Window Semantics of Data Stream and Implementation of Querying Engine

Windows of data stream had three driven models of time-based, tuple-based and batch-driven, each of which did not express the correct semantic of querying data stream. In order to solve the incompleteness of query semantics in data stream, the authors defined the full window semantic framework from three aspect of query-level, data-level and system-level in order to eliminate the semantic differences, and batch-driven model was implemented based on memory database. Finally the method was verified by experiments.