Papers from arXiv

max1836

For papers retrieved from arXiv, the arXiv number is placed in the "Publication" field. Some (at least, many?) are copies of papers that were published elsewhere.

Look at:
http://arxiv.org/abs/0708.1593

You can follow DOI to see where the paper was published.

I think it would better to put the arXiv information which identifies the document in the "Loc in Archive" field. Then the source of the published information could be entered into the normal fields.

By moving the ArXiv data you could import a new item for the DOI reference. Then you could merge the two records, (arXiv & DOI) while choosing arXiv as the master. That way you'd still know the real source of the document, but you'd have the information on the published version for citations.

aurimas

Looks like that was an intentional decision to put arXiv identifier in the publication field. But it dates back over 5 years, so maybe the reasoning no longer applies.

I'll wait for Simon or adamsmith to pitch in before changing this.

adamsmith

So, the argument for the status quo is that the working paper on arxiv is a separate publication from the journal article it ends up published as.

That's why it should be saved and - where it applies - cited differently. In other words, taking bibliographic data seriously, the DOI does _not_ apply to the arxiv paper and should not be saved with it.
That's in line with what we do with other working paper repositories such as SSRN.
I don't think this is 100% satisfactory, but I don't see a good solution that would handle this correctly.

max1836

Adam,

I understand your point and it is technically correct. The arXiv document and the "real" publication of it are two different papers.

But is my decision as a user that I want to obtain the data on the "real" publication by looking up the DOI too (I don't expect Zotero to do this...).

I can then must manually cut the arXiv information from Publication field and paste the arXiv information into the "arXiv item" which identifies the document in the "Loc in Archive" field.

Then I can load DOI doc as the "DOI Item"

Then I can merge the "Doi Item" and the "arXiv item" and select arXiv to be the master.

-----

It would just be a tad easier if the arXiv information wasn't saved in the Publication field to begin with.

sean

The arXiv ID should indeed be used as an identifier, not a publication, and in future iterations of our data model it might make sense to have more, and more flexible identifiers available. Even so, that doesn't really solve your problem, which is that you want to combine two things which really are bibliographically distinct. Reading a preprint and citing the published article may be a reasonable shortcut, but it's still inaccurate. Instead, more flexible relationships among items might do the trick, e.g. "preprint of," but there's little demand for them, and they're not on the immediate horizon.

max1836

Ok, here is what I did...

http://s564.photobucket.com/user/herb1836/media/zotero_4.jpg.html

Sort of hard to miss that the pdf is from arXiv. All(?) arXiv papers are watermarked on left!

Now if I need to look up "real" paper, I know where to go. In the meantime this "preprint" will do...

DWL-SDCA

ArXiv documents, like other working papers, often undergo several changes. If this kind of document is published in a more formal presentation, it is likely to be somewhat diffent. It can be more or less complete. The journal peer review process is likely to demand changes. The data presentation can change. The methods description can change. Tables may be added or removed. One version may support an assertion while another may not. I urge everyone to avoid thinking of the arXiv version as a preprint.

There are people who study the differences between working paper versions and the final "print" version and produce research on the evolutionary process of research publication.

max1836

DWL,

Good points. Certainly only the published version of a paper should be cited in another paper. For "background" information it seems the arXiv will meet my needs for the moment.

I'll point out that I didn't match the papers. The author did in his notes at arXiv. I was just looking for a way to easily leave some footprints to the real paper.

fbennett

The Related tab might also be useful for that.

DWL-SDCA

Max wrote

...only the published version of a paper should be cited in another paper...

I didn't intend to suggest that idea. There may be good reasons to cite something from a working paper that did not appear in the journal version.

max1836

Sorry, we're looking at this from two different angles. As you pointed out the history of how a paper developed could be of interest too. I'm sure there are other reasons...

max1836

Problem 2

Look at paper:
http://arxiv.org/abs/math/9311201

Used Zotero icon in firebox browser to create a Zotero entry.

Would it be possible to put text in "comment" line into the "extra" field in the Zotero? The zrXiv comments are usually some juicy stuff...

adamsmith

I'd be inclined to put it in a note, but that should be doable, yes.

adamsmith

pull request here.
https://github.com/zotero/translators/pull/586
adds comments as a note, also changes arXiv papers over to reports, which is what we do for other working paper archives (most closely related: SSRN, also NBER) comments welcome.

max1836

Super Fantastic Adam!

Thank you.

adamsmith

this is now live.

max1836

Adam,

Updated my translators. Sweet!

A different solution than I imagined, but I'm sure it fits the overall scheme better.

Thanks,
Herb

ijvm

Great that the handling of arXiv papers has got an update. But setting the citation type to reports is a bit odd, as in most journals arXiv citations are treated differently than a report.

A clear example: for AIP and APS (physics) journals, the RevTeX class is used when writing papers in latex, and it does a wonderful job at treating arXiv citations, which worked perfect with the exported bibtex citation from zotero... before the update. Now the arXiv is treated as a report, which RevTeX formats in a manner inconsistent with the style of the journals (including the title, writing technical report...).

Could you please consider reverting the type to article, or adjusting the bibtex export to accommodate current journal practices?

adamsmith

ugh. That's unfortunate. I don't have a great idea to solve this atm. but will think about it. noksagt, aurimas, any thoughts?

adamsmith

ticket - though no solution yet - here:
https://github.com/zotero/translators/issues/616

noksagt

Thanks for pinging me in the github ticket. I somehow missed this thread & I'll reply there.

adamsmith

changed the translator again.
1. It will now get the citation data for the published article where the item exists - while that's problematic for the reasons outlined by DWL and sean above, I think in the end noksagt (who is the "resident" physicist) is right that arXiv is very commonly used as not just a preprint, but also as a reprint server, so this makes the most sense.

2. In line with ijvm's request, I've also reverted the import for unpublished arxiv preprints to journal articles. The publication title now, somewhat awkwardly, includes the entire arxiv ID - certainly not ideal, but that will work best for both citations from Zotero and bibtex export with the fields we currently have. As soon as we have an arxiv ID field I will revisit this, of course.

ijvm

Awesome, this has been working like a charm, thanks.