Aspect-based Sentiment Analysis Method based on Multi-Feature Fusion Graph Neural Networks
International Journal of Computer Science and Information Technology · 09.10.2026 · 8 (7) · 1-8
DOI: 10.62051/ijcsit.v8n7.01 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Computer Science and Information Technology, online erschienen am 09.10.2026. Verfasst von Yunfei Chen und 3 weiteren Autor:innen. Die Arbeit nennt 40 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Aspect-based sentiment analysis (ABSA) identifies the sentiment expressed toward specific aspects in text. Graph Neural Networks (GNNs) represent nonlinear relations between words through graph structures and have become increasingly prominent in ABSA. In complex contexts, however, a single feature type rarely captures the full relation between an aspect and its associated sentiment expression. Multi-feature fusion has therefore emerged as an important research direction. This review first outlines lexicon-based, machine-learning, and deep-learning approaches to ABSA. It then examines GNN methods, with emphasis on how they model and integrate different features. We synthesize the limitations of current methods, including information redundancy, noise propagation, model complexity, and limited cross-domain adaptation. Finally, we identify promising directions for future research on textual ABSA.
Zitieren
APA
Chen, Y., He, B., Gong, W., & Bai, W. (2026). Aspect-based Sentiment Analysis Method based on Multi-Feature Fusion Graph Neural Networks. International Journal of Computer Science and Information Technology, 8(7), 1-8. https://doi.org/10.62051/ijcsit.v8n7.01
Vancouver
Chen Y, He B, Gong W, Bai W. Aspect-based Sentiment Analysis Method based on Multi-Feature Fusion Graph Neural Networks. International Journal of Computer Science and Information Technology. 2026;8(7):1-8. doi:10.62051/ijcsit.v8n7.01
BibTeX
@article{Chen2026,
author = {Chen, Yunfei and He, Bo and Gong, Wenru and Bai, Weining},
title = {Aspect-based Sentiment Analysis Method based on Multi-Feature Fusion Graph Neural Networks},
journal = {International Journal of Computer Science and Information Technology},
year = {2026},
volume = {8},
number = {7},
pages = {1--8},
doi = {10.62051/ijcsit.v8n7.01}
}
EUVIMED