EJ/JE Machine Translation System ASTRANSAC — Extensions toward Personalization

[MT Summit III, Washington, DC, USA, July 1991] EJ/JE Machine Translation System ASTRANSAC - Extensions toward Personalization Hideki Hirakawa, Hiroyasu Nogami, Shin-ya Amano R&D Center Toshiba Corporation 1 Komukaitoshiba-cho, Saiwai-ku, Kawasaki, 210 Japan E-mail: hirakawa@isl.rdc.toshiba.co.jp Abstract The demand for personal use of a translation system seems to be increasing in accordance with the improvement in MT quality. A re- cent portable and powerful engineering work- station, such as AS1000 (SPARC LT), en- ables us to develop a personal-use oriented MT system This paper describes the outline of ASTRANSAC (an English-Japanese/Japanese- English bi-directional MT system) and the ex- tensions related to the personalization of AS- TRANSAC, which have been newly made since the MT Summit II. 1 Introduction ASTRANSAC is an English-Japanese/Japanese-English machine translation system which works on a TOSHIBA engineering workstation AS1000 (SPARC LT). AS3000 and AS4000 series. Its design philosophy is, as shown in our paper for the MT Summit II [Amano et al., 1989], the improvement in overall translation efficiency, from input of the source text to output of the target text. The au- tomatization of a translation process in a translation of- fice, the so-called translation factory, where co-operative work among human translators, operators and MT sys- tems is realized, is one of the best means of achieving the highest efficiency in translating industrial documents (manuals). ASTRANSAC is designed for this particular kind of applications; in fact, that is currently the pri- mary needs for the MT system. However, we envision a future when the MT system is used in a more personal environment; that is, non-professional users utilize MT systems on their own computers. We call this the "per- sonalization" of the MT system. In such a situation, the following features are required of the MT system besides the accuracy of translation, which is the most important: * Small-size powerful computer * Full support of translation (OCR,DTP) * Easy customization (grammars and dictionaries) « Easy operation This paper mainly describes the extensions related to the personalization of ASTRANSAC, which have been newly made since the MT Summit II. 2 Outline of ASTRANSAC 2.1 Translation Method Translation part of ASTRANSAC is based on the trans- fer method. All components except for the morpho- logical analyzer and the morphological generator are language-independent; the same software is used for both English-to-Japanese and Japanese-to-English transla- tion. The subsequent sections expound the translation process. 2.1.1 Dictionaries and Morphological Analysis ASTRANSAC has three kinds of dictionaries: (1) Common word dictionary (50,000 words) (2) Technical term dictionary (maximum of 200,000 technical terms) (3) User-defined word dictionary (maximum of 200,000 words for one field) The entry words and their target equivalents in the user-defined dictionary have the highest priority in trans- lation. The morphological analyzer constructs word lat- tice for input sentences and produces a sequence of word structures, which in turn will be the input of the syn- tactic analyzer. The Japanese morphological analyzer utilizes word connection information to eliminate mor- phologically implausible word sequences. 2.1.2 Syntactic Analysis Syntactic analysis and semantic analysis …

Content truncated to 500 words