In thinking about new vocabularies, whether for subjects, predicates, or objects of triples, one should begin with a survey of what already exists. By using one another's vocabularies, we reinforce the interoperability, and therefore utility, of our data. And it saves us the time needed to invent a taxonomy.
Sets of RDF vocabularies tend to fall in two groups: (1) terms for items, persons, concepts, and other resources and (2) terms for the relations that hold between resources. The first group correspond to what many scholars call controlled vocabulary, and they frequently show up as the subjects and objects of triples. The second corresponds to the vocabularies used in ontologies (e.g., RDFS, OWL, SKOS), and frequently show up as the predicates of triples.
These projects listed below rely not merely upon a unique identification system, but one rooted in IRIs, whether URL-based or URNs.
The Standards for Networking Ancient Prosopographies project is now curating a virtual authority of ancient people using RDF exports from many other major digital prosopographies.
Other major person-data or prosopographical resources include:
It is helpful to understand something about the hierarchy of texts (such as the one adopted by Functional Requirements for Bibliographic Records [FRBR]). Vocabularies describing ancient works in the abstract (under FRBR called works) will take a different approach than ones describing manuscripts, papyri, ostraca, etc. (under FRBR called items)