The CrossRef Citation Plugin for WordPress reinvents the Structured Blogging plugin, but provides a minimal Lucene-based search interface to CrossRef's metadata for looking up citations, which is useful. I'd have preferred it if they'd just opened up that API and let people build their own interfaces, because the current plugin is awkward, limited and keeps timing out. The citation it returns is one long string, rather than MODS (say), so you can't use other citation styles (although I guess you could extract the pieces from the COinS string). It embeds COinS tags, but no other microformats, so is lacking compared to the Structured Blogging markup that was around two years ago.
On the other hand, the Structured Blogging plugin tried to do too much and stored the metadata in a non-standard way, so didn't really get taken up by bloggers. Maybe the combination of Zotero, a decent open CrossRef API and some plugin innovation (and aggregation along the lines of BPR3, but based on standard markup) could produce an acceptable compromise between ease of use and depth of metadata.
Update: I've added a "Blog This" box on the right-hand side of Scintilla's Conversations results page, which produces the same kind of markup as the CrossRef plugin or BPR3, ready to be copy-pasted into a blog post. Here's an example of the code it generates at the moment:
<div class="citation">
<span class="title"><a href="http://dx.doi.org/10.1371/journal.pmed.0020124" rev="review">Why Most Published Research Findings Are False</a></span>. <span class="author">John P. A. Ioannidis</span>, <span class="journal">PLoS Medicine</span>. <span class="year">2005</span>; <span class="volume">2</span>(<span class="issue">8</span>) <span class="page">e124</span>
<span class="Z3988" title="ctx_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Why+Most+Published+Research+Findings+Are+False&rft.jtitle=PLoS+Medicine&rft.volume=2&rft.issue=8&rft.spage=e124&rft.epage=&rft.date=2005&rft.au=John+P.+A.+Ioannidis"><!-- COinS --></span>
<a href="http://scintilla.nature.com/review" rel="tag" class="review"></a>
</div>
Comments
All fields are optional, email address will not be shown; no HTML, URLs are automatically hyperlinked.
Just to note that we are encouraging people to build their own plugins and that one of the ideas behind releasing the WordPress plugin was that others could look at it and create plugins of their own. Also that, behind the "minimal" Lucene-based interface is full Lucene query support. So, for instance, the following also works:
expression=contributorLastName:renear
To be fair, you couldn't know this just from looking at the code. More thorough documentation of this will follow.
I should also point out that it is fairly easy to retrieve the structured metadata from CrossRef with just one more call to CrossRef's OpenURL interface. Again, if there is demand, we could look into doing this behind the scenes.
As for MODS- after a little informal consultation and fairly low (read "zero") expressed interest in MODS, we decided to go with what seems to be a pragmatic approach using COinS. If there is a lot of expressed interest in MODs, we'd be happy to revisit this.
And the timeouts are a bug. We are looking into this.
Finally, I should reiterate that the point of releasing this "as is" has been to get feedback like this- so thank you. We want to make this as useful as possible.
--G
Right - using the CrossRef OpenURL interface to fetch the metadata was what I thought of as well. I've just added a link to a box on Scintilla that uses this to generate code for a blog post.
On styling, it might be time to start thinking about how it should be integrated into web apps. There's a guy that's started working on a PHP implementation of citeproc for use in Drupal, but that's the kind of thing that I'd like to see more widely implemented. The question is how best to do this from a markup and processing standpoint.
Yes, I'd like to use CSL here too. I'm not sure whether it should be at the stage where the HTML code snippet is generated (easiest) or when the HTML of the actual blog post is created (harder, but more flexible).
Wait, who is behind this plug-in? Is it the cross ref people themselves? Apparently. Geoffrey Bilder, can you give us a direct email contact where we can contact you for more info?
I am VERY interested in this development. Are you saying there's a full lucene index available for querrying cross-ref data and returning full metadata now? That's _huge_. And I can make lots of use of it.
I also thought that using cross ref data in this way required a license, and was only free up to a limited amount of traffic. Is this no longer true? I would very much like more information, and some documentation.
And yes, Geoffrey, COinS is _not_ enough. In particular with regard to multiple authors on the same article. Since COinS in particular requires you use the key-value URL-get style of OpenURL, there is NO good way to encode multiple authors in there, with structured firstname/lastname seperation. This is a big problem.
If you provided OpenURL XML ("SAP2") instead, that might be sufficient. But if you're still losing granularity on data you _have_ in CrossRef, that would be bad! I want all the granularity you've got! I think OpenURL XML would likely be sufficient though. But COinS is not.
The CrossRef OpenURL only gives the first author of a paper. Are we ever likely to see the other authors?
Sorry about the delay in response- I was at a workshop and off the grid. So to address points raised above...
1) CSL looked very interesting to us too. One question that it raised though (and we didn't really get good answers to this ) was "given that most citation styles are designed for print publications and so that somebody in meatspace can find the item in question, are they appropriate for citations in an e-only medium (i.e. a blog posting) where you already have a persistent link to the resource in question (DOI). In other words- in an e-only context should the citation style be designed more to
"jog the memory" than to enable somebody to find something? The truth is, we were really not sure how far down the "citation formatting flexibility" rathole we wanted to go. Seemed like diminishing returns.
2) Yes, it is the "CrossRef people themselves" that are behind the plugin and yes, this does mean there is a full lucene index behind the plugin, and while we appreciate the "that's _huge_", we're not really sure that it is actually that huge. We already have several mechanisms for querying CrossRef metadata and they are generally available for free use for non-commercial applications. The major point of the plugin is that it is optimized for easy end-user use instead of automated querying.
3) The problem of multiple authors with COinS is a pain, but as I mentioned above, we kind of expected that people could just use our existing OpenURL metadata interface if they wanted to retrieve the structured metadata. Zotero, I think, currently works this way.
Of course, as soon as I typed the above, I realized that our OpenURL metadata implementation didn't return multiple author's either. Doh! Anyway- a quick word with our technology director (thanks, Chuck!) and we have fixed this by enabling a third non-standard parameter which will return a more verbose XML response.
format=unixref
So, for instance, this will return multiple authors.
http://www.crossref.org/openurl/?id=doi:10.1103/PhysRev.47.777&noredirect=true&format=unixref
4) Finally, for anybody who is interested, you can reach me directly either via:
citation-plugin at crossref dot org
or
firstnameinitial + lastname at crossref dot org
Hope this helps.
--G
Before I forget and before you get too happy ;-) - I need to emphasize that the above new OpenURL parameter will only return all the authors *if* the publisher has submitted all the authors. This is a general rule with CrossRef metadata- we only return what the publisher deposited and publishers are not required to deposit much. Many publishers deposit more than is absolutely required, but a few only deposit the minimum. In short, it ain't perfect, but it's something.
Geoffrey (and Chuck), many thanks for giving us the 'unixref' XML response! I agree with Jonathan that this is potentially huge, since it allows all kinds of bibliographic tools (such as Zotero, Umlaut, JabRef, refbase, etc) to fetch *full* citation metadata for any CrossRef article.
It is my hope that you will be able to convince all CrossRef publishers that depositing full metadata will ultimately benefit them most.
Thanks again for the effort, Matthias