That is great.
I checked the numbers using the Scholia tool and was a bit disappointed. The total numbers was "only" 337 for every year. Benjamin uses different tools; he mentioned his use of Google Scholar and indeed it shows so much more.
I was really pleased with Daniel Mietchen helping out on the subject of "probiotics" and I asked him if he could run his bot for the subject of "Wikipedia" and "Wikidata". But nevermind what he decides to do, running a bot adding key words to research is not scalable when you consider the overwhelming amount of research known to Wikidata. It is not only running it one time, it is also adding key words for any and all new research entered to Wikidata.
Given that we work in a Wiki way, this is totally acceptable. We do what we can, what takes our fancy and slowly but surely new approaches, tools improve on the quality and quantity of the data that we have. When Scholia was a commercial enterprise it would be different; the exposure and use of data would be a primary concern.