Stream Prediction Using A Generative Model Based On Frequent Episodes In Event Sequences

For small join tables, REED always chooses to push them into the network if their selectivity is smaller than one. For intermediate tables, the REED query optimizer makes a decision as to whether to push the join into the network based on the estimated selectivity of the predicate (which may be learned from past performance or gathered statistics, or estimated using basic query optimization tech- niques [28]) and the average depth of sensor nodes in the network. It uses a novel algorithm to store several copies of the join table at different groups of neighboring nodes in the...
Nội dung trích xuất từ tài liệu:
Stream Prediction Using A Generative Model Based On Frequent Episodes In Event Sequences