Method and device for translating a source text into a target using modeling and dynamic programming

• US 6,182,026 B1
• Filed: 06/26/1998
• Issued: 01/30/2001
• Est. Priority Date: 06/26/1997
• Status: Expired due to Fees

• Current Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Original Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Inventors: Vogel, Stephan, Tillmann, Christoph, Ney, Hermann
• Primary Examiner(s): Isen, Forester W.
• Assistant Examiner(s): EDOUARD, PATRICK NESTOR
• Application Number:
  US09/105,855
• Time in Patent Office: 949 Days
• Field of Search: 704/2-8, 704/236, 704/241, 704/255, 704/256
• US Class Current: 704/2
• CPC Class Codes: G06F 40/42 Data-driven translation
• CPC Class Codes: G06F 40/44 Statistical methods, e.g. probability models

Abstract

For translating a word-organized source text into a word-organized target text through mapping of source words on target words, both a translation model and a language model are used. In particular, alignment probabilities are ascertained between various source word & target word pairs, whilst preemptively assuming that alignment between such word pairs is monotonous through at least substantial substrings of a particular sentence. This is done by evaluating incrementally statistical translation performance of various target word strings, deciding on an optimum target word string, and outputting the latter.

Claims

1. A method for translating a received source text into a target text through mapping source words on target words, wherein said method comprising the steps of:

ascertaining alignment probabilities between source word and target word pairs;

assuming that alignment between such word pairs is monotonous through at least substantial substrings of a particular sentence;

evaluating, using a Hidden Markov Model and accelerated dynamic programming recombination, statistical translation performance of various target word strings; and

deciding on an optimum target word string.

2. The method as claimed in claims 1, wherein the translating a received source text into a target text uses Bayesian Decision Theory.

3. The method as claimed in claims 1, wherein the translating a received source text into a target text uses a dynamic programming beam search.

4. The method as claimed in claim 1, wherein the translating a received source text into a target text uses accelerated dynamic programming recombination.

5. The method as claimed in claim 1, further including the step of reordering the word sequence of the source sentence.

6. The method as claimed in claim 1, further including the step of integrating a word reordering into a dynamic programming search.

7. The method as claimed in any of claim 1, furthermore integrating with a speech recognition method.

8. The method as claimed in claim 1, furthermore comprises a preprocessing step for executing at least one of the following reductive measures:

replacing selective word categories,

treating punctuation marks like regular words,

replacing certain politeness phrases by single words,

counteracting the phenomenon that personal pronouns may form part of an inflected verb, by splitting the verb into a verb part and a pronoun part;

recording words and phrases that have particular positions of other pronouns.

9. A device for implementing the method as claimed in claim 1.