根据Deep reinforcement learning for dialogue generation被引用的情况
18年的有：
Reward estimation for dialogue policy optimisation 最先读

Personalizing a Dialogue System With Transfer Reinforcement Learning【数据集不同，学得日常的交流方式】

Bootstrapping a Neural Conversational Agent with Dialogue Self-Play,Crowdsourcing and **On-Line** Reinforcement Learning
https://www.researchgate.net/profile/Richard_Csaky2/publication/323587007_Deep_Learning_Based_Chatbot_Models/links/5a9ef956a6fdcc22e2cb4a3a/Deep-Learning-Based-Chatbot-Models.pdf

A Study on Dialogue Reward Prediction for Open-Ended Conversational Agents【综述性】
Neural Approaches to Conversational AI【综述性】

Affective Neural Response Generation【emotional】【在对话系统中来建模情感，提出了包含情感的词向量。这篇论文的模型以 seq2seq 为背景，主要在三个点上做改进。（1）在 embedding 上加情感信息（2）改进 loss function（3）beam search 时考虑情感。】
Automatic Dialogue Generation with Expressed Emotions【seq2seq模型，加入emotion embedding】

https://zhuanlan.zhihu.com/p/39999667有兴趣可以读一读
https://helicqin.github.io/categories/NLP/