NTDH: Complex Reasoning for Comprehensive Affective Analysis

2026-08-05

AI Summary

This study addresses the challenges in unified sentiment analysis, particularly the heterogeneity of output formats—spanning continuous, ordinal, and multi-label types—and the difficulty of modeling contextual dependencies while explicitly reconciling conflicting emotional cues. To this end, the work introduces a complex reasoning framework for joint sentiment and emotion analysis, proposing the NTDH synthesis strategy to tackle correctness, fault tolerance, domain adaptability, and target leakage in data synthesis. The approach employs a unified output interface, naturalized label formulation, fault-tolerant gating mechanisms, and directional error prompting, integrated within SFT and GRPO training on Qwen3-8B. Using only 16,302 samples—approximately one-fourteenth the size of baseline datasets—it surpasses SFT on five out of six evaluation metrics, achieving a Pearson correlation coefficient of 0.862 on the EI-reg task.

Abstract

Comprehensive affective analysis is challenging for two reasons: it spans heterogeneous prediction tasks with continuous, ordinal, and multi-label outputs, and affective meaning is context-dependent, requiring conflicting cues to be reconciled rather than mapped directly to labels. Existing methods learn this mapping directly and do not model the reconciliation explicitly. We recast the task as a complex-reasoning problem, which yields one output interface across heterogeneous label spaces and a trajectory over which a verifiable reward can be optimised; to our knowledge, this is the first such treatment covering both sentiment and emotion. The obstacle is on the data side: affective reasoning traces must be synthesised, and generic synthesis is misaligned with the targets, tolerances, and phenomena of affect, and discards or leaks its failure cases. We propose NTDH, which addresses these four failures. Naturalisation sets the training answer to the gold label, so it is correct by construction. A Tolerance-aware gate checks each answer against the task's own scoring margin. Domain-aware strategies refine the reasoning using ideas from affective science. Directional Hints report only the type and direction of an error, without exposing the target. We train Qwen3-8B with SFT and then GRPO under the same tolerance used for verification (up to a more permissive construction gate on the multi-label subtask), and a component ablation quantifies the data-quality effect of each part. Using 16,302 training records, about 14x fewer than comparable instruction-tuned systems, the final policy improves over its SFT checkpoint on five of six official-test metrics and achieves the strongest EI-reg result among the compared systems, at a Pearson correlation of 0.862.

Problem

• affective analysis
• complex reasoning
• heterogeneous outputs
• context-dependent emotion
• conflicting cues

Innovation

• complex reasoning
• affective analysis
• NTDH
• tolerance-aware learning
• directional hints