基于三角形特征融合与感知注意力的方面级情感分析

doi:10.19678/j.issn.1000-3428.0070397

摘要/Abstract

摘要： 方面级情感分析(ABSA)旨在获取句法结构复杂的语句中特定方面词的情感极性。现有基于依存树与图神经网络(GNN)的模型因难以完整提取句法结构与深层语义特征,以及特征融合机制难以有效融合语义特征与句法结构,导致情感分析中句子情感极性判断的准确率较低。为此,建立基于DeBERTa的新型ABSA模型。首先,借助DeBERTa生成文本词向量,同时利用方面感知注意力机制提取方面词的特征,利用抽象语义表示获取文本句法结构,减少特征信息因提取不完整而对后续情感分析的影响;其次,构建融合句法结构与深层语义特征的新型三角形乘法机制;最后,通过三角形自注意力机制和全连接网络,将方面词的情感极性特征映射到情感分类层,使无关噪声的干扰得到有效抑制,从而提升情感极性判断的准确率。实验结果表明,相较于最新的基线模型,该模型的情感极性判断准确率和宏平均F1值分别平均提升0.93和1.39百分点,其能有效获取句子的句法结构与深层语义,且情感极性的分类准确率较高。

关键词: DeBERTa, 方面感知注意力机制, 三角形乘法机制, 三角形自注意力机制, 方面级情感分析

Abstract: Aspect-Based Sentiment Analysis (ABSA) aims to determine the sentiment polarity of specific aspect terms in sentences with complex syntactic structures. Existing models based on dependency trees and Graph Neural Networks (GNN) suffer from difficulty in fully extracting syntactic structures and deep semantic features, and their feature fusion mechanisms fail to effectively integrate semantic features with syntactic structures, resulting in low accuracy in sentence-level sentiment polarity judgment. To address this issue, a novel ABSA model based on DeBERTa is proposed. First, DeBERTa is employed to generate textual word embeddings, whereas an aspect-aware attention mechanism is utilized to extract aspect term features and abstract semantic representations are leveraged to capture textual syntactic structures, thereby reducing the impact of incomplete feature extraction on subsequent sentiment analysis. Second, a novel triangular multiplicative fusion mechanism that integrates syntactic structures with deep semantic features is constructed. Finally, through a triangular self-attention mechanism and fully connected networks, the sentiment polarity features of aspect terms are mapped to a sentiment classification layer, effectively suppressing the interference of irrelevant noise and thereby improving the accuracy of the sentiment polarity judgment. Experimental results demonstrate that, compared with the latest baseline models, the proposed model achieves an average improvement of 0.93 percentage points in sentiment polarity judgment accuracy and 1.39 percentage points in the macro-averaged F1 value. This model can effectively capture both the syntactic structures and deep semantics of sentences, yielding high accuracy in sentiment polarity classification.

Key words: DeBERTa, aspect-aware attention mechanism, triangular multiplication mechanism, triangular self-attention mechanism, Aspect-Based Sentiment Analysis (ABSA)

中图分类号:

TP391

彭李湘松, 张著洪. 基于三角形特征融合与感知注意力的方面级情感分析[J]. 计算机工程, 2026, 52(10): 106-115.

PENG Lixiangsong, ZHANG Zhuhong. Aspect-Based Sentiment Analysis Based on Triangular Feature Fusion and Perceptual Attention[J]. Computer Engineering, 2026, 52(10): 106-115.

https://www.ecice06.com/CN/Y2026/V52/I10/106

参考文献

[1] WILSON T, WIEBE J, HOFFMANN P. Recognizing contextual polarity in phrase-level sentiment analysis[C]//Proceedings of Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing. [S.l.]:ACL,2005: 347-354.
[2] BAHDANAU D, CHO K, BENGIO Y. Neural machine translation by jointly learning to align and translate[EB/OL]. [2024-08-05]. https://arxiv.org/abs/1409.0473.
[3] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
[4] TANG D Y, QIN B, LIU T. Aspect level sentiment classification with deep memory network[EB/OL]. [2024-08-05].https://arxiv.org/abs/1605.08900.
[5] ZHANG C, LI Q C, SONG D W. Aspect-based sentiment classification with aspect-specific graph convolutional networks[EB/OL]. [2024-08-05].https://arxiv.org/abs/1909.03477.
[6] LIANG S, WEI W, MAO X L, et al. BiSyn-GAT+: bi-syntax aware graph attention network for aspect-based sentiment analysis[EB/OL]. [2024-08-05].https://arxiv.org/abs/2204.03117.
[7] BANARESCU L, BONIAL C, CAI S, et al. Abstract meaning representation for sembanking[C]//Proceedings of the 7th Linguistic Annotation Workshop and Interoperability with Discourse. [S.l.]:ACL,2013: 178-186.
[8] WANG Y Q, HUANG M L, ZHU X Y, et al. Attention-based LSTM for aspect-level sentiment classification[C]//Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing.[S.l.]:ACL,2016: 606-615.
[9] XING B W, LIAO L J, SONG D D, et al. Earlier attention? aspect-aware LSTM for aspect-based sentiment analysis[EB/OL]. [2024-08-05].https://arxiv.org/abs/1905.07719.
[10] CHEN P, SUN Z Q, BING L D, et al. Recurrent attention network on memory for aspect sentiment analysis[C]//Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing. [S.l.]:ACL,2017: 452-461.
[11] GU S Q, ZHANG L P, HOU Y X, et al. A position-aware bidirectional attention network for aspect-level sentiment analysis[C]//Proceedings of the 27th International Conference on Computational Linguistics. [S.l.]:ACL,2018: 774-784.
[12] ZHOU J, CUI G Q, HU S D, et al. Graph neural networks: a review of methods and applications[J]. AI Open, 2020, 1: 57-81.
[13] 杨春霞, 吴亚雷, 闫晗, 等. 融合双图卷积与门控线性单元的方面级情感分析模型[J]. 计算机工程, 2024, 50(4): 141-149. YANG C X, WU Y L, YAN H, et al. Aspect-level sentiment analysis model combining double graph convolution and gated linear unit[J]. Computer Engineering, 2024, 50(4): 141-149. (in Chinese)
[14] 郭荣荣, 高建瓴, 徐瑞涓. 融合多特征图卷积神经网络的方面级情感分析[J]. 小型微型计算机系统, 2024, 45(5): 1039-1045. GUO R R, GAO J L, XU R J. Aspect-based sentiment analysis by fusing multi-feature graph convolutional network[J]. Journal of Chinese Computer Systems, 2024, 45(5): 1039-1045. (in Chinese)
[15] ZHANG Y H, QI P, MANNING C D. Graph convolution over pruned dependency trees improves relation extraction[EB/OL]. [2024-08-05].https://arxiv.org/abs/1809.10185.
[16] CHEN G M, TIAN Y H, SONG Y. Joint aspect extraction and sentiment analysis with directional graph convolutional networks[C]//Proceedings of the 28th International Conference on Computational Linguistics. [S.l.]:ACL,2020: 272-279.
[17] LIANG B, SU H, GUI L, et al. Aspect-based sentiment analysis via affective knowledge enhanced graph convolutional networks[J]. Knowledge-Based Systems, 2022, 235: 107643.
[18] HE P C, LIU X D, GAO J F, et al. DeBERTa: decoding-enhanced BERT with disentangled attention[EB/OL]. [2024-08-05].https://arxiv.org/abs/2006.03654.
[19] CAI D, LAM W. AMR parsing via graph-sequence iterative inference[EB/OL]. [2024-08-05].https://arxiv.org/abs/2004.05572.
[20] MA D H, LI S J, ZHANG X D, et al. Interactive attention networks for aspect-level sentiment classification[EB/OL]. [2024-08-05].https://arxiv.org/abs/1709.00893.
[21] LI R F, CHEN H, FENG F X, et al. DualGCN: exploring syntactic and semantic information for aspect-based sentiment analysis[J]. IEEE Transactions on Neural Networks and Learning Systems, 2024, 35(6): 7642-7656.
[22] XIAO Z G, WU J R, CHEN Q L, et al. BERT4GCN: using BERT intermediate layers to augment GCN for aspect-based sentiment classification[EB/OL]. [2024-08-05].https://arxiv.org/abs/2110.00171.
[23] CHEN C H, TENG Z Y, WANG Z Q, et al. Discrete opinion tree induction for aspect-based sentiment analysis[C]//Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics. [S.l.]:ACL,2022: 2051-2064.
[24] ZHANG M, ZHU Y X, LIU Z, et al. Span-level aspect-based sentiment analysis via table filling[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics. [S.l.]:ACL,2023: 9273-9284.
[25] CHAI H Y, YAO Z Y, TANG S Y, et al. Aspect-to-scope oriented multi-view contrastive learning for aspect-based sentiment analysis[C]//Proceedings of the Findings of the Association for Computational Linguistics: EMNLP 2023. [S.l.]:ACL,2023: 10902-10913.
[26] CHEN B F, OUYANG Q H, LUO Y Q, et al. S²GSL: incorporating segment to syntactic enhanced graph structure learning for aspect-based sentiment analysis[EB/OL]. [2024-08-05].https://arxiv.org/abs/2406.02902.
[27] NASEEM T, SHAH A, WAN H, et al. Rewarding smatch: transition-based AMR parsing with reinforcement learning[EB/OL]. [2024-08-05].https://arxiv.org/abs/1905.13370.
[28] ZHANG S, MA X T, DUH K, et al. AMR parsing as sequence-to-graph transduction[EB/OL]. [2024-08-05].https://arxiv.org/abs/1905.08704.
[29] ZHOU Q J, ZHANG Y, JI D H, et al. AMR parsing with latent structural information[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. [S.l.]:ACL,2020: 4306-4319.

选择文件类型/文献管理软件名称

选择包含的内容