《计算机应用研究》|Application Research of Computers

结合依存句法分析与交互注意力机制的隐式方面提取

Combining dependency syntactic parsing with interactive attention mechanism for implicit aspect extraction

免费全文下载 (已被下载 次)  
获取PDF全文
作者 汪兰兰,姚春龙,李旭,于晓强
机构 大连工业大学 a.信息科学与工程学院;b.工程训练中心,辽宁 大连 116034
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2022)01-006-0037-06
DOI 10.19734/j.issn.1001-3695.2021.06.0249
摘要 隐式方面提取对于提升细粒度情感分析的准确性具有重要意义,然而现有隐式方面提取技术在处理大规模数据时泛化能力不强。为此,提出结合依存句法分析与交互注意力机制的隐式方面提取模型。首先利用预训练语言模型BERT生成文本的初始表征,然后传递给依存句法引导的自注意力层再次处理,再将两次处理的结果经交互注意力机制进一步提取特征,最终用分类器判断句子所属的隐式方面类别。与基线BERT及其他深度神经网络模型对比,所提模型在增强的SemEval隐式方面数据集上取得了更高的<i>F</i><sub>1</sub>与AUC值,证明了模型的有效性。
关键词 方面级情感分析; 隐式方面提取; BERT; 依存句法分析; 交互注意力
基金项目 国家重点研发计划专项资助项目(2017YFC0821003-3)
辽宁省自然科学基金资助项目(20180550395)
辽宁省教育厅青年科技人才“育苗”资助项目(J2020113)
辽宁省科技厅科学研究项目(LJKZ0537)
本文URL http://www.arocmag.com/article/01-2022-01-006.html
英文标题 Combining dependency syntactic parsing with interactive attention mechanism for implicit aspect extraction
作者英文名 Wang Lanlan, Yao Chunlong, Li Xu, Yu Xiaoqiang
机构英文名 a.School of Information Science & Engineering,b.Engineering Training Center,Dalian Polytechnic University,Dalian Liaoning 116034,China
英文摘要 Implicit aspect extraction is important for improving the accuracy of fine-grained sentiment analysis. However, existing implicit aspect extraction techniques do not have strong generalization ability when dealing with large-scale data. To address the problem, this paper proposed an implicit aspect extraction model combining dependency syntactic parsing and interactive attention mechanism. First, the model generated the initial representation of the text by the pre-trained language model BERT. Then, it passed the initial representation to the self-attention layer guided by the dependency syntactic parsing. Due to the interactive attention mechanism, the model further extracted the results of the above two processes. Finally it used a classifier to determine the implicit aspect of the sentence. Compared with baseline BERT and other deep neural network models, the proposed model has achieved higher <i>F</i><sub>1</sub> and AUC on the enhanced SemEval implicit aspect dataset, which proves the effectiveness of the model.
英文关键词 aspect level sentiment analysis; implicit aspect extraction; BERT; dependency syntactic parsing; interactive attention
参考文献 查看稿件参考文献
 
收稿日期 2021/6/23
修回日期 2021/8/11
页码 37-42
中图分类号 TP391
文献标志码 A