Pattern generation

• US 8,423,348 B2
• Filed: 06/10/2006
• Issued: 04/16/2013
• Est. Priority Date: 03/08/2006
• Status: Active Grant

• Current Assignee: Trigent Software Ltd.
• Original Assignee: Trigent Software Ltd.
• Inventors: Rehberg, Charles, Satyanarayana, Krishnamurthy, Seshadri, Rengarajan, Comandur, Vasudevan, Mehta, Abhishek, Goel, Amit
• Primary Examiner(s): GUERRA-ERAZO, EDGAR X
• Application Number:
  US11/451,239
• Publication Number:
  US 20070213973A1
• Time in Patent Office: 2,502 Days
• Field of Search: 704 1- 10, 704/251, 704/255, 704/257, 704/270, 704/270.1, 704/272, 704/274, 707/776
• US Class Current: 704/9
• CPC Class Codes: G06F 40/295 Named entity recognition

Abstract

A method and system is disclosed herein for generating a plurality of equivalent sentence patterns from a declared sentence pattern for a specific language. The declared pattern is fed into a pattern selector. The pattern selector reads a predetermined library of equivalent pattern sets and selects an equivalent pattern set for the declared pattern. The selected equivalent pattern set corresponds to the declared pattern and represents a set of equivalent declared patterns. The set of equivalent declared patterns and the declared pattern are fed to a rules generator. The rules generator outputs executable semantic pattern recognition rules. The reader module, using the generated executable semantic pattern recognition rules, reads the given information source to determine the information of interest.

Claims

1. A computer-implemented method of generating a set of equivalent declared patterns from each declared pattern in a reading style of a user in a specific language, comprising:

a computer comprising one or more processors configured to;

generating a plurality of predetermined equivalent pattern sets, wherein each of said equivalent pattern sets represents a set of linguistic formations representing the grammar constraints of a given language or genre, wherein each of the said equivalent pattern sets individually resolve to the same meaning as represented in an ontology, wherein each of said equivalent pattern sets comprises a specific combination of one or more pattern variables and includes additional terms from the language, wherein each of said equivalent pattern sets comprise

a plurality of equivalent patterns, wherein each of said equivalent patterns in an equivalent pattern set is a linguistic formation conveying the same meaning and comprise a different arrangement of said specific combination of one or more of said pattern variables and includes the same or different terms from the language, wherein said pattern variables in each of said equivalent patterns are arranged in a relative linguistic sequence, and wherein each of said pattern variables in each of the equivalent patterns comprises a plurality of specified grammar constraints;

providing a plurality of declared patterns for each sentence in a given language or genre, wherein said declared patterns are determined by providing text in the given language or genre, wherein each of said declared patterns comprises a sequence of linguistic pattern units comprising identified senses of a term in each input phrase or sentence in said provided text;

feeding each of said declared patterns to a pattern selector for selecting one or more equivalent pattern sets for said declared pattern;

comparing each of said declared patterns with each pattern in each of said equivalent pattern sets, wherein said declared pattern represents arrangement of said linguistic pattern units of said declared pattern in said sequence, and wherein a match occurs when said relative linguistic sequence and said sequence of at least one of said declared patterns matches with said relative linguistic sequence of one of said equivalent patterns of an equivalent pattern set;

selecting the matching equivalent pattern set for said declared patterns from said equivalent pattern sets by said pattern selector;

mapping said pattern variables within said matched equivalent pattern set to the sense of the matching declared pattern units; and

substituting said pattern variables of each of said equivalent patterns of said matched equivalent pattern set with said matching declared pattern units to generate a set of equivalent declared patterns.

2. A computer-implemented method of applying equivalent declared patterns and equivalent name sets to generate executable semantic pattern recognition rules for use in determining the information of interest from an information source comprising:

a computer comprising one or more processors configured to;

inputting said equivalent declared patterns;

defining a specification of a target rules language format by a target rules language format module;

providing a library of equivalent name sets;

reducing said matching declared pattern units of each equivalent declared pattern to semantic items in an ontology, wherein each semantic item represents an equivalent name set, to produce equivalent name set enhanced semantic units; and

applying said specification of said target rules language format to said equivalent declared patterns containing said equivalent name set enhanced semantic units to generate executable semantic pattern recognition rules by a rules generator.

4. The method of claim 2, further comprising the step of resolving references, comprising the steps of:

identifying reference items that are words or phrases that require resolution;

determining if a language specific context contains an equivalent name set substitution for said reference item;

wherein said language specific context is created or updated for each identified pattern at run time with the role, number and any other distinguishing factors; and

substituting said equivalent name set substitution for the reference item.

6. The method of claim 2, wherein a semantic unit comprises words, sentences or paragraphs.

3. A computer-implemented method of applying executable semantic pattern recognition rules to an information source for identifying the information of interest in said information source for an end user, comprising:

a computer comprising one or more processors configured to;

providing said executable semantic pattern recognition equivalent sentence patterns;

providing said information source;

identifying the sections of the information source that require processing, wherein said step of identifying is directed by a reading plan of said end user;

substituting each term in the information source with equivalent name set enhanced semantic units, wherein said term appears in the equivalent name set of said equivalent name set enhanced semantic unit;

providing a library of equivalent name sets;

decomposing compound sentences of said equivalent name set enhanced semantic units to generate simplified equivalent name set enhanced semantic units; and

applying said executable semantic pattern recognition rules to said information

source with said substituted equivalent name set enhanced semantic units to identify the information of interest.

5. The method of claim 3, wherein the step of providing and applying a specification of the target rules language format to said equivalent patterns, further comprises the step of applying grammar logic using a grammar function signature specification, wherein said grammar function signature specification further comprises:

an instance-level equivalent name set function that is an interface definition to a function that handles the equivalent name sets;

an instance-level number function that is an interface definition to a function that handles singular and plural forms of a word;

an instance-level reference resolution function that is an interface definition to a function that handles the resolution of pronouns; and

an instance-level tense function that is an interface definition to a function that handles the different forms of a verb.