Dynamic Updating Algorithms For Local Neighborhood Rough Sets

With the rapid development of information technology,the data in life is becoming increasingly large,complex,and constantly changing.Effective processing of dynamic data in real-world environments to acquire potentially useful knowledge has been one of the important research topics in the field of data mining.The rough set models do not require any prior knowledge other than the dataset,and approximates the imprecise or uncertain knowledge through the upper and lower approximation sets,which can effectively acquire knowledge from information systems.The generalized local rough sets only need to consider the information granules within the target concept,which effectively solve overfitting problems and significantly reduce the time required to compute the approximate sets,providing an efficient method for knowledge discovery in large numerical datasets.However,when dealing with dynamic data,existing approximate set algorithms ignore existing knowledge,which result in low computational efficiency and even failure due to memory overflow.To address these issues,we propose dynamic updating algorithms for local neighborhood rough set in this paper.The main research contents include:(1)To effectively compute the approximation operators in the case of object set change,we analyze the relationships of inclusion degree between neighborhood classes and target concepts.Based on these relationships,we propose formulas for updating local neighborhood approximate sets,and provide corresponding dynamic updating algorithms.The experimental results verify the efficiency of the dynamic updating algorithms in solving local neighborhood approximate sets when object set change.(2)For the case of attribute set change in neighborhood information systems,we introduce neighborhood approximation sets into the study of dynamically updating local neighborhood approximation sets.Based on the neighborhood approximate sets,we propose dynamic updating formulas for local neighborhood approximate sets,and design the corresponding dynamic updating algorithms.The experimental results show that the dynamic updating algorithms have higher computational efficiency in computing local neighborhood approximate sets when attribute set change.