基于ERNIE-BiLSTM的社交网络文本情感分析

Sentiment Analysis of Social Network Text Based on ERNIE-BiLSTM

社交网络文本情感分析任务中,因短文本信息模糊等特点,传统的词向量模型无法更好地表示词的语义特征,当前短文本情感分类任务多以二分类研究为主,将结果分类为积极情感与消极情感,未能对分类结果更细入的划分。文中提出一种舆情情感分析的ERNIE-BiLSTM方法,实现了对用户评论情感的七种情绪分类,包括恐惧、厌恶、乐观、惊喜、感恩、悲伤和愤怒。ERNIE-BiLSTM方法利用ERNIE预训练模型获取文本的语义信息,结合BiLSTM提取文本的双向特征,最后使用softmax函数获得最终的情感分类结果。实验结果表明,ERNIE-BiLSTM方法具有87.7%的精确率、86.9%的召回率和86.8%的F1得分,比其他方法得到了有效提升。

In the text sentiment analysis task of social network,due to the characteristics of fuzzy short text information,the traditional word vector model cannot better represent the semantic features of words.The current short text sentiment classification task mainly focuses on binary classification research,classifying the results into positive and negative emotions,and failing to divide the classification results into more details.This paper proposes an ERNIE-BilSTM method for sentiment analysis of public opinion,which can classify seven emotions of user comments,including fear,disgust,optimism,surprise,gratitude,sadness and anger.The ERNIE-BilSTM method uses the pre-trained ERNIE model to obtain the semantic information of the text,combines with BiLSTM to extract the bidirectional features of the text,and finally uses softmax function to obtain the final sentiment classification results.The experimental results show that the ERNIE-BILSTM method has 87.7% precision,86.9% recall and 86.8% F1 score,which is effectively improved compared with other methods.