【摘要】 文本分类中的两个关键问题,算法和特征提取。贝叶斯算法是最有效的文本分类算法之一,但是属性间强独立性的假设在现实中并不成立,借鉴概率论中的多项式模型提出了一种改进型的贝叶斯方法;传统的特征抽取方法有词频法、互信息法、CHI统计、信息增益法等,然而上述方法对于词条的权重未作考虑,引进了权重的表征方式,给出了改进方法。由实验证明了通过以上方面的改进,文本分类的正确率得到了提高。更多还原

【Abstract】 Two important factors in text classification are discussed— algorithm and feature abstraction.The practical Bayesian algorithm has an assumption of strong independence of different properties and a modified way on polynomial is introduced.In Feature abstraction,different ways of abstracting features are discussed and a modified CHI based on word weight is introduced.At last the experiments show seen that correct rate of text classification is improved.更多还原