Estimating OLAP queries over uncertain and imprecise multidimensional data streams

In this paper, we propose a novel framework for estimating OLAP queries over uncertain and imprecise multidimensional data streams. The proposed framework introduces the following three relevant research contributions: (i) a probabilistic data stream model, which describes both precise and imprecise multidimensional data stream readings in terms of nice interval-confidence-based Probability Distribution Functions (PDF); (ii) a possible-worlds semantics for uncertain and imprecise multidimensional data streams, which is based on an innovative data-driven approach that exploits "natural" features of OLAP data, such as the presence of clusters and high correlations; (iii) an innovative approach for providing theoretically-founded estimates to OLAP queries over uncertain and imprecise multidimensional data streams that exploits the well-recognized probabilistic estimators theory. We complete our theoretical contributions throughout several case studies demonstrating the suitability of our proposed framework in the context of modern data stream applications and systems, which are more and more characterized by the presence of uncertainty and imprecision.