Meetings as Agent and Meetings as Events: maybe they can be both, and we’re conflating them because they have the same label?
“ALA Midwinter 2016” is both a publisher and an event, and probably should have two uris, one as a madsrdf:Meeting and one as a bf:Event , each with different properties describing the different aspects of the same idea.
Network Development & MARC Standards Office
LA308, Mail Stop 4402
Library of Congress
Washington DC 20540
This has been a really interesting thread to monitor. Some reactions to various discussions:
1.) I think it’s worth clarifying what happens with subclassing. (I think everyone participating understands this, but it might help tease out some problematic terms.) If one class is asserted as a subclass of another, every instance of the former is always an instance of the latter. E.g.
If: ex:Meeting rdfs:subclassOf ex:Event .
Then this statement: <Some Meeting> a ex:Meeting .
*Always* entails: <Some Meeting> a ex:Event . [Perhaps this is what was originally meant by hierarchies are “static”? Totally agree that in RDF something can exist in multiple hierarchies, but subclasses aren’t for "sometimes situations”.]
2.) Regardless of historic practice, I’m not sure I would want a Meeting to be a subclass of Agent. It’s more fitting for Meetings be treated as Events that Agents participate in.
3.) Because bf:Work and bf:Event are not (to my knowledge) asserted to be disjoint, there is nothing formal stopping us for asserting that something is both a bf:Work and bf:Event when it is the case (e.g. the performances that Tim alluded to). Depending on the Event and its relationships to other entities, it may or not BE a Work. It may or may not generate/depict/be the subject of a Work. What I’m trying to say is that because there will be so many ways we will want to refer to bf:Event they shouldn’t be pigeonholed, but there may be some Event types that we want to treat always as works (e.g. Performances).
4.) The points I made about Works/Events above apply for Contributions and Provisions and Events. I could see a case where we want to say the “event” represented as an AuthorContribution is the subject of a book. Or occasionally wanting to use schema:Event properties (I believe suggested by Amanda) to better describe a Contribution.
5.) I too, don’t understand what the Content class adds.
On 1/20/16 3:25 AM, Oddrun Ohren wrote:
Thanks Ray Denenberg for your clarifications! It might well be that life-cycle events are best kept separate from events as entities described or captured in works.
However, I still think that bf:Content (or bf:EventContent) is unnecessary, and I hope you BIBFRAME 2.0 developers will come round to the same way of thinking J.
(Concerning examples of events modeled as works, I think Tim Thompson provided several good examples. Referring to the draft proposal, perhaps the battle re-enactment event may be considered a work)
IMHO one should always think long and hard before solving any need for increased granularity by subclassing existing classes. Class hierarchies are static structures, and should express fairly stable knowledge. Therefore, I am wondering if you plan to do something about the bf:Work class and its subclassing into media specific sub-classes in BIBFRAME 2.0? As far as I can see, none of the Work subclasses has additional properties (compared to Work), a fact which in itself rather defeats the purpose of subclassing. A more flexible solution would be to introduce a property “type” or similar to Work, and offer a controlled vocabulary of work types as potential value set. A work type vocabulary would at any rate be easier to maintain through changing media types than would a set of subclasses. Moreover, it will then be possible to use other type vocabularies in domains where these are more relevant than the “recommended” one.
It so happens that I just did a short blog post on subclassing Work, albeit related to FRBR but possibly valid also for BIBFRAME.
There are indeed additional properties, they just haven't been singled out as such. Any property, like "bf:musicKey" is a de facto indicator of a sub-type (aka sub-class). BIBFRAME has a number of properties whose names begin with "cartographic..." and others that begin with "music..." So the type-specific properties exist they just haven't been organized as such (something which might be useful for folks cataloging in those areas).
I disagree that subclassing is static -- at least not in RDF. Any subject can be an instance of more than one class, and classes only have impact when operated upon, as in querying. It is my understanding that in RDF it is very convenient to operate on data using classes, much more so than indicating types using values. So there may be a practical reason for sub-typing, but it doesn't have to impose limitations, AFAIK. Anyway, it's worth thinking about.
Note also that some non-library implementations of FRBR have made use of sub-typing of WEM, some even quite extensively:
The frbrCore vocabulary introduced just a few sample subtypes:
Oddrun Pauline Ohren
> From: Oddrun Ohren
> • Not being sure how explicitly point *1.c*of the proposal is meant, I’d just like
> to point out that events may play other roles than being the *subject* of some
The line "A bf:Event will be described in the same manner as other BIBFRAME Subject Types.." is poorly worded (my fault). Probably better would be: "An event will be described in the same manner as other external resources."
For example, a person. While a bf:Person is a BIBFRAME resource, it consists of simply a label, and a link to an external description of the person (a MADS description, FOAF, VIAF, etc.). That's really all that that was trying to say: the concept of a bf:Event relies on the availability of an external description of that event. (Except that for the event, there may be some basic properties besides just the label within the BIBFRAME resource, for example date and time, but for any additional description there will have to be an external resource describing the event.)
it might be useful to
> represent life-cycle events of a work (launching, publication, recording)
> explicitly in some cases. At any rate we should take care that the Event class is
> not modelled in such a way that one specific role is assumed.
Event, as we currently envision it to be modeled, will not include these life-cycle events, we plan to model these differently. Tentatively, there will be a property with name something like bf:originationActivity and class bf:OriginationActivity, with subclasses like bf:Publication, bf:Distribution, and so on, and each of these will have properties like agent, date, place.
> I am not
> able to see what bf:Content contributes other than extra
> (unnecessary) complexity… o Firstly, it is problematic to constrain something as
> general-sounding as Content to be a capture of an Event.
We are currently considering changing the name to EventContent.
> o Secondly, if bf:depicts/bf:captures are defined as properties of both Work and
> Event (like their parent bf:subject) with expected value *any resource*
> (instances of any BIBFRAME class, including Work), there should be no need for
> bf:Content. This way, bf:depicts/bf:captures could also be used to represent the
> fact that some works capture other works (e.g. photographs of paintings).
> o Lastly, seeing that the existing subclasses of Work are more or less disjunct,
> bf:Content will create confusion, as it clearly overlaps several of the existing
These are good points and we will need to discuss them.
> It will also be possible to represent
> events as a work where appropriate, without losing the possibility to express
> information about capturing
Do you have an example of an Event that could be modelled as a Work?
Thanks much for your comments and suggestions.
--Karen Coylem: +1-510-435-8234skype: kcoylenet/+1-510-984-3600