Abstract
In big data settings, the data can often be externally sourced with little or no knowledge of its quality. In such settings, users need to be empowered with the capacity to understand the quality of data sets and implications for use, in order to mitigate the risk of making investments in datasets that will not deliver. In this paper we present an approach for detecting the completeness of high volume stream data generated by a large number of data providers. By exploiting the inherent hierarchies within database attributes, we are able to devise an efficient solution for computing query specific completeness, thereby improving user under-standing of implications of using query results based on incomplete data.