There is a SPARQL query page, with useful examples included.

In general, Wikidata is not made up of discrete "datasets". Rather, each identifiable real-world entity, event or concept has a unique Q-identifier and a listing of "properties" that apply to that entity. So, to construct a dataset, you'd query for entities that participate in some set of properties you might care about.

Thanks! I ended up here, which is about the place one needs to start https://medium.com/freely-sharing-the-sum-of-all-knowledge/w...

understanding which tags from the ontology to use and then building the query.

you need to use SparQL, on query.wikidata.org there are tons of examples