OVID export

NSBoxall

I want to cry!

I'm trying to get a reference list of ~3000 references off OVID (having done my beautiful systematic literature search) and into Zotero.

I have tried every format I can, but RIS insists upon importing every item as a webpage, which is meaningless.

Is there ANY way that this can be fixed? Should I go to the RIS file and change every TY - ELEC to TY - JOUR? But what about the references that aren't actually journal articles and are, instead, book chapters, or clinical trial information?

Are there any workarounds which won't involve me selecting all items on this page (using the URL bar translator)? Ovid only allows you to 'see' 100 at a time, so I'd have to do that 30 times. I CAN do it, I'd just really really REALLY rather not.

Many thanks!

adamsmith

I think the best solution would be to filter your results by publication type, export them separately, and search and replace individually - i.e. export all articles and replace with JOUR, export all chapter and replace with CHAP, export all books and replace with BOOK. Not sure what you'd do with clinical trials - RPRT probably?

We may eventually look at an import specifically for OVID's own data format, but it's a dumb format and I don't get why they don't just export something standard.

NSBoxall

Thanks Adam, I'm doing that now (so much fun).

Quick question because I seem unable to find the long list. Is there a list anywhere of the pt/TY's*?

I had a look at SQLlite and can see that itemtypeid=4 for typeName=journalArticle.

Can I therefore edit my ris files to say TY=4 instead of TY=JOUR?

Many thanks for your support. I understand the frustration with Ovid (oh. so. well.)

*I have things like conference proceedings/protocols/manuscripts etc. it's all very mixed up.

adamsmith

no - RIS and Zotero's database format are very much not the same.
Here's a list of RIS tags and what they mean. They don't all have clear Zotero equivalents, but most are pretty clear.
https://github.com/aurimasv/translators/wiki/RIS-Tag-Map-(narrow)

NSBoxall

Funny, I'd just found that, as I tried to start editing the ris exports.

Alas, it's seriously a mess, with abstracts from conference proceedings that have tags that don't match anything sensible (i.e. N2 instead of T2, JA instead of C3), mixed in with protocols from clinicaltrials.gov.

I know endnote and refman have both had filters written for them - to import these tagged ris files from those (particular) sources.

Is Zotero primarily used by 'other than' medical/biological science? Just wondering how many people out there are also wishing there was an easier way to import from Ovid. Or what other workarounds there are - if any. Because I've cost my client nearly 2 days already and still aren't really getting to what I want (i.e. a complete database of all work conducted in a narrow field).

adamsmith

there are a fair amount of medical researchers using Zotero. I imagine most of them relay principally on pubmed?
Also, the URL bar import is relatively good (also not perfect), so for anyone who doesn't do systematic reviews that's not really a problem.

adamsmith

BTW. which Ovid database(s) are you using?

NSBoxall

Embase 1988 to 2014 Week 03
Ovid MEDLINE(R) In-Process & Other Non-Indexed Citations and Ovid MEDLINE(R) 1946 to Present
Econlit 1886 to December 2013
EBM Reviews - Cochrane Central Register of Controlled Trials December 2013
EBM Reviews - Cochrane Database of Systematic Reviews 2005 to December 2013
EBM Reviews - Database of Abstracts of Reviews of Effects 4th Quarter 2013
EBM Reviews - Health Technology Assessment 4th Quarter 2013
EBM Reviews - NHS Economic Evaluation Database 4th Quarter 2013

I should mention that my search is "yr=2000-current" - so I'm not anticipating tags from 1886! :)

I'm wondering whether to now (well, when the licenses become available again) go back and export by publication type by datasource so that there's hopefully a LITTLE bit more systematised tagging.

The URL bar translator also has significant differences in importation capability. When you bring it in from OVID, you have the fields
Type
Title
Author (also includes institution, incorrectly)
Publication
Library catalog
Date added/modified

but when you click on the article link in Ovid and get taken to Pubmed, you bring in
Type
Title
Authors (correctly)
Abstract
Publication
Date (MM-dd,yyyy)
Journal Abbr,
Language
DOI
ISSN
Short title
Library catalog
Extra
Date added/modified

Any idea why there might be such a vast difference?

adamsmith

pubmed has great systematic data in a well defined XML format. Ovid exports less data, their XML format is very confusing and their tagged format both ugly and less systematic and complete than pubmed's.

NSBoxall

Gotcha.

Don't suppose you know of another means by which to systematically search those dbases without Ovid's involvement at all, then, do you?

I'd be far happier to rewrite my search syntax than retag 2803 outputs. :)

adamsmith

no, I fear I don't. The way to go in the long run would be to write an import translator for OVID's tagged format - still won't be as good as pubmed, but I think it could be done reasonably well. But it's a lot of work. (Alternatively, of course, you could try to convince OVID tro produce better RIS, but I wouldn't get my hopes up).

NSBoxall

But wait a second. OVID exports to xml, apparently. Can Zotero import from xml?

(this is bringing on trichotillomania; i swear! Apologies if I seem inept...I feel it!)

adamsmith

no, unfortunately not - XML is just a generic description of a format, like .txt or .html. Zotero can import several XML formats, including Pubmed XML, but not the one that OVID exports (which looks like a complete mess).

NSBoxall

*sob*

Okay, thanks. I'm back to the drawing board.

adamsmith

I mean - if this is a one time thing, you could install an Endnote trial version, import to there and then import from Endnote to Zotero which is pretty clean. I'm not 100% sure, but I believe the Endnote trial is just time limited, not in terms of number of items or so.

NSBoxall

That's also an option I'm considering. But right now, I have to wait till the licenses are all free, so I can export The One File (to rule them all) and see where that gets me.

Thanks for all your comments, I really appreciate that my messages aren't simply falling into the ether.

adamsmith

we have import from OVID's export format (Export to Endnote) working now. Any testing and/or error reports would be welcome.

eggsbrown

I just tried exporting from OVID EMBASE as and Endnote .cgi file. When I imported into zotero, the author names seemed a little funky. It wasn't noticing the difference between authors' first and last names.

aurimas

Can you paste one of the entries from the RIS file? I looked at EMBASE (not through OVID) and it looks like they're not separating last name from first name with a comma.

NSBoxall

Hey guys, I've found the same issue with the last names. Do you know of any way we can change things so they work? I'm (again) importing loads of references in a go from Ovid!

(and THANK YOU for getting it to mostly work!!)

NSBoxall

Oh - and the workaround I've been using (as I have RefMan) is:

Export from Ovid as RIS document
Change the extension from .ris to .txt and open in notepad
Find and replace all “TY – ELEC” with “TY – JOUR”
Change extension back from .txt to .ris
Import into RefMan (using RIS (add periods).cap import filter)
Export as RIS
Import into Zotero

if you fail to do this, the sirnames will all have the first initials together, i.e. Smith J Et al, instead of the more appropriate Smith et al. Quelle DISASTER!

aurimas

Have you tried using OVID's EndNote export format?

plabelle

Wow. Ok, that works. Thanks for information. I really appreciate how quickly you respond to new discussions. Much appreciated.