BinGold: Towards robust binary analysis by extracting the semantics of binary code as semantic flow graphs (SFGs)
Bibliographic record
Abstract
Binary analysis is useful in many practical applications, such as the detection of malware or vulnerable software components. However, our survey of the literature shows that most existing binary analysis tools and frameworks rely on assumptions about specific compilers and compilation settings. It is well known that techniques such as refactoring and light obfuscation can significantly alter the structure of code, even for simple programs. Applying such techniques or changing the compiler and compilation settings can significantly affect the accuracy of available binary analysis tools, which severely limits their practicability, especially when applied to malware. To address these issues, we propose a novel technique that extracts the semantics of binary code in terms of both data and control flow. Our technique allows more robust binary analysis because the extracted semantics of the binary code is generally immune from light obfuscation, refactoring, and varying the compilers or compilation settings. Specifically, we apply data-flow analysis to extract the semantic flow of the registers as well as the semantic components of the control flow graph, which are then synthesized into a novel representation called the semantic flow graph (SFG). Subsequently, various properties, such as reflexive, symmetric, antisymmetric, and transitive relations, are extracted from the SFG and applied to binary analysis. We implement our system in a tool called BinGold and evaluate it against thirty binary code applications. Our evaluation shows that BinGold successfully determines the similarity between binaries, yielding results that are highly robust against light obfuscation and refactoring. In addition, we demonstrate the application of BinGold to two important binary analysis tasks: binary code authorship attribution, and the detection of clone components across program executables. The promising results suggest that BinGold can be used to enhance existing techniques, making them more robust and practical.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.005 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".