Abstrakt
As part of LOD2.eu project and OpenData.cz initiative, we are developing an ODCleanStore framework (1) enabling management of governmental linked data and (2) providing web applications with a possibility to consume cleaned and integrated governmental linked data; the provided data is accompanied with data provenance and a quality score based on a set of policies designed by the governmental domain experts. Nevertheless, these (objective) policies fail to express subjective quality of the data as perceived by various data consumers and different situations at their hand.
In this paper, we describe how consumers can define their own situation-specific policies based on the idea of filtering certain data sources due to certain aspects in the data provenance records associated with these sources. In particular, we describe how these policies can be (1) constructed by data consumers and (2) applied as part of the data consumption process in ODCleanStore.
We are persuaded that provenance policies are an important mechanism to address the subjective dimension of data quality.