Study on mining novelty temporal pattern

The algorithm to find the novelty temporal pattern from the temporal database is presented. The basic idea of the algorithm is firstly to extract the feature sequence from a time series, then to compare the feature points of the time series with ones of the normal pattern to decide whether there is a novelty pattern in the time series. The temporal relation of time series is reserved in the feature sequence, therefore the novelty feature points are easier to be found. The algorithm has been used in the data mining of temporal database, and it works well.