期刊论文

中文

现代计算机：中旬刊

1007-1423

2016

14

52-55

10.3969/j.issn.1007-1423.2016.14.011

湖南省哲学社会科学基金（No.14YBA335）；

本校为第一机构

提出一种通过提取词向量,并利用机器学习对新闻文本进行分类的方法。首先,通过对预先收集好的新闻语料进行分析,获取到词的向量表示形式;然后通过新闻中的一些关键词提取出新闻的关键句;最后把词向量和关键句当中的关键词词性作为组合特征,采用SVM算法进行分类,得到新闻的倾向性类别。实验表明,基于组合特征的SVM文本分类方法具有较高的F值。

Proposes an approach for classifying the sentiment of news text based on SVM and Word Embedding.Firstly,word embedding is achieved by training the pre-collected news corpus with word2 vec.Secondly,key sentences are constructed from some key words in news text.At last,the word embedding and key words＇ part-of-speech are selected as combination features to apply in SVM algorithm,and then the sentiment classification of news text is obtained.Experimental results show that SVM based on combina...