信息工程大学学报2026，Vol.27Issue(4)：443-449,7.DOI:10.3969/j.issn.1671-0673.2026.04.009
二进制代码语义恢复技术研究综述
A Survey on Binary Code Semantic Recovery Technology
孙佳佳 1李泽潭 1付艳华 1朱肖雅 1尹小康 1杨启超 1刘胜利1
作者信息
摘要
Abstract
Binary code semantic recovery technology aims to reconstruct high-level semantic informa-tion such as functions,variables,and data structures in the absence of source code and debugging in-formation,to support malicious code understanding,vulnerability localization,and software component identification.Due to symbol stripping,compiler optimizations,and cross-platform differences,there coexist problems of information loss and morphological perturbation between high-level semantics and low-level implementations,leading to sparse information,heterogeneous representations,and difficul-ties in characterizing global dependencies.Representative studies are systematically reviewed in this paper.The existing methods are categorized into 3 paradigms,which include heuristic rule matching,deep learning discriminative inference,and large language model-driven generative recovery,and their evolution is traced.The development trends of mainstream datasets and evaluation metrics are further summarized.Finally,the real challenges faced by research in this field are analyzed and future re-search directions are anticipated.关键词
二进制代码/语义恢复/逆向工程/代码理解/深度学习/大语言模型Key words
binary code/semantic recovery/reverse engineering/code comprehension/deep learn-ing/large language model分类
信息技术与安全科学引用本文复制引用
孙佳佳,李泽潭,付艳华,朱肖雅,尹小康,杨启超,刘胜利..二进制代码语义恢复技术研究综述[J].信息工程大学学报,2026,27(4):443-449,7.
