Abstract: We study the problem of efficiently computing rankings of joinable attributes in data lakes. Traditional set-overlap measures produce numerous false positives in this scenario, while modern, ...