Performance of a SCFG-based language model with training data sets of increasing size

Joan Andreu Sánchez; José Miguel Benedí; Diego Linares

doi:10.1007/11492542_72

Performance of a SCFG-based language model with training data sets of increasing size

Joan Andreu Sánchez, José Miguel Benedí, Diego Linares

Polytechnic University of Valencia

Research output: Contribution to journal › Conference article › peer-review

1 Scopus citations

Abstract

In this paper, a hybrid language model which combines a word-based n-gram and a category-based Stochastic Context-Free Grammar (SCFG) is evaluated for training data sets of increasing size. Different estimation algorithms for learning SCFGs in General Format and in Chomsky Normal Form are considered. Experiments on the UPenn Treebank corpus are reported. These experiments have been carried out in terms of the test set perplexity and the word error rate in a speech recognition experiment.

Original language	English
Pages (from-to)	586-594
Number of pages	9
Journal	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	3523
Issue number	II
DOIs	https://doi.org/10.1007/11492542_72
State	Published - 2005
Event	Second Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2005 - Estoril, Portugal Duration: 07 Jun 2005 → 09 Jun 2005

Access to Document

10.1007/11492542_72

Cite this

@article{e8c035afa2c74fec9b8377f6c2717e4f,

title = "Performance of a SCFG-based language model with training data sets of increasing size",

abstract = "In this paper, a hybrid language model which combines a word-based n-gram and a category-based Stochastic Context-Free Grammar (SCFG) is evaluated for training data sets of increasing size. Different estimation algorithms for learning SCFGs in General Format and in Chomsky Normal Form are considered. Experiments on the UPenn Treebank corpus are reported. These experiments have been carried out in terms of the test set perplexity and the word error rate in a speech recognition experiment.",

author = "S{\'a}nchez, {Joan Andreu} and Bened{\'i}, {Jos{\'e} Miguel} and Diego Linares",

year = "2005",

doi = "10.1007/11492542_72",

language = "English",

volume = "3523",

pages = "586--594",

journal = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

issn = "0302-9743",

publisher = "Springer Verlag",

number = "II",

note = "Second Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2005 ; Conference date: 07-06-2005 Through 09-06-2005",

}

Performance of a SCFG-based language model with training data sets of increasing size. / Sánchez, Joan Andreu; Benedí, José Miguel; Linares, Diego.
In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 3523, No. II, 2005, p. 586-594.

Research output: Contribution to journal › Conference article › peer-review

TY - JOUR

T1 - Performance of a SCFG-based language model with training data sets of increasing size

AU - Sánchez, Joan Andreu

AU - Benedí, José Miguel

AU - Linares, Diego

PY - 2005

Y1 - 2005

N2 - In this paper, a hybrid language model which combines a word-based n-gram and a category-based Stochastic Context-Free Grammar (SCFG) is evaluated for training data sets of increasing size. Different estimation algorithms for learning SCFGs in General Format and in Chomsky Normal Form are considered. Experiments on the UPenn Treebank corpus are reported. These experiments have been carried out in terms of the test set perplexity and the word error rate in a speech recognition experiment.

AB - In this paper, a hybrid language model which combines a word-based n-gram and a category-based Stochastic Context-Free Grammar (SCFG) is evaluated for training data sets of increasing size. Different estimation algorithms for learning SCFGs in General Format and in Chomsky Normal Form are considered. Experiments on the UPenn Treebank corpus are reported. These experiments have been carried out in terms of the test set perplexity and the word error rate in a speech recognition experiment.

UR - http://www.scopus.com/inward/record.url?scp=25144434854&partnerID=8YFLogxK

U2 - 10.1007/11492542_72

DO - 10.1007/11492542_72

M3 - Conference article

AN - SCOPUS:25144434854

SN - 0302-9743

VL - 3523

SP - 586

EP - 594

JO - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

JF - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

IS - II

T2 - Second Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2005

Y2 - 7 June 2005 through 9 June 2005

ER -

Performance of a SCFG-based language model with training data sets of increasing size

Abstract

Access to Document

Other files and links

Fingerprint

Cite this