Towards Automatic Evaluation of Dialog Systems: A Model-Free Off-Policy Evaluation Approach.
Haoming JiangBo DaiMengjiao YangTuo ZhaoWei WeiPublished in: EMNLP (1) (2021)
Keyphrases
- automatic evaluation
- dialog systems
- model free
- policy evaluation
- natural language generation
- reinforcement learning
- temporal difference
- natural language
- policy iteration
- function approximation
- reinforcement learning algorithms
- least squares
- human judgments
- markov decision processes
- artificial neural networks
- dynamic programming