MSA-Thinker: Discrimination-Calibration Reasoning with Hint-Guided Reinforcement Learning for Multimodal Sentiment Analysis

Build a multimodal sentiment analyzer using hint-guided reinforcement learning to combine accuracy with explainability.

Suggested repo: thinker-sentiment

"Explainable multimodal sentiment analysis via hint-guided reinforcement learning."

Estimated effort: 80h