Multi-Stage Pre-training Enhanced by ChatGPT for Multi-Scenario Multi-Domain Dialogue Summarization.
Weixiao ZhouGengyao LiXianfu ChengXinnian LiangJunnan ZhuFeifei ZhaiZhoujun LiPublished in: EMNLP (Findings) (2023)
Keyphrases
- multistage
- multi domain
- spoken dialogue systems
- cross domain
- dynamic programming
- production system
- domain specific
- single stage
- stochastic programming
- stochastic optimization
- training examples
- lot sizing
- training set
- attack detection
- optimal policy
- heterogeneous networks
- semi supervised
- search computing
- lot streaming
- neural network
- co occurrence
- natural language
- decision trees
- feature selection
- learning algorithm