Make it clear that broader/narrower relations are all direct

[atomrab]

How do we construct broader/narrower relationships for the following set of periods?

Bronze Age
Late Bronze Age
Late Bronze Age III
Late Bronze Age IIIC

Are all the subperiods in a "broader" relationship with the top-level "Bronze Age" only? Or should each one only go up one level (so Late Bronze Age IIIC is narrower than Late Bronze Age III but not Late Bronze Age or Bronze Age)? Or should anything below a second-tier division just go up to the second tier (so Late Bronze Age IIIC and Late Bronze Age III are both narrower than Late Bronze Age, but IIIC is not described as narrower than III)?

If these relationships are inherited, so that it's apparent that any subperiod of a subperiod of a subperiod is also a subperiod of the main period, I would prefer that each subperiod only refer back one level (LBA IIIC is a subperiod of LBA III). But we should explain this somewhere, or we're going to have a lot of inconsistent submissions.

This issue is occasioned by a patch (see #221) that has relationships and expresses them as LBAIIIC<LBAIII rather than LBAIIIC<LBA.

[rybesh]

The semantics of the "broader" relationship are defined by skos:broader, which is intended to be used only for direct parent-child relations in a hierarchy. (There is another property, skos:broaderTransitive, which can point from any descendent to any ancestor in the hierarchy, but that property is intended to be automatically inferred via RDFS reasoning, i.e. not included as part of the dataset. See Transitive Hierarchies for details.)

I'll leave this issue open until we put a note in the interface about only "directly" broader/narrower relations.