A Generic NLI approach for Classification of Sentiment Associated with Therapies

TL;DR

This paper presents a novel approach using Natural Language Inference with transformer models to classify sentiment related to therapies, achieving significant improvements in F1-score over previous methods.

Contribution

It introduces an NLI-based framework for aspect-oriented sentiment classification in the medical domain, demonstrating its effectiveness on shared task data.

Findings

Best model achieved 75.22% F1-score

Outperformed mean and median team scores by 11% and 4%

Transformer-based NLI approach is effective for therapy sentiment classification

Abstract

This paper describes our system for addressing SMM4H 2023 Shared Task 2 on "Classification of sentiment associated with therapies (aspect-oriented)". In our work, we adopt an approach based on Natural language inference (NLI) to formulate this task as a sentence pair classification problem, and train transformer models to predict sentiment associated with a therapy on a given text. Our best model achieved 75.22\% F1-score which was 11\% (4\%) more than the mean (median) score of all teams' submissions.