Dry run T213930
 Define a set of representative languages
 Define a set of dimensions for the map of verifiability: e.g. article popularity, topic, etc.
 Download Training data from Featured Articles by querying for all articles belonging to the corresponding category in each language, see: https://www.wikidata.org/wiki/Q4387444
 Download Test data by randomly sampling articles for each language
 Create and Evaluate citation need models with the same logic as T186354 for each language
 Compute the proportion of unsourced content in all articles in the test set of each language
 Analyze unsourced content VS the dimensions defined above.