An Incremental Computation Scheme over Array Database

概要

論文の詳細を見る
Array database is a new kind of database, dealing with the big data storing and processing in science fields. The window aggregation is a typical aggregate query of it. However, in the current array databases, naive method is used to calculate the window aggregates, which will do much redundant computation and lead to low efficiency. In this thesis, I propose an algorithm improved with incremental computation for window aggregate queries. It buffers some intermediate aggregate results and reuses them when computing, eliminates the redundant computation. The time complexity analysis is also offered, which shows the advantages of the improved algorithm clearly. In order to test the performance, a simple array data process system is implemented, with both naive method and incremental computation algorithm implemented inside the system. The test result also shows that the improved algorithm is much better than the naive one, especially when the aggregated window is big.
2013-11-19