基于语义分析的在线评论文本情感分类算法研究

Study on Sentiment Classification Algorithm of Online Reviews Based on Semantic Analysis

针对用户评论文本情感分类过程中缺乏特征词语义分析和数据维度过高的问题,提出了一种基于语义分析的在线评论情感分类方法。利用Word2Vec工具获得词向量,通过词向量运算获取评论文本中的词与情感词典中的词之间的语义相似度,然后根据此相似度的大小选择反映正面或负面情感的词作为评论文本的关键特征。通过非负矩阵分解算法,将原始评论文本映射到一个低维的语义空间,降低评论数据维度,增强评论文本之间的语义相关度。实验表明,提出的算法具有更好的文本情感分类能力。

Aiming at the lack of semantic analysis of feature words and the high data dimension in the process of online review sentiment classification,a method of online review sentiment classification based on semantic analysis is proposed.Firstly,Word2Vec is used to obtain the word vector,and the semantic similarity between the words in the review and the words in the sentiment dictionary is obtained through the word vector operation.Then,according to the degree of semantic similarity,the words reflecting positive or negative emotions are selected as the key semantic features of the review.The original reviews are mapped into a low-dimensional semantic space by the non-negative matrix factor algorithm,which reduces the dimension of review data and enhances the semantic correlation between the reviews.Experiments show that the proposed algorithm has a good ability of sentiment classification.