Text feature selection based on water wave optimization algorithm

Abstract

This paper starts from the perspective of the entire impacts of text features to classification results, and present a method of text feature selection that based on Water Wave Optimization (WWO) algorithm, Water Wave Optimization Text Feature Selection (WWOTFS), mining text feature selection rules by using Water Wave Optimization algorithm. WWOTFS firstly selects text features by using Chi-square Test, then screens the selected text features by using WWO algorithm. WWOTFS regards a water wave as a feature selection candidate solution, the candidate solutions correspond to the water wave cluster, the accuracy rate is used for fitness function, and the method of grouping is used to lower the dimension of the water wave. The experimental results show that the effect of WWOTFS is better than the traditional feature selection method, and it has good stability.