Document Type

Dissertation

Rights

Available under a Creative Commons Attribution Non-Commercial Share Alike 4.0 International Licence

Disciplines

1.2 COMPUTER AND INFORMATION SCIENCE

Publication Details

A Thesis Presented For the Award of Master of Computer Science

Abstract

The thesis is devoted to the problem of misogyny detection in social media. In the work we analyse the difference between all offensive language and misogyny language in social media, and review the best existing approaches to detect offensive and misogynistic language, which are based on classical machine learning and neural networks. We also review recent shared tasks aimed to detect misogyny in social media, several of which we have participated in. We propose an approach to the detection and classification of misogyny in texts, based on the construction of an ensemble of models of classical machine learning: Logistic Regression, Naive Bayes, Support Vectors Machines. Also, at the preprocessing stage we used some linguistic features, and novel approaches which allow us to improve the quality of classification. We tested the model on the real datasets both English and multilingual corpora. The results we achieved with our model are highly competitive in this area and demonstrate the capability for future improvement.

DOI

https://doi.org/10.21427/d1jc-vj32

Download


Share

COinS