Pattern-Based Window: A Novel Window Operator to Support Event Detection over Data Stream

This paper proposes a novel window operator, pattern-based window, for data stream processing. By using pattern-based windows, user-specified complex events can be extracted from data streams under the framework of Continuous Query Language (CQL), which integrates the current data stream processing and event stream processing. In this paper, the pattern-based window is formally defined and the query language of pattern-based windows is designed. A simple pattern matching algorithm is introduced to show logical expression of the query language. We also give simple examples to show the usage of the query language by using pattern-based windows.