Optimizing Sarcasm Detection Through Various Input Methods in Text Data

This paper explores the intricacies of sarcasm detection using a BERT-based model, focusing on how different input methods impact detection accuracy. Despite limitations posed by a small-sized dataset, the study fine-tunes BERT to reveal valuable insights into its performance and adaptability in sarcasm detection tasks. By experimenting with various data input formats and configurations, the research highlights the strengths and weaknesses of each method, contributing to the development of more robust and accurate sarcasm detection systems in natural language processing. The findings serve as a groundwork for future research, emphasizing the importance of input methods in optimizing BERT for sarcasm detection.