Scholarlink

<p>In the dark web, fingerprinting attacks utilize traffic analysis to detect the websites that are visited by the users and affect the user’s privacy. The adversarial web user prediction is essential to improve data protection and mitigate malicious activities. However, the adversarial web user prediction is complex due to the website inquiries forced by defense systems. For solving such issues, the Siamese Convolutional ResNeXt (SCResNeXt) is proposed for adversarial dark web user prediction based on dark webpage profiling, and fingerprinting extraction. Dark web classification and crawling are the prime processes. The Onion Router (TorBot) with keywords is utilized for dark web crawling. For classifying the dark web, the Bayesian Hierarchical Neural Attention Harmonic Network (BHNAHN) is utilized. The graph model is employed to perform the dark web profiling, and the fingerprints are extracted from the Operating Systems (OS), browser, browsing sequence, caching, cookies, application layer protocol, and network conditions. The web user adversary is performed using the SCResNeXt based on dark web profiling and the extracted features. Furthermore, the SCResNeXt-based adversarial user prediction in the dark web achieved the optimal accuracy, True Positive Rate (TPR), and True Negative Rate (TNR) of 92.69%, 91.24%, and 93.24%.</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

Siamese convolutional resnext and graph model-based profiling for dark webpage fingerprinting and adversary prediction

  • 期刊论文

  • 发表日期：

    2026年2月26日

查看全文
  • Vinod Babu Bollikonda,
  • K. V. D. Kiran

摘要

In the dark web, fingerprinting attacks utilize traffic analysis to detect the websites that are visited by the users and affect the user’s privacy. The adversarial web user prediction is essential to improve data protection and mitigate malicious activities. However, the adversarial web user prediction is complex due to the website inquiries forced by defense systems. For solving such issues, the Siamese Convolutional ResNeXt (SCResNeXt) is proposed for adversarial dark web user prediction based on dark webpage profiling, and fingerprinting extraction. Dark web classification and crawling are the prime processes. The Onion Router (TorBot) with keywords is utilized for dark web crawling. For classifying the dark web, the Bayesian Hierarchical Neural Attention Harmonic Network (BHNAHN) is utilized. The graph model is employed to perform the dark web profiling, and the fingerprints are extracted from the Operating Systems (OS), browser, browsing sequence, caching, cookies, application layer protocol, and network conditions. The web user adversary is performed using the SCResNeXt based on dark web profiling and the extracted features. Furthermore, the SCResNeXt-based adversarial user prediction in the dark web achieved the optimal accuracy, True Positive Rate (TPR), and True Negative Rate (TNR) of 92.69%, 91.24%, and 93.24%.