次へ: Statistical Machine Translation
上へ: Introduction
戻る: Introduction
Pattern-based machine translation [6] was proposed in the 1960s. This form of machine translation uses the source language and target language translation patterns made manually. It has certain advantages. In particular, if the input sentence matches a translation pattern, the translated sentence will be of high quality. However, it has disadvantages as well. It cannot translate input sentences that do not match any of the prepared translation patterns. This means that to match many sentences, we either have to make many patterns or relaxed these patterns. Consequently, problem becomes either too high a translation cost or poor translation accuracy. This tradeoff has thus far been difficult to surmount, and hence, the amount of research on pattern-based machine translation has declined.
平成25年9月17日