软件导刊2025，Vol.24Issue(5)：35-39,5.DOI:10.11907/rjdk.241069
基于BERT-BiLSTM-ATT的大宗商品新闻情感分析
Sentiment Analysis of Commodity News Based on the BERT-BiLSTM-ATT Model
摘要
Abstract
To achieve sentiment analysis of commodity news,this article proposes an sentiment analysis method based on the complete sen-tence segmentation BERT-BiLSTM-ATT model to address the issues of traditional methods that lack contextual semantic relationships and ig-nore important information in text,as well as the difficulty of BERT in handling long texts.Firstly,the news text is divided into multiple parts according to the segmentation mode of the complete sentence,and then the BERT model is used to obtain the feature representation of the com-modity news text,and then the obtained text features are pooled and input to BiLSTM to extract the emotional features of the news,add the at-tention layer before the output of BiLSTM to improve the classification accuracy,and finally use the Softmax classifier to classify the extracted features.The experimental results show that the F1 index is increased by 5.4%,3.4%and 2.2%compared with BERT-LSTM,BERT-BILSTM and BERT-BiLSTM-ATT,respectively,which proves the effectiveness of the proposed method in sentiment analysis.关键词
大宗商品新闻/情感分析/长文本处理/完整句分割Key words
commodity news/sentiment analysis/long text processing/complete sentence segmentation分类
信息技术与安全科学引用本文复制引用
柳标良,谷晓燕..基于BERT-BiLSTM-ATT的大宗商品新闻情感分析[J].软件导刊,2025,24(5):35-39,5.基金项目
国家自然科学基金项目(71701020) （71701020）
