Potential Problem Data Tagging: Augmenting information systems with the capability to deal with inaccuracies

摘要

Data quality tags are a means of informing decision makers about the quality of the data they use from information systems. Unfortunately, data quality tags have not been successfully adopted despite their potential to assist decision makers. One reason for the non-adoption is that maintaining the tags is expensive and time-consuming: having a tag that represents accuracy, for example, would be massively time-consuming to measure because it requires some physical observation of reality to check the true value. We argue that a useful surrogate tag for accuracy can be created—without having to physically measure it—by counting the number of times the data has been exposed to an event that could cause it to become inaccurate. Experimental results show that the tags can help to avoid problems caused by inaccuracies, and also to help find the inaccuracies themselves.