Research And Application Of Complex Event Analysis Method Based On DBSCAN

The rapid development of information technology has caused the world's data volume to explode.Network streaming media,GPS-based car networking systems and other software applications continue to generate large amounts of data,and the formation of real-time data stream.The data generated by it contain huge economic value and social value.Due to the characteristics of big data itself,such as the speed of data generation,the disorder of data,and the limited processing,it can not rely on human inspection to excavate people's feelings Interest and can make use of the information.Therefore,the effective use of these data to mine the potential of complex data in the relationship between big events,companies,companies and even countries are profane without harm.In response to the above problems,this paper presents a method of big data analysis of complex events.Starting with the complex event model,three types of complex event relational models are defined,the big data stream processing from different data sources is transformed into the extraction and analysis of complex event relations,and the traditional DBSCAN clustering analysis algorithm is improved to suit different structures Type of data sources,and FP-Growth associated with the analysis of the associated algorithms,a new complex event big data analysis algorithm,mining data in the relationship between complex events.At the same time,the method is deployed on the Hadoop cloud platform.Combined with the MapReduce computing model,parallel DBSCAN algorithm and FP-Growth algorithm are proposed to deal with large data sets.Discrete GPS datasets and Autonomous System(AS)datasets are selected for experimental analysis to verify the accuracy of this method in different datasets.The final design and implementation of a complex event big data processing tools to further verify the feasibility of the above work.