From Prediction to Justification: Aligning Sentiment Reasoning with Human Rationale via Reinforcement Learning

2026-04-14

AI Summary

This work addresses the limited human-like explicit reasoning capability in existing aspect-based sentiment analysis (ABSA) methods, which often fail to provide causal explanations for sentiment predictions. To bridge this gap, the authors propose ABSA-R1, a framework that emulates the human cognitive process of “reasoning before predicting” by generating natural language rationales aligned with sentiment labels through reinforcement learning. The approach innovatively integrates a cognition-aligned reward model and a metacognitive monitoring-based rejection sampling strategy, enabling the model to focus on challenging instances under uncertainty and thereby enhancing reasoning consistency. Evaluated on four benchmark datasets, ABSA-R1 consistently outperforms non-reasoning baselines in both sentiment classification and aspect-sentiment triplet extraction, while significantly improving model interpretability.

Technology Category

• Knowledge Representation and Reasoning: Argumentation
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Reasoning under Uncertainty: Causality

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

While Aspect-based Sentiment Analysis (ABSA) systems have achieved high accuracy in identifying sentiment polarities, they often operate as "black boxes," lacking the explicit reasoning capabilities characteristic of human affective cognition. Humans do not merely categorize sentiment; they construct causal explanations for their judgments. To bridge this gap, we propose ABSA-R1, a large language model framework designed to mimic this ``reason-before-predict" cognitive process. By leveraging reinforcement learning (RL), ABSA-R1 learns to articulate the why behind the what, generating natural language justifications that ground its sentiment predictions. We introduce a Cognition-Aligned Reward Model (formerly sentiment-aware reward model) that enforces consistency between the generated reasoning path and the final emotional label. Furthermore, inspired by metacognitive monitoring, we implement a performance-driven rejection sampling strategy that selectively targets hard cases where the model's internal reasoning is uncertain or inconsistent. Experimental results on four benchmarks demonstrate that equipping models with this explicit reasoning capability not only enhances interpretability but also yields superior performance in sentiment classification and triplet extraction compared to non-reasoning baselines.

Problem

• Aspect-based Sentiment Analysis
• Explainable AI
• Sentiment Reasoning
• Human Rationale
• Interpretability

Innovation

• reinforcement learning
• aspect-based sentiment analysis
• natural language justification
• cognition-aligned reward model
• reasoning-before-prediction