×

Use of annotations in statistical machine translation

  • US 9,122,674 B1
  • Filed: 12/15/2006
  • Issued: 09/01/2015
  • Est. Priority Date: 12/15/2006
  • Status: Active Grant
First Claim
Patent Images

1. A method for translating a document, comprising:

  • defining translation rules associated with annotations occurring in a translation memory database, wherein the translation rules comprise an output context;

    integrating the translation rules in the translation memory database into statistical machine translation used by a translation engine, the integration including training the translation engine to recognize annotations in the translation memory database while performing statistical machine translation;

    training the statistical machine translation of the translation engine using both the translation memory database and parallel segments to modify statistical probability values;

    storing the statistical probability values modified during the training, the stored values for use in the statistical machine translation by the translation engine;

    receiving a source document in a source language, the source document comprising a string and an annotated segment of text, the annotated segment of text comprising text and an instance of an annotation;

    translating the source document as a whole using the translation engine without breaking the source document into segments that are translated according to whether a match appears in the translation memory database, the translation engine configured for;

    identifying annotations and associated annotated segments of the text in the source document to be processed using the translation rules, and strings in the source document to be translated using statistical machine translation,translating the identified strings in the source document using the statistical machine translation, andprocessing the identified instances of the annotations and the associated annotated segments of text according to the translation rules using the translation memory database, at least one of the translation rules being associated with the annotation; and

    generating a target document in a target language based on the translated source document and the translated text associated with the annotation.

View all claims
  • 3 Assignments
Timeline View
Assignment View
    ×
    ×