Method and apparatus to identify the relation of meaning between words in text expressions

• US 4,849,898 A
• Filed: 05/18/1988
• Issued: 07/18/1989
• Est. Priority Date: 05/18/1988
• Status: Expired due to Term

• Current Assignee: Management Information Technologies Incorporated
• Original Assignee: Management Information Technologies Incorporated
• Inventors: Adi, Tammam
• Primary Examiner(s): Ruggiero, Joseph
• Assistant Examiner(s): Hayes, Gail O.
• Application Number:
  US07/195,293
• Time in Patent Office: 426 Days
• Field of Search: 364/419, 364/274.8, 364/943, 364/200, 364/900
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3338 Query expansion
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 40/58 Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

A text comprehension and retrieval method and apparatus that uses letter-semantic analysis of the micro-syntax, or the syntax between the letters, in two words to measure how much two words are related as to their meanings or the human language concepts they present. Letter-semantic analysis involves assigning numerical values to the letters of a first word and a second word based on the dual characteristics of orientation and category inherent in the letters, and then analyzing those numerical values to identify the semantic relatedness of the letters of the first word to the letters of the second word. A letter semantic-matrix assigns weights to the meaningful letters to allow the application of letter semantic rules to convert the concepts represented by the letters of the words to numeric values. The numeric values represent the amount of relatedness of the first word to the second word and are used to retrieve text from documents having concepts related to a user supplied query expression.

Claims

1. A text comprehension and retrieval method for identifying text words presenting concepts in a computer coded document related to the concepts presented in a computer coded expression of query words, comprising the steps of:

calculating a first letter-semantic value for each of the query words by comparing each of the query words to itself;

calculating a plurality of second letter-semantic values for each of the text words in a plurality of blocks of text of dynamically predetermined size by comparing each of the query words to each of the text words in each of said blocks;

identifying the text words of the one of said blocks having second letter-semantic values that compare most closely to said first letter-semantic values.

2. The method of claim 1 wherein said step of calculating a first letter-semantic value and said step of calculating a plurality of second letter-semantic values includes assigning a numeric value to the relationship of the letters of a first word to the letters of a second word according to a predetermined relationship matrix.

3. The method of claim 2 wherein said step of assigning a numeric value includes reducing said first word to a first stem of more than one and less than four consonants and reducing said second word to a second stem of more than one and less than four consonants, wherein the digraphs TH, GN, NG, NC, NK, CK, CH, GH, CQ, PH, and SH are treated as a single consonant.

4. The method of claim 3 wherein said step of assigning a numeric value includes assigning an orientation number, assigning orientation weight, assigning a category number and assigning a category weight to each consonant of said first and second stems, based on the following letter semantic matrix:

space="preserve" listing-type="tabular">______________________________________ CATE- GORY WEIGHT ORIENTATION ______________________________________ 1 2 3 4 1 1 AA J V, W H 2 1 M F, PH, P.sub.2 D DH, TH 3 1 QH N, GN Q, CQ NG, NK, NC 4 2 R L B T 5 2 S Z C HC 6 2 K, CK DK TK X, KH, GH, CH 7 3 HH SH G ZH ______________________________________

and where the category weight of the letter P is 2.

5. The method of claim 4 wherein said step of assigning an orientation number includes locating each consonant of each of said stems in a column of said matrix and assigning to said each consonant the orientation number at the top of said column, said step of assigning an orientation weight includes assigning the value of 0.5 to said each consonant, said step of assigning a category number includes locating said each consonant in a row of said matrix and assigning to said each consonant the category number of said row, and said step of assigning a category weight includes assigning to said each consonant the category weight number of said row.

6. The method of claim 5 wherein said step of assigning a numeric value includes comparing the orientation numbers of each pair of adjacent consonants of said first stem to the orientation numbers of each pair of adjacent consonants of said second stem and increasing an orientational association number by the sum of the orientation weights of the pair of adjacent consonants of said first stem when said orientation numbers of each pair match.

7. The method of claim 5 wherein said step of assigning a numeric value includes comparing the category numbers of each pair of adjacent consonants of said first stem to the category numbers of each pair of adjacent consonants of said second stem and increasing a category association number by the sum of the category weights of the pair of adjacent consonants of said first stem having the same category numbers as the pair of adjacent consonants of said second stem.

8. A text comprehension retrieval apparatus for identifying text words presenting concepts in a computer coded document related to the concepts presented in a computer coded expression of query words, comprising:

means for calculating a first letter-semantic value for each of the query words by comparing each of the query words to itself;

means for calculating a plurality of second letter- semantic values for each of the text words in a plurality of blocks of text of dynamically predetermined size by comparing each of the query words to each of the text words in each of said blocks;

means for identifying the text words of the one of said blocks having second letter-semantic values that compare most closely to said first letter-semantic values.

9. The apparatus of claim 8 wherein said means for calculating a first letter-semantic value and said means for calculating a plurality of second letter-semantic values includes means for assigning a numeric value to the relationship of the letters of a first word to the letters of a second word according to a predetermined relationship matrix.

10. The apparatus of claim 9 wherein said step of assigning a numeric value includes reducing said first word to a first stem of more than one and less than four consonants and reducing said second word to a second stem of more than one and less than four consonants, wherein the digraphs TH, GN, NG, NC, NK, CK, CH, GH, CQ, PH, and SH are treated as a single consonant.

11. The apparatus of claim 10 wherein said means for assigning a numeric value includes assigning an orientation number, assigning an orientation weight, assigning a category number and assigning a category weight to each consonant of said first and second stems, based on the following letter semantic matrix:

space="preserve" listing-type="tabular">______________________________________ CATE- GORY WEIGHT ORIENTATION ______________________________________ 1 2 3 4 1 1 AA J V, W H 2 1 M F, PH, P.sup.2 D DH, TH 3 1 QH N, GN Q, CQ NG, NK, NC 4 2 R L B T 5 2 S Z C HC 6 2 K, CK DK TK X, KH, GH, CH 7 3 HH SH G ZH ______________________________________

and where the category weight of the letter P is 2.

12. The apparatus of claim 11 wherein said means for assigning an orientation number includes locating each consonant of each of said stems in a column of said matrix and assigning to said each consonant the orientation number at the top of said column, said means for assigning an orientation weight includes means for assigning the value of 0.5 to said each consonant, said means for assigning a category number includes means for locating said each consonant in a row of said matrix and assigning to said each consonant the category number of said row, and said means for assigning a category weight includes means for assigning to said each consonant the category weight number of said row.

13. The apparatus of claim 12 wherein said means for assigning a numeric value includes means for comparing the orientation numbers of each pair of adjacent consonants of said first stem to the orientation numbers of each pair of adjacent consonants of said second stem and means for increasing an orientational association number by the sum of the orientation weights of the pair of adjacent consonants of said first stem when said orientation numbers of each pair match.

14. The apparatus of claim 12 wherein said means for assigning a numeric value includes means for comparing the category numbers of each pair of adjacent consonants of said first stem to the category numbers of each pair of adjacent consonants of said second stem and means for increasing a category association number by the sum of the category weights of the pair of adjacent consonants of said first stem having the same category numbers as the pair of adjacent consonants of said second stem.

15. A method for determining a relation between a first word and a second word, comprising the steps of:

storing an array of characters that represent selected alphabetical letters;

assigning a predetermined first and second numeric position value and a predetermined first and second numeric weight value to each of said characters;

converting selected letters of the first word and selected letters of the second word into first and second numeric position values and first and second numeric weight values corresponding to said first and second numeric position values and first and second numeric weight values assigned to those of said characters representing said selected letters;

comparing the first and second numeric position values of said selected letters of the first word to the first and second numeric position values of said selected letters of the second word;

accumulating the first and second numeric weight values of those of said selected letters of the first word having first and second numeric position values corresponding to the first and second numeric position values of said selected letters of the second word, respectively; and

determining similarity between the first word and the second word in accordance with the accumulated first and second numeric weight values.

16. The method of claim 15 wherein said step of comparing includes comparing the first and second numeric position values of pairs of selected letters of the first word to the first and second numeric position values of corresponding pair of letters of the second word.

17. The method of claim 15 wherein said step of storing includes storing characters in a position of said array corresponding to a first and second meaning of the alphabetic letters represented by said character.

18. An apparatus for determining a relation between a first word and a second word, comprising:

means for storing an array of characters that represent selected alphabetical letters;

means for assigning a predetermined first and second numeric position value and a predetermined first and second numeric weight value to each of said characters;

means for converting selected letters of the first word and selected letters of the second word into first and second numeric position values and first and second numeric weight values corresponding to said first and second numeric position values and first and second numeric weight values assigned to those of said characters representing said selected letters;

means for comparing the first and second numeric position values of said selected letters of the first word to the first and second numeric position values of said selected letters of the second word;

means for accumulating the first and second numeric weight values of those of said selected letters of the first word having first and second numeric position values corresponding to the first and second numeric position values of said selected letters of the second word, respectively; and

means for determining similarity between the first word and the second word in accordance with the accumulated first and second numeric weight values.

19. The apparatus of claim 18 wherein said means for comparing includes means for comparing the first and second numeric position values of pairs of selected letters of the first word to the first and second numeric position values of corresponding pairs of letters of the second word.

20. The apparatus of claim 18 wherein said means for storing includes means for storing characters in a position of said array corresponding to a first and second meaning of the alphabetic letters represented by said character.

21. A method for determining sematic values of a word, comprising the steps of:

storing an array of characters that represent selected alphabetical letters;

assigning a predetermined first and second numeric positon value and a predetermined first and second numeric weight value to each of said characters;

converting selected letters of the word into first and second numeric position values and first and second numeric weight values corresponding to said first and second numeric position values and first and second numeric weight values assigned to those of said characters representing said selected letters.