基于概率模型的概率频繁项集挖掘方法

A Probability Model-Based Approach of Mining Probabilistic Frequent Itemsets

在很多应用中,数据本身就包括不确定性,如基于位置的服务,传感器监控系统和数据融合等等。为了解决大量的不精确信息,提出了不确定数据库的概念。在不确定数据库中挖掘概率频繁项集是一个重要的研究内容,但是,由于不确定数据库是用可能世界来解释的,会产生指数级的可能世界,使得挖掘方法效率不高。本文采用泊松二项分布来对不确定数据库的可能世界进行建模,在挖掘过程中,不计算具体的频繁概率,通过计算与最小频繁概率对应的期望支持度,用频繁概率的近似值实现概率频繁项集的挖掘,提高了概率频繁项集挖掘算法的效率,仿真实验也证明该方法具有较好的效果。

Data uncertainty is inherent in emerging applications such as location -based services, sensor mo- nitoring systems, and data integration. To handle a large amount of imprecise information, uncertain databases have been recently developed. Mining probabilistic frequent itemsets is an important research problem in un- certain database management. Since an uncertain database contains an exponential number of possible worlds, the efficiency of mining algorithms is not effective. In this paper, the mining process is modeled as a Poisson binomial distribution. In the process of mining, the exact frequent probability is not calculated. By computing expected support through minsup, the approximation of frequent probability is used to mine probabilistic fre- quent itemsets. An extensive evaluation on synthetic databases is performed to validate this method.