LineVul / README.md
starsofchance's picture
Add dataset card
fa5f897 verified
metadata
dataset_info:
  dataset_name: LineVul Dataset Splits
  description: >
    This dataset contains splits (train, validation, test) of the LineVul
    dataset, originally designed for transformer-based line-level vulnerability
    prediction in C/C++ code. It includes processed functions, vulnerability
    labels, and fixed versions of vulnerable functions.
  version: 1.0.0
  license: MIT
  splits:
    - name: train
      description: Training split of the LineVul dataset.
    - name: validation
      description: Validation split of the LineVul dataset.
    - name: test
      description: Test split of the LineVul dataset.
  features:
    - name: processed_func
      dtype: string
      description: The original function written in C/C++.
    - name: target
      dtype: int
      description: Function-level label (1 for vulnerable, 0 for non-vulnerable).
    - name: vul_func_with_fix
      dtype: string
      description: The fixed function with added/deleted lines labeled.

LineVul Dataset Splits

This dataset provides the train, validation, and test splits of the LineVul dataset, originally introduced in the paper "LineVul: A Transformer-based Line-Level Vulnerability Prediction" by Michael Fu and Chakkrit Tantithamthavorn. The dataset is designed for predicting software vulnerabilities at the line level in C/C++ code using transformer-based models. It was sourced from the LineVul replication package available at https://github.com/awsm-research/LineVul.

Dataset Description

The LineVul dataset consists of C/C++ functions with associated vulnerability labels and fixed versions. Each split (train, validation, test) contains 39 columns, but the key columns used for vulnerability prediction are:

  • processed_func (string): The original function written in C/C++.
  • target (int): A binary label indicating whether the function is vulnerable (1) or not (0).
  • vul_func_with_fix (string): The fixed version of the function, with added and deleted lines labeled.

The dataset is intended for training and evaluating models that predict vulnerabilities in software code, particularly at the line level.

Citation

The original dataset and methodology are detailed in the following paper:

@inproceedings{fu2022linevul,
  title={LineVul: A Transformer-based Line-Level Vulnerability Prediction},
  author={Fu, Michael and Tantithamthavorn, Chakkrit},
  booktitle={2022 IEEE/ACM 19th International Conference on Mining Software Repositories (MSR)},
  year={2022},
  organization={IEEE}
}

Please cite this paper if you use this dataset in your research.

License

The dataset is released under the MIT License, consistent with the original LineVul replication package.