基于贝叶斯算法的中文新闻标题分类研究

Research on Chinese News Title Classification Based on Bayesian Algorithm

摘要是对消息的一种高度概括,因此,如何有效地对摘要进行快速、准确的识别,是当前中文摘要识别领域的一个重要课题。文章提出TF-IDF和贝叶斯算法相结合的新闻分类方法,通过TF-IDF算法提取短文文本中的特征词集合,捕捉短文文本表达的语义,并计算出相应的TF-IDF值,将TF-IDF值形成特征向量作为贝叶斯算法的输入来实现新闻文本分类,最后根据错误率对预测结果进行评价。试验结果表明,该方法可以将贝叶斯方法与TF-IDF相结合,实现对信息的快速分类。

Abstract is a high level summary of messages,therefore,how to effectively identify abstracts quickly and accurately is an important topic in the current field of Chinese abstract recognition.This paper proposes a news classification method that combines TF-IDF and Bayesian algorithm.Using the TF-IDF algorithm to extract the set of feature words in short text,captures the semantics expressed in the short text,and calculates the corresponding TF-IDF values.The TF-IDF values are formed into feature vectors as input to the Bayesian algorithm to achieve news text classification.Finally,the prediction results are evaluated based on the error rate.The experimental results indicate that this method can combine Bayesian method with TF-IDF to achieve rapid classification of information.