We need to be careful about what we call fictitious. Without getting too heady, who gets to say what is fictional? Every weekend I go for a walk in the woods with my daughters up Fairy Hike Mountain. I’m pretty sure that experience is real, but you won’t
find any mention of it except for hand-drawn map in my dining room. Are places referred to in religious texts real? I don’t want our metadata taking those positions. What’s the use case? Are we really worried that someone will try to get a ticket to Narnia?
I think the safest/easiest thing we can do is to allow things to be things if we want to describe them similarly. A person can be a person and still be the alternate identity for another person. If we don’t know the person is really a pen name, then obviously
we don’t say it. If we know they are a pen name, but we don’t have a lot of details on the person/s who write as that that person, we only say what we know. Using existing ontologies:
<Work1> dcterms:creator <Person1> .
<Person1> a foaf:Person ;
Note, that if a we said Person2 (the “real” person) created Work1 we wouldn’t be able to satisfy searches for things written by Benjamin Franklin as Alice Addertongue. We would only know generally that Benjamin Franklin also wrote as Alice Addertongue.
Also, if Person1 (the alternate identity) was only a Nomen, we couldn’t use the dcterms:creator property because it has a range of dcterms:Agent. Aligning data would become more complicated than it need be.
Re: Wrongful attributions, Something erroneously attributed to someone is different, and I would suggest we not have one way to say something is either wrongfully attributed, attributed to a fictional person, or both. We could extend the PROV qualifiedAttribution
pattern (perhaps with a subclass) to say something was wrongfully attributed.
For example:
<Work1> dcterms:creator <Person1> ;
prov:qualifiedAttribution <Attribution1> .
<Person1> a foaf:Person ;
<Attribution1> a ex:WrongAttribution ;
prov:agent <Person3> .
The list of use cases isn't really about fictitious entities--it's about fictitious creator/contributor relationships, relationships which are asserted in bibliographic objects for various reasons but known to be false in real-world terms. Some of them
involve fictitious entities, but not all.
A list of use cases for fictitious entities should certainly include subject use. Presumably in FRBR-LRM, "subject" is not defined as an entity class because it is co-extensive with res ("LRM-R12, Work / has as subject / res").
A separate question would be whether fictitiousness should be an attribute of res to identify fictitious entities. I'd argue that it should be. It can arguably be combined with any of the categories defined for res and its subentities, so defining it
at the level of res makes sense.
Stephen