基于元信息的粗糙集规则增式生成方法

A META-INFORMATION-BASED METHOD FOR INCREMENTAL ROUGH SET RULE GENERATION

粗糙集方法已成为数据挖掘的重要手段,由于数据挖掘需要处理大量的数据,并且数据动态递增,因而粗糙集方法需要具备处理大数据库数据的能力,以及有效处理动态递增数据的能力。然而,当前大多数粗糙集方法缺乏这些能力。对此,作者提出基于元信息的粗糙集规则增式生成方法。元信息是数据挖掘中间结果的描述,该方法首先渐增生成当前数据场的元信息,再从元信息中推导出规则,由于元信息是可重用的,因而该方法仅处理还未处理的数据,从而减少了数据挖掘的时间,同时,元信息的可重用性为数据挖掘系统的故障恢复提供了一种手段,提高了系统的鲁棒性。

Rough set is one important method in data mining . Because of the great quantity of data which increase dynamically, rough set method should be capable of handling data in large database and processing dynamic data effectively. Most of the existed rough set methods have no such abilities. So, in this paper we present a Meta-in-formation-based method for incremental rough set rule generation. Meta-information is the description of interim result of data mining . The method incrementally generates the meta - information , and generates the rough set rule according to the meta-information. Time of data mining is decreased because meta-information is reusable and the method only processes data that have not been processed. Also, the reusability of meta-information provides a failure-recovery means in data mining system, and enhances the robustness of the system.