Aspect-Level Sentiment Analysis with Integrated Dynamic Edge-Weighted Graph Attention Network

doi:10.3778/j.issn.1673-9418.2410037

Abstract

Abstract: In the field of aspect-level sentiment analysis, existing research mainly uses static graph neural networks to model the dependency relationship between context and aspects, ignoring the dynamic nature of syntactic structure and part-of-speech information. This limitation restricts the model??s ability to capture complex semantic relationships. A dynamic edge-weighted graph attention network (DEWGAT) is proposed for aspect-level sentiment analysis. The model integrates BERT embeddings, syntactic distance matrix, aspect-related mask, and grammatical weight matrix to comprehensively capture the semantic, syntactic, and aspect information of the text. This paper dynamically calculates edge weights based on syntactic dependencies, reinforcing key emotional dependency relationships. It introduces an attention mechanism that considers aspect terms and grammatical dependencies in the graph attention network, enhancing the capture of crucial sentiment information and improving classification accuracy. Experimental results on four public benchmark datasets show that DEWGAT improves accuracy by 1.16, 0.15, and 4.85 percentage points on Twitter, Lap14, and MAMS datasets, respectively. Notably, on the MAMS dataset, the accuracy reaches 0.8937, significantly outperforming most comparison models. It demonstrates superior performance in handling complex sentiment analysis tasks. Although the DEWGAT model performs excellently on multiple datasets, it has not yet achieved end-to-end framework integration.

Key words: dynamic edge weight, graph attention network, grammatical weight matrix, aspect-level sentiment analysis

摘要： 在方面级情感分析领域，传统研究多依赖静态图神经网络来建模文本上下文与方面之间的依赖关系。然而，这些方法往往忽略了句法结构和词性信息的动态特性，限制了模型在捕捉复杂语义关系方面的能力。为此，提出了一种动态边权重图注意力网络（DEWGAT），旨在更全面地捕捉文本的语义、句法和方面信息。DEWGAT模型通过整合BERT嵌入、句法距离矩阵、方面相关掩码和语法权重矩阵，动态计算基于句法依存关系的边权重，从而强化关键情感依存关系的建模。在图注意力网络中，引入了一种新的注意力机制，该机制同时考虑了方面词和语法依存关系，显著增强了对关键情感信息的捕捉能力，提高了情感分类的准确性。在实验中，DEWGAT模型在四个公开基准数据集上进行了评估，结果显示其在Twitter、Lap14和MAMS数据集上的准确率分别提升了1.16、0.15和4.85个百分点。特别是在MAMS数据集上，DEWGAT模型的准确率达到了0.893 7，显著优于多数对比模型，展示了其在处理复杂情感分析任务中的优越性能。尽管DEWGAT模型在多个数据集上表现出色，但目前尚未实现端到端的框架集成。

关键词: 动态边权重, 图注意力网络, 语法权重矩阵, 方面级情感分析

ZHONG Zhaoman, LYU Huihui, ZHANG Yu, CUI Xinru, FAN Jidong, HUANG Zeyu. Aspect-Level Sentiment Analysis with Integrated Dynamic Edge-Weighted Graph Attention Network[J]. Journal of Frontiers of Computer Science and Technology, 2025, 19(11): 3083-3093.

仲兆满, 吕慧慧, 张渝, 崔心如, 樊继冬, 黄泽宇. 融合动态边权重图注意力网络的方面级情感分析[J]. 计算机科学与探索, 2025, 19(11): 3083-3093.

References

[1] ZARANDI A K, MIRZAEI S. A survey of aspect-based sentiment analysis classification with a focus on graph neural network methods[J]. Multimedia Tools and Applications, 2024, 83(19): 56619-56695.
[2] 商容轩, 张斌, 米加宁. 基于BRNN的政务APP评论端到端方面级情感分析方法[J]. 数据分析与知识发现, 2022, 6(S1): 364-375.
SHANG R X, ZHANG B, MI J N. End-to-end aspect-level sentiment analysis for E-government applications based on BRNN[J]. Data Analysis and Knowledge Discovery, 2022, 6(S1): 364-375.
[3] LIU B W, BLASCH E, CHEN Y, et al. Scalable sentiment classification for big data analysis using na?ve Bayes classifier[C]//Proceedings of the 2013 IEEE International Conference on Big Data. Piscataway: IEEE, 2013: 99-104.
[4] HUANG B X, CARLEY K. Parameterized convolutional neural networks for aspect level sentiment classification[C]//Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2018: 1091-1096.
[5] LIU Y F, QIN Z C, LI P Y, et al. Stock volatility prediction using recurrent neural networks with sentiment analysis[C]//Advances in Artificial Intelligence: From Theory to Practice-the 30th International Conference on Industrial Engineering and Other Applications of Applied Intelligent Systems. Cham: Springer, 2017: 192-201.
[6] CHEN C T, ZHUO R, REN J T. Gated recurrent neural network with sentimental relations for sentiment classification[J]. Information Sciences, 2019, 502: 268-278.
[7] WANG J, YU L C, LAI K R, et al. Tree-structured regional CNN-LSTM model for dimensional sentiment analysis[J]. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2020, 28: 581-591.
[8] ZHANG C, LI Q C, SONG D W. Aspect-based sentiment classification with aspect-specific graph convolutional networks[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing. Stroudsburg: ACL, 2019: 4567-4577.
[9] SUN K, ZHANG R C, MENSAH S, et al. Aspect-level sentiment analysis via convolution over dependency tree[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing. Stroudsburg: ACL, 2019: 5678-5687.
[10] ZHANG M, QIAN T Y. Convolution over hierarchical syntactic and lexical graphs for aspect level sentiment analysis[C]//Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2020: 3540-3549.
[11] YUAN L, WANG J, YU L C, et al. Graph attention net-work with memory fusion for aspect-level sentiment analysis[C]//Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing. Stroudsburg: ACL, 2020: 27-36.
[12] SONG Y , WANG J, JIANG T, et al. Attentional encoder network for targeted sentiment classification[EB/OL]. [2024-08-15]. https://arxiv.org/abs/1902.09314.
[13] YAN H, DAI J Q, JI T, et al. A unified generative framework for aspect-based sentiment analysis[C]//Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing. Stroudsburg: ACL, 2021: 2416-2429.
[14] ZHANG K, ZHANG K, ZHANG M D, et al. Incorporating dynamic semantics into pre-trained language model for aspect-based sentiment analysis[C]//Findings of the Association for Computational Linguistics: ACL 2022. Stroudsburg: ACL, 2022: 3599-3610.
[15] ZHAO P L, HOU L L, WU O. Modeling sentiment dependencies with graph convolutional networks for aspect-level sentiment classification[J]. Knowledge-Based Systems, 2020, 193: 105443.
[16] ZHANG Z, ZHOU Z L, WANG Y N. SSEGCN: syntactic and semantic enhanced graph convolutional network for aspect-based sentiment analysis[C]//Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: ACL, 2022: 4916-4925.
[17] WANG K, SHEN W Z, YANG Y Y, et al. Relational graph attention network for aspect-based sentiment analysis[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 2020: 3229-3238.
[18] YUAN L, WANG J, YU L C, et al. Syntactic graph attention network for aspect-level sentiment analysis[J]. IEEE Transactions on Artificial Intelligence, 2024, 5(1): 140-153.
[19] PHAN M H, OGUNBONA P O. Modelling context and syntactical features for aspect-based sentiment analysis[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 2020: 3211-3220.
[20] CHEN Y X, YAO J T. Sentiment analysis using part-of-speech-based feature extraction and game-theoretic rough sets[C]//Proceedings of the 2021 International Conference on Data Mining. Piscataway: IEEE, 2022: 110-117.
[21] ALI S F, MASOOD N. Evaluation of adjective and adverb types for effective Twitter sentiment classification[J]. PLoS One, 2024, 19(5): e0302423.
[22] 孔兰若. 现代汉语情感形容词研究[J]. 现代语文(语言研究版), 2013(4): 57-59.
KONG L R. A study on emotional adjectives in modern Chinese[J]. Modern Chinese (Edition of Language Research), 2013(4): 57-59.
[23] PONTIKI M, GALANIS D, PAVLOPOULOS J, et al. SemEval-2014 task 4: aspect based sentiment analysis[C]//Proceedings of the 8th International Workshop on Semantic Evaluation. Stroudsburg: ACL, 2014: 27-35.
[24] DONG L, WEI F R, TAN C Q, et al. Adaptive recursive neural network for target-dependent twitter sentiment classification[C]//Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 2014: 49-54.
[25] JIANG Q N, CHEN L, XU R F, et al. A challenge dataset and effective models for aspect-based sentiment analysis[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th Inter-national Joint Conference on Natural Language Processing. Stroudsburg: ACL, 2019: 6279-6284.