Abstract:Large language models (LLMs) have been widely applied to emotional support conversation (ESC). However, complex multi-turn support remains this http URL is because existing alignment schemes rely on sparse outcome-level signals, thus offering limited supervision for intermediate strategy decisions. To fill this gap, this paper proposes affective flow language model for emotional support conversation (AFlow), a framework that introduces fine-grained supervision on dialogue prefixes by modeling a continuous affective flow along multi-turn trajectories. AFlow can estimate intermediate utility over searched trajectories and learn preference-consistent strategy transitions. To improve strategy coherence and empathetic response quality, a subpath-level flow-balance objective is presented to propagate preference signals to intermediate states. Experiment results show consistent and significant improvements over competitive baselines in diverse emotional contexts. Remarkably, AFlow with a compact open-source backbone outperforms proprietary LMMs such as GPT-4o and Claude-3.5 on major ESC metrics. Our code is available at this https URL.
| Comments: | 19 pages, 7 figures |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2602.08826 [cs.CL] |
| (or arXiv:2602.08826v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2602.08826 arXiv-issued DOI via DataCite |
Submission history
From: Chenghui Zou [view email]
[v1]
Mon, 9 Feb 2026 15:58:50 UTC (980 KB)
[v2]
Wed, 29 Apr 2026 12:24:39 UTC (983 KB)