Artifact of SynthDroid
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
How to reproduce our results Code: sequence extraction code: Code/AFE rule synthesis code: Code/SynthDroid. Dataset: androzoo and Mystique are in Dataset. Rule: all synthesized rules under different hyperparameter settings can be found in Rule. Sequnce: all API sequence extracted from APK can be found in Sequence. Interpretability Evaluation Result: all basis are listed Detection Basis, report file shows the statistics of the results, basis file shows the rules used for detection and corresponding API sequence, automated evaluation results show the metadata evaluated by LLMs, human evaluation results show the metadata evaluated by experts. How to extract sequence from APK ? # APK_DIR contains all Mystique apks, FEATURE_DIR to save all sequence extracted from apk ./run_myst.sh * ${APK_DIR} ${FEATURE_DIR} # RES_DIR save all sequence filter from FEATURE_DIR, # This command removes redundant characters such as unnecessary Spaces, line breaks, consecutive same API calls, etc # Character * means no obfuscation python post_process.py --option 8 --n_res ${RES_DIR} --n_feature ${FEATURE_DIR} --obfuscated_type * # DATASET_DIR contains myst.json which stores the partitioning information of the myst # MALWARE_SEQ_DIR to save all extracted malicious sequence based on the ground truth # GROUND_TRUTH contains all ground truth given by Mystique dataset # This command extract all malicious API sequence annotated by Mystique python post_process.py --option 14 --n_res ${RES_DIR} --dataset ${DATASET_DIR} --seq_dir ${MALWARE_SEQ_DIR} --ground_truth ${GROUND_TRUTH} # This command merge all API sequence into one file, which can be used to benign sequence selection python post_process.py --option 15 --n_res ${RES_DIR} --n_feature ${FEATURE_DIR} --obfuscated_type * # RATE means Benign to Malicious Sequence Rate # This command pair each malicious sequence with benign sequence ./run_benign.sh ${RES_DIR} ${MALWARE_SEQ_DIR} ${RATE} # copy myst_feature_${RATE}.jsonc to SynthDroid/data directory How to synthesize rules ? cd SynthDroid # generate fact for rule synthesis # -i to specify input folder, -o to specify output folder # -t to specify malware type, default: myst, -r to specify BMR rate python3 scripts/extract.py -i data -o fact -t myst -r 1 # rule synthesis, -o to specify output folder ./main_train -o fact # Synthesize rules are in fact/myst/logic/malicious directory, which can be used for malware detection.
Fields of Study & Subjects
Advanced Malware Detection Techniques
Software Engineering Research
Security and Verification in Computing
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 17 June 2026
- Language
- EN