Security Relevant Methods of Android’s API Classification: A Machine Learning Empirical Evaluation

To have all the sub-packages, run:

$ git clone --recursive https://github.com/walbermr/android-srm-ml-evaluation.git

Install required packages:

$ apt-get -y upgrade && apt-get install -y python3 wget git g++ gcc ca-certificates gnupg software-properties-common r-base

Install conda environment:

$ conda env create -f env.yml

Install required R packages:

$ cd preprocessing && conda run -n srm python ./r_packages.py && cd ..

To replicate the results from the paper, you first must download the embedding algorithms checkpoints from the gdrive link and place it in this directory root. After extraction, it should have the following structure:

├── data
│   ├── saved_models

To perform all experiments, run on Linux:

$ ./run_pipeline.sh

On Windows, it is required to use the provided Docker image, as Auto-sklearn is not supported in Windows:

$ ./docker-run.ps1

Usage

To run SuSi, you just need to run the command below, if you get heap memory error, add -Xmx1024M . The objective is that SuSi will extract all the features needed to run the machine learning algorithms generating a static dataset instead of using the android.jar files to extract every time the classification is needed.

$ java -cp lib/weka.jar:soot.jar:soot-infoflow.jar:soot-infoflow-android.jar:bin de.ecspride.sourcesinkfinder.SourceSinkFinder android-12.jar permissionMethodWithLabel.pscout out.pscout

To run on Windows use:

$ java -cp "lib/weka.jar;soot.jar;soot-infoflow.jar;soot-infoflow-android.jar;bin" de.ecspride.sourcesinkfinder.SourceSinkFinder android-12.jar permissionMethodWithLabel.pscout out.pscout

To build SuSi you can use Eclipse directly or use:

$ javac -d bin/ -cp lib/weka.jar:soot.jar:soot-infoflow.jar:soot-infoflow-android.jar -sourcepath src/ src/de/ecspride/sourcesinkfinder/SourceSinkFinder.java

To build on windows use:

$ javac -d bin/ -cp "lib/weka.jar;soot.jar;soot-infoflow.jar;soot-infoflow-android.jar" -sourcepath src/ src/de/ecspride/sourcesinkfinder/SourceSinkFinder.java

Troubleshooting

If you get heap memory error, add the flag -Xmx1024M while executing the java commands.