Paper: Generation Under Uncertainty

ACL ID W10-4235
Title Generation Under Uncertainty
Venue International Conference on Natural Language Generation
Session Main Conference
Year 2010

We invite the research community to con- sider challenges for NLG which arise from uncertainty. NLG systems should be able to adapt to their audience and the genera- tion environment in general, but often the important features for adaptation are not known precisely. We explore generation challenges which could employ simulated environmentstostudyNLGwhichisadap- tive under uncertainty, and suggest possi- ble metrics for such tasks. It would be par- ticularly interesting to explore how differ- ent planning approaches to NLG perform in challenges involving uncertainty in the generation environment.