贝叶斯、RidgeClassifier和fastText算法在匿名新闻文本分类中的对比研究

A Comparative Study of Bayesian, Ridgeclassifier and fastText Algorithms in Anonymous News Text Classification

针对互联网时代面临的通过人工对海量新闻进行分类较为困难的问题,本文通过贝叶斯、RidgeClassifier和fastText这三种分类器进行新闻文本分类,分析比较了这三种分类器对新闻文本分类的效果。实验结果表明,三种分类器均可以实现新闻文本分类的任务,其中fastText方法在匿名新闻文本分类问题中各方面性能指标最优,本文从算法理论上分析了产生这一差异的原因。

The massive amount of news on the Internet makes manual news text classification an unrealistic task.In this paper,we classify news text by Bayesian,RidgeClassifier,and fastText classifiers and analyze and compare the effects of these three classifiers.The experiments show that these classifiers can achieve the task of news text classification.Among them,the fastText method has the best performance index in all aspects of anonymous news text classification,and the reasons for this difference are analyzed in the theoretical aspect.