Sluice resolution in English is the problem of finding antecedents of wh-fronted ellipses. Previous work has relied on handcrafted features over syntax trees that scale poorly to other languages and domains; in particular, to dialogue, which is one of the most interesting applications of sluice resolution. Syntactic information is arguably important for sluice resolution, but we show that multi-task learning with partial parsing as auxiliary tasks effectively closes the gap and buys us an additional 9% error reduction over previous work. Since we are not directly relying on features from partial parsers, our system is more robust to domain shifts, giving a 26% error reduction on embedded sluices in dialogue.
|Titel||Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies|
|Redaktører||Stephanie Lukin, Margaret Mitchell|
|Udgivelses sted||New Orleans, LA|
|Forlag||Association for Computational Linguistics|
|Status||Udgivet - 2018|
|Begivenhed||The 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies - New Orleans, USA|
Varighed: 1 jun. 2018 → 6 jun. 2018
Konferencens nummer: 16
|Konference||The 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics|
|Periode||01/06/2018 → 06/06/2018|
Rønning, O., Hardt, D., & Søgaard, A. (2018). Sluice Resolution without Hand-Crafted Features over Brittle Syntax Trees. I S. Lukin, & M. Mitchell (red.), Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Bind 2, s. 236–241). New Orleans, LA: Association for Computational Linguistics.