Improving Prompt Tuning-based Software Vulnerability Assessment by Fusing Source Code and Vulnerability Description
This is the source code to the paper "Improving Prompt Tuning-based Software Vulnerability Assessment by Fusing Source Code and Vulnerability Description". Please refer to the paper for the experimental details.
Due to the large size of the datasets, we have stored them in Google Drive: Dataset Link
if you want to use the original dataset(MegaVul), you can download it from the following link:https://github.com/Icyrockton/MegaVul
We provide a code file for crawling CVSS v3 data, and on this basis, you can crawl other data you need.
You can install the required dependency packages for our environment by using the following command: pip install - r requirements.txt.
1.Use the py file under data crawling and processing for data processing. Of course, you can directly use the dataset we have processed: Google Drive Link
2.Run prompt_code&desc.py. After running, you can retrain the model and obtain results.
3.You can find the implementation code for the RQ1-RQ4 section and the Discussion section experiments in the corresponding folders. The results obtained from the experiment are also in the corresponding folder.
You can obtain our saved model and reproduce our results through the link:Model Link.