Text segmentation and label assignment with user interaction by means of topic specific language models and topic-specific label statistics

• US 8,200,487 B2
• Filed: 11/12/2004
• Issued: 06/12/2012
• Est. Priority Date: 11/21/2003
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications Austria Gmbh (Microsoft Corporation)
• Inventors: Peters, Jochen, Matusov, Evgeny, Meyer, Carsten, Klakow, Dietrich
• Primary Examiner(s): Opsasnick, Michael N
• Application Number:
  US10/595,831
• Publication Number:
  US 20080201130A1
• Time in Patent Office: 2,769 Days
• Field of Search: 704/236, 704/9, 704/10
• US Class Current: 704/236
• CPC Class Codes: G06F 40/10 Text processing natural language analysis G06F40/20; semantic analysis G06F40/30; processing or translation of natural language G06F40/40
• CPC Class Codes: G06F 40/279 Recognition of textual entities
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

The invention relates to a method, a computer program product, a segmentation system and a user interface for structuring an unstructured text by making use of statistical models trained on annotated training data. The method performs text segmentation into text sections and assigns labels to text sections as section headings. The performed segmentation and assignment is provided to a user for general review. Additionally, alternative segmentations and label assignments are provided to the user being capable to select alternative segmentations and alternative labels as well as to enter a user defined segmentation and user defined label. In response to the modifications introduced by the user, a plurality of different actions are initiated incorporating the re-segmentation and re-labelling of successive parts of the document or the entire document. Furthermore the method comprises a learning functionality, logging and analyzing user introduced modifications for adaptation of user'"'"'s preferences and for further training of the statistical models.

Claims

1. A method for generating a structured text from an unstructured text, the method comprising acts of:

segmenting the unstructured text into text sections;

assigning, to at least one text section, a topic being indicative of content of the at least one text section, wherein the act of segmenting the unstructured text and/or the act of assigning a topic to the at least one text section uses at least one statistical model built from annotated training data;

providing to a user a first structured text comprising the at least one text section and a section heading for the at least one text section, the section heading corresponding to the topic assigned to the at least one text section;

receiving user input indicating at least one modification to the first structured text;

using a computer system to process the at least one modification received from the user to generate a second structured text; and

logging and analyzing the at least one modification received from the user to adapt the at least one statistical model.

2. The method according to claim 1, wherein the topic assigned to the at least one text section is associated with a plurality of section headings, and wherein the section heading for the at least one text section is selected from the plurality of section headings.

3. The method according to claim 2, wherein the section heading for the at least one text section is a section heading in the plurality of section headings that is most frequently selected for the topic assigned to the at least one text section.

4. The method according to claim 2, further comprising providing to the user the plurality of section headings associated with the topic assigned to the at least one text section, wherein the at least one modification comprises the user'"'"'s selection of an alternative section heading for the at least one text section from the plurality of section headings.

5. The method according to claim 2, wherein the at least one modification comprises a new section heading entered by the user to replace the section heading for the at least one text section, the new section heading being different from every section heading in the plurality of section headings associated with the topic assigned to the at least one text section.

6. The method according to claim 1, wherein the section heading is inserted into the first structured text at a first location, and wherein the at least one modification comprises moving the section heading to a second location different from the first location to re-define a boundary of the at least one text section.

7. The method according to claim 1, wherein the plurality of text sections is a first plurality of text sections, and wherein the method further comprises acts of:

re-segmenting at least a portion of the second structured text into a second plurality of text sections, without overruling the at least one modification received from the user; and

generating a third structured text comprising the second plurality of text sections and a corresponding section heading for each of the second plurality of text sections.

8. The method according to claim 1, further comprising:

identifying a text portion as being a full or partial verbalization of the section heading for the at least one text section; and

removing the text portion from the first structured text provided to the user.

9. The method according to claim 1, wherein a granularity of segmentation is controlled by the user using a customizable granularity parameter.

10. An apparatus for generating a structured text from an unstructured text, the apparatus comprising a computer system configured to:

segment the unstructured text into text sections;

assign, to at least one text section, a topic being indicative of content of the at least one text section, the topic being associated with a plurality of section headings, wherein, in segmenting the unstructured text and/or assigning a topic to the at least one text section, the computer system is configured to use at least one statistical model built from annotated training data;

provide to a user a first structured text comprising the at least one text section and a section heading for the at least one text section, the section heading being selected from the plurality of section headings associated with the topic assigned to the at least one text section;

receive user input indicating at least one modification to the first structured text;

process the at least one modification received from the user to generate a second structured text; and

log and analyze the at least one modification received from the user to adapt the at least one statistical model.

11. The apparatus according to claim 10, wherein the plurality of section headings associated with the topic assigned to the at least one text section are provided to the user in conjunction with the first structured text, and wherein the at least one modification comprises the user'"'"'s selection of an alternative section heading for the at least one text section from the plurality of section headings.

12. The apparatus according to claim 10, wherein the at least one modification comprises a new section heading entered by the user to replace the section heading inserted for the at least one text section, the new section heading being different from every section heading in the plurality of section headings associated with the topic assigned to the at least one text section.

13. The apparatus according to claim 10, wherein the section heading is inserted into the first structured text at a first location, and wherein the at least one modification comprises moving the section heading to a second location different from the first location to re-define a boundary of the at least one text section.

14. The apparatus according to claim 10, wherein the plurality of text sections is a first plurality of text sections, and wherein the computer system is further configured to:

re-segment at least a portion of the second structured text into a second plurality of text sections, without overruling the at least one modification received from the user; and

generate a third structured text comprising the second plurality of text sections and a corresponding section heading for each of the second plurality of text sections.

15. At least one article of manufacture, comprising:

at least one computer-readable storage device having encoded thereon executable instructions that, when executed by a computer system, perform a method for generating a structured text from an unstructured text, the method comprising acts of;

segmenting the unstructured text into text sections;

assigning, to at least one text section, a topic being indicative of content of the at least one text section, the topic being associated with a plurality of section headings, wherein the act of segmenting the unstructured text and/or the act of assigning a topic to the at least one text section uses at least one statistical model built from annotated training data;

providing to a user a first structured text comprising the at least one text section and a section heading for the at least one text section, the section heading being selected from the plurality of section headings associated with the topic assigned to the at least one text section;

receiving user input indicating at least one modification to the first structured text;

processing the at least one modification received from the user to generate a second structured text; and

logging and analyzing the at least one modification received from the user to adapt the at least one statistical model.

16. The at least one article of manufacture according to claim 15, wherein the section heading for the at least one text section is a section heading in the plurality of section headings that is most frequently selected for the topic assigned to the at least one text section.

17. The at least one article of manufacture according to claim 15, wherein the method further comprises providing to the user the plurality of section headings associated with the topic assigned to the at least one text section, and wherein the at least one modification comprises the user'"'"'s selection of an alternative section heading for the at least one text section from the plurality of section headings.

18. The at least one article of manufacture according to claim 15, wherein the at least one modification comprises a new section heading entered by the user to replace the section heading inserted for the at least one text section, the new section heading being different from every section heading in the plurality of section headings associated with the topic assigned to the at least one text section.

19. The at least one article of manufacture according to claim 15, wherein the section heading is inserted into the first structured text at a first location, and wherein the at least one modification comprises moving the section heading to a second location different from the first location to re-define a boundary of the at least one text section.

20. The at least one article of manufacture according to claim 15, wherein the plurality of text sections is a first plurality of text sections, and wherein the method further comprises acts of:

re-segmenting at least a portion of the second structured text into a second plurality of text sections, without overruling the at least one modification received from the user; and

generating a third structured text comprising the second plurality of text sections and a corresponding section heading for each of the second plurality of text sections.

21. A system comprising:

means for providing to a user a first structured text comprising at least one text section and a section heading for the at least one text section, the at least one text section being one of a plurality of text sections obtained from segmenting an unstructured text, the section heading being selected from a plurality of section headings associated with a topic assigned to the at least one text section, the topic being indicative of content of the at least one text section, wherein segmenting the unstructured text and/or assigning the topic to the at least one text section uses at least one statistical model built from annotated training data;

means for receiving user input indicating at least one modification to the first structured text;

means for processing the at least one modification received from the user to generate a second structured text; and

means for logging and analyzing the at least one modification received from the user to adapt the at least one statistical model.

22. The system according to claim 21, wherein the means for providing the first structured text is further configured to provide to the user via a graphical user interface the plurality of section headings associated with the topic assigned to the at least one text section, wherein the at least one modification comprises the user'"'"'s selection of an alternative section heading for the at least one text section from the plurality of section headings.

23. The system according to claim 21, wherein the at least one modification comprises a new section heading entered by the user to replace the section heading for the at least one text section, the new section heading being different from every section heading in the plurality of section headings associated with the topic assigned to the at least one text section.

24. The system according to claim 21, wherein the section heading is inserted into the first structured text at a first location, and wherein the at least one modification comprises moving the section heading to a second location different from the first location to re-define a boundary of the at least one text section.

25. A method for generating a structured text from an unstructured text, the method comprising acts of:

segmenting the unstructured text into text sections;

assigning, to at least one text section, a topic being indicative of content of the at least one text section;

identifying a text portion as being a full or partial verbalization of a section heading for the at least one text section, the section heading corresponding to the topic assigned to the at least one text section;

providing to a user a first structured text comprising the at least one text section and the section heading for the at least one text section, wherein the text portion identified as being a full or partial verbalization of the section heading is removed from the first structured text;

receiving user input indicating at least one modification to the first structured text; and

using a computer system to process the at least one modification received from the user to generate a second structured text.

26. The method according to claim 25, wherein the topic assigned to the at least one text section is associated with a plurality of section headings, and wherein the section heading for the at least one text section is a section heading in the plurality of section headings that is most frequently selected for the topic assigned to the at least one text section.

27. The method according to claim 26, further comprising providing to the user the plurality of section headings associated with the topic assigned to the at least one text section, wherein the at least one modification comprises the user'"'"'s selection of an alternative section heading for the at least one text section from the plurality of section headings.

28. The method according to claim 25, wherein the at least one modification comprises a new section heading entered by the user to replace the section heading for the at least one text section.

29. The method according to claim 25, wherein the section heading is inserted into the first structured text at a first location, and wherein the at least one modification comprises moving the section heading to a second location different from the first location to re-define a boundary of the at least one text section.

30. The method according to claim 25, wherein the plurality of text sections is a first plurality of text sections, and wherein the method further comprises acts of:

re-segmenting at least a portion of the second structured text into a second plurality of text sections, without overruling the at least one modification received from the user; and

generating a third structured text comprising the second plurality of text sections and a corresponding section heading for each of the second plurality of text sections.

31. An apparatus for generating a structured text from an unstructured text, the apparatus comprising a computer system configured to:

segment the unstructured text into text sections;

assign, to at least one text section, a topic being indicative of content of the at least one text section, the topic being associated with a plurality of section headings;

identify a text portion as being a full or partial verbalization of a section heading for the at least one text section, the section heading being selected from the plurality of section headings associated with the topic assigned to the at least one text section;

provide to a user a first structured text comprising the at least one text section and the section heading for the at least one text section, wherein the text portion identified as being a full or partial verbalization of the section heading is removed from the first structured text;

receive user input indicating at least one modification to the first structured text; and

process the at least one modification received from the user to generate a second structured text.

32. The apparatus according to claim 31, wherein the section heading for the at least one text section is a section heading in the plurality of section headings that is most frequently selected for the topic assigned to the at least one text section.

33. The apparatus according to claim 31, wherein the plurality of section headings associated with the topic assigned to the at least one text section are provided to the user in conjunction with the first structured text, and wherein the at least one modification comprises the user'"'"'s selection of an alternative section heading for the at least one text section from the plurality of section headings.

34. The apparatus according to claim 31, wherein the at least one modification comprises a new section heading entered by the user to replace the section heading for the at least one text section, the new section heading being different from every section heading in the plurality of section headings associated with the topic assigned to the at least one text section.

35. The apparatus according to claim 31, wherein the section heading is inserted into the first structured text at a first location, and wherein the at least one modification comprises moving the section heading to a second location different from the first location to re-define a boundary of the at least one text section.

36. The apparatus according to claim 31, wherein the plurality of text sections is a first plurality of text sections, and wherein the computer system is further configured to:

re-segment at least a portion of the second structured text into a second plurality of text sections, without overruling the at least one modification received from the user; and

generate a third structured text comprising the second plurality of text sections and a corresponding section heading for each of the second plurality of text sections.

37. At least one article of manufacture, comprising:

at least one computer-readable storage device having encoded thereon executable instructions that, when executed by a computer system, perform a method for generating a structured text from an unstructured text, the method comprising acts of;

segmenting the unstructured text into text sections;

assigning, to at least one text section, a topic being indicative of content of the at least one text section, the topic being associated with a plurality of section headings;

identifying a text portion as being a full or partial verbalization of a section heading for the at least one text section, the section heading being selected from the plurality of section headings associated with the topic assigned to the at least one text section;

providing to a user a first structured text comprising the at least one text section and the section heading for the at least one text section, wherein the text portion identified as being a full or partial verbalization of the section heading is removed from the first structured text;

receiving user input indicating at least one modification to the first structured text; and

processing the at least one modification received from the user to generate a second structured text.

38. The at least one article of manufacture according to claim 37, wherein the section heading for the at least one text section is a section heading in the plurality of section headings that is most frequently selected for the topic assigned to the at least one text section.

39. The at least one article of manufacture according to claim 37, wherein the method further comprises providing to the user the plurality of section headings associated with the topic assigned to the at least one text section, and wherein the at least one modification comprises the user'"'"'s selection of an alternative section heading for the at least one text section from the plurality of section headings.

40. The at least one article of manufacture according to claim 37, wherein the at least one modification comprises a new section heading entered by the user to replace the section heading for the at least one text section, the new section heading being different from every section heading in the plurality of section headings associated with the topic assigned to the at least one text section.

41. The at least one article of manufacture according to claim 37, wherein the section heading is inserted into the first structured text at a first location, and wherein the at least one modification comprises moving the section heading to a second location different from the first location to re-define a boundary of the at least one text section.

42. The at least one article of manufacture according to claim 37, wherein the plurality of text sections is a first plurality of text sections, and wherein the method further comprises acts of:

re-segmenting at least a portion of the second structured text into a second plurality of text sections, without overruling the at least one modification received from the user; and

generating a third structured text comprising the second plurality of text sections and a corresponding section heading for each of the second plurality of text sections.