SentiAgent: An agentic LLM framework for context-aware and explainable sentiment analysis.

Abstract

Sentiment analysis has become a fundamental task in Natural Language Processing (NLP) due to the rapid growth of user-generated content across social media, e-commerce, healthcare, education, and customer feedback platforms. Despite the success of transformer-based models and Large Language Models (LLMs), existing approaches often struggle with contextual ambiguity, implicit sentiment, sarcasm, commonsense reasoning, and limited explainability. To address these challenges, this paper proposes SentiAgent, a novel Agentic LLM framework for context-aware and explainable sentiment analysis. The framework employs multiple autonomous agents that collaboratively perform contextual understanding, external knowledge integration, sentiment reasoning, and explanation generation. Extensive experiments on four benchmark datasets-SST-2, IMDB Reviews, Sentiment140, and Amazon Product Reviews-demonstrate the effectiveness of the proposed framework. SentiAgent achieved 97.43% accuracy, 97.18% precision, 96.95% recall, 97.06% F1-score, and 0.99 AUC, consistently outperforming conventional machine learning, deep learning, transformer-based, and standalone LLM models. Ablation studies further validate the contribution of each autonomous agent, while qualitative analyses highlight the framework's ability to produce transparent and interpretable sentiment predictions. The proposed framework provides a scalable and trustworthy solution for next-generation explainable sentiment analysis.•Introduces SentiAgent, a collaborative Agentic LLM framework for context-aware and explainable sentiment analysis.•Combines contextual understanding, external commonsense knowledge, and autonomous reasoning to improve sentiment prediction and interpretability.•Achieves state-of-the-art performance on four benchmark datasets, demonstrating superior accuracy, robustness, and transparency over existing approaches.