Research Of Attribute Reduction Algorithm Based On Matrix In Composite Information Systems

Rough set theory is a mathematical tool for data analysis,it can be used to describe the dependencies between attributes,evaluate the importance of attributes and export decision rules in the attribute value expression model.Pawlak Rough set is very suitable for the processing of discrete data through the equivalence relation granular domain.The classical rough set model can only deal with discrete data attributes,but in real life applications,information system usually contains a variety of different types of attributes,and then many extended rough set models have been developed.Attribute reduction is one of the important research contents in rough set theory.This thesis is based on the existing attribute reduction algorithm based on positive domain,in order to improve the efficiency of some exist algorithms,the reduction algorithm of high reduction efficiency is designed by introducing the structure difference matrix,and apply it into the composite information system.In view of the dynamic change of data set in real life,the dynamic attribute reduction algorithm based on the structure difference matrix is studied in this thesis.The main work of this thesis is as follows:First of all,this thesis aims to improve the low efficiency of some attribute reduction algorithm based on the positive domain,the attribute reduction algorithm based on the structure difference matrix is proposed.This algorithm speeds up the attribute reduction process by improving the process of solving the kernel attribute set and narrowing the domain.For real-world data sets that contain multiple types of data,the classification method of various attributes is analyzed in detail,the attribute reduction algorithm based on the structure difference matrix is applied to the composite information system.In this thesis,the attribute reduction process of the algorithm is explained in detail with an example,and the validity of the algorithm is verified by experiments on 6 data sets of UCI database.Then,in terms of the ever-changing data set in the actual application,In this thesis,the change of structural differential matrix is studied fully when the data set is dynamic,and designed a dynamic attribute reduction algorithm based on structural differential matrix.The validity of the algorithm is analyzed theoretically combined with examples,and the validity of the algorithm is verified by the data set Chess in UCI database.