Using Semantic Web Resources for Data Quality Management

摘要

The quality of data is a critical factor for all kinds of decision-making and transaction processing. While there has been a lot of research on data quality in the past two decades, the topic has not yet received sufficient attention from the Semantic Web community. In this paper, we discuss (1) the data quality issues related to the growing amount of data available on the Semantic Web, (2) how data quality problems can be handled within the Semantic Web technology framework, namely using SPARQL on RDF representations, and (3) how Semantic Web reference data, e.g. from DBPedia, can be used to spot incorrect literal values and functional dependency violations. We show how this approach can be used for data quality management of public Semantic Web data and data stored in relational databases in closed settings alike. As part of our work, we developed generic SPARQL queries to identify (1) missing datatype properties or literal values, (2) illegal values, and (3) functional dependency violations. We argue that using Semantic Web datasets reduces the effort for data quality management substantially. As a use-case, we employ Geonames, a publicly available Semantic Web resource for geographical data, as a trusted reference for managing the quality of other data sources.