LT4SG@SMM4H24: Tweets Classification for Digital Epidemiology of Childhood Health Outcomes Using Pre-Trained Language Models

TL;DR

This paper explores the use of pre-trained language models, specifically RoBERTa-large and BERTweet-large ensembles, for classifying tweets reporting children's medical disorders to support digital epidemiology.

Contribution

It introduces an ensemble approach with BERTweet-large that outperforms a single RoBERTa-large model in classifying relevant tweets.

Findings

BERTweet-large ensemble achieves an F1-score of 0.938.

Ensembling improves test performance over individual models.

The approach outperforms the benchmark classifier by 1.18%.

Abstract

This paper presents our approaches for the SMM4H24 Shared Task 5 on the binary classification of English tweets reporting children's medical disorders. Our first approach involves fine-tuning a single RoBERTa-large model, while the second approach entails ensembling the results of three fine-tuned BERTweet-large models. We demonstrate that although both approaches exhibit identical performance on validation data, the BERTweet-large ensemble excels on test data. Our best-performing system achieves an F1-score of 0.938 on test data, outperforming the benchmark classifier by 1.18%.