Today’s Web pages more frequently contain structured information by means of semantically-rich embedded data. These data are currently used by search engines to provide an improved semantic, structured search result. However, very often these data are conflicting when present on different Web pages. Data consumers do not receive any support in handling (recognizing, interpreting, correcting) structured data. The more frequent structured data will be available on the Web, the more critical is the problem in recognizing and handling conflicting structured data. This paper presents FactCheck as an approach for the detection and resolution of conflicting structured data on the Web. We applied our solution to websites and APIs using JSON-LD and Microdata. First experiments with a prototypical implementation of FactCheck show that the resolution of conflicting data has high potential to significantly improve the quality of information on the Web. First evaluation reveals positive effects through the additional information indicators provided by FactCheck and indicates an impact on users of FactCheck at various levels. The resulting improvement of the data quality on websites will be of benefit to all data consumers who are depending on data on the Web as well as cognitive computing services and its building blocks.

Top