Enhancing Sentiment Analysis Accuracy Amidst Sarcasm Challenges with Aspect-based Machine Learning for Detection

Abstract

This research aims to enhance sentiment analysis accuracy by addressing the challenges posed by sarcasm through the detection and interpretation of aspectual negation and amplification indicators. Our proposed methodology, SARC-AIM (Sarcasm Aspect Interpretation Model), combines an aspect-based approach with machine learning techniques. It leverages linguistic insights and computational methods to identify aspectual negation and amplification indicators within sarcastic utterances. Through a comprehensive evaluation on benchmark datasets, SARC-AIM demonstrates its efficacy in accurately detecting and interpreting aspectual negation and amplification, thereby improving sentiment analysis performance amidst sarcasm challenges. SARC-AIM introduces a novel approach that integrates linguistic insights with computational methods, offering a unique solution to the challenges of sarcasm detection in sentiment analysis.Keywords: Sarcasm detection, Aspectual negation, Sentiment analysis, Contextual embeddings, Multi-head attention.