US 7,599,921 B2Granted

System and method for improved name matching using regularized name forms

AssigneeInternational Business Machines Corporation

InventorsDavid Edward Biesenbach, Richard Theodore Gillam, Frankie Elizabeth Patman Maguire, Leonard Arthur Shaefer, Jr., Charles Kinston Williams

Life of this patent

Estimating the term…

patentgrep

Loading patent…

Abstract

A system and method for improved name matching using regularized name forms is presented. A regularization rule engine uses culture-specific regularization rules to iteratively convert candidate names and query names to a canonical form, which are regularized candidate names and regularized query names, respectively. The regularization rules are context-sensitive or context-free rules that pertain to a name's originating culture. Subsequently, a name search engine compares the regularized query name with the regularized candidate names and identifies the regularized candidate names that meet a particular regularization matching threshold. In turn, name search engine selects the candidate names that correspond to the identified regularized candidate names and provides the selected candidate names to a user.

Claims17

Dependent claims

A computer-implemented method comprising:

retrieving, by a processor, a candidate name from memory;
identifying, by the processor, a cultural classification that corresponds to the candidate name;
retrieving, by the processor, one or more culture-specific regularization rules from the memory corresponding to the cultural classification;
applying, by the processor, one or more of the culture-specific regularization rules to the candidate name, resulting in a regularized candidate name, wherein the applying further comprises:
determining that a first regularization rule included in the one or more culture-specific regularization rules applies to the candidate name;
generating a first iteration regularized candidate name by applying the first regularized rule to the candidate name;
determining that a second regularization rule included in the one or more culture-specific regularization rules applies to the candidate name;
and
generating the regularized candidate name by applying the second regularized rule to the first iteration regularized candidate name;
storing the regularized candidate name in the memory;
comparing, by the processor, the regularized candidate name with a regularized query name;
determining, by the processor, that the comparison meets a regularization matching threshold, which indicates a potential match between the regularized candidate name and the regularized query name;
and
in response to determining that comparison meets the regularization matching threshold, providing the candidate name to the user.

11 more elements

A computer program product stored in computer memory, comprising functional descriptive material that, when executed by an information handling system, causes the information handling system to perform actions that include:

retrieving a candidate name;
identifying a cultural classification that corresponds to the candidate name;
retrieving one or more culture-specific regularization rules from corresponding to the cultural classification;
applying one or more of the culture-specific regularization rules to the candidate name, resulting in a regularized candidate name, wherein the applying further comprises:
determining that a first regularization rule included in the one or more culture-specific regularization rules applies to the candidate name;
generating a first iteration regularized candidate name by applying the first regularized rule to the candidate name;
determining that a second regularization rule included in the one or more culture-specific regularization rules applies to the candidate name;
and
generating the regularized candidate name by applying the second regularized rule to the first iteration regularized candidate name;
storing the regularized candidate name;
comparing the regularized candidate name with a regularized query name;
determining that the comparison meets a regularization matching threshold, which indicates a potential match between the regularized candidate name and the regularized query name;
and
in response to determining that comparison meets the regularization matching threshold, providing the candidate name to the user.

11 more elements

An information handling system comprising:

one or more processors;
a memory accessible by the processors;
one or more nonvolatile storage devices accessible by the processors;
and
a set of instructions stored in the memory, wherein one or more of the processors executes the set of instructions in order to perform actions of:
retrieving a candidate name from one of the nonvolatile storage areas;
identifying a cultural classification that corresponds to the candidate name;
retrieving one or more culture-specific regularization rules corresponding to the cultural classification from one of the nonvolatile storage areas;
applying one or more of the culture-specific regularization rules to the candidate name, resulting in a regularized candidate name, wherein the applying further comprises:
determining that a first regularization rule included in the one or more culture-specific regularization rules applies to the candidate name;
generating a first iteration regularized candidate name by applying the first regularized rule to the candidate name;
determining that a second regularization rule included in the one or more culture-specific regularization rules applies to the candidate name;
and
generating the regularized candidate name by applying the second regularized rule to the first iteration regularized candidate name;
and
storing the regularized candidate name in one of the nonvolatile storage areas;
comparing, by the processor, the regularized candidate name with a regularized query name;
determining, by the processor, that the comparison meets a regularization matching threshold, which indicates a potential match between the regularized candidate name and the regularized query name;
and
in response to determining that comparison meets the regularization matching threshold, providing the candidate name to the user.

17 more elements

Dependencies are inferred from the claim text.