Abstract
With the limited memory and time, a fast and effective clustering can’t be achieved for massive, high-speed data stream, so this paper mainly studies the key method of data stream clustering under the restriction of resource, and then proposes a dynamic data stream clustering algorithm (D-DStream) based on wavelet network and density, which uses sliding window to process data stream. Firstly, apply wavelet network to compress data stream and build a much smaller synopsis data structure to save major characteristics of data stream, then cluster with two-phase density clustering algorithm. The results of experiment show that the D-DStream algorithm can successfully solve clustering problems caused by STREAM or others, also has high time efficiency and high clustering quality.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have