language (including misogynistic language) in them. In this article we survey approaches proposed in the literature to solve the problem of misogynistic message recognition. These include classical machine learning models like Sup-port Vector Machine, Naive Bayes, Logistic Regression and ensembles of different classical machine learning models and deep neural networks such as Long Short-term memory and Convolutional Neural Networks. We consider results of experiments with these models in different languages: English, Spanish and Italian tweets. The survey describes some features which help to identify misogynistic tweets and some challenges which aim was to create misogyny language classifiers. The survey includes not only models which help to identify misogyny language, but also systems which help to recognize a target of an offense (an individual or a group of persons).