hindbr Goto Github PK

HINDBR: Heterogeneous Information Network Based Duplicate Bug Report Prediction (ISSRE'2020).

@inproceedings{xiao2020hindbr,
  title={HINDBR: Heterogeneous Information Network Based Duplicate Bug Report Prediction},
  author={Xiao, Guanping and Du, Xiaoting and Sui, Yulei and Yue, Tao},
  booktitle={2020 IEEE 31st International Symposium on Software Reliability Engineering (ISSRE)},
  pages={195--206},
  year={2020},
  organization={IEEE}
}

Run the following command to install software dependencies (Python 3.6, tensorflow-gpu==1.15.0, Keras==2.3.1)

pip3 install -r requirements.txt

Note, HINDbr is developed and tested in a DELL Precision Tower with a 3.60 GHz Intel i9-9900K CPU, 32 GB memory, 512GB SSD and 2TB HDD storage, and an NVIDIA RTX 2080Ti GPU, running Ubuntu 18.04. We didn't test it in other hardware and software environments.

Download the bug reports and arrange the directory level as the following depiction. Otherwise, you need to change the directory setting from data/xmlfile_path.setting.

  AnyDir   
    |  
    |----BugData  
    |      |----eclipse  
    |      |----freedesktop  
    |      |----...  
    |----HINDbr 
    |      |----data
    |      |----...

Download the bug data first. Then, run the command to generate bug report heterogeneous information network from bug reports. You can change the project's name in PROJECT from data_generation_br_hin.py.

python3 data_generation_br_hin.py

Generated bug report HINs are stored in data/bug_report_hin.

HIN2Vec embedding:
Download the hin_embedding that we have trained and put it into data/pretrained_embeddings/hin2vec/. Or, you can train it by the command.

python3 hin2vec_training.py

Word2Vec embedding:
Download the word_embedding that we have trained and put it into data/pretrained_embeddings/word2vec/. Or, you can train it by the command.

python3 word2vec_training.py

Run the command to generate bug groups. You can change the project's name in PROJECT from data_generation_bug_groups.py.

python3 data_generation_bug_groups.py

Generated bug groups are stored in data/bug_report_groups.

Run the command to generate model training and test data. You can change the project's name in PROJECT from data_generation_model_training.py.

python3 data_generation_model_training.py

Generated bug pairs are stored in data/model_training.

Run the command to generate before-JIT and after-JIT pairs for the model evaluation. You can change the project's name in PROJECT from data_generation_before_after_jit.py.

python3 data_generation_before_after_jit.py

Generated before-JIT and after-JIT bug pairs are stored in data/before_after_jit_data.

Run the command to perform HINDbr and Baseline model training and test. You can change the setting of the training projects from run_hindbr_training.sh for HINDbr, or run_baseline_training.sh for Baseline model.

bash run_hindbr_training.sh

bash run_baseline_training.sh

Trained models are stored in directory output/trained_model and the training history and test results are stored in output/training_history.

Run the command to perform before-JIT and after-JIT evaluation. You can change the setting of the training projects from run_jit_evaluation.sh.

bash run_jit_evaluation.sh

Evaluation results are stored in output/training_history.