Experiments with Language Models for Word Completion and Prediction in Hebrew

Yaakov HaCohen-Kerner; Asaf Applebaum; Jacob Bitterman

doi:10.1007/978-3-319-10888-9_44

Experiments with Language Models for Word Completion and Prediction in Hebrew

Yaakov HaCohen-Kerner, Asaf Applebaum, Jacob Bitterman

Research output: Contribution to journal › Article › peer-review

2 Scopus citations

Abstract

In this paper, we describe various language models (LMs) and combinations created to support word prediction and completion in Hebrew. We define and apply 5 general types of LMs: (1) Basic LMs (unigrams, bigrams, trigrams, and quadgrams), (2) Backoff LMs, (3) LMs Integrated with tagged LMs, (4) Interpolated LMs, and (5) Interpolated LMs Integrated with tagged LMs. 16 specific implementations of these LMs were compared using 3 types of Israeli web newspaper corpora. The foremost keystroke saving results were achieved with LMs of the most complex variety, the Interpolated LMs Integrated with tagged LMs. Therefore, we conclude that combining all strengths by creating a synthesis of all four basic LMs and the tagged LMs leads to the best results.

Original language	English
Pages (from-to)	450-462
Number of pages	13
Journal	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	8686
DOIs	https://doi.org/10.1007/978-3-319-10888-9_44
State	Published - 2014
Externally published	Yes

Bibliographical note

Publisher Copyright:
© Springer International Publishing Switzerland 2014.

Keywords

Hebrew
Keystroke savings
Language models
Word completion
Word prediction

Access to Document

10.1007/978-3-319-10888-9_44

Cite this

@article{01b42df446c34d1486c9690fb290da6d,

title = "Experiments with Language Models for Word Completion and Prediction in Hebrew",

abstract = "In this paper, we describe various language models (LMs) and combinations created to support word prediction and completion in Hebrew. We define and apply 5 general types of LMs: (1) Basic LMs (unigrams, bigrams, trigrams, and quadgrams), (2) Backoff LMs, (3) LMs Integrated with tagged LMs, (4) Interpolated LMs, and (5) Interpolated LMs Integrated with tagged LMs. 16 specific implementations of these LMs were compared using 3 types of Israeli web newspaper corpora. The foremost keystroke saving results were achieved with LMs of the most complex variety, the Interpolated LMs Integrated with tagged LMs. Therefore, we conclude that combining all strengths by creating a synthesis of all four basic LMs and the tagged LMs leads to the best results.",

keywords = "Hebrew, Keystroke savings, Language models, Word completion, Word prediction",

author = "Yaakov HaCohen-Kerner and Asaf Applebaum and Jacob Bitterman",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2014.",

year = "2014",

doi = "10.1007/978-3-319-10888-9_44",

language = "אנגלית",

volume = "8686",

pages = "450--462",

journal = "Lecture Notes in Computer Science",

issn = "0302-9743",

publisher = "Springer Verlag",

}

Experiments with Language Models for Word Completion and Prediction in Hebrew. / HaCohen-Kerner, Yaakov; Applebaum, Asaf; Bitterman, Jacob.
In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 8686, 2014, p. 450-462.

Research output: Contribution to journal › Article › peer-review

TY - JOUR

T1 - Experiments with Language Models for Word Completion and Prediction in Hebrew

AU - HaCohen-Kerner, Yaakov

AU - Applebaum, Asaf

AU - Bitterman, Jacob

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2014.

PY - 2014

Y1 - 2014

N2 - In this paper, we describe various language models (LMs) and combinations created to support word prediction and completion in Hebrew. We define and apply 5 general types of LMs: (1) Basic LMs (unigrams, bigrams, trigrams, and quadgrams), (2) Backoff LMs, (3) LMs Integrated with tagged LMs, (4) Interpolated LMs, and (5) Interpolated LMs Integrated with tagged LMs. 16 specific implementations of these LMs were compared using 3 types of Israeli web newspaper corpora. The foremost keystroke saving results were achieved with LMs of the most complex variety, the Interpolated LMs Integrated with tagged LMs. Therefore, we conclude that combining all strengths by creating a synthesis of all four basic LMs and the tagged LMs leads to the best results.

AB - In this paper, we describe various language models (LMs) and combinations created to support word prediction and completion in Hebrew. We define and apply 5 general types of LMs: (1) Basic LMs (unigrams, bigrams, trigrams, and quadgrams), (2) Backoff LMs, (3) LMs Integrated with tagged LMs, (4) Interpolated LMs, and (5) Interpolated LMs Integrated with tagged LMs. 16 specific implementations of these LMs were compared using 3 types of Israeli web newspaper corpora. The foremost keystroke saving results were achieved with LMs of the most complex variety, the Interpolated LMs Integrated with tagged LMs. Therefore, we conclude that combining all strengths by creating a synthesis of all four basic LMs and the tagged LMs leads to the best results.

KW - Hebrew

KW - Keystroke savings

KW - Language models

KW - Word completion

KW - Word prediction

UR - http://www.scopus.com/inward/record.url?scp=84921810347&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-10888-9_44

DO - 10.1007/978-3-319-10888-9_44

M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???

AN - SCOPUS:84921810347

SN - 0302-9743

VL - 8686

SP - 450

EP - 462

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

ER -

Experiments with Language Models for Word Completion and Prediction in Hebrew

Abstract

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this