HYBRIDIZING RULE-BASED AND EXAMPLE-BASED APPROACHES IN MACHINE AIDED TRANSLATION SYSTEM

摘要

Two distinct approaches have emerged over decades of research in the machine translation area. One group of researchers believes in encoding large complex linguistic knowledge bases in the form of rules and invoking them to achieve translation. Another group of researchers believes that the real answer to MT lies in exploiting statistical information from the corpora and in using the bi-lingual corpora as examples as they believe that building such large knowledge bases is impractical and can never be presumed to be complete. Their critics believe that the problem of MT is too complex to be handled by statistics and examples. An obvious practical approach is to hybridize the rule-based and corpus-based methodologies to get the best out of these.