At present the major approaches include rule-based machine translation which heavily relies on linguistic analysis and representation at various linguistic levels, and example-based machine translation and statistical machine translation, both of which follow a more general corpus-based approach and make use of parallel corpora as a primary resource.