Record-Linkage from a Technical Point of View
Publication details
Abstract
Record linkage is used for preparing sampling frames, deduplication of lists and combining information on the same object from two different databases. If the identifiers of the same objects in two different databases have error free unique common identifiers like personal identification numbers (PID), record linkage is a simple file merge operation. If the identifiers contain errors, record linkage is a challenging task. In many applications, the files have widely different numbers of observations, for example a few thousand records of a sample survey and a few million records of an administrative database of social security numbers. Available software, privacy issues and future research topics are discussed.
Keywords: Record-Linkage, Data-mining, Privacy preserving protocols