METHOD AND APPARATUS FOR CROSS-LINGUAL COMMUNICATION

• US 20090204386A1
• Filed: 04/15/2009
• Published: 08/13/2009
• Est. Priority Date: 09/05/2003
• Status: Active Grant

• Current Assignee: ZAMA Innovations LLC
• Original Assignee: Barton Friedland, Gerald Richard Cain, Mark Seligman, Mike Dillinger
• Inventors: Cain, Gerald Richard, Seligman, Mark, Friedland, Barton, Dillinger, Mike
• Granted Patent: US 8,239,207 B2
• Time in Patent Office: Days
• US Class Current: 704/2
• CPC Class Codes: G06F 40/268 Morphological analysis
• 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 13/00 Speech synthesis; Text to speech systems
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications

Abstract

A system and method for a highly interactive style of speech-to-speech translation is provided. The interactive procedures enable a user to recognize, and if necessary correct, errors in both speech recognition and translation, thus providing robust translation output than would otherwise be possible. The interactive techniques for monitoring and correcting word ambiguity errors during automatic translation, search, or other natural language processing tasks depend upon the correlation of Meaning Cues and their alignment with, or mapping into, the word senses of third party lexical resources, such as those of a machine translation or search lexicon. This correlation and mapping can be carried out through the creation and use of a database of Meaning Cues, i.e., SELECT. Embodiments described above permit the intelligent building and application of this database, which can be viewed as an interlingua, or language-neutral set of meaning symbols, applicable for many purposes. Innovative techniques for interactive correction of server-based speech recognition are also described.

Claims

1. A method enabling cross linguistic communication through speech and text machine translation, comprising method operations of:

receiving a first and a second language selection;

receiving an expression in the first language;

presenting the expression in the first language for verification;

translating the verified expression into an expression in the second language;

confirming the meaning of terms within the verified expression in the first language;

back-translating the expression in the second language to a back-translated expression in the first language; and

verifying the back-translated expression in the first language.

2. The method of claim 1, wherein the method operation of confirming the meaning of terms within the verified expression in the first language includes,

receiving a selection of one of the terms;

accessing a database supplying Meaning Cues associated with the one of the terms, each of the Meaning Cues representing a sense of the one of the terms; and

selecting one of the Meaning Cues.

3. The method of claim 2, wherein the method operation of selecting one of the multipleprofiles includes,

determining if one of the Meaning Cues represents a preferred word sense;

and if one of the Meaning Cues represents a preferred word sense then the method includes,

selecting the preferred word sense via a relevant Meaning Cue.

4. The method of claim 1, wherein the first and the second language selection are received as voice commands.

5. The method of claim 1, wherein the expression in the first language and the expression in the second language are communicated as both a textual expression and an audio expression.

6. The method of claim 2, wherein each of the multiple profiles is a list of synonyms associated with the sense.

7. A method for mapping word senses across lexical resources, comprising method operations of:

selecting a target term from a first lexical resource;

identifying a restricted range of possible matches to the target term from a second lexical resource;

calculating a semantic distance between possible matches to the target term and the target term;

ranking the possible matches to the target term according to relevance; and

associating the target term from the first lexical resource with one of the possible matches having a lowest semantic distance from the target term.

8. The method of claim 7, wherein the method operation of identifying possible matches to the target term from a second lexical resource includes,

identifying word senses of the second lexical resource based on the target term.

9. The method of claim 7, wherein the method operation of identifying possible matches to the target term from a second lexical resource includes,

identifying word senses of the second lexical resource based on Profile terms associated with the target term.

10. The method of claim 7, wherein the method operation of identifying possible matches to the target term from a second lexical resource includes,

identifying word senses of the second lexical resource based on definitions of candidate senses.

11. The method of claim 9, wherein the Profile terms are selected from the group consisting of synonym, hypernyms, hyponyms, and neighboring semantic terms.

12. The method of claim 7, wherein the method operation of identifying possible matches to the target term from a second lexical resource includes,

identifying word senses of the second lexical resource based on component words of the target term.

13. The method of claim 7, wherein the semantic distance represents a degree of similarity between the target term and each of the possible matches.

14. The method of claim 7, wherein the method operation of ranking the possible matches to the target term according to relevance includes,

assigning a weight to each of the possible matches based upon a technique used to associate the target term and respective possible matches.

15. The method of claim 7, wherein the method operation of ranking the possible matches to the target term according to relevance includes,

determining the ranking by a matching technique selected from a group consisting of a target term match, a profile match, a definition match, or a target term component match;

and

extracting a first set of candidates from the target term match technique and a second set of candidates from remaining matching techniques.

16. A computer readable medium having program instructions for mapping word senses across lexical resources, comprising:

program instructions for selecting a target term from a first lexical resource;

program instructions for identifying a restricted range of possible matches to the target term from a second lexical resource;

program instructions for calculating a semantic distance between possible matches to the target term and the target term;

program instructions for ranking the possible matches to the target term according to relevance; and

program instructions for associating the target term from the first lexical resource with one of the possible matches having a lowest semantic distance from the target term.

17. The computer readable medium of claim 16, wherein the program instructions for identifying possible matches to the target term from a second lexical resource includes,

program instructions for identifying word senses of the second lexical resource based on the target term.

18. The computer readable medium of claim 16, wherein the program instructions for identifying possible matches to the target term from a second lexical resource includes,

program instructions for identifying word senses of the second lexical resource based on Profile terms associated with the target term.

19. The computer readable medium of claim 16, wherein the program instructions for identifying possible matches to the target term from a second lexical resource includes,

program instructions for identifying word senses of the second lexical resource based on definitions of candidate senses.

20. The computer readable medium of claim 18, wherein the Profile terms are selected from the group consisting of synonym, hypernyms, hyponyms, and neighboring semantic terms.

21. The computer readable medium of claim 16, wherein the program instructions for identifying possible matches to the target term from a second lexical resource includes,

program instructions for identifying word senses of the second lexical resource based on component words of the target term.

22. The computer readable medium of claim 16, wherein the semantic distance represents a degree of similarity between the target term and each of the possible matches.

23. The computer readable medium of claim 16, wherein the program instructions for ranking the possible matches to the target term according to relevance includes,

program instructions for assigning a weight to each of the possible matches based upon a technique used to associate the target term and respective possible matches.

24. The computer readable medium of claim 16, wherein the program instructions for ranking the possible matches to the target term according to relevance includes,

program instructions for determining the ranking by a matching technique selected from a group consisting of a target term match, a profile match, a definition match, or a target term component match;

and

program instructions for extracting a first set of candidates from the target term match technique and a second set of candidates from remaining matching techniques.