Rough set reduction has been used as an important preprocessing step for pattern recognition,machine learning and big data analysis.As the classical rough set model can just be used to evaluate categorical features,a neighborhood rough set model is introduced to deal with numerical datasets.Traditional hill-climbing search approaches to neighborhood rough set reduction have difficulties to find optimal reducts.And the current stochastic search strategies,such as GA,ACO and PSO,provide a more robust solution but at the expense of increased computational effort.It is necessary to investigate fast and effective search methods.In this paper,we define a knowledge representation structure called power set tree(PS-tree),which is an order tree representing the power set,and each possible reduct is mapped to a node of the tree.We develop a tree search framework for reduction question solving by the PS-tree.Furthermore,we propose four tree search methods based on PS-tree,which are depth-first,breadth-first,uniform-cost and A* search methods.Some experiments on UCI datasets are designed to compare with the four tree search methods.Experiment results demonstrate that our tree search methods are effective and efficient.