Polite Dialogue Generation Without Parallel Data
Tong Niu Mohit Bansal UNC Chapel Hill {tongn,
Abstract
Stylistic dialogue response generation, with valuable applications in personality-based conversational agents, is a challenging task because the response needs to be fluent, contextually-relevant, as well as paralinguistically accurate. Moreover, parallel datasets for regular-to-stylistic pairs are usually unavailable. We present three weakly-supervised models that can generate diverse polite (or rude) dialogue responses without parallel data. Our late fusion model (Fusion) merges the decoder of an encoder-attention-decoder dialogue model with a language model trained on stand-alone polite utterances. Our label-fine-tuning (LFT) model prepends to each source sequence a politeness-score scaled label (predicted by our state-of-the-art politeness classifier) during training, and at test time is able to generate polite, neutral, and rude responses by simply scaling the label embedding by the corresponding score. Our reinforcement learning model (Polite-RL) encourages politeness generation by assigning rewards proportional to the politeness classifier score of the sampled response. We also present two retrieval-based polite dialogue model baselines. Human evaluation validates that while
中文速览
让对话系统说话有礼貌（或粗鲁）看起来简单，实则很难——既要回应得体、语境连贯，又要把礼貌这种复杂的语言风格自然地融进去，而且现实中根本没有"普通话↔礼貌话"的平行对照语料。研究者提出了三种弱监督生成模型：Fusion模型把对话解码器与在礼貌语料上训练的语言模型做后期融合；LFT模型在训练时给每条输入前加一个由礼貌分类器打分、连续缩放的标签嵌入，测试时只需调整这个数值就能控制输出的礼貌程度；Polite-RL模型则用强化学习，以分类器给出的礼貌得分作为奖励来引导生成。人工评测表明，LFT和Polite-RL这两种方法在显著提升回复礼貌度的同时，能保持与基线Seq2seq模型相当的对话质量，而Fusion和检索式方法虽然也能让回复更礼貌，但对话相关性明显下降。这项工作的意义在于：它只需要一个风格分类器、无需平行语料，就能以模块化方式把礼貌乃至任何情感/人格风格注入对话系统，为构建更自然、更负责任的个性化对话智能体提供了实用框架。
原文 arXiv:1805.03162;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1805.03162v1