A Manually-Curated Dataset of Fixes to Vulnerabilities of Open-Source Software

Link to actual data https://github.com/copernico/msr2019 [repository] https://github.com/copernico/msr2019/blob/master/dataset/vul... [csv dump]

[seriously, it should not be this hard to find that within the PDF...]

Really cool! Nearly three years ago I wrote a post [0] on here asking if anyone had something like this, thinking you could use it to train an AI to recognize vulnerable portions of code. More details on that in the linked post.

Really cool someone went through this effort; would love to see more people build on it and try to make productive use of the data.

[0] https://news.ycombinator.com/item?id=11573547