As Wikidata's structure is fairly flat we need to either trust a record's identification patterns either belong together, or that they are separated somehow (usually the next BOF in the sequence). Presently Siegfried contains some heuristics (to be described) but having Wikidata describe this vs. Roy work it out would be ideal.
To get an idea about what Roy can currently process, run ./roy build -wikidatadebug.
ross-spencer
commented
3 years ago
Description of problem
As Wikidata's structure is fairly flat we need to either trust a record's identification patterns either belong together, or that they are separated somehow (usually the next BOF in the sequence). Presently Siegfried contains some heuristics (to be described) but having Wikidata describe this vs. Roy work it out would be ideal.
To get an idea about what Roy can currently process, run
./roy build -wikidatadebug.Permalink