ClarifyDelphi: Reinforced Clarification Questions with Defeasibility Rewards for Social and Moral Situations.
Valentina PyatkinJena D. HwangVivek SrikumarXiming LuLiwei JiangYejin ChoiChandra BhagavatulaPublished in: ACL (1) (2023)
Keyphrases
- social networks
- social interaction
- real world
- markov decision processes
- reinforcement learning
- social media
- social behavior
- knowledge sharing
- question answer
- online communities
- virtual communities
- social learning
- social web
- social norms
- user generated
- bandit problems
- social issues
- multiarmed bandit
- deontic logic
- answer questions
- current situation
- social networking
- genetic algorithm