Determining similarity of unfielded names using feature assignments

Patent metadata

Field	Value
Publication number	US-9251140-B2
Application number	US-201313939718-A
Country	US
Kind code	B2
Filing date	Jul 11, 2013
Priority date	Dec 3, 2012
Publication date	Feb 2, 2016
Grant date	Feb 2, 2016

Abstract

Provided are techniques for comparing names. A first phrase score is obtained by comparing a name phrase in a first name to a name phrase in a second name. A second phrase score is obtained by comparing another name phrase in the first name to another name phrase in the second name. An overall score is generated based on the obtained first phrase score and the obtained second phrase score. The overall score is updated based on comparing features of the first name with features of the second name.

First claim

The invention claimed is:

1. A method for comparing names, comprising:
receiving, into a computer, with a processor of the computer, a first name for comparison;
for each pair of names including the first name and a second name selected from a set of names:
identifying features of the first name that comprise semantic relationships that include information about cultural derivation, hierarchical relationships that describe dependency relationships between multiple words that form a first name phrase within the first name, position information for each of the words in the first name phrase, and position information for name phrases in the first name;
identifying features of the second name that comprise semantic relationships that include information about cultural derivation, hierarchical relationships that describe dependency relationships between multiple words that form a second name phrase within the second name, position information for each of the words in the second name phrase, and position information for name phrases in the second name;
obtaining a first phrase score by comparing the first name phrase in the first name to the second name phrase in the second name;
obtaining a second phrase score by comparing another name phrase in the first name to another name phrase in the second name;
generating an overall score for the pair of names based on the obtained first phrase score and the obtained second phrase score; and
updating the overall score for the pair of names based on comparing features of the first name with features of the second name; and

providing, with the computer, a list of the second names from the set of second names that matched the first name in order of the updated score for each pair of names.

2. The method of claim 1 , further comprising:
obtaining additional phrase scores; and
selecting highest phrase scores from among the first phrase score, the second phrase score, and the additional phrase scores for generating the overall score.

3. The method of claim 1 , further comprising:
adjusting the first phrase score for at least one of an incompatible feature and a missing dependency feature.

4. The method of claim 1 , further comprising:
adjusting the first phrase score based on comparing features of the first name with the features of the second name.

5. The method of claim 1 , wherein software is provided as a service in a cloud environment.

Assignees

• IBM

Inventors

• Patman Maguire Frankie E

Classifications

• G06F40/295 Primary

  Named entity recognition · CPC title

• G06F40/30 Primary

  Semantic analysis · CPC title

• G06F17/2785 Primary

  Physics · mapped topic

• G06F17/278

  Physics · mapped topic