A Dual-Module Information Fusion Aspect-Level Sentiment Classification Model

doi:10.1109/ICECAI62591.2024.10675168

	A Dual-Module Information Fusion Aspect-Level Sentiment Classification Model
	He, Bowen; Li, Qiang; Zhang, Yihua; Zhang, Yongfu
	2024
会议名称	5th International Conference on Electronic Communication and Artificial Intelligence, ICECAI 2024
会议录名称	2024 5th International Conference on Electronic Communication and Artificial Intelligence, ICECAI 2024
页码	19-23
会议日期	May 31, 2024 - June 2, 2024
会议地点	Hybrid, Shenzhen, China
会议录编者/会议主办者	IEEE
出版者	Institute of Electrical and Electronics Engineers Inc.
摘要	Aspect-level sentiment categorization is the task of identifying sentiments or opinions expressed about particular aspects or entities in a given text. It involves analyzing the sentiment of different aspects or features in a text, such as product features in a customer review or a specific topic in a social media post. The goal is to categorize the sentiment of each aspect, e.g., positive, negative or neutral. In this paper, we propose a BD-MGCN model via bimodular information fusion for addressing the shortcomings of traditional sentiment classification models in feature extraction and fusion. The model includes a BERT pre-trained multi-granularity convolution (B-MGCN) module based on BERT pre-training and a syntactic relation-based graph convolution (Dep-GCN) module with feature fusion via the attention mechanism. The experimental results demonstrate that the model performs well in terms of prediction accuracy and F1 value, and can effectively mine and fuse feature information to improve the performance of sentiment classification. © 2024 IEEE.
关键词	Classification (of information) Data fusion BERT pre-training Classification models Customer review Features fusions Multi-granularity Multi-granularity convolution Pre-training Product feature Sentiment classification Social media
DOI	10.1109/ICECAI62591.2024.10675168
收录类别	EI
语种	英语
EI入藏号	20244217193933
EI主题词	Information fusion
EI分类号	1106.2 ; 716.1 Information Theory and Signal Processing ; 903.1 Information Sources and Analysis
原始文献类型	Conference article (CA)
文献类型	会议论文
条目标识符	http://ir.lzufe.edu.cn/handle/39EH0E1M/38157
专题	信息工程与人工智能学院
通讯作者	Li, Qiang
作者单位	School of Information Engineering, Lanzhou University of Finance and Economics, Gansu, Lanzhou, China
通讯作者单位	信息工程与人工智能学院
推荐引用方式 GB/T 7714	He, Bowen,Li, Qiang,Zhang, Yihua,et al. A Dual-Module Information Fusion Aspect-Level Sentiment Classification Model[C]//IEEE:Institute of Electrical and Electronics Engineers Inc.,2024:19-23.