Hierarchical Construction And Attribute Reduction Of Uncertainty Measures In Multi-source Fuzzy Neighborhood System

The rapid development of information technology has caused the explosive growth of uncertain data,which is increasingly challenging people's data collection and analysis capabilities.Faced with the uncertainty of information systems brought about by big data,the effective construction of uncertainty measures has become an important topic in the study of data mining and knowledge acquisition.Fuzzy set theory,rough set theory and information entropy theory,as powerful mathematical tools to deal with uncertain information,have proposed many practical and effective reduction algorithms.Based on a branch of fuzzy sets—intuitionistic fuzzy sets,this thesis studies the hierarchical construction and attribute reduction of equivalence classes,as well as the model construction of similarity measures in intuitionistic fuzzy decision-making and intuitionistic fuzzy multi-granularity decision-making.main tasks as follows:(1)The(?,?)-level Cut-Set R_?~? under the binary intuitionistic fuzzy compatibility relation in the intuitionistic fuzzy information system and the(?,?)-level Cut-Set X_?~?under the binary intuitionistic decision goal concept It has a good division property.This paper uses granular computing and Bayes'theorem to construct the three-level granularity and three-way information measurement of the decision table of the intuitionistic fuzzy information system.For the decision table,according to the different form structure and system granularity,it is divided into the Mcro-Bottom,the Meso-Middle and the Macro-Top structure,and then different information measures are obtained through Bayes'theorem.This paper analyzes the reasons for the monotonicity of the hierarchical granular structure in the traditional sense,and improves the conditional probability of the micro-bottom layer,and then realizes the evolution from the Mcro-Bottom to the Meso-Middle layer to the Macro-Top step by step in an integrated way,the improved three-layer granular structure is monotonic in granulation and systematic in evolution.Finally,the structure and result information are effectively explained through algorithms and examples,and the attribute reduction of the information is further realized.(2)In view of the fact that the existing intuitionistic fuzzy similarity measures have some"counter-intuitive"situations under certain semantics,this paper constructs a new similarity measure under the conditions of comprehensive consideration of membership,non-membership and hesitation.Proof of feasibility and analysis of its advantages and disadvantages have been carried out.Using the similarity measure,the intuitionistic fuzzy similarity class is defined in the intuitionistic fuzzy information system,and the intuitionistic fuzzy decision model is constructed.Then,a multi-granularity intuitionistic fuzzy decision model is constructed in the multi-source intuitionistic fuzzy information system,its related properties are discussed,and reasonable results are given in the form of examples.