A Hierarchical dirichlet process model for joint part-of-speech and morphology induction

Kairit Sirts; Tanel Alumäe

A Hierarchical dirichlet process model for joint part-of-speech and morphology induction

Kairit Sirts, Tanel Alumäe

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

Abstract

In this paper we present a fully unsupervised nonparametric Bayesian model that jointly induces POS tags and morphological segmentations. The model is essentially an infinite HMM that infers the number of states from data. Incorporating segmentation into the same model provides the morphological features to the system and eliminates the need to find them during preprocessing step. We show that learning both tasks jointly actually leads to better results than learning either task with gold standard data from the other task provided. The evaluation on multilingual data shows that the model produces state-of-the-art results on POS induction.

Original language	English
Title of host publication	NAACL HLT 2012
Subtitle of host publication	The Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies : Proceedings of the Conference
Place of Publication	Stroudsburg, PA
Publisher	Association for Computational Linguistics
Pages	407-416
Number of pages	10
ISBN (Print)	9781937284206
Publication status	Published - 2012
Externally published	Yes
Event	Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies - Montreal, Canada Duration: 3 Jun 2012 → 8 Jun 2012

Conference

Conference	Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies
City	Montreal, Canada
Period	3/06/12 → 8/06/12

Access to Document

http://www.aclweb.org/anthology/N12-1045

Cite this

Sirts, K., & Alumäe, T. (2012). A Hierarchical dirichlet process model for joint part-of-speech and morphology induction. In NAACL HLT 2012: The Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies : Proceedings of the Conference (pp. 407-416). Association for Computational Linguistics. http://www.aclweb.org/anthology/N12-1045

Sirts, Kairit ; Alumäe, Tanel. / A Hierarchical dirichlet process model for joint part-of-speech and morphology induction. NAACL HLT 2012: The Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies : Proceedings of the Conference. Stroudsburg, PA : Association for Computational Linguistics, 2012. pp. 407-416

@inproceedings{b0078b605588431aac22187804666c9b,

title = "A Hierarchical dirichlet process model for joint part-of-speech and morphology induction",

abstract = "In this paper we present a fully unsupervised nonparametric Bayesian model that jointly induces POS tags and morphological segmentations. The model is essentially an infinite HMM that infers the number of states from data. Incorporating segmentation into the same model provides the morphological features to the system and eliminates the need to find them during preprocessing step. We show that learning both tasks jointly actually leads to better results than learning either task with gold standard data from the other task provided. The evaluation on multilingual data shows that the model produces state-of-the-art results on POS induction.",

author = "Kairit Sirts and Tanel Alum{\"a}e",

year = "2012",

language = "English",

isbn = "9781937284206",

pages = "407--416",

booktitle = "NAACL HLT 2012",

publisher = "Association for Computational Linguistics",

note = "Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies ; Conference date: 03-06-2012 Through 08-06-2012",

}

Sirts, K & Alumäe, T 2012, A Hierarchical dirichlet process model for joint part-of-speech and morphology induction. in NAACL HLT 2012: The Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies : Proceedings of the Conference. Association for Computational Linguistics, Stroudsburg, PA, pp. 407-416, Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies, Montreal, Canada, 3/06/12. <http://www.aclweb.org/anthology/N12-1045>

A Hierarchical dirichlet process model for joint part-of-speech and morphology induction. / Sirts, Kairit; Alumäe, Tanel.
NAACL HLT 2012: The Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies : Proceedings of the Conference. Stroudsburg, PA: Association for Computational Linguistics, 2012. p. 407-416.

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

TY - GEN

T1 - A Hierarchical dirichlet process model for joint part-of-speech and morphology induction

AU - Sirts, Kairit

AU - Alumäe, Tanel

PY - 2012

Y1 - 2012

N2 - In this paper we present a fully unsupervised nonparametric Bayesian model that jointly induces POS tags and morphological segmentations. The model is essentially an infinite HMM that infers the number of states from data. Incorporating segmentation into the same model provides the morphological features to the system and eliminates the need to find them during preprocessing step. We show that learning both tasks jointly actually leads to better results than learning either task with gold standard data from the other task provided. The evaluation on multilingual data shows that the model produces state-of-the-art results on POS induction.

AB - In this paper we present a fully unsupervised nonparametric Bayesian model that jointly induces POS tags and morphological segmentations. The model is essentially an infinite HMM that infers the number of states from data. Incorporating segmentation into the same model provides the morphological features to the system and eliminates the need to find them during preprocessing step. We show that learning both tasks jointly actually leads to better results than learning either task with gold standard data from the other task provided. The evaluation on multilingual data shows that the model produces state-of-the-art results on POS induction.

M3 - Conference proceeding contribution

SN - 9781937284206

SP - 407

EP - 416

BT - NAACL HLT 2012

PB - Association for Computational Linguistics

CY - Stroudsburg, PA

T2 - Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies

Y2 - 3 June 2012 through 8 June 2012

ER -