Method and apparatus for highlighting and categorizing documents using coded word tokens

• US 5,526,443 A
• Filed: 11/09/1995
• Issued: 06/11/1996
• Est. Priority Date: 10/06/1994
• Status: Expired due to Term

• Current Assignee: Fuji Xerox Company Limited (Xerox Holdings Corp.), Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Fuji Xerox Company Limited (Xerox Holdings Corp.), Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Nakayama, Takehiro
• Primary Examiner(s): Couso, Jose L.
• Assistant Examiner(s): DO, ANH HONG
• Application Number:
  US08/556,436
• Time in Patent Office: 215 Days
• Field of Search: 382/229, 382/224, 382/173, 395/600, 364/419.08
• US Class Current: 382/229
• CPC Class Codes: G06F 16/313 Selection or weighting of terms for indexing
• CPC Class Codes: G06F 16/338 Presentation of query results

Abstract

Highlighting and categorization of documents is carried out by using word tokens which represent words appearing in a document. Elimination of certain unimportant word tokens is first completed, after which the remaining words of the document are ranked according to their word token appearance rates. These rates are then used to highlight frequently appearing words in the document which indicate the document'"'"'s topic. The document can also be categorized using document profiles developed from the word tokens.

Claims

1. A method for highlighting and categorizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the method comprising the steps of:

eliminating predetermined character shape code classes from said sequence of word tokens;

removing predetermined common function word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list;

determining word token frequency appearance rates for the word tokens of the reduced sequence;

ranking said frequency of appearance rates;

determining nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates;

highlighting words of the document corresponding to the nth or more most frequently appearing word tokens.

2. The method according to claim 1, further comprising the step of categorizing the document into one of a plurality of pre-existing categories.

3. The method according to claim 2 wherein the categorizing step comprises the steps of:

determining a token profile for said document;

measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

categorizing the document into at least one of said preexisting categories based on the at least one distance measured.

4. The method according to claim 1, wherein said removing step comprises the step of further removing numerical word tokens from said sequence of word tokens using an optional token list.

5. The method according to claim 1, wherein the highlighting step comprises the step of underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

6. A method for highlighting and categorizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the method comprising the steps of:

eliminating predetermined character shape code classes from said sequence of word tokens;

removing predetermined common function word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list;

determining word token frequency appearance rates for the word tokens of the reduced sequence;

ranking said frequency of appearance rates;

determining nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates;

highlighting words of the document corresponding to the nth or more most frequently appearing word tokens; and

categorizing the document into one of a plurality of pre-existing categories.

7. The method according to claim 6 wherein the categorizing step comprises the steps of:

determining a token profile for said document;

measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

categorizing the document into at least one of said preexisting categories based on the at least one distance measured.

8. The method according to claim 6 wherein said removing step comprises the step of further removing numerical word tokens from said sequence of word tokens using an optional token list.

9. The method according to claim 6 wherein the highlighting step comprises the step of underlining said words of the document corresponding to the nth or more most-frequently-appearing word tokens.

10. A method for highlighting and categorizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the method comprising the steps of:

eliminating predetermined character shape code classes from said sequence of word tokens;

removing predetermined common function word tokens and numerical word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list comprising an optional token list;

determining word token frequency appearance rates for the word tokens of the reduced sequence;

ranking said frequency of appearance rates;

determining nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates; and

highlighting words of the document corresponding to the nth or more most frequently appearing word tokens.

11. The method according to claim 10, further comprising the step of categorizing the document into one of a plurality of pre-existing categories.

12. The method according to claim 11 wherein the categorizing step comprises the steps of:

determining a token profile for said document;

measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

categorizing the document into at least one of said preexisting categories based on the at least one distance measured.

13. The method according to claim 10, wherein the highlighting step comprises the step of underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

14. A method for highlighting and characterizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the method comprising the steps of:

eliminating predetermined character shape code classes from said sequence of word tokens;

removing predetermined common function word tokens and numerical word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern machine technique and a stop token list comprising an optional token list;

determining word token frequency appearance rates for the word tokens of the reduced sequence;

ranking said frequency of appearance rates;

determining nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates;

highlighting words of the document corresponding to the nth or more most frequently appearing word tokens; and

categorizing the document into one of a plurality of pre-existing categories.

15. The method according to claim 14 wherein the categorizing step comprises the steps of:

determining a token profile for said document;

measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

categorizing the document into at least one of said preexisting categories based on the at least one distance measured.

16. The method according to claim 14, wherein the highlighting step comprises the step of underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

17. A method for highlighting and characterizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the method comprising the steps of:

eliminating predetermined character shape code classes from said sequence of word tokens;

removing predetermined common function word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list;

determining word token frequency appearance rates for the word tokens of the reduced sequence;

ranking said frequency of appearance rates;

determining nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates; and

categorizing the document into one of a plurality of pre-existing categories.

18. The method according to claim 17 wherein the categorizing step comprises the steps of:

determining a token profile for said document;

measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

categorizing the document into at least one of said preexisting categories based on the at least one distance measured.

19. The method according to claim 17, further comprising the step of highlighting words of the document corresponding to the nth or more most frequently appearing word tokens.

20. The method according to claim 19, wherein said highlighting step comprises the step of underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

21. The method according to claim 17, wherein said removing step comprises the step of further removing numerical word tokens from said sequence of word tokens using an optional token list.

22. A method for highlighting and categorizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the method comprising the steps of:

eliminating predetermined character shape code classes from said sequence of word tokens;

removing predetermined common function word tokens and numerical word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list comprising an optional token list;

determining word token frequency appearance rates for the word tokens of the reduced sequence;

ranking said frequency of appearance rates;

determining nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates; and

categorizing the document into one of a plurality of pre-existing categories.

23. The method according to claim 22 wherein the categorizing step comprises the steps of:

determining a token profile for said document;

measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

categorizing the document into at least one of said preexisting categories based on the at least one distance measured.

24. The method according to claim 22, further comprising the step of highlighting words of the document corresponding to the nth or more most frequently appearing word tokens.

25. The method according to claim 24, wherein the highlighting steps comprises the step of underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

26. An apparatus for highlighting and categorizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the apparatus comprising:

means for eliminating predetermined character shape code classes from said sequence of word tokens;

means for removing predetermined common function word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list;

means for determining word token frequency appearance rates for the word tokens of the reduced sequence;

means for ranking said frequency of appearance rates;

means for determining the nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates; and

means for highlighting words of the document corresponding to the nth or more most frequently appearing word tokens.

27. The apparatus according to claim 26, further comprising means for categorizing the document into one of a plurality of pre-existing categories.

28. The apparatus according to claim 27, wherein the means for categorizing comprises:

means for determining a token profile for said document;

means for measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

means for categorizing the document into at least one of said pre-existing categories based on the at least one distance measured.

29. The apparatus according to claim 26, wherein said means for removing comprises means for further removing numerical word tokens from said sequence of word tokens using an optional token list.

30. The apparatus according to claim 26, wherein said means for highlighting comprises means for underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

31. An apparatus for highlighting and categorizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the apparatus comprising:

means for eliminating predetermined character shape code classes from said sequence of word tokens;

means for removing predetermined common function word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list;

means for determining word token frequency appearance rates for the word tokens of the reduced sequence;

means for ranking said frequency of appearance rates;

means for determining nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates;

means for highlighting words of the document corresponding to the nth or more most frequently appearing word tokens; and

means for categorizing the document into one of a plurality of pre-existing categories.

32. The apparatus according to claim 31 wherein the means for categorizing comprises:

means for determining a token profile for said document;

means for measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

means for categorizing the document into at least one of said pre-existing categories based on the at least one distance measured.

33. The apparatus according to claim 31, wherein said means for removing comprises means for further removing numerical word tokens from said sequence of word tokens using an optional token list.

34. The apparatus according to claim 31, wherein said means for highlighting comprises means for underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

35. An apparatus for highlighting and categorizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the apparatus comprising:

means for eliminating predetermined character shape code classes from said sequence of word tokens;

means for removing predetermined common function word tokens and numerical word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list comprising an optional token list;

means for determining word token frequency appearance rates for the word tokens of the reduced sequence;

means for ranking said frequency of appearance rates;

means for determining the nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates; and

means for highlighting words of the document corresponding to the nth or more most frequently appearing word tokens.

36. The apparatus according to claim 35, further comprising means for categorizing the document into one of a plurality of pre-existing categories.

37. The apparatus according to claim 36 wherein the means for categorizing comprises:

means for determining a token profile for said document;

means for measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

means for categorizing the document into at least one of said pre-existing categories based on the at least one distance measured.

38. The apparatus according to claim 35, wherein said means for highlighting comprises means for underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

39. An apparatus for highlighting and categorizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the apparatus comprising:

means for eliminating predetermined character shape code classes from said sequence of word tokens;

means for removing predetermined common function word tokens and numerical word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list comprising an optional token list;

means for determining word token frequency appearance rates for the word tokens of the reduced sequence;

means for ranking said frequency of appearance rates;

means for determining the nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates;

means for highlighting words of the document corresponding to the nth or more most frequently appearing word tokens; and

means for categorizing the document into one of a plurality of pre-existing categories.

40. The apparatus according to claim 39 wherein the means for categorizing comprises:

means for determining a token profile for said document;

means for measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

means for categorizing the document into at least one of said pre-existing categories based on the at least one distance measured.

41. The apparatus according to claim 39, wherein said means for highlighting comprises means for underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

42. An apparatus for highlighting and categorizing images from a document using a sequence of word tokens, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the apparatus comprising:

means for eliminating predetermined character shape code classes from said sequence of word tokens;

means for removing predetermined common function word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list;

means for determining word token frequency appearance rates for the word tokens of the reduced sequence;

means for ranking said frequency of appearance rates;

means for determining the nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates; and

means for categorizing the document into one of a plurality of pre-existing categories.

43. The apparatus according to claim 42 wherein the means for categorizing comprises:

means for determining a token profile for said document;

means for measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

means for categorizing the document into at least one of said pre-existing categories based on the at least one distance measured.

44. The apparatus according to claim 42, further comprising means for highlighting words of the document corresponding to the nth or more most frequently appearing word tokens.

45. The apparatus according to claim 44, wherein said means for highlighting comprises means for underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.

46. The apparatus according to claim 42, wherein said means for removing comprises means for further removing numerical word tokens from said sequence of word tokens using an optional token list.

47. An apparatus for highlighting and categorizing images from a document using a sequence of word tokens representing words of the document, the word tokens comprising character shape code classes, each word of the document being represented by only one word token, the apparatus comprising:

means for eliminating predetermined character shape code classes from said sequence of word tokens;

means for removing predetermined common function word tokens and numerical word tokens from said sequence of word tokens to form a reduced sequence of word tokens using a pattern matching technique and a stop token list comprising an optional token list;

means for determining word token frequency appearance rates for the word tokens of the reduced sequence;

means for ranking said frequency of appearance rates;

means for determining the nth or more most frequently appearing word tokens based on the ranked frequency of appearance rates; and

means for categorizing the document into one of a plurality of pre-existing categories.

48. The apparatus according to claim 47 wherein the means for categorizing comprises:

means for determining a token profile for said document;

means for measuring at least one distance between the token profile and at least one reference profile developed from reference data; and

means for categorizing the document into at least one of said pre-existing categories based on the at least one distance measured.

49. The apparatus according to claim 47, further comprising means for highlighting words of the document corresponding to the nth or more most frequently appearing word tokens.

50. The apparatus according to claim 49, wherein said means for highlighting comprises means for underlining said words of the document corresponding to the nth or more most frequently appearing word tokens.