Vulnerability Management Chaining: An Integrated Framework for Efficient Cybersecurity Risk Prioritization

Naoyuki Shimizu, Masaki Hashimoto · 2025

Read Vulnerability Management Chaining: An Integrated Framework for Efficient Cybersecurity Risk Prioritization by Naoyuki Shimizu, Masaki Hashimoto (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-01220.txt.

Introduction

Naoyuki Shimizu and Masaki Hashimoto presented a framework in 2025 for ranking which software flaws to fix first. The count of known flaws grows fast, and single scoring systems either flag too many as urgent or miss flaws that attackers are already using.

Their method chains three public systems in a decision tree. A first filter keeps flaws known to be exploited, or likely to be, and a second step grades severity, which lets teams safely downgrade the rest. Tested on more than 28,000 real flaws, it was eighteen times more efficient and kept most coverage. Urgent work fell by about ninety-five percent. All data used are free, so any team can adopt it.

Catalog record

CREATORNaoyuki Shimizu, Masaki Hashimoto
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE65 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-01220

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Vulnerability Management Chaining: An Integrated Framework for Efficient Cybersecurity Risk Prioritization Naoyuki Shimizu, Masaki Hashimoto Abstract Cybersecurity teams face an overwhelming vulnerability crisis: with 25,000+ new CVEs disclosed annually, traditional CVSS-based prioritization requires addressing approximately 57% of all vulnerabilities while correctly identifying only 20% of those actually exploited. We propose Vulnerability Management Chaining, an integrated decision tree framework combining historical exploitation evidence (KEV), predictive threat modeling (EPSS), and technical impact assessment (CVSS) to transform vulnerability management from reactive patching to strategic threat-driven prioritization. Experimental validation using 28,377 real-world vulnerabilities demonstrates approximately 14-18 fold efficiency improvements while maintaining 85+ percent coverage of actual threats. Organizations can reduce urgent remediation workload by approximately 95% (from 16,000 to 850 vulnerabilities). The integration identifies 57 additional exploited vulnerabilities that neither KEV nor EPSS captures individually. Our framework uses exclusively open-source data, democratizing advanced vulnerability management regardless of budget or expertise. This research establishes the first empirically validated methodology for systematic vulnerability management integration, with immediate applicability across diverse organizational contexts. Keywords Vulnerability Management \cdot CVSS \cdot EPSS \cdot KEV \cdot Threat Intelligence \cdot Cybersecurity Risk Assessment \cdot Decision Trees \cdot Security Prioritization 1 Introduction The exponential growth of cybersecurity threats and the increasing complexity of modern IT infrastructures have made effective vulnerability management one of the most critical challenges facing organizations today. As software vulnerabilities continue to be discovered and disclosed at an unprecedented rate, security teams face challenges in allocating their limited resources effectively to address the most critical risks. 1.1 The Vulnerability Management Challenge Modern organizations face an increasingly difficult vulnerability landscape. The number of Common Vulnerabilities and Exposures (CVE) published annually has grown substantially, with data from the National Vulnerability Database showing consistent year-over-year increases [ 1 ] . This growth trend shows no signs of slowing, placing enormous pressure on security teams to evaluate, prioritize, and remediate an ever-expanding attack surface. Compounding this challenge is the fundamental asymmetry between attackers and defenders. While attackers need only to find and exploit a single vulnerability to achieve their objectives, defenders must identify and address all potential security weaknesses across their entire infrastructure. This asymmetry, combined with widespread shortages of cybersecurity professionals, has created a situation where traditional comprehensive approaches to vulnerability management are no longer sustainable. The shift toward remote work following the COVID-19 pandemic has further exacerbated these challenges. Organizations now expose more services to the internet than ever before, with VPN gateways, remote desktop services, and cloud applications becoming primary attack vectors. High-profile ransomware attacks have repeatedly demonstrated that attackers frequently exploit known vulnerabilities in internet-facing services as their initial access vector. 1.2 Limitations of Current Approaches The Common Vulnerability Scoring System (CVSS) has served as the de facto standard for vulnerability management since its introduction. CVSS provides a standardized method for rating the severity of security vulnerabilities based on their technical characteristics, including attack complexity, required privileges, and potential impact on confidentiality, integrity, and availability. However, CVSS-based vulnerability management has come under increasing criticism for its fundamental inefficiency. Research has consistently shown that only a small fraction of high-CVSS vulnerabilities are ever exploited in practice. Studies indicate that fewer than 20% of vulnerabilities rated as "Critical" or "High" by CVSS are actually observed being exploited in real-world attacks [ 2 ] , yet these high-severity ratings are assigned to approximately 57% of all published vulnerabilities [ 3 ] . This mismatch creates a significant operational burden, leading researchers to question whether it’s "time to change the CVSS" [ 3 ] and describing the system as "ubiquitous and broken" [ 4 ] . Security teams following CVSS-based prioritization must address thousands of vulnerabilities that pose little real-world risk. Simultaneously, they may overlook vulnerabilities that attackers actively exploit but receive lower CVSS scores due to complex attack vectors or authentication requirements. Government agencies have recognized these limitations, with CISA implementing binding operational directives that mandate specific remediation timelines [ 5 , 6 ] . These directives represent a shift toward evidence-based vulnerability management, requiring federal agencies to prioritize vulnerabilities with confirmed exploitation evidence regardless of their CVSS scores. Furthermore, CVSS was designed to measure technical severity rather than risk in the broader sense. The Forum of Incident Response and Security Teams (FIRST), which maintains CVSS, explicitly states that CVSS base scores should not be used alone for risk assessment and recommends incorporating additional contextual factors [ 7 ] . However, in practice, most organizations rely primarily on CVSS base scores due to the complexity and resource requirements of implementing more comprehensive risk assessment frameworks. 1.3 Emerging Threat Intelligence Approaches Recognizing the limitations of traditional vulnerability management, several threat intelligence-based approaches have emerged to provide more effective prioritization. In 2021, the U.S. Cybersecurity and Infrastructure Security Agency (CISA) introduced the KEV catalog, which identifies vulnerabilities that have been observed being exploited in real-world attacks. KEV represents a paradigm shift from theoretical severity assessment to evidence-based prioritization, requiring federal agencies to remediate KEV-listed vulnerabilities within specified timeframes regardless of their CVSS scores. Developed by researchers and maintained by FIRST, EPSS uses machine learning to predict the probability that a vulnerability will be exploited within the next 30 days. EPSS incorporates numerous data sources, including threat intelligence feeds, proof-of-concept availability, and social media discussions, to generate probabilistic exploitation scores. While both KEV and EPSS represent significant advances over pure CVSS-based prioritization, each has important limitations. KEV provides high-confidence information about past exploitation but cannot predict future threats. EPSS offers forward-looking predictions but may miss vulnerabilities that are already being exploited. Additionally, EPSS is explicitly not recommended for use as a standalone risk assessment tool, particularly for vulnerabilities where exploitation has already been confirmed. 1.4 Research Motivation and Objectives This research is motivated by the observation that existing vulnerability management approaches—whether based on technical severity (CVSS), confirmed exploitation (KEV), or exploitation prediction (EPSS)—each capture important but incomplete aspects of vulnerability risk. Organizations need a practical framework that combines the strengths of these approaches while mitigating their individual limitations. Our primary research objective is to develop and validate an integrated vulnerability management methodology that improves efficiency over traditional CVSS-based approaches by focusing resources on vulnerabilities with confirmed or predicted exploitation, maintains comprehensive coverage to ensure that critical vulnerabilities are not overlooked, leverages existing open-source data to enable broad adoption without requiring proprietary threat intelligence, and provides actionable guidance through clear decision criteria rather than abstract scoring systems. 1.5 Proposed Approach and Contributions We propose Vulnerability Management Chaining, a decision tree-based framework that systematically combines CVSS, EPSS, and KEV to achieve more effective vulnerability prioritization. Our approach employs a two-stage evaluation process: first assessing threat likelihood using KEV and EPSS data, then evaluating vulnerability characteristics using CVSS to determine appropriate response priorities. The key insight underlying our approach is that vulnerabilities should be prioritized based on both threat likelihood and vulnerability characteristics. A vulnerability that is being actively exploited (high threat) but requires significant privileges and has limited impact (low vulnerability severity) may warrant different treatment than one with theoretical high impact but no evidence of exploitation. Our primary contributions include an integrated framework design that develops a practical decision tree framework systematically combining three major vulnerability management approaches, addressing the limitations of using any single method

Read the complete text