Activity Number:
|
200
|
Type:
|
Topic Contributed
|
Date/Time:
|
Tuesday, August 13, 2002 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Survey Research Methods*
|
Abstract - #301526 |
Title:
|
Improving EM Algorithm Estimates for Record Linkage Parameters
|
Author(s):
|
William Yancey*+
|
Affiliation(s):
|
U.S. Census Bureau
|
Address:
|
4600 Silver Hill Road, Washington, District of Columbia, 20233-9100, U.S.A.
|
Keywords:
|
record linkage ; EM algorithm
|
Abstract:
|
The EM algorithm can be used to estimate conditional probabilities for matching field patterns for the Fellegi-Sunter model for record linkage. The algorithm is based on a latent class model for the record pairs where one of the classes is the set of true matches. If the number of true match pairs in the data set is too small, then the EM algorithm cannot detect the correct latent class. We consider methods for enriching the density of matches in the set of examined record pairs in order to obtain improved EM algorithm estimates for the record linkage conditional probability parameters.
|