Scholarlink

<p>Online reviews play a key role in deciding purchases by the customers. However, the proliferation of fake reviews has made it hard to differentiate between real and fake reviews. Manual classification of reviews is subjective, and traditional methods are limited in accuracy. To address this issue, we propose a novel Parallel Hybrid k-means and Henry Gas Solubility Optimization (H-KHGSO) algorithm for detecting fake reviews in social media. The H-KHGSO algorithm capitalizes on the strengths of k-means clustering to initialize the initial population and Henry Gas Solubility Optimization (HGSO) to find optimal cluster heads. Leveraging Hadoop’s distributed computing capabilities, our parallel approach enhances computational efficiency and accelerates the detection process. To validate the effectiveness of H-KHGSO, experiments on three distinct spam review datasets, including SSR, MR, and YHRR, were investigated against state-of-the-art clustering methods, including PSO, DE, GA, CS, GWOK, and k-means. The comparative analysis confirmed the superiority of the H-KHGSO algorithm.</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 Novel Parallel Hybrid k-Means and HGSO Based Approach for Detecting Fake Reviews in Social Media

  • 期刊论文

  • 发表日期：

    2025年5月3日

查看全文
  • Rajendra Purohit,
  • K. R. Chowdhary,
  • S. D. Purohit,
  • Raju Pal,
  • Harish Sharma

摘要

Online reviews play a key role in deciding purchases by the customers. However, the proliferation of fake reviews has made it hard to differentiate between real and fake reviews. Manual classification of reviews is subjective, and traditional methods are limited in accuracy. To address this issue, we propose a novel Parallel Hybrid k-means and Henry Gas Solubility Optimization (H-KHGSO) algorithm for detecting fake reviews in social media. The H-KHGSO algorithm capitalizes on the strengths of k-means clustering to initialize the initial population and Henry Gas Solubility Optimization (HGSO) to find optimal cluster heads. Leveraging Hadoop’s distributed computing capabilities, our parallel approach enhances computational efficiency and accelerates the detection process. To validate the effectiveness of H-KHGSO, experiments on three distinct spam review datasets, including SSR, MR, and YHRR, were investigated against state-of-the-art clustering methods, including PSO, DE, GA, CS, GWOK, and k-means. The comparative analysis confirmed the superiority of the H-KHGSO algorithm.