Scholarlink

<p>Online debates are often reduced to binary categories, such as favor, against, and none, by using traditional opinion analysis methods. This oversimplification fails to capture the nuances of opinions, such as the intensity of feelings or the degree to which a stance varies in favor or against. To address this limitation, a new method, SFBiLT, is proposed, which leverages a fuzzy logic-based BiLSTM model optimized using the Teacher-Learning-Based Optimization (TLBO) technique to provide more nuanced categories for opinions. This approach enables a more accurate understanding of people’s views, particularly in the context of ordered-class classification. The performance of the proposed model is evaluated on two new datasets and the benchmark SemEval-2016 TaskA dataset. The SFBiLT model demonstrates significant improvements over State-of-the-Art (SOTA) models, achieving accuracy and F1-score improvements of 6.11% and 0.0824 on Dataset 1, 8.44% and 0.0833 on Dataset 2, and 1.46% and 0.0356 on the SemEval-2016 TaskA benchmark.</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

Enhanced Stance Detection Using Fuzzy Logic-based BiLSTM with TLBO Optimization

  • 期刊论文

  • 发表日期：

    2026年6月26日

查看全文
  • Km Poonam,
  • Tene Ramakrishnudu

摘要

Online debates are often reduced to binary categories, such as favor, against, and none, by using traditional opinion analysis methods. This oversimplification fails to capture the nuances of opinions, such as the intensity of feelings or the degree to which a stance varies in favor or against. To address this limitation, a new method, SFBiLT, is proposed, which leverages a fuzzy logic-based BiLSTM model optimized using the Teacher-Learning-Based Optimization (TLBO) technique to provide more nuanced categories for opinions. This approach enables a more accurate understanding of people’s views, particularly in the context of ordered-class classification. The performance of the proposed model is evaluated on two new datasets and the benchmark SemEval-2016 TaskA dataset. The SFBiLT model demonstrates significant improvements over State-of-the-Art (SOTA) models, achieving accuracy and F1-score improvements of 6.11% and 0.0824 on Dataset 1, 8.44% and 0.0833 on Dataset 2, and 1.46% and 0.0356 on the SemEval-2016 TaskA benchmark.