Pseudo-labelling Enhanced Media Bias Detection

TL;DR

This paper introduces a pseudo-labelling data augmentation technique that enhances media bias detection accuracy by effectively utilizing noisy, unlabeled datasets through weak supervision.

Contribution

It presents a novel pseudo-labelling based method for improving media bias detection using noisy distant supervision data.

Findings

Improved accuracy in biased news detection models

Effective utilization of noisy unlabeled data

Simple yet impactful data augmentation approach

Abstract

Leveraging unlabelled data through weak or distant supervision is a compelling approach to developing more effective text classification models. This paper proposes a simple but effective data augmentation method, which leverages the idea of pseudo-labelling to select samples from noisy distant supervision annotation datasets. The result shows that the proposed method improves the accuracy of biased news detection models.