Scholarlink

<p>As people increasingly express opinions, offer feedback, and share suggestions on websites, e-forums, and blogs, consumers have come to rely heavily on online product reviews before making purchases or using services. Some spammers deliberately manipulate reviews to either enhance or discredit products, leading consumers to make misguided decisions based on these misleading reviews. To address this issue, a hybrid approach combining TF-IDF (term frequency-inverse document frequency) with the shuffled frog leaping algorithm (SFLA) is proposed. This approach aims to reduce the high dimensionality of feature sets and select optimized subsets of features. Finally, the sentiments from online reviews are classified into genuine and misleading by using Extreme Gradient Boosting (XGBoost) classifier. By enhancing review credibility, the proposed approach helps customers identify authentic feedback and improves classification accuracy. The performance of the proposed approach is assessed and validated with existing studies based on accuracy (%), precision (%) and recall (%). The results indicate that this hybrid feature selection approach achieves an optimized subset of features, leading to superior classification accuracy.</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

A semantic driven model for extraction of text using TF-IDF, SFLA and XGBoost

  • 期刊论文

  • 发表日期：

    2025年5月13日

查看全文
  • R. B. Madhumala,
  • B. Vineetha,
  • M. Rajani Shree,
  • Riya Sanjesh,
  • B. R. Charanraj

摘要

As people increasingly express opinions, offer feedback, and share suggestions on websites, e-forums, and blogs, consumers have come to rely heavily on online product reviews before making purchases or using services. Some spammers deliberately manipulate reviews to either enhance or discredit products, leading consumers to make misguided decisions based on these misleading reviews. To address this issue, a hybrid approach combining TF-IDF (term frequency-inverse document frequency) with the shuffled frog leaping algorithm (SFLA) is proposed. This approach aims to reduce the high dimensionality of feature sets and select optimized subsets of features. Finally, the sentiments from online reviews are classified into genuine and misleading by using Extreme Gradient Boosting (XGBoost) classifier. By enhancing review credibility, the proposed approach helps customers identify authentic feedback and improves classification accuracy. The performance of the proposed approach is assessed and validated with existing studies based on accuracy (%), precision (%) and recall (%). The results indicate that this hybrid feature selection approach achieves an optimized subset of features, leading to superior classification accuracy.