Learning Character-level Representations for Part-of-Speech Tagging

Cicero Dos Santos, Bianca Zadrozny
; Proceedings of the 31st International Conference on Machine Learning, PMLR 32(2):1818-1826, 2014.

Abstract

Distributed word representations have recently been proven to be an invaluable resource for NLP. These representations are normally learned using neural networks and capture syntactic and semantic information about words. Information about word morphology and shape is normally ignored when learning word representations. However, for tasks like part-of-speech tagging, intra-word information is extremely useful, specially when dealing with morphologically rich languages. In this paper, we propose a deep neural network that learns character-level representation of words and associate them with usual word representations to perform POS tagging. Using the proposed approach, while avoiding the use of any handcrafted feature, we produce state-of-the-art POS taggers for two languages: English, with 97.32% accuracy on the Penn Treebank WSJ corpus; and Portuguese, with 97.47% accuracy on the Mac-Morpho corpus, where the latter represents an error reduction of 12.2% on the best previous known result.

Cite this Paper


BibTeX
@InProceedings{pmlr-v32-santos14, title = {Learning Character-level Representations for Part-of-Speech Tagging}, author = {Cicero Dos Santos and Bianca Zadrozny}, booktitle = {Proceedings of the 31st International Conference on Machine Learning}, pages = {1818--1826}, year = {2014}, editor = {Eric P. Xing and Tony Jebara}, volume = {32}, number = {2}, series = {Proceedings of Machine Learning Research}, address = {Bejing, China}, month = {22--24 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v32/santos14.pdf}, url = {http://proceedings.mlr.press/v32/santos14.html}, abstract = {Distributed word representations have recently been proven to be an invaluable resource for NLP. These representations are normally learned using neural networks and capture syntactic and semantic information about words. Information about word morphology and shape is normally ignored when learning word representations. However, for tasks like part-of-speech tagging, intra-word information is extremely useful, specially when dealing with morphologically rich languages. In this paper, we propose a deep neural network that learns character-level representation of words and associate them with usual word representations to perform POS tagging. Using the proposed approach, while avoiding the use of any handcrafted feature, we produce state-of-the-art POS taggers for two languages: English, with 97.32% accuracy on the Penn Treebank WSJ corpus; and Portuguese, with 97.47% accuracy on the Mac-Morpho corpus, where the latter represents an error reduction of 12.2% on the best previous known result.} }
Endnote
%0 Conference Paper %T Learning Character-level Representations for Part-of-Speech Tagging %A Cicero Dos Santos %A Bianca Zadrozny %B Proceedings of the 31st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2014 %E Eric P. Xing %E Tony Jebara %F pmlr-v32-santos14 %I PMLR %J Proceedings of Machine Learning Research %P 1818--1826 %U http://proceedings.mlr.press %V 32 %N 2 %W PMLR %X Distributed word representations have recently been proven to be an invaluable resource for NLP. These representations are normally learned using neural networks and capture syntactic and semantic information about words. Information about word morphology and shape is normally ignored when learning word representations. However, for tasks like part-of-speech tagging, intra-word information is extremely useful, specially when dealing with morphologically rich languages. In this paper, we propose a deep neural network that learns character-level representation of words and associate them with usual word representations to perform POS tagging. Using the proposed approach, while avoiding the use of any handcrafted feature, we produce state-of-the-art POS taggers for two languages: English, with 97.32% accuracy on the Penn Treebank WSJ corpus; and Portuguese, with 97.47% accuracy on the Mac-Morpho corpus, where the latter represents an error reduction of 12.2% on the best previous known result.
RIS
TY - CPAPER TI - Learning Character-level Representations for Part-of-Speech Tagging AU - Cicero Dos Santos AU - Bianca Zadrozny BT - Proceedings of the 31st International Conference on Machine Learning PY - 2014/01/27 DA - 2014/01/27 ED - Eric P. Xing ED - Tony Jebara ID - pmlr-v32-santos14 PB - PMLR SP - 1818 DP - PMLR EP - 1826 L1 - http://proceedings.mlr.press/v32/santos14.pdf UR - http://proceedings.mlr.press/v32/santos14.html AB - Distributed word representations have recently been proven to be an invaluable resource for NLP. These representations are normally learned using neural networks and capture syntactic and semantic information about words. Information about word morphology and shape is normally ignored when learning word representations. However, for tasks like part-of-speech tagging, intra-word information is extremely useful, specially when dealing with morphologically rich languages. In this paper, we propose a deep neural network that learns character-level representation of words and associate them with usual word representations to perform POS tagging. Using the proposed approach, while avoiding the use of any handcrafted feature, we produce state-of-the-art POS taggers for two languages: English, with 97.32% accuracy on the Penn Treebank WSJ corpus; and Portuguese, with 97.47% accuracy on the Mac-Morpho corpus, where the latter represents an error reduction of 12.2% on the best previous known result. ER -
APA
Santos, C.D. & Zadrozny, B.. (2014). Learning Character-level Representations for Part-of-Speech Tagging. Proceedings of the 31st International Conference on Machine Learning, in PMLR 32(2):1818-1826

Related Material