Synopses for probabilistic data over large domains

摘要: Many real world applications produce data with uncertainties drawn from measurements over a continuous domain space. Recent research in the area of probabilistic databases has mainly focused on managing and querying discrete which is limited to small number values (i.e. order 10). When size increases, current methods fail due their nature explicitly storing each value/probability pair. Such are not capable extending use continuous-valued attributes. In this paper, we provide scalable, accurate, space efficient synopsis for uncertain attributes defined domain. Our construction all error-aware ensure that our provides an accurate representation underlying given budget. Additionally, able approximate query results error bounds.We extensive experimental evaluation show proposed improve upon state art terms time accuracy. particular, can be constructed O(N2) (where N tuples database). We also demonstrate ability answer variety interesting queries set reduced by up magnitude previous state-of-the-art method.