The lack of quality of stored data is reflected by violations of integrity constraints. Answers to queries in databases containing bad quality information usually cannot be trusted. Nevertheless, many answers given by such databases may still be useful, as long as they are derived from data the quality of which is sufficiently high. We formalize our intuition of answers that have quality on the basis of 'causes'. A cause of an answer is a minimal excerpt of the database that explains why the answer has been given. Thus, an answer has quality if the overlap of its causes with the causes of integrity violation is empty. Even if that overlap is not empty, but is sufficiently low, an answer may have sufficient quality. The amount of causes in the overlaps of causes of answers and integrity violations can be sized by quality metrics. © 2013 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Decker, H. (2013). Answers that have quality. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7972 LNCS, pp. 543–558). https://doi.org/10.1007/978-3-642-39643-4_39
Mendeley helps you to discover research relevant for your work.