Abstract
The purpose of object matching in conflation is to identify corresponding objects in different data sets that represent the same real-world entity. This article presents an improved linear object matching approach, named the optimization and iterative logistic regression matching (OILRM) method, which combines the optimization model and logistic regression model to obtain a better matching result by detecting incorrect matches and missed matches that are included in the result obtained from the optimization (Opt) method for object matching in conflation. The implementation of the proposed OILRM method was demonstrated in a comprehensive case study of Shanghai, China. The experimental results showed the following. (1) The Opt method can determine most of the optimal one-to-one matching pairs under the condition of minimizing the total distance of all matching pairs without setting empirical thresholds. However, the matching accuracy and recall need to be further improved. (2) The proposed OILRM method can detect incorrect matches and missed matches and resolve the issues of one-to-many and many-to-many matching relationships with a higher matching recall. (3) In the case where the source data sets become more complicated, the matching accuracy and recall based on the proposed OILRM method are much better than those based on the Opt method.
Acknowledgments
The authors are grateful to Professor Michael Goodchild and Ms. Linna Li for inspiring our interest in the issue of geospatial data conflation based on the theory of optimization and logistic regression when the first author was a visiting professor in the University of California at Santa Barbara and NCGIA in 2008–2009. The authors greatly appreciate the valuable comments of the anonymous reviewers.
Funding
The work described in this article was substantially supported by the National Natural Science Foundation of China (Project No. 41325005 and 41171352), the National High-tech Research and Development Program (Project No. 2012AA12130), the Fund of Academic Leader Program of Shanghai (Project No. 12XD1404900), and the Kwang-Hua Fund for College of Civil Engineering, Tongji University.