Texterkennung als Herausforderung bei der Digitalisierung von Tabellen

Am Beispiel des Projektes für belgische historische Zählungen (KU Leuven Libraries Economics and Business)

Abstract

Censuses have been taking place for more than 5000 years. They were originally carried out for the purposes of tax collection and the military, but later also used for scientific research. The first censuses which, from the start, were also available for research were carried out in Belgium in 1846 under the direction of Adolphe Quetelet. After Quetelet’s famous first censuses for research, many subsequent censuses were taken. Since the analysis of these censuses is very time-consuming due to their extent and format, it makes sense to convert them into digital form by means of electronic text recognition. The KU Leuven Libraries Economics and Business (Belgium) is currently working on a project aimed at offering the printed editions of the Belgian industrial censuses between 1846 and 1947 as Excel spreadsheets. This article addresses the challenges involved and describes the procedures.