System and method of generating automated document analysis tools

• US 20070055653A1
• Filed: 09/02/2005
• Published: 03/08/2007
• Est. Priority Date: 09/02/2005
• Status: Abandoned Application

• Current Assignee: Finetooth Enterprises, Inc.
• Original Assignee: Finetooth Enterprises, Inc.
• Inventors: Lagarde, James, Guerra Currie, Anne-Marie, Diedrick, Scott, Maycotte, Higinio, Fricke, Christian
• Application Number:
  US11/218,693
• Publication Number:
  US 20070055653A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/353 into predefined classes

Abstract

A method of generating an automated document analyst is disclosed and includes receiving a plurality of source documents including text strings and performing an automated computer executable build operation on the plurality of source documents with respect to at least one target field associated with data to be extracted from the plurality of source documents. Further, the method includes performing a linguistic analysis, a statistical analysis, and a document structure analysis on an output file produced as a result of performing the automated computer executable build operation.

Claims

1. A method of generating an automated document analyst, the method comprising:

receiving a plurality of source documents including text strings;

performing an automated computer executable build operation on the plurality of source documents with respect to at least one target field associated with data to be extracted from the plurality of source documents; and

performing a linguistic analysis on an output file produced as a result of performing the automated computer executable build operation.

2. The method of claim 1, wherein the linguistic analysis includes at least one of the following:

a lexical analysis, a semantic analysis, a pragmatic analysis, a syntactic analysis, and a discourse analysis.

3. The method of claim 1, further comprising performing a statistical analysis with respect to the output file.

4. The method of claim 3, wherein the statistical analysis includes at least one of the following:

a lexical frequency analysis and a clustering analysis.

5. The method of claim 1, further comprising performing a document structure analysis on the output file.

6. The method of claim 5, wherein the document structure analysis includes at least one of the following:

a section analysis, a table structure analysis, a document format analysis, and a document level discourse analysis.

7. The method of claim 1, further comprising processing the automated text-based document analyst based on a plurality of dictionary files to create a pre-production automated text-based document analyst.

8. The method of claim 7, further comprising performing further processing of the pre-production automated text-based document analyst based on a plurality of patterns identified by performing at least one of the following:

a linguistic analysis, a statistical analysis, and a document structure analysis.

9. The method of claim 8, further comprising performing additional processing on the pre-production automated text-based document analyst based on desired data formats and desired data extractions.

10. The method of claim 9, further comprising performing a set of normalization rules with respect to the pre-production automated text-based document analyst with respect to desired data formats and data extraction.

11. The method of claim 10, further comprising testing the pre-production automated text-based document analyst using a set of test documents to determine a tested accuracy measure.

17. The method of claim 11, wherein the tested accuracy measure is based on a substantially randomized testing procedure.

18. The method of claim 11, wherein the tested accuracy measure is a precision rate.

19. The method of claim 18, wherein the precision rate is greater than 85 percent.

20. The method of claim 18, wherein the precision rate is greater than 90 percent.

21. The method of claim 18, wherein the precision rate is greater than 95 percent.

12. The method of 11, further comprising modifying the pre-production automated text-based document analyst after determining that the tested accuracy measure is below a threshold.

13. The method of claim 12, further comprising classifying the pre-production automated text-based document analyst as a production automated text-based document analyst after determining that the tested accuracy measure is above a threshold.

14. The method of claim 13, further comprising documenting the tested accuracy measure associated with the production automated text-based document analyst.

15. The method of claim 14, further comprising storing the production automated text-based document analyst in a library of automated text-based document analysts and storing the tested accuracy measure associated with the production automated text-based document analyst.

16. The method of claim 15, wherein the library of automated text-based document analysts includes at least a first automated text-based document analyst associated with a first document type and at least a second automated text-based document analyst associated with a second document type.

22. A system for generating at least one virtual analyst, the system comprising:

a data build module;

a data analysis module coupled to the data build module;

a development module coupled to the data analysis module; and

a test module, wherein the test module determines a performance metric associated with a test of a pre-production automated text-based document.

23. The system of claim 22, wherein the performance metric is an accuracy measurement.

24. The system of claim 22, wherein the performance metric is a precision measurement.

25. The system of claim 22, wherein the test module provides a production automated text-based document analyst when the test accuracy measure is greater than a threshold.

26. The system of claim 25, wherein the test module returns the pre-production automated text-based document analyst to the data analysis module when the test accuracy measure is below a threshold.

27. The system of claim 22, wherein the data build module performs an automated computer executable build operation on a plurality of source documents with respect to at least one target field associated with data to be extracted from the plurality of source documents.

28. The system of claim 27, wherein the data analysis module comprises a linguistic analysis module that performs a linguistic analysis on an output file received from the data build module, wherein the output file is a result of the automated computer executable build operation.

29. The system of claim 28, wherein the linguistic analysis includes at least one of the following:

a lexical analysis, a semantic analysis, a pragmatic analysis, a syntactic analysis, and a discourse analysis.

30. The system of claim 27, wherein the data analysis module further comprises a statistical analysis module that performs a statistical analysis with respect to the output file.

31. The system of claim 30, wherein the statistical analysis includes at least one of the following:

a lexical frequency analysis and a clustering analysis.

32. The system of claim 27, wherein the data analysis module further comprises a document structure analysis module that performs a document structure analysis on the output file.

33. The system of claim 32, wherein the document structure analysis includes at least one of the following:

a section analysis, a table structure analysis, a document format analysis, and a document level discourse analysis.

34. The system of claim 22, wherein the development module receives an automated text-based document analyst from the data analysis module and processes the automated text-based document analyst based on a plurality of dictionary files to create a pre-production automated text-based document analyst.

35. The system of claim 34, wherein the development module further processes the pre-production automated text-based document analyst based on a plurality of patterns identified by at least one of the following:

a linguistic analysis module, a statistical analysis module, and a document structure analysis module.

36. The system of claim 35, wherein the development module further processes the pre-production automated text-based document analyst based on desired data formats and desired data extractions.

37. The system of claim 36, wherein the development module applies a set of normalization rules with respect to the pre-production automated text-based document analyst with respect to desired data formats and data extraction.

38. The system of claim 22, wherein the production automated text-based document analyst is stored within a library that includes at least two production automated text-based document analysts.

39. A library system comprising:

at least a first automated text-based document analyst associated with a first document type; and

at least a second automated text-based document analyst associated with a second document type, wherein the first automated text-based document analyst and the second automated text-based analyst have a precision rate that is greater than 85 percent.

40. The system of claim 39, wherein the first automated text-based document analyst and the second automated text-based analyst have a precision rate that is greater than 90 percent when processing documents having a particular document type.

41. The system of claim 40, wherein the first automated text-based document analyst and the second automated text-based analyst have a precision rate that is greater than 95 percent when processing documents having a particular document type.

42. The system of claim 39, wherein the first automated text-based document analyst and the second automated text-based analyst are generated based on an output file that results from an automated computer executable build operation performed on a plurality of source documents with respect to at least one target field associated with data to be extracted from the plurality of source documents.

43. The system of claim 42, wherein the first automated text-based document analyst and the second automated text-based analyst are also generated based on a linguistic analysis performed with respect to the output file.

44. The system of claim 43, wherein the first automated text-based document analyst and the second automated text-based analyst are further generated based on a statistical analysis performed with respect to the output file.

45. The system of claim 44, wherein the first automated text-based document analyst and the second automated text-based analyst are further generated based on a document structure analysis performed with respect to the output file.

46. The system of claim 39, wherein the first automated text-based document analyst and the second automated text-based analyst are tested to determine whether an accuracy measure is above a predetermined threshold.

47. The system of claim 46, wherein the first automated text-based document analyst and the second automated text-based analyst are modified when the accuracy measure is not above the predetermined threshold.

48. The system of claim 39, wherein the first document type is different from the second document type.

49. The system of claim 48, wherein the first document type and the second document type are selected from the group including:

contracts, medical files, clinical files, legal files, insurance files, and government files.