| Summary: | 碩士 === 明新科技大學 === 資訊管理研究所 === 96 === In many applications such as network traffic analysis, Web click stream mining, and on-line transaction analysis, the data we need to process is not static, but a continuous dynamic data stream. In order to efficiently process continuous queries, algorithms are restricted to making only one pass over the data. Due to the difference in the essences, query processing in the data stream processing system is more complicated than that in the traditional relational database system. At present, researches about data stream query processing focus on online auctions and traffic monitoring. The study on transactional data stream query processing has not been proposed yet. In this thesis, we design the data stream query language based on the sliding window model. We focus on transactional data streams, and consider “window join” operation and “window aggregate” operation on multiple data streams and relations. We use the notion of “summary information” to achieve the goal of “one-pass” processing, and propose an efficient approach to process data stream queries. Besides, we find that the optimization techniques used in the traditional relational database system can not be directly applied on the data stream processing system. This thesis presents three query execution plans and investigates the techniques of query optimization, in order to efficiently process the query of transactional data streams.
|
|---|
