Summarizing Cross Language Malware Using Unified Code Representation In Web Applications
Year: 2026
Type: article
Abstract: By distributing malicious logic across JavaScript and WebAssembly components, cross-language malware in web applications can evade detectors that rely on language-specific features. Such strategies disconnect malicious semantics across code artifacts, causing detectors that analyze only one language view to miss behavior that is expressed jointly across JavaScript and WebAssembly. To address this challenge, we study the static detection of JavaScript and WebAssembly cross-language malware when the relevant code artifacts are available for inspection. We design a malware detection approach that represents JavaScript and WebAssembly code using a unified code representation and summarizes the resulting graph semantics with graph embeddings. Compared with existing semantics-based detectors and antivirus solutions, our approach achieves strong recall and competitive F1-score under a controlled JavaScript andWebAssembly cross-language transformation setting, making it suitable as a recall-oriented first-stage detector. The results demonstrate the potential of using unified code representation for static detection of JavaScript and WebAssembly cross-language malware in web applications.
Authors: Anis Yusof, Jiahao Liu, Adrian Tang, Zhenkai Liang
Institution: National University of Singapore
Language: English
Subfield: Software
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $2,645
OpenAlex ID: W7225318276