Please use this identifier to cite or link to this item:
Ehrenfeld, Wilfried
Year of Publication: 
Series/Report no.: 
IWH Technical Reports 02/2015e
The primary objective of record linkage is the merger of different data sets on the basis of an unique identifier. The cases at hand are mostly company data sets from databanks with company characteristics (e.g. BvD Amadeus/Dafne), patent data sets (e.g. Patstat or DPMA) and funding data sets (e.g. BMBF funding catalog). These data sets shall be merged on the basis of the company names. Due to the fact that company names have varying notations in different databases - for example the corporate structure – a harmonization and standardization is necessary. The routines described here implement the record linkage pre-cleaning (RLPC). They are used to create record linkage compatible names (RLName) from given (actor) names (Name). This includes converting special characters to ASCII characters, identifying corporate structures, isolating and separating bracketed expressions. The result is an expression which allows for a comparison with other names. Following this pre-cleaning, record linkage systems can be used to merge several data sets that have been pretreated in the same way.
Abstract (Translated): 
Übergeordnetes Ziel des Record-Linkage ist die Zusammenführung verschiedener Datensätze anhand eines eindeutigen Identifizierungsmerkmals. In den uns vorliegenden Fällen handelt es sich primär um Unternehmensdatensätze aus Datenbanken mit Unternehmensmerkmalen (z. B. BvD Amadeus/Dafne), Patentdatensätze (z. B. Patstat oder DPMA) sowie Förderdatensätze (z. B. BMBF Förderkatalog). Diese Datensätze sollen über den Namen der Unternehmen verknüpft werden. Da in der Praxis Unternehmensnamen in verschiedenen Datenbasen uneinheitlich geschrieben werden – beispielsweise die Unternehmensform – ist eine Harmonisierung und Standardisierung notwendig.
Document Type: 
Research Report
Social Media Mentions:

Files in This Item:

Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.