A Research Of Multi-granularity Knowledge Acquisition Algorithm Based On VPRS

With the rapid development of computer communication technology and the Internet,the amount of information in various fields has increased rapidly over time,and human beings have entered the era of large data.Although people are surrounded by massive data,they have no ability to analyze and processing,facing with the situation of lack of knowledge.Therefore,knowledge acquisition has become a hotspot in recent years.As a means of knowledge acquisition,rough set theory has unique advantages in dealing with uncertain information.However,when the noise or information missing exists in the data,the rough set theory based on the exact classification has a poor performance,and the variable precision rough set provides a better solution for this kind of problem.As a model for dealing with large-scale,complex problems,granular computing processes information through the form of information granule.The key idea is to transform the problem into a number of sub problems,and then solve the problem in each granularity space.Finally,the solutions of multiple sub problems can be combined to form the solution of the original problem.Granular computing from multi-granularity analysis to solve the problem not only reduces the difficulty of solving the original problem,but also to meet the human desire to analyze the requirements from multiple perspectives.Some researchers have combined the multi-granularity idea and the rough set theory to propose a multi-granularity rough set,which provides a powerful tool for solving the knowledge acquisition from massive data.Based on the research of variable precision rough set and multi-granularity rough set theory,this paper focuses on the attribute reduction and rule extraction of information system,mainly from the following aspects:1.The granulation idea of granular computing is used to the granulation of condition equivalence class and decision equivalence class.We use granular matrix to represent condition information granule and decition information granule,and present a ? granular relation matrix based on VPRS.In essence,this matrix reflects the probability inclusion relation between condition information granule and decision information granule,which is the theoretical basis of attribute reduction and rule acquisition.2.In the study of attribute reduction,we first propose a multi-granularity attribute reduction algorithm based on variable precision normal domain.Then we analyze the shortcomings of this algorithm,and propose an improved multi-granularity attribute reduction algorithm based on variable precision lower approximation.3.For the rule extraction of decision information system,we first use the matrix operation to solve in different granularity space,and through the definition of heuristic information to sort the granularity space,in order to reduce the search space.Then,a multi-granularity rule acquisition algorithm for complete decision table based on variable precision rough set is proposed.The algorithm is applicable to consistent and inconsistent decision table,which improves the generalization of the algorithm.Finally,we propose a multi-granularity rule acquisition algorithm for incomplete decision table based on variable precision rough set,and then we analyze the two algorithms and test them on UCI dataset.The results show that the algorithm is correct and effective.4.A multi-granularity knowledge acquisition system platform based on variable precision rough set is designed in this paper.And the platform integrates the proposed algorithm is convenient for the user to select the appropriate algorithm for knowledge acquisition.