通过对比式LLM微调对齐反馈通道和对话上下文表示
Aligning Backchannel and Dialogue Context Representations via Contrastive LLM Fine-Tuning
AI总结：
本文提出双阶段框架，通过对话转录微调大语言模型获取上下文表示，并学习对话上下文和反馈通道表示的联合嵌入空间，实验表明学习的投影显著提升了上下文-反馈通道检索性能。
摘要：
反馈通道（如`yeah'、`mhm'和`right'）是短且非中断的反馈信号，其词形和语调共同传达语用意义。尽管先前研究多集中于预测反馈通道时间，但词-语调形式与意义的关系仍待探索。本文提出双阶段框架：首先，通过对话转录微调大语言模型以获得丰富的上下文表示；其次，学习对话上下文和反馈通道实现的联合嵌入空间。我们通过三元相似性判断（语调和跨词性）及上下文-反馈通道适配任务评估对齐情况。实验结果表明，学习的投影相比先前方法显著提升了上下文-反馈通道检索性能。此外，结果揭示反馈通道形式对扩展对话上下文高度敏感，且学习的嵌入表示比原始WavLM特征更贴近人类判断。
英文摘要：
Backchannels (e.g., `yeah', `mhm', and `right') are short, non-interruptive feedback signals whose lexical form and prosody jointly convey pragmatic meaning. While prior computational research has largely focused on predicting backchannel timing, the relationship between lexico-prosodic form and meaning remains underexplored. We propose a two-stage framework: first, fine-tuning large language models on dialogue transcripts to derive rich contextual representations; and second, learning a joint embedding space for dialogue contexts and backchannel realizations. We evaluate alignment with human perception via triadic similarity judgments (prosodic and cross-lexical) and a context-backchannel suitability task. Our results demonstrate that the learned projections substantially improve context-backchannel retrieval compared to previous methods. In addition, they reveal that backchannel form is highly sensitive to extended conversational context and that the learned embeddings align more closely with human judgments than raw WavLM features.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
