arXiv：cs.CL· Hang Ren, Peidong Wang, Yichen Gao, Wen Zhang, Jiayang Yu, Xiaocui Yang, Shi Feng·· 16 小时前
FACES：面向多模态情感支持对话的细粒度门控语义对齐框架
FACES: Fine-grained Gated Semantic Alignment for Multimodal Emotional Support Conversation
AI 导读
FACES 是一个面向多模态情感支持对话（MESC）的细粒度门控语义对齐框架，通过整合多源奖励构建多维语义与功能对齐信号，并用非补偿性门控融合机制减少跨维度奖励抵消。在受控回复长度协议下，FACES 将基于 MLLM 的平均评测分从 3.61 提升至 4.04，对 SOTA 取得 80.22% 的两两胜率，人工评测偏好率为 75.3%。代码与数据集已公开。
正文
Abstract:In recent years, with advances in multimodal interaction technologies and the growing demand for intelligent emotional support, Multimodal Emotional Support Conversation (MESC) has emerged as an important task. However, existing systems still suffer from shallow responses and inappropriate support strategies due to weak semantic feedback, cross-dimension reward compensation, and coarse reasoning supervision. To address these challenges, we propose FACES, a fine-grained gated semantic alignment framework for MESC. Specifically, FACES integrates multi-source rewards to construct multidimensional semantic and functional alignment signals, employs a non-compensatory gated fusion mechanism to reduce cross-dimension reward offset, and leverages structured Chain-of-Thought (CoT) anchors to perform token-level credit assignment. Under a controlled response-length protocol, FACES improves the average MLLM-based evaluation score from 3.61 to 4.04 and achieves an 80.22% pairwise win rate over SOTA. Human evaluation under the same protocol yields a 75.3% preference rate. Results on three external datasets further support transferability. The code and dataset are available at this https URL.
| Comments: | 27 pages, including references and appendices |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.20051 [cs.CL] |
| (or arXiv:2610.20051v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.20051 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Hang Ren [view email]
[v1]
Wed, 14 Oct 2026 11:54:06 UTC (12,664 KB)
来源：arXiv：cs.CL · arxiv.org