Enterprise Software

Reasoning About Record Matching Rules

Download Now Free registration required

Executive Summary

Record matching is the problem for identifying tuples in one or more relations that refer to the same real-world entity. This problem is also known as record linkage, merge-purge, and duplicate detection and object identification. The need for record matching is evident. In data integration it is necessary to collate information about an object from multiple data sources. In data cleaning it is critical to eliminate duplicate records. In master data management one often needs to identify links between input tuples and master data. The need is also highlighted by payment card fraud, which cost $4.84 billion worldwide in 2006. In fraud detection it is a routine process to cross-check whether a card user is the legitimate card holder.

  • Format: PDF
  • Size: 605.6 KB