Our second submission

Today I submitted our second paper.

This one turns the lens on the data itself. DGA detection leans on a "benign" list as the clean half of its dataset — most often Tranco, a research ranking of popular domains. But that list isn't actually clean: malicious and DGA-generated domains leak into it, quietly contaminating the ground truth a detector is supposed to learn from. Our paper is about detecting and cleansing that contamination in our DGA dataset, Tranco especially.

This one meant a lot to me — it was a chance to resolve the questions our previous work had left me with.

Grateful again to my dear Chaeri, who was with me through this one too, and to Prof. Jeong. Hope it goes well — again.