Toward Probabilistic Data Stream Management Systems

摘要

The inherent imprecision of data in many applications motivates us not to ignore uncertainty of values and to support it as first-class concept. Data stream and probabilistic data have been recently considered noticeably in isolation. Nevertheless, there are many applications including sensor data management and object monitoring systems which need both of issues in tandem. Therefore, we intend to show how we could extend a traditional Data Stream Management System (DSMS) into a Probabilistic Data Stream Management System (PDSMS) which can understand and deal with uncertainty from input data admission to final query result generation. In this paper, after considering requirements of PDSMSs, we focus on essential aspects of Probabilistic Data Models and finally consider benchmarking PDSMSs as our future trend.