Presentation 2000/1/21
Clustering Word Histories for Language Model
Keikichi Miyai, Youichi Yamashita,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper describes a new language model based on sharing word probabilities for similar word history. A clustering technique divides word histories into some history classes using similarity of the word history in terms of output probability distribution of next words. This model can estimate a word output probability for unknown word context. Three types of history clustering are tried in order to interpolate word output probabilities. Evaluation tests verified that the proposed model reduced the model parameters with keeping the speech recognition accuracy.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) language model / clustering / word history / smoothing
Paper # SP99-143
Date of Issue

Conference Information
Committee SP
Conference Date 2000/1/21(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Speech (SP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Clustering Word Histories for Language Model
Sub Title (in English)
Keyword(1) language model
Keyword(2) clustering
Keyword(3) word history
Keyword(4) smoothing
1st Author's Name Keikichi Miyai
1st Author's Affiliation Ritsumeikan University()
2nd Author's Name Youichi Yamashita
2nd Author's Affiliation Ritsumeikan University
Date 2000/1/21
Paper # SP99-143
Volume (vol) vol.99
Number (no) 577
Page pp.pp.-
#Pages 8
Date of Issue