FLAT is a novelunsupervised graphical model, called FSPN . It uses independent and conditional factorization to adaptively model different levels of attributes . FLAT achieves 1 to 5 orders of magnitude better accuracy, 1 to 3 orders ofmagnitude faster . faster probability computation speed (around 0.2ms) and 1 to 2 ordersof magnitude lower storage cost (only tens of KB). FLAT supports efficient online probability computation in nearliner time on the underlying FSPn model, and provides effective offline modelconstruction. It can estimate cardinality for both single table queries andmulti-table join queries. FLAT can estimate Cardinality for . both single tables queries and multi-table joins queries. It is also possible to estimate cardinalities for both table queries. For more information about FLAT, please visit http://www.flat.com/flat-report/flAT.org/report/glachenary-matmatrocessure-matryryry.

Author(s) : Rong Zhu, Ziniu Wu, Yuxing Han, Kai Zeng, Andreas Pfadler, Zhengping Qian, Jingren Zhou, Bin Cui

Links : PDF - Abstract

Code :

Keywords : flat - queries - table - estimate - cardinality -

Leave a Reply

Your email address will not be published. Required fields are marked *