Learning Morphology of Natural Language as a Finite-state Grammar

Javad Nouri; Roman Yangarber

doi:10.1007/978-3-319-68456-7_4

Learning Morphology of Natural Language as a Finite-state Grammar

Javad Nouri, Roman Yangarber

Avdelningen för datavetenskap

Forskningsoutput: Kapitel i bok/rapport/konferenshandling › Konferensbidrag › Vetenskaplig › Peer review

Sammanfattning

We present algorithms that learn to segment words in morphologically
rich languages, in an unsupervised fashion. Morphology of many languages can
be modeled by finite state machines (FSMs). We start with a baseline MDL-
based learning algorithm. We then formulate well-motivated and general linguistic
principles about morphology, and incorporate them into the algorithm as heuristics,
to constrain the search space. We evaluate the algorithm on three highly-inflecting
languages. Evaluation of segmentation shows gains in performance compared to
the state of the art. We conclude with a discussion about how the learned model
relates to a morphological FSM, which is the ultimate goal.

Originalspråk	engelska
Titel på värdpublikation	Statistical Language and Speech Processing : 5th International Conference, SLSP 2017, Le Mans, France, October 23-25, 2017, Proceedings
Redaktörer	Nathalie Camelin, Yannick Estève, Carlos Martín-Vide
Utgivningsort	Cham
Förlag	Springer International Publishing AG
Utgivningsdatum	27 sep. 2017
Sidor	44-57
ISBN (tryckt)	978-3-319-68455-0
ISBN (elektroniskt)	978-3-319-68456-7
DOI	https://doi.org/10.1007/978-3-319-68456-7_4
Status	Publicerad - 27 sep. 2017
MoE-publikationstyp	A4 Artikel i en konferenspublikation
Evenemang	International Conference on Statistical Language and Speech Processing - Le Mans, Frankrike Varaktighet: 23 okt. 2017 → 25 okt. 2017 Konferensnummer: 5

Publikationsserier

Namn	Lecture Notes in Artificial Intelligence
Förlag	Springer International Publishing AG
Volym	10583
ISSN (tryckt)	0302-9743
ISSN (elektroniskt)	1611-3349

Vetenskapsgrenar

113 Data- och informationsvetenskap

Åtkomst av dokument

10.1007/978-3-319-68456-7_4

Revita: Language learning and AI
Yangarber, R., Katinskaia, A., Hou, J., Furlan, G. & Kylliäinen, I. P.
Projekt: Forskningsprojekt
LLL: Language Learning Lab
Yangarber, R., Katinskaia, A., Hou, J., Furlan, G. & Kylliäinen, I. P.
Projekt: Forskningsprojekt

Citera det här

Nouri, J., & Yangarber, R. (2017). Learning Morphology of Natural Language as a Finite-state Grammar. I N. Camelin, Y. Estève, & C. Martín-Vide (Red.), Statistical Language and Speech Processing: 5th International Conference, SLSP 2017, Le Mans, France, October 23-25, 2017, Proceedings (s. 44-57). (Lecture Notes in Artificial Intelligence; Vol. 10583). Springer International Publishing AG. https://doi.org/10.1007/978-3-319-68456-7_4

Nouri, Javad ; Yangarber, Roman. / Learning Morphology of Natural Language as a Finite-state Grammar. Statistical Language and Speech Processing: 5th International Conference, SLSP 2017, Le Mans, France, October 23-25, 2017, Proceedings. redaktör / Nathalie Camelin ; Yannick Estève ; Carlos Martín-Vide. Cham : Springer International Publishing AG, 2017. s. 44-57 (Lecture Notes in Artificial Intelligence).

@inproceedings{d252d6d2dc6d4c0fac495815b5d3764a,

title = "Learning Morphology of Natural Language as a Finite-state Grammar",

abstract = "We present algorithms that learn to segment words in morphologicallyrich languages, in an unsupervised fashion. Morphology of many languages canbe modeled by finite state machines (FSMs). We start with a baseline MDL-based learning algorithm. We then formulate well-motivated and general linguisticprinciples about morphology, and incorporate them into the algorithm as heuristics,to constrain the search space. We evaluate the algorithm on three highly-inflectinglanguages. Evaluation of segmentation shows gains in performance compared tothe state of the art. We conclude with a discussion about how the learned modelrelates to a morphological FSM, which is the ultimate goal.",

keywords = "113 Computer and information sciences, finite-state automata, unsupervised morphology induction, minimum description length principle, MDL",

author = "Javad Nouri and Roman Yangarber",

year = "2017",

month = sep,

day = "27",

doi = "10.1007/978-3-319-68456-7_4",

language = "English",

isbn = "978-3-319-68455-0",

series = "Lecture Notes in Artificial Intelligence",

publisher = "Springer International Publishing AG",

pages = "44--57",

editor = "Nathalie Camelin and Yannick Est{\`e}ve and Carlos Mart{\'i}n-Vide",

booktitle = "Statistical Language and Speech Processing",

address = "Switzerland",

note = "International Conference on Statistical Language and Speech Processing, SLSP ; Conference date: 23-10-2017 Through 25-10-2017",

}

Nouri, J & Yangarber, R 2017, Learning Morphology of Natural Language as a Finite-state Grammar. i N Camelin, Y Estève & C Martín-Vide (red), Statistical Language and Speech Processing: 5th International Conference, SLSP 2017, Le Mans, France, October 23-25, 2017, Proceedings. Lecture Notes in Artificial Intelligence, vol. 10583, Springer International Publishing AG, Cham, s. 44-57, International Conference on Statistical Language and Speech Processing, Le Mans, Frankrike, 23/10/2017. https://doi.org/10.1007/978-3-319-68456-7_4

Learning Morphology of Natural Language as a Finite-state Grammar. / Nouri, Javad; Yangarber, Roman.

Statistical Language and Speech Processing: 5th International Conference, SLSP 2017, Le Mans, France, October 23-25, 2017, Proceedings. red. / Nathalie Camelin; Yannick Estève; Carlos Martín-Vide. Cham : Springer International Publishing AG, 2017. s. 44-57 (Lecture Notes in Artificial Intelligence; Vol. 10583).

Forskningsoutput: Kapitel i bok/rapport/konferenshandling › Konferensbidrag › Vetenskaplig › Peer review

TY - GEN

T1 - Learning Morphology of Natural Language as a Finite-state Grammar

AU - Nouri, Javad

AU - Yangarber, Roman

N1 - Conference code: 5

PY - 2017/9/27

Y1 - 2017/9/27

N2 - We present algorithms that learn to segment words in morphologicallyrich languages, in an unsupervised fashion. Morphology of many languages canbe modeled by finite state machines (FSMs). We start with a baseline MDL-based learning algorithm. We then formulate well-motivated and general linguisticprinciples about morphology, and incorporate them into the algorithm as heuristics,to constrain the search space. We evaluate the algorithm on three highly-inflectinglanguages. Evaluation of segmentation shows gains in performance compared tothe state of the art. We conclude with a discussion about how the learned modelrelates to a morphological FSM, which is the ultimate goal.

AB - We present algorithms that learn to segment words in morphologicallyrich languages, in an unsupervised fashion. Morphology of many languages canbe modeled by finite state machines (FSMs). We start with a baseline MDL-based learning algorithm. We then formulate well-motivated and general linguisticprinciples about morphology, and incorporate them into the algorithm as heuristics,to constrain the search space. We evaluate the algorithm on three highly-inflectinglanguages. Evaluation of segmentation shows gains in performance compared tothe state of the art. We conclude with a discussion about how the learned modelrelates to a morphological FSM, which is the ultimate goal.

KW - 113 Computer and information sciences

KW - finite-state automata

KW - unsupervised morphology induction

KW - minimum description length principle

KW - MDL

U2 - 10.1007/978-3-319-68456-7_4

DO - 10.1007/978-3-319-68456-7_4

M3 - Conference contribution

SN - 978-3-319-68455-0

T3 - Lecture Notes in Artificial Intelligence

SP - 44

EP - 57

BT - Statistical Language and Speech Processing

A2 - Camelin, Nathalie

A2 - Estève, Yannick

A2 - Martín-Vide, Carlos

PB - Springer International Publishing AG

CY - Cham

T2 - International Conference on Statistical Language and Speech Processing

Y2 - 23 October 2017 through 25 October 2017

ER -

Nouri J, Yangarber R. Learning Morphology of Natural Language as a Finite-state Grammar. I Camelin N, Estève Y, Martín-Vide C, redaktörer, Statistical Language and Speech Processing: 5th International Conference, SLSP 2017, Le Mans, France, October 23-25, 2017, Proceedings. Cham: Springer International Publishing AG. 2017. s. 44-57. (Lecture Notes in Artificial Intelligence). doi: 10.1007/978-3-319-68456-7_4

Learning Morphology of Natural Language as a Finite-state Grammar

Sammanfattning

Publikationsserier

Vetenskapsgrenar

Åtkomst av dokument

Projekt

Revita: Language learning and AI

LLL: Language Learning Lab

Citera det här