Research On Multi-attribute Decision Making Of Interval-valued Information System Based On Rough Set

Multi-attribute decision making is one of the most important parts of modern decisionscience. The essence of it is to classify and rank a group of objects through some way using theexisting information of decision making, then to choose the best. As modern societies becomemore complex and fuzzy and for the uncertainty of the thinking mode of human beings, thedecision information is often given with interval-valued numbers.Compared with the other methods, the rough set has two major advantages. First, the methodavoids the problem of determining weights which is important but hard and doesn’t have a unifiedmethod. This paper uses the theory of rough set to reduce the number of attributes avoiding weightdetermination. Secondly, rough set is a very systematic and comprehensive theory which cancover each process of decision making. The attribute reduction and rule extraction even the finalranking are all based on the rough set theory.The content of this paper is mainly about the attribute reduction and the decision ruleextraction of the multi-attribute decision making of interval-valued information system. First, thepaper reviews the existing methods about the multi-attribute decision making problem ofinterval-valued information system, and summaries the merits of the relevant theoretical resultsand the problems which still need to be improved. Then, the paper divides the interval-valuedinformation system into two different kinds, one with decision attributes and the other without that,and does further study of the two new systems based on the rough set theory. At last, builds theattribute reduction models and the rule extraction models about the two systems and verifies themodels by some examples.The main innovation of this paper is as follows:(1) Analyze the shortcomings of the existing indiscernibility relation, the dominance relationand the dominance relation, propose a dominance relation specially for themulti-attribute decision making of interval-valued information system and then verify it.(2) Build the attribute reduction model and the ranking model for the interval-valued systemwithout decision attribute based on the dominance relation and the rough settheory, and verify the models by some examples.(3) Build the attribute reduction model and the rule extraction model for the interval-valuedsystem with decision attribute based on the dominance relation and the roughset theory, and verify the models by some examples.