This repository includes the code and experimental data in our paper entitled "DreamLoc: A Deep Relevance Matching based Framework for Bug Localization". It can be used to localize bug files based on bug reports.
- python 3.7.1
- pandas 0.24.2
- gensim 3.7.2
- gitpython 3.1.1
- scikit-learn 0.20.1
- pytorch 1.3.1
- lizard 1.17.3
- numpy 1.17.4
- sent2vec
- GPU with CUDA support is also needed
Install the dependent packages via pip:
$ pip install pandas==0.24.2 gensim==3.7.2 GitPython==3.1.1 scikit-learn==0.20.1 lizard==1.17.3 numpy==1.17.4
Install pytorch according to your environment, see https://pytorch.org/.
Install Sent2vec according to the documentation.
mkdir data
- download the dataset and pre-trained models(password: 8v4r) into the floder
data
cd data
tar xvf data.tar
cd src
- run
python pipeline.py tomcat
- run
python dream_loc.py --project tomcat --rmm_dense_dim 100 --irff_dense_dim 20 --fusion_dense_dim 100 --k_max_pool 3 --lr 0.001
- run
python dream_loc.py --project tomcat --rmm_dense_dim 100 --irff_dense_dim 20 --fusion_dense_dim 100 --k_max_pool 3 --just_test