Real-time spoken language translation using associative processors

Authors:
Kozo Oi;Eiichiro Sumita;Osamu Furuse;Hitoshi Iida;Tetsuya Higuchi
Affiliations:
ATR Interpreting Telecommunications Research Laboratories, Hikaridai, Seika, Souraku, Kyoto, Japan;ATR Interpreting Telecommunications Research Laboratories, Hikaridai, Seika, Souraku, Kyoto, Japan;ATR Interpreting Telecommunications Research Laboratories, Hikaridai, Seika, Souraku, Kyoto, Japan;ATR Interpreting Telecommunications Research Laboratories, Hikaridai, Seika, Souraku, Kyoto, Japan;Electrotechnical Laboratory, Umezono, Tsukuba, Ibaraki, Japan
Venue:
ANLC '94 Proceedings of the fourth conference on Applied natural language processing
Year:
1994

Citing 4
Cited 2

A framework of a mechanical translation between Japanese and English by analogy principle

Proc. of the international NATO symposium on Artificial and human intelligence
\Phi DM-Dialog: An Experimental Speech-to-Speech Dialog Translation System

Computer
IXM2: a parallel associative processor

ISCA '91 Proceedings of the 18th annual international symposium on Computer architecture
Cooperation between transfer and analysis in example-based framework

COLING '92 Proceedings of the 14th conference on Computational linguistics - Volume 2

Review Article: Example-based Machine Translation

Machine Translation
Incremental speech translation

Incremental speech translation

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper proposes a model using associative processors (APs) for real-time spoken language translation. Spoken language translation requires (1) an accurate translation and (2) a realtime response. We have already proposed a model, TDMT (Transfer-Driven Machine Translation), that translates a sentence utilizing examples effectively and performs accurate structural disambiguation and target word selection. This paper will concentrate on the second requirement. In TDMT, example-retrieval (ER), i.e., retrieving examples most similar to an input expression, is the most dominant part of the total processing time. Our study has concluded that we only need to implement the ER for expressions including a frequent word on APs. Experimental results show that the ER can be drastically speeded up. Moreover, a study on communications between APs demonstrates the scalability against vocabulary size by extrapolation. Thus, our model, TDMT on APs, meets the vital requirements of spoken language translation.