E-Services Translation Utilizing Machine Translation and Translation Memory

Abstract

A system and method for translating data from a source language to a target language is provided wherein machine generated target translation of a source sentence is compared to a database of human generated target sentences. If a matching human generated target sentence is found, the human generated target sentence may be used instead of the machine generated sentence, since the human generated target sentence is more likely to be a well-formed sentence than the machine generated sentence. The system and method does not rely on a translation memory containing pairs of sentences in both source and target languages, and minimizes the reliance on a human translator to correct a translation generated by machine translation.

What is claimed is: 
     
         1 . A machine translation method comprising:
 receiving source data in a source language for translation to a target language;   searching a translation memory including source strings of the source language that have already been translated into the target language and corresponding target strings of the target language, the searching to identify a match between a first phrase of the source data and a phrase of a first source string of the source language;   extracting a first target phrase from a target string corresponding to the first source string;   searching the translation memory to identify a match between a second phrase of the source data and a phrase of a second source string of the source language;   extracting a second target phrase from a target string corresponding to the second source string;   building translated data including the first target phrase and second target phrase;   searching the translation memory for a target string matching the translated data; and   sending the matching target string for subsequent review by a translator if the matching target string is found.   
     
     
         2 . The method of  claim 1 , wherein the second source string is different from the first source string. 
     
     
         3 . The method of  claim 1 , further comprising aligning sentences of the translation memory from source strings of the source language to target strings of the target language. 
     
     
         4 . The method of  claim 1 , further comprising aligning phrases of source strings in the translation memory to phrases to phrases from target strings the in the translation memory. 
     
     
         5 . The method of  claim 1 , further comprising aligning words of source strings in the translation memory to words from target strings in the target language. 
     
     
         6 . A machine translation system comprising:
 a translation memory for storing source strings of a source language that have already been translated to target strings of a target language;   a portal for receiving source data in the source language from a user and providing a translation of the source data to the user; and   a machine translation module for:
 matching phrases of the source data to phrases of source strings; 
 building translated data from phrases of the target strings corresponding to matched phrases of the source strings; 
 searching the translation memory for a matching target string; and 
 providing the matching target string as a translation of the source data to the portal. 
   
     
     
         7 . The machine translation system of  claim 6 , wherein matching phrases of the source data to phrases of source strings comprises:
 matching a first phrase of the source data to a phrase of a first source string to identify a first target phrase corresponding to the matched phrase of the first source string; and   matching a second phrase of the source data to a phrase of a second source string to identify a second target phrase corresponding to the matched phrase of the second source string, the translated data built from phrases including the first target phrase and the second target phrase.   
     
     
         8 . The machine translation system of  claim 7 , wherein the first source string is a different string from the second source string. 
     
     
         9 . The machine translation system of  claim 6 , wherein the translation memory includes a glossary. 
     
     
         10 . The machine translation system of  claim 6 , further comprising means for translating the source data using a glossary. 
     
     
         11 . The machine translation system of  claim 6 , wherein the sentences of source strings and target strings in the translation memory are aligned. 
     
     
         12 . The machine translation system of  claim 6 , wherein the phrases of source strings and target strings in the translation memory are aligned. 
     
     
         13 . The machine translation system of  claim 6 , wherein the words of source strings and target strings in the translation memory are aligned. 
     
     
         14 . The machine translation system of  claim 6 , wherein the machine translation module adds the source data and matching string to the translation memory. 
     
     
         15 . A machine translation method for translating data from a source language to a target language, the method comprising:
 receiving source data in a source language for translation to a target language;   matching phrases of the source data to phrases of source strings of the source language that have already been translated to phrases of target strings of the target language, the source and target strings stored in a translation memory;   building translated data from phrases of the target strings corresponding to matched phrases of the source strings;   searching the translation memory for a matching target string; and   sending the matching target string for subsequent review by a translator if the matching target string is found.   
     
     
         16 . The method of  claim 15 , wherein matching phrases of the source data to phrases of source strings comprises:
 matching a first phrase of the source data to a phrase of a first source string to identify a first target phrase corresponding to the matched phrase of the first source string; and   matching a second phrase of the source data to a phrase of a second source string different from the first source string to identify a second target phrase corresponding to the matched phrase of the second source string, the translated data built from phrases including the first target phrase and the second target phrase.   
     
     
         17 . The method of  claim 15 , further comprising using the translation memory to align sentences from the source language to the target language. 
     
     
         18 . The method of  claim 15 , further comprising using the translation memory to align phrases from the source language to the target language. 
     
     
         19 . The method of  claim 15 , further comprising using the translation memory to align words from the source language to the target language. 
     
     
         20 . The method of  claim 15 , wherein the source data is a glossary for translation.