Identifying epidemic related Tweets using noisy learning

TL;DR

This paper explores using noisy learning techniques to automatically generate weak supervision signals for identifying epidemic-related tweets, reducing manual annotation effort and achieving high classification performance.

Contribution

It introduces a method to leverage noisy learning for epidemic tweet classification, bypassing manual annotation and demonstrating high accuracy in a complex setting.

Findings

Models trained with noisy data achieved over 90% performance.

Weak supervision effectively handles class imbalance and multi-classification.

Noisy learning reduces annotation effort while maintaining high accuracy.

Abstract

Supervised learning algorithms are heavily reliant on annotated datasets to train machine learning models. However, the curation of the annotated datasets is laborious and time consuming due to the manual effort involved and has become a huge bottleneck in supervised learning. In this work, we apply the theory of noisy learning to generate weak supervision signals instead of manual annotation. We curate a noisy labeled dataset using a labeling heuristic to identify epidemic related tweets. We evaluated the performance using a large epidemic corpus and our results demonstrate that models trained with noisy data in a class imbalanced and multi-classification weak supervision setting achieved performance greater than 90%.