Upgrade StorageYear not correct after import in zotero from arxiv
When importing a paper that was updated years later like
Li et al. 2015 (https://arxiv.org/abs/1511.05493) it appears in Zotero as from 2017, as that was the last date where the paper was updated on arxiv.
The arxiv page states: (Submitted on 17 Nov 2015 (v1), last revised 22 Sep 2017 (this version, v4))
How can I systematically fix these errors in my library?
Currently I have to manually go to google scholar, look for the paper and import the reference into zotero from there.
I.e. from here:
https://scholar.google.com/scholar?cluster=16266567510296342081&hl=en&as_sdt=2005
Li et al. 2015 (https://arxiv.org/abs/1511.05493) it appears in Zotero as from 2017, as that was the last date where the paper was updated on arxiv.
The arxiv page states: (Submitted on 17 Nov 2015 (v1), last revised 22 Sep 2017 (this version, v4))
How can I systematically fix these errors in my library?
Currently I have to manually go to google scholar, look for the paper and import the reference into zotero from there.
I.e. from here:
https://scholar.google.com/scholar?cluster=16266567510296342081&hl=en&as_sdt=2005
Original date: 2015
If we ask the author, how he would like to be cited, I am sure he will agree and the arxiv API provided citation reflects that.
Why would Zotero force me into this ?
From a citation point of view, having something that has been updated in 2017 cited as 2015 seems quite misleading. If we're going to have dates in citations, shouldn't they reflect when what we're using is actually published? A "2015" paper that contains citations (and presumably information) from 2017 seems very problematic to me.
Where the arXiv entry does properly link to a published version, we do grab that metadata -- and that makes total sense to me -- but since it's just a link in a note here, that's not possible.
(FWIW, you can just click on the first version of a paper and then import in Zotero and that does import the original date)
To me a good way would be to have the original date on arxiv works and then add somewhere a remark that says 'version 4, updated 2017-Sep-22', the same way that hyperlinks commonly display a 'date retrieved' and the same way the arxiv papers carry a stamp.
https://groups.google.com/forum/#!topic/ml-news/OO-ZgVpPDQ4
https://iclr.cc/archive/www/2016.html
https://dblp.org/db/conf/iclr/iclr2016.html
Are the Zotero record publication dates errors?
@adamsmith is certainly correct about misleading dates. An item appears on arxiv with a date; then it is replaced with a revised version with a later date (and often again with a third version with a still later date). Unless someone wishes to comment upon the differences among the 3 items there is no reason to use the "original" date of a changed publication. Did you cite the original or the revised version? I don't see any reason not to cite the actual version that was used to support the assertion in any author's manuscript. Unless one has made and retained a copy of the original, in most cases one cannot know if the original supports the assertion. (Maybe the supporting statement was added as the paper was revised and was not in the first posting of the paper.) Please cite the version you read not some version you may not have seen.
I just think what most people need - let's say in addition - is a way to get to the meta data that they need to cite something eventually, independent of which version of a paper they happened to download to their personal drives, open in a webbrowser.
In practice, it can be difficult though to access the published version of a paper, while it is trivial to find the exact same paper on arxiv. Additionally both the authors and the readers would probably prefer if the published version was referenced.
Here is a hacky and stupid work-around that did it for me:
- Import all PDFs from Arxiv to mendeley,
- Click 'Update Details' on those, which will mostly figure out when they were eventually published and change the dates and details accordingly.
- Export mendeley entries to bibtex
- Import bibtex file into zotero from clipboard
- Unpin all the citation keys and reshape them with the great better bibtex plugin.
- Resolve duplicates by letting the mendeley entry be the master version.
Why is no one crying foul on the way mendeley or google scholar *incorrectly* report dates and versions for imported work on a bigger scale?
I agree with everything being said above, but it still seems - independent to what the world *should* look like - to be a minority position serving a minority need.