System and method of providing a spoken dialog interface to a website

• US 8,060,369 B2
• Filed: 07/31/2009
• Issued: 11/15/2011
• Est. Priority Date: 12/18/2002
• Status: Expired due to Fees

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: AT&T Intellectual Property II LP (AT&T, Inc.)
• Inventors: Bangalore, Srinivas, Feng, Junlan, Rahim, Mazin G.
• Primary Examiner(s): Smits, Talivaldis Ivars
• Application Number:
  US12/533,300
• Publication Number:
  US 20090292529A1
• Time in Patent Office: 837 Days
• Field of Search: 704/257, 704/260, 704/270.1, 704/275
• US Class Current: 704/257
• CPC Class Codes: G10L 13/08 Text analysis or generation of parameters for speech synthesis out of text, e.g. grapheme to phoneme translation, prosody generation or stress or intonation determination
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: H04M 3/4936 Speech interaction details speech recognition per se G10L15/00

Abstract

Disclosed is a system and method for training a spoken dialog service component from website data. Spoken dialog service components typically include an automatic speech recognition module, a language understanding module, a dialog management module, a language generation module and a text-to-speech module. The method includes converting data from a structured database associated with a website to a structured text data set and a structured task knowledge base, extracting linguistic items from the structured database, and training a spoken dialog service component using at least one of the structured text data, the structured task knowledge base, or the linguistic items. The system includes modules configured to implement the method.

Claims

1. A method for training a spoken dialog service component, the method comprising:

converting, via a processor, data from a structured database associated with a website to a structured text data set and a structured task knowledge base;

extracting, via a processor, linguistic items from the structured database; and

training, via a processor, a spoken dialog service component using at least one of the structured text data set, the structured task knowledge base, and the linguistic items.

2. The method of claim 1, wherein the spoken dialog service component is selected from the group consisting of a language understanding module, a language generation module, a dialog manager, and an automatic speech recognition engine.

3. The method of claim 1, wherein the structured text data set comprises a hierarchical tree.

4. The method of claim 3, wherein the hierarchical tree comprises a plurality of non-leaf tree nodes having a node alias that is human understandable.

5. The method of claim 4, wherein each non-leaf tree node further comprises a verbose description document and a concise summary.

6. The method of claim 5, wherein the verbose description document and concise summary are used for information matching and help prompt construction during dialog execution.

7. The method of claim 3, wherein the hierarchical tree further comprises at least one leaf node corresponding to a webpage.

8. The method of claim 1, further comprising encoding each webpage in the website as a sequence of feature vectors.

9. The method of claim 8, further comprising detecting boundaries between information units on each webpage in the website to yield detected boundaries.

10. The method of claim 9, further comprising classifying information blocks organized according to the detected boundaries between information units into classified information blocks, wherein the classified information blocks are utilized for generating a spoken dialog interface to the website.

11. The method of claim 1, wherein the linguistic items comprise named-entities.

12. The method of claim 11, wherein the structured text data set further comprises nominal, verbal, and adjectival phrases.

13. A system for training a spoken dialog service component, the system comprising:

a first module configured to control a processor to convert semi-structured heterogeneous web data on a website to a structured text data set and a structured task knowledge base;

a second module configured to control a processor to extract linguistic items from the semi-structured heterogeneous web data; and

a third module configured to control a processor to train a spoken dialog service component using at least one of the structured text data set, the structured task knowledge base, and the linguistic items.

14. The system of claim 13, wherein the spoken dialog service component is selected from the group consisting of a language understanding module, a language generation module, a dialog manager, an automatic speech recognition engine, and a text-to-speech synthesizer.

15. The system of claim 13, wherein the structured text data set comprises a hierarchical tree.

16. The system of claim 15, wherein the hierarchical tree comprises a plurality of non-leaf tree nodes having a node alias that is human understandable.

17. The system of claim 16, wherein each non-leaf tree node further comprises a verbose description document and a concise summary.

18. The system of claim 17, wherein the verbose description document and concise summary are used for information matching and help prompt construction during dialog execution.

19. A computer-readable storage medium storing instructions executable on a processor and usable to train a spoken dialog service component, the instructions causing the processor to perform the steps:

converting semi-structured heterogeneous web data on a website to a structured text data set and a structured task knowledge base;

extracting linguistic items from the semi-structured heterogeneous web data; and

training a spoken dialog service component using at least one of the structured text data set, the structured task knowledge base, and the linguistic items.

20. The computer-readable storage medium of claim 19, further comprising the steps:

encoding each webpage in the website as a sequence of feature vectors;

detecting boundaries between information units on each webpage in the website to yield detected boundaries; and

classifying information blocks organized according to the detected boundaries between information units into classified information blocks, wherein the classified information blocks are utilized for generating a spoken dialog interface to the website.