Remodel representation of Series

[osma]

I'm still not happy with the way series are modelled. Currently, it's the Work that is part of a Series (which is also a Work). I think that the Instance should instead be the entity that's part of a Series Work, but I need to check first with RDA modelling experts about their views.

[osma]

An additional complication is that ISSN is really an instance-level attribute for series (the same series published in print and electronically get different ISSNs), except for ISSN-L which is for the series-work level...

[osma]

The CreativeWorkSeries type is a bit unnecessary in the current model. According to the Schema.org definition, it "serves largely just to organize these more specific and practical subtypes". I think we could just use Periodical everywhere and skip CreativeWorkSeries entirely.

[osma]

After discussing this with Clément Oury (ISSN centre) I think that Series should just be a single entity, not split into a Work and Instance. The ISSN identifies this entity and it could be linked to the Linked Data that the ISSN Centre is going to publish very soon.

[sfolsom]

Last year LD4L worked on a Series/Serial model that didn't reach approval status, and thus not shared publicly.We've considered how to get <www.loc.gov/aba/pcc/conser/> more involved, but capacity issues keep us from picking this up right now. That said, I'd love to compare notes.

[osma]

@sfolsom I don't have anything very specific in mind, except what I said above. I think it's worth waiting to see how the ISSN centre has modelled their linked data that they're going to publish a few weeks from now. There is a little bit of information in a slide deck ("ISSN Register as Linked Data") linked from this European BIBFRAME Workshop wiki page. In my understanding they consider each ISSN as a single resource, that is both a bf:Work and a bf:Instance at the same time.

[osma]

There are two quite different cases to consider:

1. When the record represents a Work that is itself a series
2. When the record represents a Work that is part of a series

In case 1 we currently create both a Work and an Instance, since that's what marc2bibframe2 does. In case 2 we only create a Work (although marc2bibframe2 also creates an instance we ignore it).

I think that in both cases we should create just the Work (actually a subtype that is a CreativeWorkSeries). Obviously it should have the name and ISSN, but in case 1 we probably also need to assert all the Instance-related properties that come from the MARC record.

[osma]

It's the Instance that should belong to a Series (just like in BIBFRAME after the conversion from MARC), not the Work. There may be multiple Instances of a Work and some of them may belong to different series, but the Work itself is series-agnostic.