It seems that annotation learning has been desactivated at some point. For example, on 0.0.6.5.1, a term put in stop list on some analysis can appear again in an other analysis as a map term. Do you confirm ? If yes, can we plan to restore the community learning of terms relevance ?