Scholarlink

<p>Sentiment polarity is a key aspect of analyzing and determining the opinions and emotions of the user that are obtained in the form of text sentiments, which are expressed as positive, negative, or neutral sentiment forms. Recently, sentiment polarity detection has created massive growth and attention in online platforms. However, the existing methods fails to learn the complementarity between external knowledge and linguistic information hidden in the texts, resulting in suboptimal results. Consequently, the research proposes the Incremental self-configuring knowledge graph-based Bidirectional Long Short-Term Memory Model with Hierarchical Pounce Algorithm (HiPA-ISKTM) to detect the sentiment polarity from the text input. In addition, the proposed method exploited the combination of TF-IDF, Hybrid word2vec features, and Graph-based features to represent the text features and applied to the HiPA-ISKTM model, leading to improved detection accuracy. Hierarchical Pounce Algorithm (HiPA) harnessing the unique traits of harris hawks and grey wolf is used to fine-tune the HiPA-ISKTM model and generate significant output. Extensive experiments show that the HiPA-ISKTM method provides excellent results, attaining accuracy, specificity, precision, recall, and F1-score of 94.67%, 94.62%, 94.64%,94.72%, and 94.68% respectively, on training with the restaurant review dataset.</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

Sentiment Polarity Classification Using Incremental Self-Configuring Knowledge Graph-Based Bidirectional Long Short Term Memory Classifier with Hierarchical Pounce Optimization

  • 期刊论文

  • 发表日期：

    2025年11月26日

查看全文
  • Tanya Shruti,
  • Kamlesh Lakhwani

摘要

Sentiment polarity is a key aspect of analyzing and determining the opinions and emotions of the user that are obtained in the form of text sentiments, which are expressed as positive, negative, or neutral sentiment forms. Recently, sentiment polarity detection has created massive growth and attention in online platforms. However, the existing methods fails to learn the complementarity between external knowledge and linguistic information hidden in the texts, resulting in suboptimal results. Consequently, the research proposes the Incremental self-configuring knowledge graph-based Bidirectional Long Short-Term Memory Model with Hierarchical Pounce Algorithm (HiPA-ISKTM) to detect the sentiment polarity from the text input. In addition, the proposed method exploited the combination of TF-IDF, Hybrid word2vec features, and Graph-based features to represent the text features and applied to the HiPA-ISKTM model, leading to improved detection accuracy. Hierarchical Pounce Algorithm (HiPA) harnessing the unique traits of harris hawks and grey wolf is used to fine-tune the HiPA-ISKTM model and generate significant output. Extensive experiments show that the HiPA-ISKTM method provides excellent results, attaining accuracy, specificity, precision, recall, and F1-score of 94.67%, 94.62%, 94.64%,94.72%, and 94.68% respectively, on training with the restaurant review dataset.