| With the popularity of internet, more and more people pay attention to the management of data stream. It is more important to calculate aggregate functions in data stream. In database management system, an aggregate function is defined a function that returns a single result through calculating a group of values. In this paper, we still made use of the definition. Resolving calculation of aggregate functions in data stream has meaning realism for monitoring, counting, detecting data stream and network.The main contribution of this paper is as follow:(1)While the data type of input data is a numerical type, this paper proposed a new exact aggregation method as little as possible stored historical data for the max function and the min function. This method is an incremently calculational method based on sliding windwos of data stream. It took O(N) time to deal with an input data list of length N. It took O(N) space under the worst circumstances or O(M) (It preassigned M B memory.) space under the best circumstances to deal with an input data list of length N. And we analyzed and proofed the new method through mathematic theory. Moreover, we checked the new method by some experiments. At last, we realized an exact aggregation method — an incremently calculational method of aggregate functions, for example count, sum, avg, stdev, stdep, var, avrp.(2)While the data type of input data is a string type, this paper realized a statistical method based on general general suffix tree model for frequent string. This method didn't require any training corpus, and directly classifying and counting accepted texts throught frequent string. It took O(N) time and O(N) space to deal with the text or string of length N. And this paper realized a new spam detection method on backbone network through the exact aggregation method for input data that belong to the string type. This method represented the text of Email with general suffix tree structure. And when a new... |
PDF Full Text Request