DWReCO at CheckThat! 2023: Enhancing Subjectivity Detection through Style-based Data Sampling

TL;DR

This paper presents a method for improving subjectivity detection by augmenting training data with style-based samples generated by GPT-3, demonstrating effectiveness across multiple languages.

Contribution

The study introduces style-based data sampling using GPT-3 to address class imbalance in subjectivity detection across English, German, and Turkish.

Findings

Style-based oversampling outperforms paraphrasing in English and Turkish.

Generated style-based data improves model performance across all three languages.

GPT-3's style-based generation has limitations in non-English languages.

Abstract

This paper describes our submission for the subjectivity detection task at the CheckThat! Lab. To tackle class imbalances in the task, we have generated additional training materials with GPT-3 models using prompts of different styles from a subjectivity checklist based on journalistic perspective. We used the extended training set to fine-tune language-specific transformer models. Our experiments in English, German and Turkish demonstrate that different subjective styles are effective across all languages. In addition, we observe that the style-based oversampling is better than paraphrasing in Turkish and English. Lastly, the GPT-3 models sometimes produce lacklustre results when generating style-based texts in non-English languages.