The Processing Of Data Stream Uncertain Information Based On Granularity

With the rapid development and increasing popularity of Networks、Database aswell as the Multi-Media technologies，the information resources are growing in anexplosive way. And there is a new kind of data named data stream in many fields, suchas traffic monitoring, telecoms records, web page visits, sensor readings, and so on.However, there are very different between the data stream and traditional static data.To the data stream, their continuous arrival in multiple, rapid, time-varying, possiblyunpredictable and unbounded streams appears to yield some fundamentally newresearch problems. And the research on data stream can be divided into two differentkinds of categories, that is data stream management system and the knowledgediscovery of data stream. This paper mainly focuses on the knowledge discovery ofdata stream.The uncertainty is an inherent character of object. There are all kinds of uncertaininformation existing in practical applications. The task of uncertain knowledge miningis to explore the valuable information which hidden in uncertain data and search somevaluable pattern. The uncertain object exists in the data collection, data preprocessingand even the process of data mining within the data stream.The Granular Computing is a new kind technology of intelligent informationprocessing. It contains all of the theories, methods and technology which relate to thegranularity. This theory mainly focus onthe uncertain information processing and incomplete information processing. Thenature of the granular computing is to simulate the process of human brain obtains anddeals with problems. It aims at selecting the appropriate granularity to lessen thecomplexity of original problems to find a better and approximate solution.This paper mainly takes a deep research on the uncertain information process of datastream with the help of granularity theory based on the existing research results. Itconstructs the self-drive dynamic granularity model to lessen the complexity of datastream environment based on the divide and conquer theory,and with the aid of cloudmodel and fuzzy theory to deal with the uncertain information problems within datastream mining process which leads by the introduction of granularity concept. And itsultimate purpose is to analyze the uncertain information qualitatively and quantitatively to find the valuable knowledge hidden in the data accurately.