Integrating Historical Census Data in the Semantic Web

TY - CONF

T1 - Integrating Historical Census Data in the Semantic Web

AU - Meroño-Peñuela, Albert

AU - Ashkpour, Ashkan

PY - 2014

Y1 - 2014

N2 - Historical censuses are one the most consulted, reliable and large scale statistical data sources available, describing the demographic, social and economic history of a nation. To answer their research ques- tions, researchers often need to query long time series of census data. However, such longitudinal queries are typically hampered by the scarce integration of the historical censuses, demanding manual and knowledge intensive harmonization and restructuring in order to obtain meaning- ful comparisons over time. The challenges are even harder if provenance microdata is lost. In this paper we describe the methdology followed in CEDAR5, a project of the Computational Humanities Programme6, to provide solutions to these data-issues in the Dutch historical censuses (1795-1971). Our proposal builds on top of Linked Data and the Resource Description Framework (RDF) technology, allowing us to transform the original census tables into a graph of Linked Census Data. With such a graph, every census data-item can be interlinked on the Web with other hubs of historical socioeconomic and demographic information. By fol- lowing the Linked Data principles, our aim is two-fold. On the one hand, we show how the integration of our own historical census data is im- proved by linking them to the network of Linked Historical Datasets on the Web. On the other hand, we envisage new historical classifications (like demographical structures, housing types, occupational classes and statuses, or religious denominations) coming out of our harmonization process, which are not published yet on the Web on a standard manner and could improve the interoperability of other datasets.

AB - Historical censuses are one the most consulted, reliable and large scale statistical data sources available, describing the demographic, social and economic history of a nation. To answer their research ques- tions, researchers often need to query long time series of census data. However, such longitudinal queries are typically hampered by the scarce integration of the historical censuses, demanding manual and knowledge intensive harmonization and restructuring in order to obtain meaning- ful comparisons over time. The challenges are even harder if provenance microdata is lost. In this paper we describe the methdology followed in CEDAR5, a project of the Computational Humanities Programme6, to provide solutions to these data-issues in the Dutch historical censuses (1795-1971). Our proposal builds on top of Linked Data and the Resource Description Framework (RDF) technology, allowing us to transform the original census tables into a graph of Linked Census Data. With such a graph, every census data-item can be interlinked on the Web with other hubs of historical socioeconomic and demographic information. By fol- lowing the Linked Data principles, our aim is two-fold. On the one hand, we show how the integration of our own historical census data is im- proved by linking them to the network of Linked Historical Datasets on the Web. On the other hand, we envisage new historical classifications (like demographical structures, housing types, occupational classes and statuses, or religious denominations) coming out of our harmonization process, which are not published yet on the Web on a standard manner and could improve the interoperability of other datasets.

KW - census data

KW - linked data

KW - dutch history

M3 - Abstract

ER -