A Local Discrete Text Data Mining Method in High-Dimensional Data Space

Abstract Aiming at the problems of low accuracy, long time required, and large memory consumption traditional data mining methods, a local discrete text method in high-dimensional space is proposed. First all, through preparation preprocessing step, we obtain minimum divergence maximize dimension to meet demand for space; second, use information gain mine pre-processed establish an objective function highest gain; finally, functions established preparation, preprocessing, are combined form multi-objective optimization problem realize mining. The simulation experiment results show that our effectively reduces improves accuracy mining, where it also consumes less memory, indicating can solve multiple improve effect.