基于短语结构和词语词性相结合的情感分类方法

针对传统文本分类方法忽略词语间的语义特征的问题,并为了改善输入文本的表示质量,提出一种基于短语结构和词语词性相结合的情感分类方法.该方法首先通过短语结构优化分词,可以更好地提取文本特征;其次利用Word2vec工具训练词语和词性相结合的文本语料库得到词向量模型,解决了Word2vec无法识别一词多义的问题;最后通过SVM算法对文本进行情感分类.实验结果表明,该算法能够提高文本情感分类的正确性.该方法对舆情监控、股票市场行情预测和了解消费者对产品的偏好等具有较高的实用性.

Sentiment Classification Based on Combination of Phrase Structure and Word Parts of Speech

A sentiment classification based on the combination of phrase structure and word parts of speech is proposed to improve the quality of the input text and the semantic features between words. This method firstly optimizes word segmentation through phrase structure to extract text features better, and then Word2vec is used to train the text corpus combined of words and parts of speech to obtain a word vector model to solve the problem that Word2vec cannot recognize the word polysemy. Finally, the text is subjected to emotional classification through SVM algorithm. The experimental results show that the algorithm can improve the correctness of text sentiment classification. This method has high practicability for monitoring public opinion, forecasting the stock market, and understanding consumers'' product preference.