Chinese spam filtering based on graph regularized MNMF algorithm
Zunxiong Liu, Huang Zhi-qiang, Shujuan Zheng, Fei Shi · Jisuanji yingyong yanjiu · 2013
Text e-mail data represented by vector space model(VSM) are high dimensionality.This situation is not conducive to construct e-mail filtering model.Therefore,dimensionality reduction need be performed.MNMF could simultaneously achieve dimensionality reduction and e-mail classification,and graph regularized NMF could keep the geometrical structure of the data space.Based on the above two improved NMF models,this paper put forward GMNMF algorithm,and designed an iterative solution algorithm.Using GMNMF algorithm and other related algorithm do Chinese spam filtering experiments.The experimental results show that the model of the proposed algorithm is superior to models of other good algorithms.