Use of graph theory measures to identify errors in record linkage

Area of Science:

• Data Science
• Computer Science
• Graph Theory

Background:

• High-quality record linkage is crucial for data analysis.
• Current quality assurance methods are manual and time-consuming.
• Need for automated, efficient techniques to detect linkage errors.

Purpose of the Study:

• To evaluate the effectiveness of graph theory in identifying record linkage errors.
• To compare graph theory methods against traditional thresholding techniques.
• To explore potential for improving data quality and reducing delivery time.

Main Methods:

• Applied various graph theory techniques to two linked datasets.
• Utilized known truth sets for validation.
• Compared error identification performance against a standard thresholding method.

Main Results:

• Graph theory techniques demonstrated potential in identifying groups of records with linkage errors.
• Performance comparison indicated promise compared to threshold setting.
• Further investigation is warranted to optimize these graph theory approaches.

Conclusions:

• Graph theory offers a promising avenue for enhancing record linkage quality assurance.
• Development of efficient graph theory methods can lead to faster delivery of high-quality datasets.
• Continued research is essential for practical implementation and broader application.