An Empirical Study of Smoothing Techniques for Language Modeling

An Empirical Study of Smoothing Techniques for Language Modeling (1996)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Stanley F. Chen , Joshua Goodman

Citations:

631 - 19 self

Versions

Version History

Metadata Version 3

User correction supplied by SystemCorrections

Datum	Value	Source
TITLE	An Empirical Study of Smoothing Techniques for Language Modeling	INFERENCE
AUTHOR NAME	Stanley F. Chen	user correction - Legacy Corrections
AUTHOR AFFIL	Harvard University; Aiken Computation Laboratory	user correction - Legacy Corrections
AUTHOR ADDR	33 Oxford St.; Cambridge, MA 02138	user correction - Legacy Corrections
AUTHOR NAME	Joshua Goodman	user correction - Legacy Corrections
AUTHOR AFFIL	Harvard University; Aiken Computation Laboratory	user correction - Legacy Corrections
AUTHOR ADDR	33 Oxford St.; Cambridge, MA 02138	user correction - Legacy Corrections
ABSTRACT	We present an extensive empirical comparison of several smoothing techniques in the domain of language modeling, including those described by Jelinek and Mercer (1980), Katz (1987), and Church and Gale (1991). We investigate for the rst time how factors such as training data size, corpus (e.g., Brown versus Wall Street Journal), and n-gram order (bigram versus trigram) aect the relative performance of these methods, which we measure through the cross-entropy of test data. In addition, we introduce two novel smoothing techniques, one a variation of Jelinek-Mercer smoothing and one a very simple linear interpolation technique, both of which outperform existing methods. 1 Introduction Smoothing is a technique essential in the construction of n-gram language models, a staple in speech recognition (Bahl, Jelinek, and Mercer, 1983) as well as many other domains (Church, 1988; Brown et al., 1990; Kernighan, Church, and Gale, 1990). A language model is a probability distribution over string...	user correction - Legacy Corrections
YEAR	1996	user correction - Legacy Corrections
VENUE TYPE	TECHREPORT	INFERENCE
PAGES	310--318	INFERENCE
VOLUME	13	INFERENCE
TECH	Technical Report TR-10-98	INFERENCE
CITATIONS	20 found	ParsCit 1.0