Asm2Seq: Explainable Assembly Code Functional Summary Generation for Reverse Engineering and Vulnerability Analysis

Reverse engineering is the process of understanding inner working a software system without having source code. It critical for firmware security validation, vulnerability research, and malware analysis. However, it often requires significant amount manual effort. Recently, data-driven solutions were proposed to reduce effort by identifying code clones on assembly or level. analysts still have understand matched develop an functionality, assumed that such candidate always exists. This research bridges gap introducing problem summarization. Given as input, we propose machine-learning-based can produce human-readable summarizations functionalities in context We generate first function summary dataset leverage encoder-decoder architecture. With attention mechanism, possible what aspects had largest impact generating summary. Our experiment shows solution achieves high accuracy Bilingual Evaluation Understudy (BLEU) score. Finally, performed case studies real-life CVE cases better method’s performance practical implications.