2.3.4 Example-Based Machine Translation
After Makoto Nagao, a professor of natural language processing, had realised that "Man does not translate a simple sentence by doing deep linguistic analysis”[7] many developers followed him. Additionaly, he is of the opinion humans divide a sentence into fragmental phrases and piece the fragmental translations together into one sentence in the target language. You can say, his fundamental idea is translation by analogy which is the principle of so-called example-based machine translation systems. In their dictionaries not only words and their corresponding translations are included but also compound nouns, terms and idioms. These large dictionaries make example-based systems also able to detect and translate word pairs, collocations and idioms. Although using these fragments of human translations, dictionary entries, practice shows that this system is not a significant improvement of previous translation systems.