Enhanced metrics for temporal dimensions toward assessing Linked Data: A case study of Wikidata

Abstract

The development of Linked Open Data (LOD) has transformed the web into a decentralized extensive database. With this dataflow, ensuring quality level became very quickly a crucial issue. Many approaches were proposed using metrics and dimensions inherited from traditional relational databases to assess LOD quality. However, classic metrics used to measure dimensions like Currency, Volatility and Timeliness are unreliable knowing they use exclusively temporal metadata. We believe that assessing data “freshness” could emphasize their validity, considering the assumption that current data is tightly correlated with its credibility. In this paper, we propose an upgraded version for time-related metrics to make them more efficient and more suitable for assessing linked data. Using Wikidata as study case and Wikipedia as reference to “current” data, we implement our approach and compare it to the most relevant approach to find out the crucial part played by a reliable Currency and Volatility to determine data up-to-dateness.

Keywords

• Data quality
• Wikidata
• Currency
• Volatility
• Timeliness
• Linked data