Semantic Attribute system in Machine Translations

Masahiro MIYAZAKI*1, Satoru IKEHARA*2, AkioYOKOO*3, Satoshi SHIRAI*3

*1 Faculty of Engineering, Niigata University
8050 Ikarashi 2-nocho, Niigata 950-21, Japan
miyazaki@info.eng.niigata-u.ac.jp
*2 Faculty of Engineering, Tottori University
4-101 Koyamacho-minami, Tottori 680, Japan
ikehara@ike.tottori-u.ac.jp
*3 NTT Communication Sciense Laboratories
1-1 Hikarinooka, Yokosuka 239, Japan
{ayokoo,shirai}@cslab.kecl.ntt.co.jp

Abstract:

This paper presents semantic attribute system with 3,000 attributes classified by the viewpoint of conceptialization of the object, for semantic analysis in machine translation. A semmantic word dictionary for knowledge related to semantic use of word and a semantic structure dictionary for knowledge related to meanings of expression structures are connected by the semantic attributes. We gave semantic attributes for 400,000 words of modern Japanese stored in the semmantic word dictionary for machine translation.



Keywords:

machine translation, semantic analysis, thesaurus, semantic words dictionary



[ Technical Report of IEICE, NLC97-12, pp.29-36 (July, 1997). ]