Scholarlink

<p>Vulnerabilities in digital systems present an ever-evolving threat landscape of cyber risk, leading to unauthorized access, execution of arbitrary code, denial of service, and disclosure of sensitive information. Our study proposes text mining and deep learning-based Vulnerability Assessment and Risk Management (VARM) model to assess, quantify, and mitigate the cyber risk perpetuated by vulnerabilities. Drawing from the Protection Motivation Theory and the Cyber Kill Chain, the first module, Vulnerability Risk Assessment, evaluates the risk of cyber-attacks. This module utilizes the LDA topic modeling technique to discern which aspects of the Confidentiality-Integrity-Availability/Authenticity (CIA) security triad are compromised by exploiting different characteristics of vulnerabilities. Simultaneously, the module employs co-occurrence network analysis to comprehend the presence of correlated cyber risks. By integrating and providing this information to the Long Short-Term Memory, a recurrent neural network architecture, the module categorizes cyber risk into three types of cyber-attacks: Distributed Denial of Service, Malware, and other cyber-attacks. The second module, Vulnerability Risk Quantification, quantifies the cyber risk emanating from vulnerabilities in terms of estimated losses. Finally, guided by the Rational Choice Theory and NIST-driven vulnerability management processes, the Vulnerability Risk Mitigation module recommends a comprehensive risk mitigation strategy commensurate with the risk and severity of cyber-attacks. Any residual risk can then be transferred to cyber insurers.</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

Vulnerability Assessment and Risk Management Using Text Mining and Deep Learning Approach

  • 期刊论文

  • 发表日期：

    2025年9月9日

查看全文
  • Swati Jain,
  • Arunabha Mukhopadhyay

摘要

Vulnerabilities in digital systems present an ever-evolving threat landscape of cyber risk, leading to unauthorized access, execution of arbitrary code, denial of service, and disclosure of sensitive information. Our study proposes text mining and deep learning-based Vulnerability Assessment and Risk Management (VARM) model to assess, quantify, and mitigate the cyber risk perpetuated by vulnerabilities. Drawing from the Protection Motivation Theory and the Cyber Kill Chain, the first module, Vulnerability Risk Assessment, evaluates the risk of cyber-attacks. This module utilizes the LDA topic modeling technique to discern which aspects of the Confidentiality-Integrity-Availability/Authenticity (CIA) security triad are compromised by exploiting different characteristics of vulnerabilities. Simultaneously, the module employs co-occurrence network analysis to comprehend the presence of correlated cyber risks. By integrating and providing this information to the Long Short-Term Memory, a recurrent neural network architecture, the module categorizes cyber risk into three types of cyber-attacks: Distributed Denial of Service, Malware, and other cyber-attacks. The second module, Vulnerability Risk Quantification, quantifies the cyber risk emanating from vulnerabilities in terms of estimated losses. Finally, guided by the Rational Choice Theory and NIST-driven vulnerability management processes, the Vulnerability Risk Mitigation module recommends a comprehensive risk mitigation strategy commensurate with the risk and severity of cyber-attacks. Any residual risk can then be transferred to cyber insurers.