Abstract
This paper proposes performing natural language generation using genetic algorithms to address two issues: the difficulty in controlling global textual features which arise from a large number of interdependent local decisions, and the difficulty in applying conventional NLG wisdom in domains where the communicative goal lacks sufficient detail. It presents details of an implemented system that embodies the aforementioned proposal, and discusses the results of an empirical study conducted using the system.
Original language | English |
---|---|
Title of host publication | Proceedings of the 22nd Pacific Asia Conference on Language, Information, and Computation |
Subtitle of host publication | (PACLIC 2008) |
Pages | 272-281 |
Number of pages | 10 |
Publication status | Published - 2008 |
Event | 22nd Pacific Asia Conference on Language, Information, and Computation (PACLIC 2008) - Cebu, Philippines Duration: 20 Nov 2008 → 22 Nov 2008 |
Conference
Conference | 22nd Pacific Asia Conference on Language, Information, and Computation (PACLIC 2008) |
---|---|
Country/Territory | Philippines |
City | Cebu |
Period | 20/11/08 → 22/11/08 |
Bibliographical note
22nd Pacific Asia Conference on Language, Information, and Computation (PACLIC 2008), Cebu, Philippines, 20-22 November 2008.Keywords
- natural language generation
- genetic algorithms
- creative language