I have reproduced the code, but the results are not as good as those in the paper. There are two points of doubt. First, during the graph simplification process, a large amount of vulnerability code information is deleted. Why does the performance still improve? Second, the EA-GCN mentioned in the paper is not shown in the published source code. Please publish it. Thank you!
I have reproduced the code, but the results are not as good as those in the paper. There are two points of doubt. First, during the graph simplification process, a large amount of vulnerability code information is deleted. Why does the performance still improve? Second, the EA-GCN mentioned in the paper is not shown in the published source code. Please publish it. Thank you!