一种数据流中的频繁模式挖掘算法

Algorithm of frequent-patterns mining in data stream

时序数据流的无限性、流动性和不规则性使得传统的频繁模式挖掘算法难以适用。针对时序数据流的特点,提出了一类特殊非规则数据流频繁模式挖掘的新算法。新算法采用时序数据分段的思想,逐段挖掘局部频繁模式,然后依据局部频繁模式有效地挖掘出所有的全局频繁模式。将新算法应用于电信领域的收入保障项目之中,结果表明,新算法具有良好的性能,能有效发现挖掘时序数据流中的频繁模式。

The limitlessness, mobility, and irregularity of time series data stream make the traditional frequent-pattern mining algorithms difficult to extend to the mining problem of time series data stream. According to the characteristics of time series data stream, a new algorithm for mining the frequent-pattern from a kind of special irregular data stream was proposed, in which, time series data stream was partitioned firstly, and then the local frequent items were mined step by step. Finally, the global frequent items could be mined efficiently based on these local frequent items. After applying the new algorithm in the revenue assurance project of telecommunication field, the results show that the new algorithm has good performance, and can mine frequent-patterns effectively from the irregular data stream of telecommunication field.