SPEDS: A framework for mining sequential patterns in evolving data streams

Data streams have attracted considerable attention in recent years. A growing number of applications generates streams of data. The continuous generation of new elements in a data stream imposes additional constraints on the methods used for mining such data. For example, memory usage is restricted, the infinitely flowing original dataset cannot be scanned multiple times, and current results should be available on demand. In many cases, evolution of sequential patterns is more interesting than sequential patterns themselves. Data evolution is one of the most challenging problems in mining sequential patterns in data streams. Hence, in this paper a new framework for mining sequential patterns in evolving data streams is introduced. The proposed framework guarantees no false negatives and imposes an upper bound of the support of false positives. Analytical analysis and simulation results are carried out to prove the correctness and scalability of the proposed framework.