earticle

영어

Actually, large database is not simply considered as a stream database because of streaming data is not only containing huge data volumes, but distributed, continuous, rapid, time varying. Therefore, the general techniques may not suit for streams exactly. Accuracy responses required of approximated answers is more important in stream processing for the similarity search. Therefore, we perform data reduction across synopsis data structure and to batch processing in a particular relevance way on the data stream computation model over sliding windows. Focus on similarity search in streaming environment, D-skyline method proposed in this paper concern useful aggregate as a preprocessing phase instead of original dataset repeatedly processing manner, in order to efficiently optimize both in space usage and error control. Our experimental evaluation would show the detailed effect on approximated analysis by using different kinds of skyline methods, then effectiveness and efficiency of our approach.