ABSTRACT
Various services, such as information retrieval and information extraction, using natural language processing technologies trained by huge corpora have become available. In the field of machine translation (MT), corpus-based machine translations, such as statistical machine translation (SMT) (Brown et al. 1993: 263–311) and example-based machine translation (EBMT), are typical applications of using a large volume of data in real business situations. Thanks to the availability of megadata, current machine translation systems have the capacity to produce quality translations for some specific language pairs. Yet there are still people who doubt the usefulness of machine translation, especially when it applies to translation among different families of languages, such as those of Japanese and English. A study was conducted to examine the types of machine translation systems which are useful (Fuji et al. 2001), by simulating the retrieval and reading of web pages in a language different from one’s mother tongue. Research on the technologies which make MT systems more useful in real world situations, however, is scanty.
