METHOD AND SYSTEM FOR IMPROVING MACHINE TRANSLATION ACCURACY USING TRANSLATION MEMORY

A combination of Machine Translation (MT) techniques are used with a `Hybrid` Translation Memory (HTM)database. The HTM database stores malformed translated sentences matched with their correct translations. Each sentence that is translated with MT is analyzed by the MT algorithms to determine a value for the accuracy of the translation. If the accuracy is greater than or equal to a preset tolerance level, the sentence is considered to be translated sufficiently well. Therefore, no database need be checked for all the sufficiently translated sentences. However, if the accuracy is below the tolerance, an HTM database is checked for the same bad translation and if found, replaced with the matching correct translation.