Scholarlink

<p>Twitter sentiment analysis plays a vital role in understanding public opinion, emotions, and emerging trends from large-scale user-generated social media content. However, existing research faces limitations such as handling noisy data, understanding sarcasm, and effectively processing the nuanced and informal language often used on the platform. This research aims to accurately gauge public sentiment on Twitter through an innovative approach. The proposed model, Crossbred Analogous Tree-Knowledge Deep (CAT-KD), is composed of five key components for comprehensive sentiment analysis. The Crossbred Node Feature Encoder (CNFD) captures intricate node representations, while the Arbor Graph Attention Network (A-GAT) models the hierarchical structure of Twitter post trees. Simultaneously, the Contextual Graph Attention Network (C-GAT) focuses on user graph modeling to capture user behavior and interactions. A post-user fusion layer facilitates seamless integration between post and user information, enhancing contextual understanding. Additionally, the Knowledge Aware Attention Network (K-A2N) further refines the model’s performance by incorporating external knowledge. The final sentiment classification is conducted using a Deep Neural Network (DNN) classifier, ensuring accurate and nuanced sentiment analysis. Furthermore, to prove the efficacy of our proposed CAT-KD model, the model is evaluated on benchmark datasets including Sentiment140 (1.6&#xa0;million tweets), Airline Twitter dataset (14,641 tweets), and T4SA (1.17&#xa0;million tweets). Experimental results demonstrate that CAT-KD achieves an accuracy of 94.83% and an F1-score of 94.12%, outperforming baseline models with improvements of up to 0.62% in accuracy and 2.46% in F1-score on Twitter datasets, confirming its effectiveness in handling complex sentiment patterns.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Integrating crossbred node feature encoding with tree and analogous graph attention networks for comprehensive twitter analysis

  • 期刊论文

  • 发表日期：

    2026年7月30日

查看全文
  • Dharshini B.S,
  • Rakesh Kumar Mahendran

摘要

Twitter sentiment analysis plays a vital role in understanding public opinion, emotions, and emerging trends from large-scale user-generated social media content. However, existing research faces limitations such as handling noisy data, understanding sarcasm, and effectively processing the nuanced and informal language often used on the platform. This research aims to accurately gauge public sentiment on Twitter through an innovative approach. The proposed model, Crossbred Analogous Tree-Knowledge Deep (CAT-KD), is composed of five key components for comprehensive sentiment analysis. The Crossbred Node Feature Encoder (CNFD) captures intricate node representations, while the Arbor Graph Attention Network (A-GAT) models the hierarchical structure of Twitter post trees. Simultaneously, the Contextual Graph Attention Network (C-GAT) focuses on user graph modeling to capture user behavior and interactions. A post-user fusion layer facilitates seamless integration between post and user information, enhancing contextual understanding. Additionally, the Knowledge Aware Attention Network (K-A2N) further refines the model’s performance by incorporating external knowledge. The final sentiment classification is conducted using a Deep Neural Network (DNN) classifier, ensuring accurate and nuanced sentiment analysis. Furthermore, to prove the efficacy of our proposed CAT-KD model, the model is evaluated on benchmark datasets including Sentiment140 (1.6 million tweets), Airline Twitter dataset (14,641 tweets), and T4SA (1.17 million tweets). Experimental results demonstrate that CAT-KD achieves an accuracy of 94.83% and an F1-score of 94.12%, outperforming baseline models with improvements of up to 0.62% in accuracy and 2.46% in F1-score on Twitter datasets, confirming its effectiveness in handling complex sentiment patterns.