The Approximations Update For Dynamic Interval-Valued Information System

With the development of the information technology industry and continued to achieve novel success.The size of data and the period of data update are undergoing unprecedented changes.How to deal with the massive,ruleless,unstable and rapid updating data and take a effective way to data analysis and discover the knowledge which hidden in the data are the theme of the information explosive era.They are the new opportunities and challenges faced by people in the field of intelligent information processing.The theory of rough sets was proposed by Pawlak in 1982 and has been generalized applied as an effective mathematical tool to deal with the problem of expressing and dealing with uncertainty.The foundation of the theory is the classification mechanism which established by the equivalence relation.Through the analysis and reasoning of the data to discover the implicit knowledgeand reveal potential mathematical laws.More and more data processing models are needed to adapt to the current research of data science.This study is based on rough sets theory under this background.Several new double quantitative rough set models are investigated in ordered information systems and the approaches of incremental updating approximations when the object set of an interval-valued information system is dynamic.The main innovations in this thesis are listed in the following.1.The graded rough set model and variable precision rough set model are combined in ordered information system based on their approximate operators.Three new types of double quantitative rough set models are constructed and their basic mathematical properties are studied systematically.The method of accurate characterization the rough set regions and the relationship among of these models are obtained.Finally,a series of case study are conducted to show the process of caculation and verify the rationality and validity of the propsed models.2.The approaches of incremental updating approximations are studied when the the object set of an interval-valued ordered information system is dynamic.Two incremental algorithms are designed for inserting or deleting object based on the proposed method,respectively.Acording to the designed algorithms and traditional computational way a chain of experiments were carried out on several UCI datasets.The results indicate that the proposed approaches with higher computational efficiency.3.Based on the logical ‘disjunction' and ‘conjunction' operations of interval numbers,a new similarity measure is defined and a similarity binary relation is constructed based on the measure.Then,a rough set model is established according to the cover of universe that induced by the new binary relation.Furthermore,we researched the generalized method of incremental updating approximation for dynamic object set interval-valued decision information system.Finally,the theoretical value and practical application of the proposed method are expounded through the study of the real cases.