Text abstraction method and apparatus

• US 5,691,708 A
• Filed: 08/14/1995
• Issued: 11/25/1997
• Est. Priority Date: 08/14/1995
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: Lotus Development Corporation (International Business Machines Corporation)
• Inventors: Batchelder, Edward Marcus, Salas, R. Pito
• Primary Examiner(s): Zimmerman, Brian
• Assistant Examiner(s): Beaulieu, Yonel
• Application Number:
  US08/514,809
• Time in Patent Office: 834 Days
• Field of Search: 340/825.44, 340/825.22, 340/825.07, 340/825.52, 455/41, 455/151.1, 455/151.2, 455/152.1, 379/57, 379/58
• US Class Current: 340/7.29
• CPC Class Codes: G06F 16/345 Summarisation for human users

Abstract

A text message abstraction system and method allowing for the selection of important pieces of information based upon a number of criteria and for the removal of unnecessary components of a message. The abstraction system of the present invention is provided with an input message, a command set and a maximum message length and attempts to generate a message conveying all of the important information contained within the text message but within the maximum message length constraint.

Claims

1. A method for abstracting input text into output text having a given maximum length, said method comprising the steps of:

dividing said input text into a plurality of chunks, said chunks having at least one word;

assigning a significance to said words of a chunk;

assigning a significance to at least some of said plurality of chunks based upon the significance of words contained within said chunks; and

selecting a subset of said chunks for inclusion in said output text based upon the relative significances of said chunks.

2. The method of claim 1 wherein said selection step comprises selecting the most significant chunks for inclusion in said output text.

3. The method of claim 1 wherein said step of assigning a significance to at least some of said plurality of chunks further comprises the step of classifying at least some of said plurality of chunks based upon predetermined criteria and wherein said significance assigned to a word is dependent upon the classification of the chunk containing said word.

4. The method of claim 3 wherein the significance of a chunk is determined according to the collective significance of the words contained within that chunk.

5. The method of claim 4 wherein the significance of said chunk is further determined according to predetermined chunk criteria.

6. The method of claim 3 wherein said classifications available for said chunks comprise MAIL HEADER, TEXT and PUNCTUATION.

7. The method of claim 6 wherein said MAIL HEADER classification further comprises SIGNIFICANT and INSIGNIFICANT subclassifications and said TEXT classification further comprises FIRST, LAST, ONLY and OTHER subclassifications.

8. The method of claim 1 wherein at least some of said chunks represent a sentence in said input text.

9. The method of claim 1 wherein the significance of said words is determined according to the collective significances corresponding to each occurrence of said word.

10. The method of claim 9 wherein particular words always have a zero significance.

11. The method of claim 9 wherein particular words always have a booted significance.

12. The method of claim 9 wherein in a case where the collective significance corresponding to a word is below a given threshold, that word is assigned a zero significance.

13. The method of claim 1 wherein control of said abstraction process is interactive.

14. The method of claim 1 wherein control of said abstraction process is accomplished in batch.

15. The method of claim 6 wherein PUNCTUATION chunks are discarded.

16. The method of claim 6 wherein secondary MAIL HEADERs are discarded.

38. The method of claim 1, wherein said input text comprises an electronic mail message.

39. The method of claim 3, wherein said predetermined criteria include location within said input text.

40. The method of claim 5, wherein said predetermined chunk criteria include position of said chunk in said input text.

17. A method for abstracting a textual message consisting of a plurality of words, said abstraction comprising the steps of:

dividing said textual message into a plurality of chunks, said chunks having at least one word;

classifying said chunks according to predetermined criteria;

assigning a significance value to said words according to the classification of the chunk containing said words;

assigning a significance value to said chunks according to the total significance values for the words contained in said chunks;

writing said chunks to an output buffer in order of significance of said chunks wherein most significant chunks are written to said output buffer prior to any less significant chunks.

18. The method of claim 17 further comprising the step of transmitting the chunks contained in said output buffer after said output buffer is filled.

19. The method of claim 17 further comprising the step of abbreviating said words.

20. The method of claim 19 wherein said words are abbreviated according to an external file containing original words and corresponding abbreviations.

21. The method of claim 19 wherein said step of abbreviating said words includes the step of removing the vowels from said words.

22. The method of claim 17 wherein predetermined words are not written to said output buffer even if such words are contained in a chunk which is written to said output buffer.

23. The method of claim 17 wherein parenthetical statements are discarded in the event that the significance of a chunk containing said parenthetical statement is not diminished beyond a predetermined threshold when said parenthetical statement is eliminated from said chunk.

41. The method of claim 17, wherein said textual message comprises an electronic mail message.

42. The method of claim 17, wherein said predetermined criteria include position in said textual message.

24. A paging system comprising:

an input source for receiving textual information;

a memory operatively connected to said input source for storing said received textual information;

text abstraction means for abstracting said received textual information; and

a controller communicating with said memory and with said text abstraction means for controlling the operation of said paging system;

an output buffer for storing said abstracted textual information.

25. The paging system of claim 24 further comprising an RF transmission unit for transmitting said abstracted textual information and a paging receiver for receiving said abstracted textual information.

26. The paging system of claim 24 wherein said text abstraction means further comprises:

division means for dividing said received textual information into a plurality of chunks, said chunks having at least one word;

means for assigning a significance to said words of a chunk;

means for assigning a significance to said chunks based upon the significance of the words contained within said chunks;

selection means for selecting a subset of said chunks for inclusion in said output buffer based upon the relative significance of said chunks; and

means for writing said selected chunks into said output buffer.

27. The paging system of claim 26 wherein chunks are written to said output buffer in order of decreasing significance.

28. The paging system of claim 26 wherein said means for assigning a significance to said words of a chunk further comprises means for classifying each chunk based upon predetermined criteria.

29. The paging system of claim 28 wherein said means for assigning a significance to said words of a chunk assigns a significance to said word according to the classification of the chunk containing said word.

30. The paging system of claim 29 wherein said means for assigning a significance to each of said chunks assigns a significance to each of said chunks according to the collective significance of the words contained within that chunk.

31. The paging system of claim 30 wherein the significance of each said chunk is further determined according to that chunk'"'"'s position in said received textual information.

43. The paging system of claim 24, wherein said textual information comprises an electronic mail message.

44. The paging system of claim 28, wherein said predetermined criteria include location of said chunk within said received textual information.

32. A device for abstracting a text message, said device comprising:

input means for receiving said text message;

means for dividing said text message into a plurality of chunks, said chunks having at least one word;

means for classifying said chunks according to predetermined criteria;

means for assigning a significance value to said words according to the classification of the chunk containing said words;

means for assigning a significance value to said chunks according to the total significance values for the words contained in said chunks;

an output buffer; and

means for writing said chunks to said output buffer in order of decreasing significance.

33. The device of claim 32 further comprising means for transmitting the chunks contained in said output buffer after said output buffer is filled.

34. The device of claim 33 further comprising means for abbreviating said words.

35. The device of claim 34 wherein said words are abbreviated according to an external file containing original words and corresponding abbreviations.

36. The device of claim 35 wherein said words are abbreviated by removing the vowels from said words.

37. The device of claim 32 further comprising means for discarding a parenthetical statement in the event that the significance of a chunk containing said parenthetical statement is not diminished beyond a predetermined threshold when said parenthetical statement is eliminated from said chunk.

45. The device of claim 32, wherein said text message comprises an electronic mail message.

46. The device of claim 32, wherein said predetermined criteria include position of said chunk in said text message.