Machine Translation Using Vector Space Representations

Abstract

Disclosed herein are methods, articles of manufacture, and systems for translating text. Such a method includes generating a conceptual representation space based on a plurality of source-language documents and a plurality of target-language documents. The method also includes generating, in the conceptual representation space, respective representations of a new source-language document and each of a plurality of dictionaries. The method further includes selecting a first dictionary from the plurality of dictionaries responsive to a similarity between the representation of the new source-language document and the representation of the first dictionary. The method still further includes translating, by using the first dictionary, a term in the new source-language document into a target-language term.

1 . A computer-implemented method for translating text, comprising:
 generating a conceptual representation space based on a plurality of source-language documents and a plurality of target-language documents;   generating, in the conceptual representation space, respective representations of a new source-language document and each of a plurality of dictionaries;   selecting a first dictionary from the plurality of dictionaries responsive to a similarity between the representation of the new source-language document and the representation of the first dictionary; and   translating, by using the first dictionary, a term in the new source-language document into a target-language term.   
     
     
         2 . The method of  claim 1 , wherein generating a representation of each of the plurality of dictionaries comprises:
 concatenating terms in each of the plurality of dictionaries into a single text object; and   generating a representation of the single text object in the conceptual representation space.   
     
     
         3 . The method of  claim 1 , wherein generating a representation of each of the plurality of dictionaries comprises:
 subdividing each of the plurality of dictionaries into conceptually cohesive segments;   generating a text object for each conceptually cohesive segment; and   generating a representation of each text object in the conceptual representation space.   
     
     
         4 . The method of  claim 1 , wherein the conceptual representation space is a Latent Semantic Indexing (LSI) space. 
     
     
         5 . The method of  claim 1 , further comprising:
 determining respective similarities between the representation of the new source-language document and the representation of each of the plurality of dictionaries.   
     
     
         6 . The method of  claim 5 , wherein the similarity between the representation of the new source-language document and the representation of the first dictionary is greater than the other similarities. 
     
     
         7 . A computer-program product comprising a computer-readable storage medium having instructions stored thereon that, if executed by a computing device, cause the computing device to perform a method for translating text, the method comprising:
 generating a conceptual representation space based on a plurality of source-language documents and a plurality of target-language documents;   generating, in the conceptual representation space, respective representations of a new source-language document and each of a plurality of dictionaries;   selecting a first dictionary from the plurality of dictionaries responsive to a similarity between the representation of the new source-language document and the representation of the first dictionary; and   translating, by using the first dictionary, a term in the new source-language document into a target-language term.   
     
     
         8 . The computer-program product of  claim 7 , wherein generating a representation of each of the plurality of dictionaries comprises:
 concatenating terms in each of the plurality of dictionaries into a single text object; and   generating a representation of the single text object in the conceptual representation space.   
     
     
         9 . The computer-program product of  claim 7 , wherein generating a representation of each of the plurality of dictionaries comprises:
 subdividing each of the plurality of dictionaries into conceptually cohesive segments;   generating a text object for each conceptually cohesive segment; and   generating a representation of each text object in the conceptual representation space.   
     
     
         10 . The computer-program product of  claim 7 , wherein the conceptual representation space is a Latent Semantic Indexing (LSI) space. 
     
     
         11 . The computer-program product of  claim 7 , wherein the method further comprises:
 determining respective similarities between the representation of the new source-language document and the representation of each of the plurality of dictionaries.   
     
     
         12 . The computer-program product of  claim 11 , wherein the similarity between the representation of the new source-language document and the representation of the first dictionary is greater than the other similarities. 
     
     
         13 . A computing system, comprising:
 a memory; and   a processor coupled to the memory, wherein the processor is configured to execute a method for translating text, the method comprising:
 generating a conceptual representation space based on a plurality of source-language documents and a plurality of target-language documents; 
 generating, in the conceptual representation space, respective representations of a new source-language document and each of a plurality of dictionaries; 
 selecting a first dictionary from the plurality of dictionaries responsive to a similarity between the representation of the new source-language document and the representation of the first dictionary; and 
 translating, by using the first dictionary, a term in the new source-language document into a target-language term. 
   
     
     
         14 . The computing system of  claim 13 , wherein generating a representation of each of the plurality of dictionaries comprises:
 concatenating teens in each of the plurality of dictionaries into a single text object; and   generating a representation of the single text object in the conceptual representation space.   
     
     
         15 . The computing system of  claim 13 , wherein generating a representation of each of the plurality of dictionaries comprises:
 subdividing each of the plurality of dictionaries into conceptually cohesive segments;   generating a text object for each conceptually cohesive segment; and   generating a representation of each text object in the conceptual representation space.   
     
     
         16 . The computing system of  claim 13 , wherein the conceptual representation space is a Latent Semantic Indexing (LSI) space. 
     
     
         17 . The computing system of  claim 13 , wherein the method further comprises:
 determining respective similarities between the representation of the new source-language document and the representation of each of the plurality of dictionaries;   
     
     
         18 . The computing system of  claim 17 , wherein the similarity between the representation of the new source-language document and the representation of the first dictionary is greater than the other similarities.