We want to run an AB test comparing recommender systems. Our metric for comparison is the fraction of accepted/published suggestions. Although we can already track when a suggestion is accepted and published, we need to add the following logging tables to run the test:
suggestions_schown: one log event per rendered suggestion
rank: position in the list of suggestions at time of rendering
suggestion_actions: one log event for every interaction with a suggestion
action: remove, save, start
rank: position in the list of recommendations at time of click
Additionally, when a user starts translating a suggestion, store the suggestion_id in log.ContentTranslationCTA and wikishared.cx_translations.