| In recent years, in the field of sensor networks, Internet monitoring, financialtransactions, there has been demand for data stream analysis application. However, thetraditional data processing method can not solve this data stream analysis applicationsvery well, so data stream analysis techniques for these applications in the academic andindustry fields are subject to a high degree of attention. With increasing demand forcomplex data stream analysis, key technologies in complex data stream analysis areincreasing important. To solve complex data stream analysis, the existing solutions usuallydesign specific synopsis construction data analysis algorithms. But for more and morecomplex data stream analysis applications, this method, with which each correspondinganalysis algorithm is developed to solve the specific problem, is clearly not a good way tomeet the demand. Abetter way is to reuse some of the common analysis algorithms, whichare used by a combination with other algorithms to perform complex data stream analysis.However, data exchange between two operators is a key technology, which has to beresolved.Data stream analysis requires efficient performance of time and space, for which anefficient and universal data exchange solution is designed. Data exchange uses a starexchange mode based on pattern matching, in which intermediate data model, describes ofthe data model, pattern matching and describes of mode conversion are several key issues.In this paper, considering the different characteristics between complex data and relationaldata in data stream analysis, a metadata model of intermediate data is designed; while adescribe solution of the data model based on XML tags is designed; two key points ofpattern matching are given: structure match and type match; and in order to get the bestmatch, the problem of matching names is considered. Finally, taking advantage of thecharacteristics of the actual data which can be pre-allocated based on the data model, asimple and efficient mode conversion description is designed.This paper introduces the real-time data stream analysis platform, which is designedto achieve the study, and the experiment of data exchange, which is proved to be anefficient and universal solution to data exchange in combined analysis problems forcomplex data stream analysis. |
