SYSTEM AND METHOD FOR ENRICHING SPOKEN LANGUAGE TRANSLATION WITH DIALOG ACTS

• US 20100131260A1
• Filed: 11/26/2008
• Published: 05/27/2010
• Est. Priority Date: 11/26/2008
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: AT&T Intellectual Property I LP (AT&T, Inc.)
• Inventors: Bangalore, Srinivas, Rangarajan Sridhar, Vivek Kumar
• Granted Patent: US 8,374,881 B2
• Time in Patent Office: Days
• US Class Current: 704/3
• CPC Class Codes: G06F 40/35 Discourse or dialogue representation
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30
• CPC Class Codes: G06F 40/58 Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

Disclosed herein are systems, computer-implemented methods, and tangible computer-readable media for enriching spoken language translation with dialog acts. The method includes receiving a source speech signal, tagging dialog acts associated with the received source speech signal using a classification model, dialog acts being domain independent descriptions of an intended action a speaker carries out by uttering the source speech signal, producing an enriched hypothesis of the source speech signal incorporating the dialog act tags, and outputting a natural language response of the enriched hypothesis in a target language. Tags can be grouped into sets such as statement, acknowledgement, abandoned, agreement, question, appreciation, and other. The step of producing an enriched translation of the source speech signal uses a dialog act specific translation model containing a phrase translation table.

Claims

1. A computer-implemented method of enriching spoken language translation with dialog acts, the method comprising:

receiving a source speech signal;

tagging dialog acts associated with the received source speech signal using a classification model, dialog acts being domain independent descriptions of an intended action a speaker carries out by uttering the source speech signal;

producing an enriched hypothesis of the source speech signal incorporating the tagged dialog acts; and

outputting a version of the enriched hypothesis translated into a target language.

2. The computer-implemented method of claim 1, wherein tags are grouped into sets.

3. The computer-implemented method of claim 2, wherein sets include statement, acknowledgement, abandoned, agreement, question, and appreciation.

4. The computer-implemented method of claim 1, wherein outputting the version of the enriched hypothesis uses a dialog act specific translation model containing a phrase translation table.

5. The computer-implemented method of claim 4, the method further comprising:

appending to each phrase translation table associated with a particular dialog act specific translation model those entries from a complete model that are not present in the phrase table of the dialog act specific translation model; and

weighting appended entries by a factor α

.

6. The computer-implemented method of claim 4, wherein the dialog act specific translation model is a bag-of-words translation model.

7. The computer-implemented method of claim 1, wherein the source speech signal is part of a dialog turn having multiple sentences, the method further comprising:

segmenting the source speech signal;

tagging dialog acts in each segment using a maximum entropy model; and

producing an enriched hypothesis of each segment incorporating the tagged dialog acts.

8. The computer-implemented method of claim 1, the method further including annotating tagged dialog acts.

9. The computer-implemented method of claim 1, wherein the classification model is a maximum entropy model.

10. A system for enriching spoken language translation with dialog acts, the system comprising:

a module configured to receive a source speech signal;

a module configured to tag dialog acts associated with the received source speech signal using a classification model, dialog acts being domain independent descriptions of an intended action a speaker carries out by uttering the source speech signal;

a module configured to produce an enriched hypothesis of the source speech signal incorporating the tagged dialog acts; and

a module configured to output a version of the enriched hypothesis translated into a target language.

11. The system of claim 10, wherein tags are grouped into sets.

12. The system of claim 11, wherein sets include statement, acknowledgement, abandoned, agreement, question, and appreciation.

13. The system of claim 10, wherein the module configured to output the version of the enriched hypothesis uses a dialog act specific translation model containing a phrase translation table.

14. The system of claim 13, the system further comprising:

a module configured to append to each phrase translation table associated with a particular dialog act specific translation model those entries from a complete model that are not present in the phrase table of the dialog act specific translation model; and

a module configured to weight appended entries by a factor α

.

15. The system of claim 13, wherein the dialog act specific translation model is a bag-of-words translation model.

16. The system of claim 10, wherein the source speech signal is part of a dialog turn having multiple sentences, the system further comprising:

a module configured to segment the source speech signal;

a module configured to tag dialog acts in each segment using a maximum entropy model; and

a module configured to produce an enriched hypothesis of each segment incorporating the tagged dialog acts.

17. A tangible computer-readable medium storing a computer program having instructions for enriching spoken language translation with dialog acts, the instructions comprising:

receiving a source speech signal;

tagging dialog acts associated with the received source speech signal using a classification model, dialog acts being domain independent descriptions of an intended action a speaker carries out by uttering the source speech signal;

producing an enriched hypothesis of the source speech signal incorporating the tagged dialog acts; and

outputting a version of the enriched hypothesis translated into a target language.

18. The tangible computer-readable medium of claim 17, wherein tags are grouped into sets.

19. The tangible computer-readable medium of claim 18, wherein sets include statement, acknowledgement, abandoned, agreement, question, and appreciation.

20. The tangible computer-readable medium of claim 17, wherein outputting the version of the enriched hypothesis uses a dialog act specific translation model containing a phrase translation table.