Bitte verwenden Sie diesen Link, um diese Publikation zu zitieren, oder auf sie als Internetquelle zu verweisen: https://hdl.handle.net/10419/268428 
Autor:innen: 
Erscheinungsjahr: 
2023
Schriftenreihe/Nr.: 
ZEW Discussion Papers No. 23-001
Verlag: 
ZEW - Leibniz-Zentrum für Europäische Wirtschaftsforschung, Mannheim
Zusammenfassung: 
The SearchEngine is an open source project providing an integrated framework for diverse matching activities, especially the linkage of large scale firm data by fuzzy criteria like company names and addresses. At its core, it utilizes an efficient candidate retrieval mechanism implementing a word respectively token driven heuristic. Every record in one table becomes a search term to retrieve similar candidate records in the base table according to a search strategy replacing blocking strategies of conventional matching efforts. Because similarity is inherently established by the candidate selection, it is only required to filter false positives by using the meta data export file derived from the matching heuristic to implement a machine learning approach. This paper discusses the general foundation of the heuristic and the algorithm while two detailed walkthroughs of company linkages show practical examples.
Schlagwörter: 
data linkage
firm matching
entity resolution
machine learning
JEL: 
C81
C88
Dokumentart: 
Working Paper

Datei(en):
Datei
Größe
1.52 MB





Publikationen in EconStor sind urheberrechtlich geschützt.