Leveraging Sentiment for Offensive Text Classification

TL;DR

This paper investigates whether incorporating sentiment analysis improves offensive text classification, demonstrating that sentiment features enhance model performance on the OLID dataset.

Contribution

It introduces a method of augmenting offensive text classifiers with sentiment predictions, showing improved accuracy over baseline models.

Findings

Sentiment features boost classification accuracy.

Pre-trained sentiment models improve offensive text detection.

Sentiment augmentation outperforms baseline models.

Abstract

In this paper, we conduct experiment to analyze whether models can classify offensive texts better with the help of sentiment. We conduct this experiment on the SemEval 2019 task 6, OLID, dataset. First, we utilize pre-trained language models to predict the sentiment of each instance. Later we pick the model that achieved the best performance on the OLID test set, and train it on the augmented OLID set to analyze the performance. Results show that utilizing sentiment increases the overall performance of the model.