Natural language based search engine for handling pronouns and methods of use therefor

• US 7,555,475 B2
• Filed: 09/09/2005
• Issued: 06/30/2009
• Est. Priority Date: 03/31/2005
• Status: Active Grant

• Current Assignee: Jiles Incorporated
• Original Assignee: Jiles Incorporated
• Inventors: Fischer, Gordon H., Flowers, John S., Quiroga, Martin A.
• Primary Examiner(s): Lewis, Cheryl
• Application Number:
  US11/223,169
• Publication Number:
  US 20060224570A1
• Time in Patent Office: 1,390 Days
• Field of Search: 707/100, 707 2- 5, 715/200, 704/8, 704/9
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 16/90332 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 40/247 Thesauruses; Synonyms
• CPC Class Codes: G06F 40/295 Named entity recognition
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching

Abstract

There is provided a sentence module that handles pronouns in sentences. Each pronoun is replaced by one or more nouns. These replaced nouns are used to form statements that populate the structured representation, in order to produce precise answers to queries, as part of a search engine application.

Claims

1. A method for replacing pronouns in word groups derived from

predetermined collections of words, comprising:

isolating at least one pronoun in each word group; and

,

determining at least one noun to replace each of the isolated pronouns, by evaluating and assigning a weight value to each of the nouns with respect to their proximity to each pronoun that has been isolated, in the predetermined collection of words.

2. The method of claim 1, wherein the predetermined collection of words is selected from the group consisting of at least one sentence, at least one phrase, or at least one question.

3. The method of claim 1, wherein each word group includes a pair of words.

4. A method for creating statements for providing at least one response to at least one query comprising:

creating an array having at least one pronoun;

parsing a series of words arranged in a predetermined order into a parsed output, the parsed output including a first collection of words;

comparing the pronoun array with pronouns from the parsed output to identify pronouns;

replacing each identified pronoun in the first collection of words with a noun corresponding to the matched pronoun to form a second collection of words; and

assigning concept identifiers, if corresponding concept identifiers exist for each word in at least one database.

5. The method of claim 4, wherein the words in the second collection of words are ordered in groups in accordance with predetermined relationships of the first collection of words.

6. The method of claim 5, wherein the groups include pairs.

7. The method of claim 4, wherein the concept identifiers are paired in accordance with their respective ordered word pairs to form concept link identifiers.

8. The method of claim 7, wherein the concept link identifiers are arranged into statements.

9. The method of claim 4, wherein replacing each matched pronoun in the first collection of words with a noun corresponding to the matched pronoun includes, extracting nouns and pronouns from the parsed output, arranging the extracted nouns and pronouns into an ordered format, and, analyzing the positions of the nouns with respect to the pronoun in the ordered format to evaluate and assign a weight value to the noun for replacing each pronoun in order to form the second collection of words.

10. A method for producing statements from paired words comprising:

obtaining related groups of words from a first parse of a predetermined collection of words;

isolating the nouns and pronouns from the first parse into a second parse;

comparing predetermined pronouns against the nouns and pronouns of the second parse for pronoun matches;

determining at least one noun for each matched pronoun;

replacing each matched pronoun with the at least one determined noun in the corresponding related group of words; and

assigning concept identifier links to the resulting word pairs.

11. The method of claim 10, wherein each related group of words includes a pair of words.

12. The method of claim 10, wherein the predetermined collection of words is selected from the group consisting of at least one sentence, at least one phrase, or at least one question.

13. The method of claim 12, wherein the predetermined collection of words is in natural language.

14. In a search engine architecture, a module for handling pronouns, comprising:

at least one storage media for storing a list of predetermined pronouns; and

,

a processor programmed to;

obtain related groups of words from a first parse of a predetermined series of words

isolate the nouns and pronouns from the first parse into a second parse;

compare predetermined pronouns from the at least one storage media against the nouns and pronouns of the second parse for pronoun matches;

determine at least one noun for each matched pronoun; and

,

replace each matched pronoun with the at least one noun in the corresponding related group of words; and

assign concept identifier links to the resulting word pairs.

15. The module of claim 14, wherein the processor programmed to obtain related groups of words includes being programmed to obtain related pairs of words.

16. The module of claim 15, wherein the predetermined series of words is selected from the group consisting of at least one sentence, at least one phrase, or at least one question.

17. The module of claim 16, wherein the at least one predetermined series of words is in natural language.

18. The module of claim 14, wherein the processor programmed to determine the at least one noun for each pronoun includes analyzing the second parse based on the order of the matched pronoun with respect to at least one noun.

19. A method for providing at least one response to at least one query in natural language, comprising:

populating a data store by obtaining documents from at least a portion of a corpus;

isolating sentences from the documents, parsing the sentences into linked pairs of words in accordance with predetermined relationships;

identifying pronouns in each of the linked pairs of words;

replacing each of the identified pronouns with at least one corresponding noun;

assigning concept identifiers to each word of the linked pair of words;

assigning concept link identifiers to each pair of concept identifiers corresponding to each linked pair of words; and

,

combining valid concept link identifiers for each sentence into a statement.

20. The method of claim 19, additionally comprising:

receiving an inputted query in natural language;

parsing the query into linked pairs of words;

assigning concept identifiers to each word of each linked pair of words;

assigning concept link identifiers to each pair of concept identifiers corresponding to each linked pair of words; and

,

combining valid concept link identifiers into a query statement.

21. The method of claim 20, additionally comprising:

analyzing the query statement and the statements in the data store for matches between concept link identifiers;

isolating statements in the data store having at least one concept link identifier that matches at least one concept link identifier in the query statement; and

,

providing at least one sentence corresponding to at least one isolated statement in the data store as a response to the natural language query.

22. The method of claim 21, additionally comprising:

providing access to at least one document from which the at least one sentence, corresponding to the at least one matched statement, was isolated.

23. A computerized method for creating concept links from a set of concept groups derived from data from a corpus at a processor, the processor embodying a program of instructions executable by the processor to perform method steps, the method steps, comprising:

analyzing the words in word groups corresponding to each of the concept groups, for words that are pronouns;

replacing each of the pronouns in each of the word groups with nouns corresponding to each of the pronouns; and

,

assigning concept identifier links to the resulting word pairs.

24. The method of claim 23, wherein the concept groups include concept pairs.

25. The method of claim 23, wherein the word groups include word pairs.

26. A system for providing at least one response to a received query, comprising:

at least one storage media for storing concept identifiers, concept link identifiers, pronouns, and at least one noun corresponding to each of the pronouns, extracted from a corpus, and statements derived from the corpus; and

,

a processor in communication with the at least one storage media, the processor programmed to;

create related pairs of words from the predetermined series of words from documents of the corpus, each of the related pairs of words including one word at a first position and one word at a second position;

identify pronouns by matching pronouns in each of the related pairs of words with pronouns in the at least one storage media;

replace each of the identified pronouns with at least one corresponding noun;

assign concept identifiers to each word in each pair of words;

create pairs of concept identifiers by applying the assigned concept identifiers to each word in the related pairs of words;

assign concept link identifiers to each pair of concept identifiers;

combine all of the valid concept link identifiers into a statement; and

store the statement in the at least one storage media.

27. The system of claim 26, wherein the processor is additionally programmed to:

obtain a plurality of concept link identifiers arranged as a statement from a query entered into the system, the plurality of concept link identifiers defining a query statement;

arrange all of the concept link identifiers of the query statement into a master set, where N is the number of concept link identifiers in the master set; and

,

create a power set from the master set including, creating a plurality of subsets from the master set, the plurality of subsets defining members of the power set, the power set including at least one member of N concept link identifiers and at least N members of one concept link identifier.

28. The system of claim 27, wherein the processor is additionally programmed to:

analyze at least one statement stored in the at least one storage media with the members of the power set, including, determining matches of the concept link identifiers in the at least one stored statement with all of the concept link identifiers in each member of the power set.

29. The system of claim 28, wherein the processor is additionally programmed to:

isolate the at least one stored statement, if the at least one stored statement has concept link identifiers that match all of the concept link identifiers in a member of the power set.

30. The system of claim 29, wherein the processor is additionally programmed to:

select at least one of the at least one stored statement that has been isolated, to provide at least one response to the query.

31. The system of claim 26, wherein the at least one storage media includes a structured representation of the corpus.

32. The system of claim 26, wherein the predetermined series of words is selected from the group consisting of at least one sentence, at least one phrase, or at least one question.

33. A computerized method for conducting a natural language search at a processor, the processor embodying a program of instructions executable by the processor to perform method steps, the method steps comprising:

detecting pronouns in word groups obtained from a corpus of data, and replacing the pronouns in the word groups with nouns from the corpus of data to form substituted statements;

creating word groups from an inputted query;

isolating word groups from the substituted statements that match at least one word group from the query; and

,

providing at least one response to the inputted query based on the matching word groups.

34. The method of claim 33, wherein providing at least one response includes determining a total weight value for each response based on weight values of each of the matching word groups.

35. The method of claim 34, wherein the weight values for the word groups whose pronouns were replaced is equal to or less then the weight values for the word groups that did not include pronouns.