Relationship analysis system and method for semantic disambiguation of natural language

• US 7,739,102 B2
• Filed: 10/07/2004
• Issued: 06/15/2010
• Est. Priority Date: 10/08/2003
• Status: Active Grant

• Current Assignee: Howard J. Bender
• Original Assignee: Howard J. Bender
• Inventors: Bender, Howard J.
• Primary Examiner(s): Smits; Talivaldis Ivars
• Assistant Examiner(s): Colucci; Michael C
• Application Number:
  US10/571,760
• Publication Number:
  US 20070083359A1
• Time in Patent Office: 2,077 Days
• Field of Search: 704/9, 704/4, 704/273, 704/238, 704/2, 707/3, 706/14
• US Class Current: 704/9
• CPC Class Codes: G06F 16/36 Creation of semantic tools, e.g. ontology or thesauri
• 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

Abstract

A system, method, and computer program product for domain-independent natural language understanding, including at least one of forming pairs of words and/or phrases in a sentence, wherein each word and/or phrase is paired with every other word and/or phrase; determining meanings for the words and/or phrases; assigning numeric codes that uniquely identify semantic concepts to those word and/or phrase meanings; comparing the numeric code of each word and/or phrase with each numeric code of the other word and/or phrase in the pair; selecting the pairs with the best relationships; combining highly-related pairs with other highly-related pairs to form longer groups of words; exchanging numeric codes for the longer groups with numeric codes having a weaker relationship to determine if the exchanged numeric codes provide an overall stronger relationship; and forming longer and longer groups until the sentence is understood.

Claims

1. A computer storage medium comprising computer executable instructions for performing a computer-implemented method for domain-independent natural language understanding, comprising the steps of:

forming pairs of words and/or phrases in independent phrases in a sentence, wherein each word and/or phrase is paired with every other word and/or phrase;

determining meanings for the words and/or phrases;

assigning a numeric code from the pre-existing class and category structure that uniquely identifies semantic concepts to each meaning;

adjusting the numeric codes using a weighting component;

storing the adjusted numeric codes that uniquely identify semantic concepts in a computer memory;

comparing the adjusted numeric codes that uniquely identify semantic concepts of each word and/or phrase with each adjusted numeric code that uniquely identifies the semantic concept of the other word and/or phrase in the pair;

comparing syntactically identical sentences with a potential idiom that changes adjusted numeric codes that uniquely identify semantic concepts depending on the context using an ontology;

determining whether a potential idiom is identified and employing appropriate class and category structure to calculate a first adjusted numeric code total, wherein said first adjusted numeric code total corresponds to the best word and/or phrase pairings;

determining an overall-adjusted numeric code total based on said first adjusted numeric code total by comparing remaining words in a sentence and before selecting the pairs of adjusted numeric codes that uniquely identify semantic concepts with the best relationships based on numerical comparisons;

selecting the pairs of adjusted numeric codes that uniquely indentify semantic concepts with the best relationships based on numerical comparisons;

storing the pairs in the computer memory;

combining highly-related pairs with other highly-related pairs to form longer groups of adjusted numeric codes that uniquely identify semantic concepts;

exchanging adjusted numeric codes that uniquely identify semantic concepts for the longer groups with adjusted numeric codes that uniquely identify semantic concepts having a weaker relationship to determine if the exchanged adjusted numeric codes that uniquely identify semantic concepts provide an overall stronger relationship based on numerical comparisons;

forming longer and longer groups until the adjusted numeric codes that uniquely identify semantic concepts are selected in context for all words and phrases in the sentence; and

relating the adjusted numeric codes that uniquely identify semantic concepts to words and phrases.

2. The method of claim 1, further comprising employing the method for performing language-independent semantic analysis.

3. The method of claim 1, wherein the forming step includes forming legitimate pairs of the words and/or the phrases in the independent phrases in the sentence.

4. The method of claim 1, wherein selecting step includes selecting a same word, with different meanings (expressed as numeric codes that uniquely identify semantic concepts), multiple times, if more than one pair has a strong relationship.

5. The method of claim 1, further comprising analyzing the words and/or phrases of the groups to determine possible meanings based on a numeric class and category structure.

6. The method of claim 1, further comprising employing the method for performing machine translation.

7. The method of claim 1, further comprising employing the method for performing concept determination.

8. The method of claim 7, further comprising employing the concept determination for implementing message routing, and/or a search tool for finding text referring to a topic of interest.

9. The method of claim 1, further comprising employing the method for performing multi-language text analysis.

10. The method of claim 9, further comprising employing a graphical user interface for performing the multi-language text analysis.

11. The method of claim 10, further comprising providing in the graphical user interface a search topic field for entering a search topic, a topic clarification field for entering a word for clarifying the search topic, means for selecting a directory to search, means for selecting a language for the multi-language text analysis, means for starting the multi-language text analysis, and means for viewing a report for listing a text matching the search topic.

12. A computer program product comprising one or more computer-readable instructions configured to cause one or more computer processors to execute the steps recited in claim 1.

13. A computer system comprising one or more computer processors configured to execute the steps recited in claim 1.

14. A system for domain-independent natural language understanding, comprising:

means for forming pairs of words and/or phrases in independent phrases in a sentence, wherein each word and/or phrase is paired with every other word and/or phrase;

means for determining meanings for the words and/or phrases by assigning numeric codes from the pre-existing class and category structure that uniquely identify semantic concepts;

means for adjusting the numeric codes using a weighting component;

means for storing the adjusted numeric codes that uniquely identify semantic concepts;

means for comparing the adjusted numeric codes that uniquely identify semantic concepts of each word and/or phrase with each adjusted numeric code that uniquely identifies the semantic concept of the other word and/or phrase in the pair;

comparing syntactically identical sentences with a potential idiom that changes adjusted numeric codes that uniquely identify semantic concepts depending on the context using an ontology;

means for determining whether a potential idiom is identified and employing appropriate class and category structure to calculate a first adjusted numeric code total, wherein said first adjusted numeric code total corresponds to the best word and/or phrase pairings;

means for determining an overall-adjusted numeric code total based on said first adjusted numeric code total by comparing remaining words in a sentence and selecting the pairs adjusted numeric codes that uniquely identify semantics concepts with the best relationships based on numerical comparisons;

means for selecting the pairs with the best relationships based on numerical comparisons;

means for storing the pairs;

means for combining highly-related pairs with other highly-related pairs to form longer groups of adjusted numeric codes that uniquely identify semantic concepts;

means for exchanging adjusted numeric codes that uniquely identify semantic concepts for the longer groups with adjusted numeric codes that uniquely identify semantic concepts having a weaker relationship to determine if the exchanged adjusted numeric codes that uniquely identify semantic concepts provide an overall stronger relationship based on numerical comparisons;

means for forming longer and longer groups until the adjusted numeric codes that uniquely identify semantic concepts are selected in context for all words and phrases in the sentence; and

means for relating the adjusted numeric codes that uniquely identify semantic concepts to words and phrases.

15. The system of claim 14, wherein the system is configured for performing language-independent semantic analysis.

16. The system of claim 14, wherein the forming means includes means for forming legitimate pairs of the words and/or the phrases in the independent phrases in the sentence.

17. The system of claim 14, wherein selecting means includes means for selecting a same word, with different adjusted numeric codes that uniquely identify semantic concepts, multiple times, if more than one pair has a strong relationship.

18. The system of claim 14, further comprising means for analyzing the words and/or phrases of the groups to determine possible adjusted numeric codes that uniquely identify semantic concepts based on a pre-existing numeric class and category structure.

19. The system of claim 14, wherein the system is configured for performing machine translation.

20. The system of claim 14, wherein the system is configured for performing concept determination.

21. The system of claim 20, wherein the concept determination is configured for implementing message routing, and/or a search tool for finding text referring to a topic of interest.

22. The system of claim 14, further comprising employing the system for performing multi-language text analysis.

23. The system of claim 22, further comprising a graphical user interface for performing the multi-language text analysis.

24. The system of claim 23, wherein the graphical user interface includes a search topic field for entering a search topic, a topic clarification field for entering a word for clarifying the search topic, means for selecting a directory to search, means for selecting a language for the Multilanguage text analysis, means for starting the multi-language text analysis, and means for viewing a report for listing a text matching the search topic.

25. The system of claim 14, wherein the means for determining, the means for comparing, the means for selecting, the means for combining, the means for exchanging, and the means for forming comprise one or more devices of a computer system.

26. The system of claim 14, wherein the means for determining, the means for comparing, the means for selecting, the means for combining, the means for exchanging, and the means for forming comprise one or more computer-readable instructions stored on a computer readable medium.

27. A computer program product for domain-independent natural language understanding, including one or more computer readable instructions embedded on a computer readable medium and configured to cause one or more computer processors to perform the steps of:

forming pairs of words and/or phrases in independent phrases in a sentence, wherein each word and/or phrase is paired with every other word and/or phrase;

determining meanings for the words and/or phrases;

assigning a numeric code from the pre-existing class and category structure that uniquely identifies the semantic concept to each meaning;

adjusting the numeric codes using a weighting component;

storing the adjusted numeric codes that uniquely identify semantic concepts in a computer memory;

comparing the adjusted numeric codes that uniquely identify semantic concepts of each word and/or phrase with each adjusted numeric code that uniquely identifies the semantic concept of the other word and/or phrase in the pair;

comparing syntactically identical sentences with a potential idiom that changes adjusted numeric codes that uniquely identify semantic concepts depending on the context using an ontology;

determining whether a potential idiom is identified and employing appropriate class and category structure to calculate a first adjusted numeric code total, wherein said first adjusted numeric code total corresponds to the best word and/or phrase pairings;

determining an overall-adjusted numeric code total based on said first adjusted numeric code total by comparing remaining words in a sentence and selecting the pairs of adjusted numeric codes that uniquely identify semantic concepts with the best relationships based on numerical comparisons;

selecting the pairs of adjusted numeric codes that uniquely indentify semantic concepts with the best relationships based on numerical comparisons would be considered a better relationship than codes 4 and 10;

storing the pairs in the computer memory;

combining highly-related pairs with other highly-related pairs to form longer groups of adjusted numeric codes that uniquely identify semantic concepts;

exchanging adjusted numeric codes that uniquely identify semantic concepts for the longer groups with adjusted numeric codes that uniquely identify semantic concepts having a weaker relationship to determine if the exchanged adjusted numeric codes that uniquely identify semantic concepts provide an overall stronger relationship based on numerical comparisons;

forming longer and longer groups until the adjusted numeric codes that uniquely identify semantic concepts are selected in context for all words and phrases in the sentence; and

relating the adjusted numeric codes that uniquely identify semantic concepts to words and phrases.