基于深度学习框架的短文本情感分析方法研究

Research on Emotional Analysis of Short Text Based on Deep Learning Framework

随着电子商务的迅猛发展,越来越多的人们喜欢在网上购买商品,商品评论数据也急剧增加,这些评论中含有用户对商品的情感倾向,海量的评论加重了人工管理。本文对商品评论内容进行情感分析,有效帮助商家了解消费者对商品的认可程度,从而提高商品与服务质量。本文通过对Word2vec计算文本词向量,最后应用Keras下的LSTM对商品评论文本进行情感分类,实验验证了该方法在互联网商品评论中的有效性和准确性,取得了较好的实验结果。

With the rapid development of e-commerce,more and more people like to buy goods on the Internet,and the data of commodity reviews have increased dramatically.These reviews contain users'emotional tendencies towards commodities,and a large number of comments have aggravated the manual management.This paper makes an emotional analysis of the content of commodity reviews to help businessmen effectively understand the degree of consumer recognition of commodities,thereby improving the quality of commodities and services.This paper calculates the text word vector by Word2vec,and finally uses LSTM under Keras to classify the emotion of the comment text.The experiment verifies the validity and accuracy of this method in Internet comment,and achieves good experimental results.