We need to put section recommendations in front of users and measure (what to be measured is to be defined) the effect of providing section recommendations. This experiment may likely be done as part of the upcoming editathons in collaboration with Florence et al. or it can be a larger scale controlled experiment. The details to be discussed, but either way, we need a tool for exposing these recommendations.
Pau has already provided designs for a GapFinder like interface that can surface articles to be expanded, and how section recommendations can be exposed to the user in T160177.
Timelines: Ideally, the development of the tool starts towards the end of September 2017 and finishes a couple of weeks after that. That will give us enough time to experiment, document, and submit the result of the experiment as part of T174736