Natural Language Generation
We are engaged in research to develop improved natural language generation (NLG) capabilities for virtual human dialogue systems. One research focus has been to develop techniques to reduce the development costs and linguistic expertise that are required to incorporate a productive grammar-based generation capability into implemented virtual human systems (DeVault et al, SigDial 2008; DeVault et al, INLG 2008). Another approach, used in some projects, employs a statistical classification framework to select appropriate utterances at run-time from a set of complete, pre-authored character utterances.
Publications
-
Full TextPractical Grammar-Based NLG from Examples. In: Proceedings of the Fifth International Natural Language Generation Conference. Salt Fork, Ohio: Association for Computational Linguistics; 2008. p. 77-85.
-
Full TextMaking Grammar-Based Generation Easier to Deploy in Dialogue Systems. In: Proceedings of the 9th SIGdial Workshop on Discourse and Dialogue. Columbus, Ohio: Association for Computational Linguistics; 2008. p. 198-207.
-
Full TextNL Generation for Virtual Humans in a Complex Social Environment. In: Working Notes AAAI Spring Symposium on Natural Language Generation in Spoken and Written Dialogue.; 2003.
-
Full TextSemantics and Pragmatics of Questions and Answers for Dialogue Agents. In: proceedings of the International Workshop on Computational Semantics.; 2003. p. 380-94.