I see "features" as separate from "statements" in the proposed data model. Will they be modelled as property-value pairs or a different data structure?
Over at d:Wikidata:Property proposal/Lexemes, a number of properties are being proposed, like "person", "gender", "number", which will fit into the "features" component of the Lexeme data model.
Hello Deryck, I hope I understand your question right.
The so-called features are for example: the lemma, the language of the lemma, the language of the lexeme, the lexical category. In the forms, the representation and its language. These pieces of information are not represented by triples, but it's going to be a simple field (a bit like the label and description in items). Some of these fields will have autocompletion from Wikidata items.
If you want to look at what it will look like, you can try the demo system (information is not necessarily correctly modeled there, it's mostly a sandbox try the interface)
Let us know if you have further questions :)
Yes that makes sense. We aren't separating grammatical features by category (or properties).
Will the lemma (the Lexeme itself) have a "grammatical features" field? I only see that forms have " grammatical features" but it seems that the Lexeme doesn't. For example, how do we represent the fact that "chien" (fr) is masculine regardless of form?
No, the grammatical features are only included in the Forms. If you want to indicate something about the lexeme, you can decide to have a dedicated property and add it in a statement.
