Library of existing spoken dialog data for use in generating new natural language spoken dialog systems

• US 20060149554A1
• Filed: 01/05/2005
• Published: 07/06/2006
• Est. Priority Date: 01/05/2005
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: AT&T Corporation (AT&T, Inc.)
• Inventors: Gibbon, David Crawford, Shahraray, Behzad, Hakkani-Tur, Dilek Z., Liu, Zhu, Renger, Bernard S., Tur, Gokhan, Begeja, Lee, Fabbrizio, Giuseppe Di
• Granted Patent: US 8,478,589 B2
• Time in Patent Office: Days
• US Class Current: 704/275
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/28 Constructional details of speech recognition systems
• CPC Class Codes: G10L 25/48 specially adapted for particular use

Abstract

A machine-readable medium may include a group of reusable components for building a spoken dialog system. The reusable components may include a group of previously collected audible utterances. A machine-implemented method to build a library of reusable components for use in building a natural language spoken dialog system may include storing a dataset in a database. The dataset may include a group of reusable components for building a spoken dialog system. The reusable components may further include a group of previously collected audible utterances. A second method may include storing at least one set of data. Each one of the at least one set of data may include ones of the reusable components associated with audible data collected during a different collection phase.

Claims

1. A machine-readable medium comprising:

a plurality of reusable components for building a spoken dialog system, the reusable components including a plurality of previously collected audible utterances.

2. The machine-readable medium of claim 1, wherein the reusable components are organized into a plurality of datasets.

3. The machine-readable medium of claim 2, wherein each of the datasets includes data pertaining to a particular industrial sector.

4. The machine-readable medium of claim 2, wherein each of the datasets comprises:

at least one collection of data, each one of the at least one collection of data includes data collected during a different data collection phase.

5. The machine-readable medium of claim 1, further comprising:

a first set of instructions and data for implementing at least one natural language understanding model,

a second set of instructions and data for implementing at least one automatic speech recognition module, and

a third set of instructions and data for implementing at least one named entity detection/extraction grammar or model.

6. The machine-readable medium of claim 1, wherein:

the audible utterances are associated with at least one occurrence of utterance data stored on the medium, the at least one occurrence of utterance data including information indicating to which of the associated audible utterances a label applies or to which of the associated audible utterances a label does not apply.

7. The machine-readable medium of claim 6, wherein one of the at least one occurrence of utterance data further comprises:

transcribed data associated with at least some of the associated audible utterances, and

labeled data providing information regarding a label associated with the one of the at least one occurrence of utterance data.

8. The machine-readable medium of claim 6, wherein the at least one occurrence of utterance data further comprises information indicating an associated category.

9. The machine-readable medium of claim 2, wherein each of the datasets is stored in an XML database.

10. The machine-readable medium of claim 2, wherein each of the datasets is stored in a relational database.

11. A machine-implemented method to build a library of reusable components for use in building a natural language spoken dialog system, comprising:

storing a dataset in a database, the dataset including a plurality of reusable components for building a spoken dialog system, the reusable components including a plurality of previously collected audible utterances.

12. The machine-implemented method of claim 11, further comprising:

storing a plurality of datasets, wherein each of the datasets further comprises a plurality of reusable components pertaining to one of a different industrial sector or a different category.

13. The machine-implemented method of claim 11, wherein:

the dataset includes a plurality of sets of ones of the reusable components, each of the sets including data collected during a respective data collection phase.

14. The machine-implemented method of claim 11, wherein the storing a dataset in a database further comprises:

storing a set of data including ones of the reusable components associated with data collected during a data collection phase.

15. The machine-implemented method of claim 14, wherein the storing a set of data comprises:

storing a plurality of label information items, and

storing a plurality of occurrences of utterance data, each of the occurrences of utterance data being associated with one of the stored label information items and including a plurality of guideline utterance items, each of the guideline utterance items being one of a positive utterance item to which the one of the stored label information applies, or a negative utterance item to which the one of the stored label information does not apply.

16. The machine-implemented method of claim 15, further comprising:

assigning an attribute to one of the label information items.

17. The machine-implemented method of claim 16, wherein the attribute is a category attribute or a verb attribute.

18. The machine-implemented method of claim 15, wherein at least some of the occurrences of guideline utterance items include transcription data.

19. The machine-implemented method of claim 16, wherein the attribute indicates whether the one of the labeled information items is generic, reusable, or specific to a given application.

20. The machine-implemented method of claim 11, wherein the storing a dataset in a database further comprises:

storing information regarding a natural language understanding model;

storing information regarding an automatic speech recognition module; and

storing information regarding a named entity grammar.

21. The machine-implemented method of claim 11, wherein the database is a relational database.

22. The machine-implemented method of claim 21, wherein the database is an XML database.

23. A method to build a library of reusable components for use in building a natural language spoken dialog system, comprising:

storing at least one set of data, each one of the at least one set of data including ones of the reusable components associated with audible data collected during a different collection phase.

24. The method of claim 23, wherein the storing at least one set of data further comprises:

storing call-type information in each one of the at least one set of data.

25. The method of claim 24, wherein the storing at least one set of data further comprises:

storing utterance data in each one of the at least one set of data, and

associating the call-type information with the utterance data stored in respective ones of the at least one set of data.

26. The method of claim 23, wherein the storing at least one set of data further comprises:

storing a plurality of sets of data.

27. The method of claim 23, further comprising:

storing a plurality of sectors in a database, the storing a plurality of sectors comprising;

storing, in each one of the sectors, at least one set of data, each one of the at least one set of data including ones of the reusable components associated with audible data collected during a different collection phase.

28. The method of claim 27, wherein each one of the sectors comprises information pertaining to a different industrial sector.