Abstracting system for index search machine

• US 3,947,825 A
• Filed: 05/07/1975
• Issued: 03/30/1976
• Est. Priority Date: 04/13/1973
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Cassada, Thomas Edward
• Primary Examiner(s): Zache, Raulfe B.
• Application Number:
  US05/575,318
• Time in Patent Office: 328 Days
• Field of Search: 340/172.5, 197/19, 271/9
• US Class Current: 1/1
• CPC Class Codes: G06F 16/30 of unstructured textual data document management systems G06F16/93
• CPC Class Codes: G06K 17/00 Methods or arrangements for effecting co-operative working between equipments covered by two or more of main groups G06K1/00 - G06K15/00, e.g. automatic card files incorporating conveying and reading operations
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching

Abstract

The invention concerns a system of English language abstracting used to increase the search rate for an Index Search machine that, as an example makes use of magnetic record cards each having in a typical case 50 tracks for recording of information. To increase the rate at which groups of words can be compared, an abstract of each group of words is generated. On the magnetic card the text or groups of words are recorded on tracks 2 through 50 and the abstracts of these groups are recorded in track 1. When searching for a particular group of words, an abstract is generated for the group being sought. This abstract is then compared to the abstracts on track 1 of each card. If any of the abstracts match, the corresponding group of words on the card are searched in detail. For abstracts that do not match there is no need to search the corresponding group of words. Therefore, the need to search every group of words is eliminated, thus increasing the search rate.

The abstracts are generated from the first and third characters of each word in the group of words. Each character to be used in the generation of the abstract is encoded into one of three groups which are equally probable over the distribution of characters in the English Language. The abstract for a group of words consists of the number of characters that fall in each of the three groups.

Claims

1. Apparatus for generating at least one abstract that is useful during information searching and retrieval procedures, comprising:

1. means for sensing information comprising individual words in a selected language, each word comprising one or more individual characters;

2. means for categorizing selected ones only of the characters in said words into predefined character groups that are based on a probability distribution of characters in the language selected;

3. means maintaining a count of the number of characters categorized into each of said predefined character groups; and

4. means for storing said count as an abstract of said information.

2. The apparatus of claim 1 further comprising:

5. means for providing an inquiry word of information comprising at least an individual inquiry word in a selected language said word comprising one or more individual characters;

6. means for categorizing selected ones only of the characters in said inquiry word into predefined character groups that are based on a probability distribution of characters in the language selected;

4. The apparatus of claim 1 wherein said count maintaining means has a limited count capacity of sufficient size for counting an average number of characters in each character group, and further comprising:

5. recognizing means for recognizing when the character count for any predefined character group exceeds said limited count capacity; and

6. force search means controlled by said recognizing means for generating and recording a force search indication rather than an abstract in order to signify that a search of the related information is required.

5. The apparatus of claim 1 further comprising:

5. means for generating and recording an abstract control indication to signify the storage of one or more abstracts for use in determining whether an abstract search or a detailed search of the information is necessary.

6. The apparatus of claim 1 wherein said words of information are stored in storage areas in groups designated index groups, each index group normally occupying a predetermined area of storage but occasionally exceeding said predetermined area, and further comprising:

5. means for determining when an index group exceeds said predetermined area of storage; and

6. means for recording an area skip code for areas storing the excess of information in any index group.

7. The apparatus of claim 1 further comprising:

5. means for categorizing the selected characters into said character groups based on the physical location of characters in their related words, such as first and third character locations of each word.

3. count maintaining means for maintaining a count of the number of characters in said inquiry word categorized into each of said predefined character groups;

8. means for comparing said inquiry count with said abstract count; and

9. means providing an indication of a match or mismatch of said abstract and inquiry counts.

8. Apparatus for generating at least one abstract that is useful during information searching and retrieval procedures, comprising:

1. means for sensing an inquiry of information comprising at least an individual inquiry word in a selected language, said inquiry word comprising one or more individual characters;

2. means for categorizing selected ones only of the characters in said inquiry word into predefined character groups that are based on a probability distribution of characters in the language selected,

3. means for maintaining a count of the number of characters categorized into each of said predefined character groups; and

4. means for utilizing said count as an inquiry during searching of said information.

11. The apparatus of claim 8 further comprising:

6. means for providing an inquiry of information comprising at least an individual inquiry word in a selected language said word comprising one or more individual characters;

7. means for categorizing selected ones only of characters in said inquiry into predefined character groups that are based on a probability distribution of characters in the language selected;

8. means for maintaining a count of the number of characters in said inquiry categorized into each of said predefined character groups;

9. means for reconverting said abstract count from compressed form to a character group count form like said inquiry count;

10. means for comparing said inquiry count with said abstract count; and

11. means providing an indication of a match or mismatch of said abstract and inquiry counts.

12. The apparatus of claim 8 wherein each abstract count comprises three non-compatible count characters, and further comprising:

6. means in said converting means for compressing each abstract count into two apparatus - compatible characters.

15. The apparatus of claim 12 wherein each index group is recorded in an individual storage area and wherein a particular storage area is set aside for storage of abstracts, and further comprising:

4. means for locating each index storage area for accessing of the index stored therein;

5. means for locating said particular storage area; and

6. means for sensing each abstract in said particular storage area.

16. The apparatus of claim 12 wherein a force search indication is recorded in place of any abstract count that exceeds a predetermined number higher than an average number of characters in each character group, and further comprising:

4. means for recognizing said force search indication; and

5. means controlled by said recognizing means for searching the entire related index group for comparison with said inquiry.

19. The apparatus of claim 15, further comprising:

5. means for categorizing individual characters into said character groups based on the physical location of said characters in their related words, such as first and third character locations of each word.

20. The apparatus of claim 15 wherein said count maintaining means has count capacity of sufficient size for counting an average number of characters in each character group, and further comprising:

means for recognizing when the character count for any predefined character group exceeds said limited count capacity; and

means controlled by said recognizing means for generating and recording a force search indication rather than an abstract in order to signify that a search of the related information is required.

21. The apparatus of claim 15 further comprising:

means for generating and recording an abstract control indication to signify the storage of one or more abstracts for use in determining whether an abstract search or a detailed search of the information is necessary.

22. The apparatus of claim 15 wherein said words of information are stored in storage areas in groups designated index groups, each index group normally occupying a predetermined area of storage but occasionally exceeding said predetermined area, and further comprising:

means for determining when an index group exceeds said predetermined area of storage; and

means for recording an area skip code for areas storing the excess of information in any index group.

23. The apparatus of claim 15 further comprising:

5. means for providing an inquiry word of information comprising at least an individual word in a selected language said word comprising one or more individual characters;

6. means for categorizing selected ones of said characters in said inquiry word into predefined character groups that are based on a probability distribution of characters in the language selected;

25. The apparatus of claim 20 further comprising:

means responsive to a match of said inquiry count and a said abstract count for manifesting the related information, as by printing said related information on a document.

26. The apparatus of claim 15 wherein information is stored on said record medium in index groups, each having index words representative of original documents and wherein a separate area is allotted on said medium for storage of abstracts and further comprising:

means for accessing each index group on said record medium in order to categorize characters;

means operable to maintain an abstract count for each index group;

and means for recording each said abstract count in said allotted area on said record medium.

27. The apparatus of claim 22 wherein said record medium has a plurality of storage tracks thereon with each index group being recorded in an individual one of said storage tracks and wherein a particular track is set aside as said allotted area, and further comprising:

means for positioning each track on said record medium for accessing of the index stored therein;

means for positioning said particular track; and

means for recording each abstract in said particular track.

28. The apparatus of claim 23 wherein said record medium is a magnetic record medium and further comprising:

means for recording information on said medium by magnetic patterns.

29. The apparatus of claim 23 wherein each index group normally occupies one storage track but occassionally exceeds one track and further comprising:

means for determining when an index group exceeds one track of storage; and

means for recording a track skip code for tracks storing the excess of information in any index group.

31. The apparatus of claim 26 further comprising:

means responsive to a match of said inquiry count and a said abstract count for manifesting the related information, as by printing said related information on a document.

32. The apparatus of claim 26 wherein said record medium has a plurality of storage tracks thereon with each index group being recorded in an individual one of said storage tracks and wherein a particular track is set aside as said allotted area, and further comprising:

means for positioning each track on said record medium for accessing of the index stored therein;

means for positioning said particular track, and

means for sensing each abstract in said particular track.

33. The apparatus of claim 26 wherein said record medium is a magnetic record medium and further comprising:

means for sensing information on said medium by detection of patterns on said medium.

34. The apparatus of claim 26 wherein index groups may exceed the capacity of an area and wherein areas storing excess information are designated by an area skip indication in place of an abstract count, and further comprising:

means responsive to an area skip indication for skipping to the next abstract count during any searching and comparing of said abstract counts with said inquiry count.

37. The method of Claim 31 further comprising:

6. sensing by said machine information inquiry signals representative of an inquiry of information comprising at least an individual inquiry word in a selected language, said word comprising one or more individual inquiry characters;

7. developing in said machine character inquiry signals representative of said at least an individual inquiry character;

39. The method of claim 32 further comprising:

12. manifesting in said machine the related information, as by printing said related information on a document, responsive to a match of said inquiry count and said abstract count.

40. The method of claim 31 further comprising:

6. generating and recording in said machine an abstract control indication to signify the storage of one or more abstracts for use in determining whether an abstract search or a detailed search of the information is necessary.

41. The method of claim 31 wherein said words of information are stored in storage areas in groups designated index groups, each index group normally occupying a predetermined area of storage but occasionally exceeding said predetermined area, and further comprising:

6. determining in said machine when an index group exceeds said predetermined area of storage; and

7. recording in said machine an area skip code for areas storing the excess of information in any index group.

42. The method of claim 31 where step (3) further comprises:

3a. categorizing in said machine the selected character signals into said character groups based on the physical location of the related characters, such as first and third character locations of each word.

43. The method of claim 31 wherein said machine includes counters, each having a count capacity of sufficient size for counting an average number of character signals in each character group, and further comprising:

6. recognizing in said machine when the character count for any predefined character group exceeds said count capacity, and

45. The method of claim 37, further comprising:

8. recognizing in said machine said force search indication; and

9. searching in said machine for the related information.

46. The method of claim 31 further comprising:

6. developing in said machine abstract count signals from said abstract count, and

7. transferring by said machine information signals and abstract count signals between said machine and a record medium.

47. The method of claim 39 wherein said information is stored on said record medium in index groups each having index words representative of original documents and wherein a particular storage area is allotted for storage of abstracts, and an abstract count is stored for each said index group, said abstract count being based on individual characters categorized into predefined character groups based on a probability distribution of said characters in a selected language, and further comprising:

49. The method of claim 39 wherein said record medium has a plurality of storage tracks thereon with each index group being recorded in an individual one of said storage tracks and wherein a particular track is set aside as said allotted area, and further comprising:

8. positioning in said machine each track on said record medium for accessing of the index stored therein;

9. positioning in said machine said particular track; and

10. recording in said machine each abstract in said particular track.

50. The method of claim 41 wherein said record medium is a magnetic record medium and further comprising:

11. recording in said machine information on said medium by magnetic patterns.

51. The method of claim 41 wherein each index group normally occupies one storage track but occasionally exceeds one track and further comprising:

9. Apparatus for generating at least one abstract that is useful during information searching and retrieval procedures, comprising:

1. means for sensing information comprising individual words in a selected language, each word comprising one or more individual characters;

2. means for categorizing selected ones only of the characters in said words into predefined character groups that are based on a probability distribution of characters in the language selected;

3. means for maintaining a count of the number of characters categorized into each of said predefined character groups;

4. means for converting said count to a compressed apparatus-compatible abstract code form;

10. and means for storing said compressed abstract code as an abstract of said information.

13. The apparatus of claim 10 further comprising:

6. means for generating and recording a special mark in each apparatus-compatible character to signify that it represents an abstract.

14. Apparatus for searching and retrieving information wherein said information is stored on a record medium in index groups each having index words representative of original documents and wherein a separate area is allotted for storage of abstracts, and an abstract count is stored for each said index group, said abstract count being based on individual characters categorized into predefined character groups based on a probability distributiion of said characters in a selected language;

said apparatus comprising;

1. means for sensing said stored abstract counts from said record medium;

2. means for comparing an inquiry count with said sensed abstract counts; and

3. means for providing an indication of a match or mismatch of said abstract and inquiry counts.

17. Apparatus for generating at least one abstract that is useful during searching and retrieval procedures involving information stored on a record medium comprising:

1. means for sensing information comprising individual words in a selected language stored on said record medium, each word comprising one or more individual characters;

18. means for categorizing selected ones only of the characters in said words into predefined character groups that are based on a probability distribution of characters in the language selected;

3. means maintaining a count of the number of characters categorized groups; and

4. means for storing said count on said record medium as an abstract of said information.

24. means for maintaining a count of the number of characters in said inquiry word categorized into each of said predefined character groups;

8. means for sensing said abstract count stored on said record medium;

9. means for comparing said inquiry count with said sensed abstract count; and

10. means providing an indication of a match or mismatch of said abstract and inquiry counts.

30. Apparatus for searching and retrieving information stored on a record medium wherein said information is stored on said record medium in index groups, each having index words representative of original documents and wherein a separate area is allotted on said medium for storage of abstracts, and said record medium storing an abstract count for each said index group, said abstract count being based on individual characters in an index group that comply with predefined character groups based on a probability distribution of said characters in a selected language;

said apparatus comprising;

1. means for sensing said abstract count stored on said record medium;

2. means for comparing an inquiry count with said sensed abstract count; and

3. means providing an indication of a match or mismatch of said abstract and inquiry counts.

35. A method for generating in an information processing machine at least one abstract that is useful during information searching and retrieval procedures, comprising:

36. sensing by said machine information signals representative of information comprising individual words in a selected language, each word comprising one of more individual character;

2. developing in said machine, signals representative of characters in said system;

3. categorizing in said machine, selected ones only of the individual character signals into predefined character groups that are based on a probability distribution of the characters represented by said signals in the language selected;

4. maintaining an abstract count by said machine of the number of character signals categorized into each of said predefined character groups; and

5. storing said abstract count in said machine as an abstract of said information.

38. categorizing in said machine selected ones only of the individual character inquiry signals into predefined character groups that are based on a probability distribution of the characters represented by said signals in the language selected;

9. maintaining in said machine an inquiry count of the number of character inquiry signals categorized into each of said predefined character groups;

10. comparing in said machine said inquiry count with said abstract count; and

11. providing from said machine indication signals representative of a match or mismatch or said abstract and inquiry counts.

44. generating and recording in said machine a force search indication rather than an abstract when count capacity is exceeded in order to signify that a search of the related information is required.

54. The method of claim 44 further comprising:

4. manifesting in said machine the related information, as by printing said related information on a document, responsive to a match of said inquiry count and a said abstract count.

55. The method of claim 44 wherein said record medium has a plurality of storage tracks thereon with each index group being recorded in an individual one of said storage tracks and wherein a particular track is set aside as said allotted area, and further comprising:

4. positioning in said machine each track on said record medium for accessing of the index stored therein;

5. positioning in said machine said particular track, and

6. sensing in said machine each abstract in said particular track.

56. The method of claim 44 wherein said record medium is a magnetic record medium and further comprising:

4. sensing in said machine information on said medium by detection of magnetic patterns on said medium.

69. The method of claim 54 further comprising:

7. providing in said machine an inquiry of information comprising at least an individual word in a selected language, said word comprising at least an individual inquiry character;

8. developing in said machine inquiry signals representative of said at least an individual inquiry character;

9. categorizing in said machine individual character inquiry signals into predefined character groups that are based on a probability distribution of the characters represented by said inquiry signals in the language selected;

10. maintaining in said machine an inquiry count of the number of character inquiry signals categorized into each of said predefined character groups;

11. reconverting in said machine said abstract count from compressed form to a character group count form like said inquiry count;

12. comparing in said machine said inquiry count with said abstract count; and

13. providing from said machine indication signals representative of a match or mismatch of said abstract and inquiry counts.

70. The method of claim 55 wherein each abstract count comprises three non-compatible count characters, and further comprising:

5a. compressing in said machine each abstract count into two apparatus-compatible characters.

71. The method of claim 56 further comprising:

7. generating and recording in said machine a special mark in each apparatus-compatible character to signify that it represents an abstract.

48. locating in said machine each index storage area for accessing of the index group stored therein;

9. locating in said machine said particular storage area; and

10. sensing in said machine each abstract count in said particular storage area.

59. The method of claim 48 further comprising:

6. providing in said machine information inquiry signals representative of an inquiry of information comprising at least an individual word in a selected language said word comprising one or more individual inquiry characters;

52. determining in said machine when an index group on said record medium exceeds one track of storage;

and

12. recording on said record medium in said machine a track skip code for tracks storing the excess of information in any index group.

66. The method of claim 52 further comprising:

manifesting in said machine the index group information related to matched abstract count signals, as by printing said related information on a document, and in response to a match of said inquiry count said signal and abstract count signals.

53. A method for searching and retrieving in an information processing machine, information stored on a record medium in response to an inquiry represented by an inquiry count wherein said information is stored on said record medium in index groups, each having index words representative of original documents and wherein a separate area is allotted on said medium for storage of abstracts, and said record medium storing an abstract count for each said index group, said abstract count being based on individual characters in an index group that comply with predefined character groups based on a probability distribution of said characters in a selected language;

said apparatus comprising;

1. sensing in said machine an abstract count stored on said record medium;

2. comparing in said machine said inquiry count with a sensed abstract count; and

3. providing from said machine an indication of a match or mismatch of said abstract and inquiry counts.

57. A method for generating in an information processing machine at least one abstract that is useful during searching and retrieval procedures involving information stored on a record medium, comprising:

58. sensing by said machine information signals representative of information comprising individual words in a selected language stored on said record medium, each word comprising one or more individual characters;

2. developing in said machine character signals representative of characters in said system;

3. categorizing in said machine selected ones of the individual character signals into predefined character groups that are based a probability distribution of the characters represented by said signals in the language selected;

4. maintaining in said machine an abstract count of the number of character signals categorized into each of said predefined character groups; and

5. storing said count by said machine on said record medium as an abstract of said information.

60. categorizing in said machine selected ones of the individual character inquiry signals into predefined character groups that are based on a probability distribution of the characters represented by said character signals in the language selected;

8. maintaining in said machine an inquiry count of the number of character inquiry signals categorized into each of said predefined character groups;

9. sensing by said machine said abstract count stored on said record medium;

10. comparing by said machine said inquiry count with said sensed abstract count; and

11. providing from said machine indication signals representative of a match or mismatch of said abstract and inquiry count signals.

61. A method for generating in an information processing machine at least one abstract that is useful during searching and retrieval procedures involving information stored on a record medium, comprising:

1. providing by said machine information inquiry signals representative of information comprising at least an individual word in a selected language, said word comprising one or more individual characters;

62. developing character inquiry signals representative of selected ones only of said individual inquiry characters;

3. categorizing in said machine individual character inquiry signals into predefined character groups that are based on a probability distribution of the characters represented by said inquiry signals on said record medium in the language selected;

4. maintaining in said machine an inquiry count of the number of character inquiry signals categorized into each of said predefined character groups; and

5. referencing in said machine said inquiry count as an inquiry during searching of information stored on said record medium by said machine.

63. A method for generating in an information processing machine at least one abstract that is useful during information searching and retrieval procedures, comprising:

1. sensing by said machine information inquiry signals representative of information comprising at least an individual word in a selected language, said word comprising one or more individual characters;

2. developing in said machine character inquiry signals representative of selected ones of said individual inquiry characters;

64. categorizing in said machine individual character inquiry signals into predefined character groups that are based on a probability distribution of the characters represented by said inquiry signals in the language selected;

4. maintaining in said machine an inquiry count of the number of character inquiry signals categorized into each of said predefined character groups; and

5. referencing in said machine said inquiry count as an inquiry during searching of said information by said machine.

65. A method for searching and retrieving information in an information processing machine wherein said information is stored in index groups with individual characters categorized into predefined character groups and with each index group being represented by an abstract count that is based on a probability distribution of said characters in a selected language, comprising:

1. sensing by said machine abstract count signals representative of said stored abstract counts;

2. comparing in said machine an inquiry count signal with said abstract count signals; and

3. providing from said machine indication signals representative of a match or mismatch of said abstract and inquiry count signals.

67. A method for generating in an information processing machine at least one abstract that is useful during information searching and retrieval procedures, comprising:

1. sensing by said machine information signals representative of information comprising individual words in a selected language, each word comprising at least an individual character;

2. developing in said machine signals representative of characters in said system;

3. categorizing in said machine individual character signals into predefined character groups that are based on a probability distribution of the characters represented by said signals in the language selected;

4. maintaining an abstract count by said machine of the number of character signals categorized into each of said predefined character groups;

5. converting in said machine said abstract count to a compressed abstract count form; and

68. storing said compressed abstract count in said machine as an abstract of said information.