Knowledge Acquisition Of Interval-valued Information Systems Based On Dual-parameters

The value of an attribute about some object in information systems may be the range, not often a single numerical value, we name this information system as interval-valued information system. As an important model for data description, interval-valued information system has been widely applied in construction, scientific research, stock analysis, fault detection and intelligent decision-making. Therefore, the knowledge acquisitions for interval-valued information systems have received common concern from experts and scholars.This thesis includes the following aspects:(1) In this thesis, we define some basic operations of interval numbers, and then we construct a similarity of two intervals by the operation of interval numbers. We propose dual-parameter(α,β) tolerance relation for interval-valued information systems through similarity a and similar rate parametersβwhich can better controlled the similarity of the objects, then define the upper and lower approximations under the dual-parameter tolerance relation, and discuss the relative theories and properties for interval-valued information system based on the new binary relations.(2) Based on dual-parameter (α,β) tolerance relation, we introduce a knowledge granularity for interval-valued information systems, and obtain a new method and thought to characterize the uncertainty for interval-valued information systems. Besides, we define the importance of attribute in interval-valued information systems, and analyze the properties of knowledge granularity and the importance of attribute in the interval-valued information systems.(3) Attribute reduction and rule acquisition are two important aspects of the rough set theory. In this thesis, based on the attribute importance of interval-valued information systems, we present an attribute reduction method for interval-valued information systems. We obtain decision rules and propose a rule acquisition algorithm for interval-valued decision information systems on the basis of the condition attributes reduction.Based on rough set theory, this thesis makes the preliminary exploration and research on knowledge acquisition for interval-valued information system, and illustrates the feasibility of attribute reduction and rule extraction by examples.