Skip to content

Latest commit

 

History

History
24 lines (14 loc) · 802 Bytes

README.md

File metadata and controls

24 lines (14 loc) · 802 Bytes

SBR_Experiment

The purpose of this experiment is to find unlabelled security bugs among thousands of bug reports via active learning and ranking. This experiment simulates the review process under the specified SBR(Security Bugs Report) recall rate.

Structure

  • encoder/: Different method to transform words into vectors
  • model/: Training different models and predict the probability of SBR
  • experiment/: Experiments to run
  • RQ*.py: Runner

Configure

All configurations are in BaseExperiment class, and we use ExperimentFactory to build its instance from a list of configurations dictionary.

Running

pip install numpy pandas sklearn alive_progress
python3 RQ*.py