A novel distance metric function, called SDD (slope duration distance) is proposed, for two local patterns from data streams, which is accord with the intuition of human being to shape similarity, can automatically adapt to the change of local pattern length, and provides a facility of incremental computing. Moreover, we presente an efficient model MLCDS (monitoring local correlations among data streams) to capture the correlation of data streams by their local patterns, based on PLA (piecewise linear approximation) and novel grid structure proposed. MLCDS-dramatically decreases the computatio...