Abstract: Similarity queries are fundamental part of modern data mining application. But traditional ctuery algorithms can not be applied on data stream, which is an unbounded sectuence of data elements generated at a rapid rate. We proposed a novel approach for computing similarity over multi data streams based on wavclet sliding window model. The basic idea is to divide sliding window into equally-sized basic windows and represent the data elements of a basic window using wavelet coefficients, then form wavelet synopses window. As a result, queries toward data streams can be converted to queries toward such wavelet synopses. This algorithm takes advantage of the merit of wavelet decomposition for linear computing and achieves superior runtime performance. The extensive experiments verified the effectiveness of our algorithm.

Key words: Data stream, Similarity query, Sliding window, Wavelet decomposition