Reduction Algorithm Based On Predicate Logic And Contains The Set Of Classification Rules

The classification is an important task in data mining. There are a lot of redundant rules in classification rule set which are extracted by classification rule mining methods, so that the classification efficiency and understandability of the classification rule set are effected seriously. So, the classification rule set reducing has very important theory meaning and application value. In this paper , the post-processing methods of classification rule set are studied by using the predicate logic and the including set. The main research works are as follows:First, a reducing algorithm (RMCRPL) of classification rule set is presented based on predicate logic. Firstly, the classification rule set is described by using predicate logic, so that the rule set is changed into a predicate formula set. Secondly, the classification rule set is reduced by using logic reasoning in predicate formula, so that the redundant rules are eliminated. In the end,the experiment results validate that the algorithm is effective and feasible by taking the celestial spectrum data.Second, reducing algorithms (MMLS and RRPABIRS) of the classification rule set are presented based on the including set. Firstly,making use of classification relation between rule and data, the including set of any classification rule is extracted. Secondly, MMIS and RRPABIRS reducing algorithms are presented by dealing with the including set ong by one. In the end,the experiment results validate that the algorithms is effective and feasible by taking the celestial spectrum data.