Methods and systems for enabling speech-based internet searches

• US 6,934,675 B2
• Filed: 06/14/2001
• Issued: 08/23/2005
• Est. Priority Date: 06/14/2001
• Status: Expired due to Term

• Current Assignee: Tamiras Per Pte Ltd LLC (Intellectual Ventures LLC)
• Original Assignee: Avaya Incorporated
• Inventors: Brown, Michael K., Glinski, Stephen C.
• Primary Examiner(s): MCFADDEN, SUSAN IRIS
• Application Number:
  US09/879,892
• Publication Number:
  US 20020194004A1
• Time in Patent Office: 1,531 Days
• Field of Search: 704/270, 704/275, 704/256, 704/257, 704/248, 704/246, 704/2, 704/9, 704/10, 707/3, 707/6
• US Class Current: 704/9
• CPC Class Codes: G06F 16/3335 Syntactic pre-processing, e.g. stopword elimination, stemming
• CPC Class Codes: G10L 15/197 Probabilistic grammars, e.g. word n-grams
• CPC Class Codes: G10L 2015/228 of application context

Abstract

Merged “grammars” derived from statistical indicators (e.g., N-grams and cohorts) are used to enable speech-based, Internet searches.

Claims

1. A method for enabling a speech-based, Internet search for generating alternative web sites, comprising:

generating a cohort in a cohort generation unit;

generating a language in a generation unit;

generating a merged grammar by combining the cohort and the language in a merging unit,

wherein the cohort is a cohort probability indicating a number of times a word will occur in a first range, wherein the language is a language probability indicating a number of times a word will occur in a second range, the first range being greater than the second range.

2. The method as in claim 1 wherein the language comprises an N-gram language.

3. The method as in claim 1 wherein the language comprises a finite state language.

4. The method as in claim 1 wherein the language comprises a context free grammar.

5. The method as in claim 1 wherein the language comprises a context sensitive language.

6. The method as in claim 1 wherein the grammar comprises a phoneme grammar.

7. The method as in claim 1 wherein the grammar comprises a word grammar.

8. The method as in claim 1 wherein generating the cohorts and language further comprises eliminating cohorts and parts of the language which are associated with web sites having low traffic flow.

9. The method as in claim 1 further comprising deleting repetitive cohorts or repetitive parts of the language.

10. The method as in claim 1 further comprising generating the cohorts and language from phonemes.

11. The method as in claim 1 further comprising:

comparing a phoneme associated with a spoken word against the merged grammar.

12. The method as in claim 11 wherein the merged grammar comprises one or more synonyms.

13. The method as in claim 11 wherein the merged grammar comprises one or more conjugates.

14. The method as in claim 11 wherein the merged grammar comprises part of speech identifiers.

15. The method as in claim 11 further comprising generating information correlated to the spoken word.

16. The method as in claim 15 wherein the information comprises a web site address.

17. The method as in claim 1, further comprising the steps of generating distances between words, and ignoring words associated with any distance which exceeds a threshold distance.

18. A system for enabling a speech-based, Internet search for generating alternative web sites, comprising:

a cohort generator adapted to generate cohorts;

a language generator adapted to generate a language; and

a merging unit adapted to generate a merged grammar by combining the cohorts and the languages,

wherein the cohort is a cohort probability indicating a number of times a word will occur in a first range, wherein the language is a language probability indicating a number of times a word will occur in a second range, the first range being greater than the second range.

19. The system as in claim 18 wherein the language comprises an N-gram language.

20. The system as in claim 18 wherein the language comprises a finite state language.

21. The system as in claim 18 wherein the language comprises a context free language.

22. The system as in claim 18 wherein the language comprises a context sensitive language.

23. The system as in claim 18 wherein the grammar comprises a phoneme grammar.

24. The system as in claim 18 wherein the grammar comprises a word grammar.

25. The system as in claim 18 further comprising a statistical unit adapted to eliminate cohorts and parts of the language associated with web sites having low traffic flow.

26. The system as in claim 18 further comprising an optimization unit adapted to delete repetitive cohorts or repetitive parts of the language.

27. The system as in claim 18 further comprising:

a recognition unit adapted to compare a phoneme associated with a spoken word against the merged grammar.

28. The system as in claim 27 wherein the grammar comprises one or more synonyms.

29. The system as in claim 27 wherein the grammar comprises one or more conjugates.

30. The system as in claim 27 wherein the grammar comprises part of speech identifiers.

31. The system as in claim 27 wherein the recognition unit is further adapted to generate information correlated to the spoken word.

32. The system as in claim 31 wherein the information comprises a web site address.

33. The system as in claim 18, wherein the merging unit is adapted to generate distances between words and to ignore words associated with any distance which exceeds a threshold distance.

34. A programmed medium for enabling a speech-based, Internet search for generating alternative web sites, adapted to:

generate cohorts in a cohort generating unit;

generate a language in a language generating unit; and

generate a merged grammar by combining the cohorts and language in a merging unit,

wherein the cohort is a cohort probability indicating a number of times a word will occur in a first range, wherein the language is a language probability indicating a number of times a word will occur in a second range, the first range being greater than the second range.

35. The programmed medium as in claim 34 wherein the language comprises an N-gram language.

36. The programmed medium as in claim 34 wherein the language comprises a finite state language.

37. The programmed medium as in claim 34 wherein the language comprises a context free language.

38. The programmed medium as in claim 34 wherein the language comprises a context sensitive language.

39. The programmed medium as in claim 34 further adapted to eliminate cohorts and parts of the language associated with web sites having low traffic flow.

40. The programmed medium as in claim 34 further adapted to delete repetitive cohorts or repetitive parts of the language.

41. The programmed medium as in claim 34 further adapted to compare a phoneme associated with a spoken word against the merged grammar.

42. The programmed medium as in claim 41 wherein the grammar comprises one or more synonyms.

43. The programmed medium as in claim 41 wherein the grammar comprises one or more conjugates.

44. The programmed medium as in claim 41 wherein the grammar comprises one or more part of speech identifiers.

45. The programmed medium as in claim 41 further adapted to generate information correlated to the spoken word.

46. The programmed medium as in claim 45 wherein the information comprises a web site address.

47. The programmed medium as in claim 34 wherein the medium comprises a magnetic storage medium.

48. The programmed medium as in claim 34 wherein the medium comprises a CD.

49. The programmed medium as in claim 34, wherein the medium comprises a digital storage device.

50. The programmed medium as in claim 34, wherein the merging unit is adapted to generate distances between words and to ignore words associated with any distance which exceeds a threshold distance.