Sorry – not meaning to draw incorrect conclusions about what Bibo can do – just based my comments on the representations shown here. So to generalise my comments – I think the representation of volumes/issues as distinct entities identified by their own URIs is probably useful if it can be achieved.

I’m less sure in this particular example that a FRBRised representation is adding anything, and further I’m concerned about the ability to construct an accurate FRBRised representation from the information available in this example.

In general I’d tend toward Bibo over FaBio to be honest just on the basis that Bibo seems (to me) to have more momentum and support behind it (in fact, this is the first time I’ve come across FaBio, whereas Bibo has been on my radar for sometime)

What does seem to me to be more useful in the FaBio version is the identification of both journal issue and volume as separate entities (presumably eventually with their own URIs to identify them). With Bibo to answer the query ‘give me all the items that belong to this journal issue’ looks like it would have to rely on consistent practice in recording the volume and issue numbers, whereas in FaBio you could just draw out all the articles that were part of the issue (if I’m reading this correctly).

I think, again, you all are drawing the wrong conclusions. Bibo does have classes to describe issues. So if you have the data, or the algorithms to reliably generate that information, you can certainly encode it. Again: it really depends on the use case. My priority in the examples I’ve written is to get citation data encoded in reasonably structured RDF. So in that context, I don’t think it’s practical to ask, say, a Zotero user to help catalog journal issues. But we leave room for that in Bibo if you want.

Just maybe we’ll want an FRBR ‘Item’ to represent the fact we have multiple physical copies of something but this is far down the list — if it will be needed at all.

Lastly I’d request more examples if possible:) — e.g. could we have a standard fiction title (Harry Potter would be good because we have lots of printings, different editions, translations etc).

First a quick question. In the Bibo example you haven’t included the link to the ‘web version’ – is there any reason for this?

I think to some extent the example here isn’t complex enough to test which representation is better – and further, I’m not sure you really have enough data to really be sure the FaBio representation is quite what it claims.

The ‘webArticle’ link, actually links to a metadata record, with links to further representations – e.g. PDF and HTML. The PDF seems (to me) to be a faithful rendition of the printed version of the article – complete with page numbers (which in FaBio are assumed to only apply to the printArticle), while the HTML is clearly a different type of representation of the same content (and no page numbers of course). If you were to represent this, it would start to look a bit more interesting maybe (although hard to feel that excited about it for me). Perhaps more to the point would be if there was a pre-print version and post-print version of the article – but in this case that doesn’t seem to be true, and in any case the FaBio modelling may well breakdown, as presumably the pre-print isn’t part of the journal? As Dan points out – it can get ridiculously complex.

What does seem to me to be more useful in the FaBio version is the identification of both journal issue and volume as separate entities (presumably eventually with their own URIs to identify them). With Bibo to answer the query ‘give me all the items that belong to this journal issue’ looks like it would have to rely on consistent practice in recording the volume and issue numbers, whereas in FaBio you could just draw out all the articles that were part of the issue (if I’m reading this correctly). While you’d think that consistency in Volume/Issue numbers might be pretty good, with this example I note that the volume number printed on the PDF is ‘E65′ not ’65’ as in the metadata – so already we see confusion arising… Once you get into things that are named rather than numbered (e.g. assigning month titles, or abbreviations thereof, to issues) it will clearly get worse.

On that basis I tend to prefer the FaBio for the additional structure in the volume/issue area, but feel that the addition of the frbr ’embodiment’ adds (for this item at least) unnecessary complexity. Not sure how easy it will be to reconcile this?

The other thing I would say is that we designed BIBO to be usable in more complex modeling, much as you’ve used PRISM in your alternate example. So another way to have presented the trade-offs is to look at the basic modeling using BIBO, and to ask what additional complexity would need to be added to give a FRBR-like view, with what payoffs.