Linqiu Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Linqiu Liu has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
Jian Wang
Independent Researcher Hangzhou Zhejiang China
1 shared publicationJin Qian
State Grid Zhejiang electric power company, Hangzhou, Zhejiang, China
1 shared publicationJun Luo
State Grid Zhejiang electric power company, Hangzhou, Zhejiang, China
1 shared publicationLibing Xu
State Grid Zhejiang electric power company, Hangzhou, Zhejiang, China
1 shared publicationLingzhi Wang
Department of Computer Science Northwestern University Evanston Illinois USA
1 shared publication- Qingyang Zeng
Zhejiang University, Hangzhou, Zhejiang, China
1 shared publication
Latest Publications
- Open Access
- 8 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 71418-71428
Closing the Gap: A Comprehensive Benchmark of Static and Dynamic WebShell Detection Systems
With the proliferation of web technologies, web servers have become high-value targets for cyberattacks. A key attack vector involves the deployment of WebShells—malicious scripts that enable unauthorized remote access. Despite growing research on WebShell detection, most studies focus on novel methods and offer limited evaluations against existing open-source tools. To the best of our knowledge, this work presents the first systematic benchmarking and analysis of open-source WebShell detectors. We evaluate both static and dynamic detection tools using a dataset of 2,188 diverse PHPWebShells, examining tool performance across accuracy, false positives, and runtime efficiency. Furthermore, we propose a novel static detection approach that leverages Large Language Models (LLMs) to automatically generate regex-based detection rules. Our LLM-generated ruleset achieves the highest detection rate among all evaluated static methods at 92%, while maintaining an exceptionally low false positive rate of 0.49%. These results demonstrate the potential of LLMs as an effective aid in crafting high-quality detection signatures. Through our comprehensive evaluation, we provide actionable insights and recommendations tailored to real-world usage scenarios, empowering developers and security practitioners to select and build appropriate detection solutions based on practical constraints and goals.
- 8 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 71418-71428
Open AccessClosing the Gap: A Comprehensive Benchmark of Static and Dynamic WebShell Detection Systems
With the proliferation of web technologies, web servers have become high-value targets for cyberattacks. A key attack vector involves the deployment of WebShells—malicious scripts that enable unauthorized remote access. Despite growing research on WebShell detection, most studies focus on novel methods and offer limited evaluations against existing open-source tools. To the best of our knowledge, this work presents the first systematic benchmarking and analysis of open-source WebShell detectors. We evaluate both static and dynamic detection tools using a dataset of 2,188 diverse PHPWebShells, examining tool performance across accuracy, false positives, and runtime efficiency. Furthermore, we propose a novel static detection approach that leverages Large Language Models (LLMs) to automatically generate regex-based detection rules. Our LLM-generated ruleset achieves the highest detection rate among all evaluated static methods at 92%, while maintaining an exceptionally low false positive rate of 0.49%. These results demonstrate the potential of LLMs as an effective aid in crafting high-quality detection signatures. Through our comprehensive evaluation, we provide actionable insights and recommendations tailored to real-world usage scenarios, empowering developers and security practitioners to select and build appropriate detection solutions based on practical constraints and goals.
- Open Access
- 1 January 2026
- Published by Institution of Engineering and Technology (IET) in Digital Twins and Applications
- Vol. 3 (1)
PARIS: A Practical, Adaptive Trace‐Fetching and Real‐Time Malicious Behaviour Detection System
The escalating sophistication of cyber‐attacks and the widespread utilisation of stealth tactics have led to significant security threats globally. As digital twin systems are increasingly deployed on endpoints and edge nodes, they have become attractive targets for sophisticated malware attacks. However, existing endpoint detection and response (EDR) solutions often incur substantial runtime overhead, making real‐time deployment in resource‐constrained digital twin environments challenging. Nevertheless, the existing static detection methods exhibit limited coverage, and traditional dynamic monitoring approaches encounter challenges in bypassing evasion techniques. Thus, it has become imperative to implement nuanced and dynamic analysis to achieve precise behaviour detection in real time. There are two pressing concerns with current dynamic malware behaviour detection solutions. First, data collection and processing introduce significant overhead, limiting real‐time deployment on end hosts. Second, these approaches often treat malware as a single entity, overlooking varied behaviours within one instance. To address these gaps, we propose PARIS, an adaptive trace‐fetching, lightweight and real‐time malicious behaviour detection system. PARIS monitors malicious behaviour with Event Tracing for Windows (ETW) and selectively collects maliciousness‐related APIs or call stacks, significantly reducing collection overhead. As a result, it can monitor a wider range of APIs and detect more intricate attack behaviours. We implemented a prototype of PARIS and evaluated system overhead, comparative behaviour recognition accuracy and the impact of different models and parameters. Results show that PARIS reduces over 98.8% of data compared to raw ETW traces, thereby lowering memory, bandwidth and CPU overhead whilst maintaining a detection accuracy of 93.6%. Furthermore, breakdown evaluation shows that 80% of the memory and bandwidth savings and the complete CPU usage reduction come from our adaptive trace‐fetching collector. The lightweight design of PARIS makes it particularly suitable for protecting digital twin endpoints and edge nodes, where computational resources are limited and real‐time responsiveness is critical.
- 1 January 2026
- Published by Institution of Engineering and Technology (IET) in Digital Twins and Applications
- Vol. 3 (1)
Open AccessPARIS: A Practical, Adaptive Trace‐Fetching and Real‐Time Malicious Behaviour Detection System
The escalating sophistication of cyber‐attacks and the widespread utilisation of stealth tactics have led to significant security threats globally. As digital twin systems are increasingly deployed on endpoints and edge nodes, they have become attractive targets for sophisticated malware attacks. However, existing endpoint detection and response (EDR) solutions often incur substantial runtime overhead, making real‐time deployment in resource‐constrained digital twin environments challenging. Nevertheless, the existing static detection methods exhibit limited coverage, and traditional dynamic monitoring approaches encounter challenges in bypassing evasion techniques. Thus, it has become imperative to implement nuanced and dynamic analysis to achieve precise behaviour detection in real time. There are two pressing concerns with current dynamic malware behaviour detection solutions. First, data collection and processing introduce significant overhead, limiting real‐time deployment on end hosts. Second, these approaches often treat malware as a single entity, overlooking varied behaviours within one instance. To address these gaps, we propose PARIS, an adaptive trace‐fetching, lightweight and real‐time malicious behaviour detection system. PARIS monitors malicious behaviour with Event Tracing for Windows (ETW) and selectively collects maliciousness‐related APIs or call stacks, significantly reducing collection overhead. As a result, it can monitor a wider range of APIs and detect more intricate attack behaviours. We implemented a prototype of PARIS and evaluated system overhead, comparative behaviour recognition accuracy and the impact of different models and parameters. Results show that PARIS reduces over 98.8% of data compared to raw ETW traces, thereby lowering memory, bandwidth and CPU overhead whilst maintaining a detection accuracy of 93.6%. Furthermore, breakdown evaluation shows that 80% of the memory and bandwidth savings and the complete CPU usage reduction come from our adaptive trace‐fetching collector. The lightweight design of PARIS makes it particularly suitable for protecting digital twin endpoints and edge nodes, where computational resources are limited and real‐time responsiveness is critical.