Classifiers and text mining: application to a specific context

Abstract

[Abstract]: The constant growth of social networks has not only brought us new ways of interacting with each other, but has also given way to a severe increase in negative behaviors: hate speech, racism, gender harassment, cyberbullying, etc. Manually trying to detect this kind of behaviours in millions of daily social media posts is out of the question. The solution lies in developing intelligent systems to automate such detection tasks. As the nature of these texts is completely subjective, this problem falls under the field of sentiment analysis, which aims to systematically identify and study affective states and subjective information in textual data using natural language processing techniques. In particular, this project is focused on the research of different machine learning techniques related to natural language processing, in order to automate and perform a reliable detection and classification of sexist-related behaviours in social media texts. We will tackle the task of adequately processing the extracted data from social media, as well as researching various text classification techniques and models that we will use to develop and evaluate a variety of classifiers.