Linguistic object model

• US 7,171,352 B2
• Filed: 09/14/2004
• Issued: 01/30/2007
• Est. Priority Date: 04/23/2004
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Parkinson, David J., Cipollone, Domenic J., Calcagno, Michael V., Olsen, Mari J. B., Chang, Su Chin, Shahani, Ravi C.
• Primary Examiner(s): Azad; Abul
• Assistant Examiner(s): Shortledge; Thomas E.
• Application Number:
  US10/940,483
• Publication Number:
  US 20050251382A1
• Time in Patent Office: 868 Days
• Field of Search: 704/9, 704/7, 704/10, 704/1, 707/100, 707/101, 707/102, 707/103.R, 717/104
• US Class Current: 704/9
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 8/20 Software design
• CPC Class Codes: G06F 8/315 Object-oriented languages
• CPC Class Codes: G10L 15/18 using natural language modelling
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: Y10S 707/99944 Object-oriented database structure

Abstract

A linguistic object model is adapted to model semantic elements of a natural language. The linguistic object model includes a set of abstract representations of semantic elements of an utterance. Each abstract representation in the set corresponds to a natural language type that is independent of any particular language.

Claims

1. A linguistic object model that is encoded on a computer-readable medium and is adapted to be utilized by a computer to model semantic elements of a natural language, the linguistic object model comprising:

a set of types for modeling semantics of natural language, the set of types being independent of any particular natural language, and wherein the set of types further comprises a Denoter type for modeling user-defined categories of input strings, and wherein a data member describing the Denoter comprises a record of which user-defined category of input strings in a Denoter object of type Denoter is filled at runtime by an analysis engine to fulfill a match between a type of the set of types and an element of a natural language utterance.

2. The linguistic object model of claim 1 wherein the set of types comprises:

an Entity type for modeling semantics of noun phrases.

3. The linguistic object model of claim 1 wherein the set of types comprises:

an Entity type for modeling semantics of adjective phrases.

4. The linguistic object model of claim 2 and further comprising:

a Restriction type for modeling properties of semantic elements;

wherein the Entity hosts the Restriction.

5. The linguistic object model of claim 1 wherein the set of types comprises:

a Frame type for modeling semantics of events.

6. The linguistic object model of claim 5 wherein the event comprises a semantic event that can be expressed either as verbs or nouns.

7. The linguistic object model of claim 5 wherein the Frame type comprises one or more head words corresponding to a syntactic head of a phrase modeled by the Frame type.

8. The linguistic object model of claim 5 wherein a head word of a Frame is defined by a developer of a natural language software application to create new Frame types.

9. The linguistic object model of claim 5 wherein a developer codes an application-specific Frame to inherit from a Frame type of the set of types.

10. The linguistic object model of claim 5 further comprising:

a Restriction for modeling properties of semantic elements; and

wherein the Frame hosts the Restriction.

11. The linguistic object model of claim 1 wherein the set of types comprises:

a Restriction adapted to model properties of semantic elements.

12. The linguistic object model of claim 1 wherein the set of types comprises:

Entities adapted to model semantic objects;

Frames adapted to model events and associated relationships between one or more objects; and

Restrictions adapted to model properties of and relationships between other entities, Frames or restrictions.

13. The linguistic object model of claim 1 wherein the set of types comprises:

a Named Entity type adapted to model named Entities, the Named Entity type derived from a separately defined class.

14. The linguistic object model of claim 1 wherein the set of types comprises:

a Named Entity type adapted to model named Entities, the Named Entity type accessed directly using an interface.

15. The linguistic object model of claim 1 wherein each type in the set of types corresponds to semantically meaningful elements of a natural language utterance.

16. A linguistic object model that is encoded on a computer-readable medium and is adapted to be utilized by a computer to model semantic elements of a natural language, the linguistic object model comprising:

a set of types representative of elements of natural language utterances, each type corresponding to semantically meaningful elements of an utterance, and wherein the set of types further comprises a Denoter type for modeling user-defined categories of input strings, and wherein a data member describing the Denoter comprises a record of which user-defined category of input strings in a Denoter object of type Denoter is filled at runtime by an analysis engine to fulfill a match between a type of the set of types and an element of a natural language utterance.

17. The linguistic object model of claim 16 wherein correspondence between each type and semantically meaningful elements of an utterance is literal.

18. The linguistic object model of claim 16 wherein the set of types are implemented as a software class library.

19. The linguistic object model of claim 16 wherein the set of types comprises:

an Entity type adapted to model semantic entities.

20. The linguistic object model of claim 16 wherein the set of types comprises:

a Frame type adapted to model semantic relationships.

21. The linguistic object model of claim 16 wherein the set of types comprises:

a Restriction type adapted to model properties of semantic entities and semantic relationships.

22. The linguistic object model of claim 21 wherein the Restriction type models relationships between combinations of Entities, Frames, and other Restrictions.

23. The linguistic object model of claim 16 further comprising:

classes of linguistic expressions that the set of types model;

wherein the set of types and the classes of expressions are independent of an application domain.

24. The linguistic object model of claim 16 wherein the set of types are independent of an application domain.

25. The linguistic object model of claim 16 further comprising:

an analysis engine adapted to interact with the set of types and a natural language input to map the semantic elements of the natural language input to the set of types, the analysis engine producing one or more LOM objects corresponding to the mapping.

26. The linguistic object model of claim 25 wherein the one or more LOM objects comprises at least one LOM object generated without directly mapping the semantic elements to a type of the set of types.

27. The linguistic object model of claim 16 wherein the Denoter type is adapted to model user-defined strings, and wherein the Denoter type associates a privileged data member to another type of the set of types.

28. A computer-implemented method of modeling natural language comprising:

categorizing linguistic expressions into classes corresponding to elements of an utterance; and

deriving a set of types adapted to represent those classes as abstract objects, wherein the set of types further comprises a Denoter type for modeling user-defined categories of input strings, and wherein a data member describing the Denoter comprises a record of which user-defined category of input strings in a Denoter object of type Denoter is filled at runtime by an analysis engine to fulfill a match between a type of the set of types and an element of a natural language utterance.

29. The method of claim 28 and further comprising:

compiling the set of types into a software class library.

30. The method of claim 28 wherein the set of types comprises:

an Entity of type entity model semantics of phrases.

31. The method of claim 30 wherein the Entity models the semantics of noun phrases.

32. The method of claim 30 wherein the Entity models the semantics of adjective phrases.

33. The method of claim 30 wherein the set of types further comprises:

a Frame for modeling semantics of events; and

wherein the Entity hosts the Frame.

34. The method of claim 30 wherein the Entity corresponds to another type that can be used by an analysis engine to map Entities to natural language words.

35. The method of claim 28 further comprising:

a Frame for modeling semantics of events.

36. The method of claim 35 wherein the event comprises a semantic event that can be expressed either as verbs or nouns.

37. The method of claim 35 wherein the Frame corresponds to another type that can be used by an analysis engine to map Frames to natural language words.

38. The method of claim 35 wherein a head word of a Frame is defined by a developer to create new Frame types.

39. The method of claim 35 wherein a developer codes an application-specific Frame to inherit from a Frame.

40. The method of claim 35 further comprising:

a Restriction models properties of semantic elements; and

wherein the Frame hosts the Restriction.

41. The method of claim 28 wherein the set of types comprises:

a Restriction adapted to model properties of semantic elements.

42. The method of claim 28 wherein the set of types comprises:

a Named Entity type adapted to model named Entities, the Named Entity type accessed directly via an appropriate interface.

43. The method of claim 28 wherein the set of types comprises:

Entities for modeling semantic objects;

Frames for modeling events and associated relationships between one or more objects; and

Restrictions for modeling properties of and relationships between other entities, Frames or restrictions.

44. The method of claim 28 wherein the set of types comprises:

a Named Entity type adapted to model named Entities, the Named Entity type derived from a separately defined class.

45. A computer readable medium storing computer readable data structures that are adapted to be utilized by a computer to model natural language, comprising:

a linguistic object model comprising a set of types for modeling semantics of natural language, the set of types being independent of any particular application, and wherein the set of types further comprises a Denoter type for modeling user-defined categories of input strings, and wherein a data member describing the Denoter comprises a record of which user-defined category of input strings in a Denoter object of type Denoter is filled at runtime by an analysis engine to fulfill a match between a type of the set of types and an element of a natural language utterance.

46. The computer readable medium of claim 45, wherein the set of types of the linguistic object model comprises:

an Entity type adapted to model semantic entities;

a Frame type adapted to model events and associated relationships; and

a Restriction type adapted to model properties of and relationships between Entities, Frames, or other restrictions.

47. The computer readable medium of claim 45 wherein the set of types of the linguistic object model comprises:

an Entity type for modeling the semantics of phrases.

48. The computer readable medium of claim 47 wherein the set of types of the linguistic object model wherein the Entity models the semantics of noun phrases.

49. The computer readable medium of claim 47 wherein the Entity models the semantics of adjective phrases.

50. The computer readable medium of claim 47 wherein the set of types of the linguistic object model further comprises:

a Frame for modeling semantics of events; and

wherein the Entity hosts the Frame.

51. The computer readable medium of claim 47 wherein the Entity includes a data member that can be used by an analysis engine to map Entities to natural language words.

52. The computer readable medium of claim 45 wherein the set of types comprises:

a Frame for modeling semantics of events.

53. The computer readable medium of claim 52 wherein the event comprises a semantic event that can be expressed either as verbs or nouns.

54. The computer readable medium of claim 45 wherein the set of types comprises:

a Restriction adapted to model properties of semantic elements.

55. The computer readable medium of claim 45 wherein the set of types comprises:

a Named Entity type adapted to model named Entities, the Named Entity type derived from a separately defined class.

56. The computer readable medium of claim 45 wherein the set of types comprises:

a Named Entity type adapted to model named entities, the Named Entity type being implemented from an interface.