How to train ApSIC for language-specific rules to reduce false-positives


As we all know, there is a high rate of false positives flagged in ApSIC due to specific language rules.

What is the best strategy to “train” ApSIC to language-specific rules so that the false positives are reduced?

As to not recreate the wheel, I am also wondering if there are any “open source” language specific ApSIC “projects”/ect already created to exchange. Furthermore, any resources to use when “training” ApSIC.

Thank you!