Print

Print


OWL 2 does not require inverse properties to be explicitly named.

You can just use something like "inverse(contributedTo)" just about anywhere you can use a named property.

When rendering to RDF, assertions using inverse properties are flipped, so the RDF for an assertion using the inverse of a property is the same as that for an assertion using the uninverted property, with the subject and object reversed (so information about which direction the assertion is made is lost).

Using a separate named inverse property is unlikely to improve performance ; depending on the implementation you may end up searching on both the  forward subject, and the inverse object, or storing each assertion twice. It's not likely to make a big performance difference either way.

Triple stores usually have an index on predicate+object+subject, as well as subject+predicate+object; since inverse properties cannot involve literals, retrieval is often possible directly from the index.

Simon

On Oct 17, 2014 11:31 AM, "Steven Folsom" <[log in to unmask]> wrote:
Does BIBFRAME have an official position on inverse properties?

We would like to be able explicitly say that a person 'contributedTo' a work or instance (the subject of the triple) and not have to rely on querying for the person as object.

There are other areas of the vocabulary that are one-way where we may want to define inverse properties; I scoured the listserv to see if BIBFRAME had a guiding philosophy or stated decisions for these decisions, but related threads only point out where the vocabulary may or may not have reciprocity.

Thanks for any clarity,
Steven

--
Steven Folsom
Discovery Metadata Librarian. Cornell University