Scholarlink

<p>Advancements in data collection, storage, and inference techniques have enabled the creation of massive databases containing sensitive and personal information about individuals. This data is collected by private companies, government agencies, and healthcare institutions for various mutual benefits. The underlying motivations include investigations, mandatory disclosures, surveillance, the right to information, and more. However, even when data is formally anonymized, the dissemination of such information poses significant risks to user privacy due to the potential for data leakage. The development of sophisticated data mining techniques has further amplified concerns about privacy breaches. In our previous work, we proposed a hybrid approach based on the principles of K-anonymity and L-diversity to safeguard user privacy. While effective to an extent, this approach falls short in defending against semantic similarity attacks. To address this limitation, we introduce a novel integrated approach called “Hybrid KLT”, which combines K-anonymity, L-diversity, and T-closeness. This method aims to enhance privacy protection on social networking platforms while maintaining the utility of the published data. We evaluated the proposed approach using a real-world Twitter dataset and benchmarked its performance against a state-of-the-art technique in the field. The results demonstrate that Hybrid KLT is both effective and efficient in preserving user privacy.</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 solution: threat to privacy leakage in online social networks through semantic similarity attacks

  • 期刊论文

  • 发表日期：

    2025年7月29日

查看全文
  • Amardeep Singh,
  • Monika Singh

摘要

Advancements in data collection, storage, and inference techniques have enabled the creation of massive databases containing sensitive and personal information about individuals. This data is collected by private companies, government agencies, and healthcare institutions for various mutual benefits. The underlying motivations include investigations, mandatory disclosures, surveillance, the right to information, and more. However, even when data is formally anonymized, the dissemination of such information poses significant risks to user privacy due to the potential for data leakage. The development of sophisticated data mining techniques has further amplified concerns about privacy breaches. In our previous work, we proposed a hybrid approach based on the principles of K-anonymity and L-diversity to safeguard user privacy. While effective to an extent, this approach falls short in defending against semantic similarity attacks. To address this limitation, we introduce a novel integrated approach called “Hybrid KLT”, which combines K-anonymity, L-diversity, and T-closeness. This method aims to enhance privacy protection on social networking platforms while maintaining the utility of the published data. We evaluated the proposed approach using a real-world Twitter dataset and benchmarked its performance against a state-of-the-art technique in the field. The results demonstrate that Hybrid KLT is both effective and efficient in preserving user privacy.