Konferensartikel

Active learning for sense annotation

Héctor Martínez Alonso
University of Copenhagen, Copenhagen, Denmark

Barbara Plank
University of Copenhagen, Copenhagen, Denmark

Anders Johannsen
University of Copenhagen, Copenhagen, Denmark

Anders Søgaard
University of Copenhagen, Copenhagen, Denmark

Ladda ner artikel

Ingår i: Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015, May 11-13, 2015, Vilnius, Lithuania

Linköping Electronic Conference Proceedings 109:31, s. 245-249

NEALT Proceedings Series 23:31, p. 245-249

Visa mer +

Publicerad: 2015-05-06

ISBN: 978-91-7519-098-3

ISSN: 1650-3686 (tryckt), 1650-3740 (online)

Abstract

This article describes a real (non-synthetic) active-learning experiment to obtain supersense annotations for Danish. We compare two instance selection strategies and evaluate their performance during the annotation process, across domains for the final resulting system, as well as against in-domain adjudicated data.

Nyckelord

Inga nyckelord är tillgängliga

Referenser

Jørg Asmussen and Jakob Halskov. 2012. The clarin dk reference corpus. In Sprogteknologisk Workshop. Yee Seng Chan and Hwee Tou Ng. 2007. Domain adaptation with active learning for word sense disambiguation. In ACL.

Aron Culotta and Andrew McCallum. 2005. Reducing labeling effort for structured prediction tasks. In AAAI.

Hal Daum´e, John Langford, and Daniel Marcu. 2009. Search-based structured prediction. Machine learning, 75(3):297–325.

David D Lewis and Jason Catlett. 1994. Heterogenous uncertainty sampling for supervised learning. In ICML.

Diego Marcheggiani and Thierry Artieres. 2014. An experimental comparison of active learning strategies for partially labeled sequences. In EMNLP.

H´ector Mart´inez Alonso, Anders Johannsen, Anders Søgaard, Sussi Olsen, Anna Braasch, Sanni Nimb, Nicolai Hartvig Sørensen, and Bolette Sandford Pedersen. 2015. Supersense tagging for danish. In Nodalida.

Barbara Plank, Anders Johannsen, and Anders Søgaard. 2014. Importance weighting and unsupervised domain adaptation of POS taggers: a negative result. In EMNLP.

Piyush Rai, Avishek Saha, Hal Daume, and Suresh Venkatasubramanian. 2010. Domain adaptation meets active learning. InWorkshop on Active Learning for NLP, NAACL.

Tobias Scheffer and Stefan Wrobel. 2001. Active learning of partially hidden markov models. In In Proceedings of the ECML/PKDD Workshop on Instance Selection.

Burr Settles and Mark Craven. 2008. An analysis of active learning strategies for sequence labeling tasks. In EMNLP.

Min Xiao and Yuhong Guo. 2013. Online active learning for cost sensitive domain adaptation. In CoNLL.

Citeringar i Crossref