Research On Dynamic Optimal Scale Selection In Multi-scale Set-valued Decision Tables

Optimal scale selection is crucial for knowledge discovry in multi-scale decision tables(MDTs).Set-valued decision tables are the generalized versions of single-valued decision information systems and can also be the multi-scale property There are few studies on the optimal scale selection in multi-scale set-valued decision tables(MSDTs),which methods need to consider consistent and inconsistent situations respectively.In addition,in the current era of big data,data is constantly changing,Existing researches do not consider the optimal scale selection in a dynamic multi-scale set-valued decision table.To address above two issues,we conduct research as follows:(1)Aiming at the problem of optimal scale selection in MSDTs,we proposes an optimal scale selection method based on the undetermined degree for MSDTs.This method does not need to consider consistent and inconsistent situations respectively.Firstly,we propose an MSDT model under dominance relations and investigate its characteristics.Secondly,a sequential three-way decision model is established in MSDT.Through reasoning and analyzing the changing trends of the three-way decision at different scales,the optimal scale selection method(OSS-MSDT)based on the undetermined degree is proposed.(2)Aiming at the problem of optimal scale selection in MSDTs with the increments of the attributes,we develop increm ental algorithms to accelerate optimal scale selection.Firstly,we dynamically model an MSDT with the increments of the attributes.Secondly,we update the dominance classes and uncertain decision regions at different scales.Then,we propose two incremental algorithms for adding an attribute and adding multiple attributes,respectively.Finally,a series of comparative expeeiments with the non-incremental algorithm(OSS-MSDT)verifies the efficiency of the incremental algorithm.(3)Aiming at the problem of optimal scale selection in MSDTs with the increments of the objects,we develop incremental algorithms to accelerate optimal scale selection.Firstly,we dynamically model an MSDT with the increments of the objects.Seondly,we update the decision equivalence classes,the dominance classes and uncertain decision regions at different scales.Then,we propose two incremental algorithms for adding an object and adding multiple objects,respectively.Finally,a series of comparative experiment,with the non-incremental algorithm(OSS-MSDT)verifies the efficiency of the increm ental algorithm.(4)Aiming at the problem of optimal scale selection in MSDTs with the increments of the attribute values,we decelop increm ental algorithms to accelerate optimal scale selection.Firstly,we dynamically model an MSDT with the incremeuts of the attribute values.Secondly,we update attribute value range,the dominance classes and uncertain decision regions at different scales.Then,we propose two incremental algorithms for adding an attribute value and adding multiple attribute values,respectively.Finally,a series of comparative experiments with the nonincremental algorithm(OSS-MSDT)verifies the efficiency of the incremental algorithm.The research in this paper can not only develop the related research on multiscale decision tables,but also provide a new idea for solving uncertain problems.