--- layout: post status: publish published: true title: Converting Zotero Documents to Mendeley wordpress_id: 462 wordpress_url: http://www.martineve.com/?p=462 date: !binary |- MjAxMC0xMi0wOCAwODoyOToyMCArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0wOCAwODoyOToyMCArMDEwMA== categories: - Technology - Open Access - Academia tags: - zotero - Mendeley - OpenOffice - Word - Reference Management comments: - id: 4013 author: Tweets that mention Converting Zotero Documents to Mendeley | Martin Paul Eve -- Topsy.com author_email: '' author_url: http://topsy.com/www.martineve.com/2010/12/08/converting-zotero-documents-to-mendeley/?utm_source=pingback&utm_campaign=L2 date: !binary |- MjAxMC0xMi0wOCAxMTo0MTo0NiArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0wOCAxMTo0MTo0NiArMDEwMA== content: ! '[...] This post was mentioned on Twitter by Doctoral School, Martin Eve. Martin Eve said: New blog post: Converting Zotero Documents to Mendeley http://martineve.com/?p=462 #phdchat #phd [...]' - id: 4029 author: adam.smith author_email: karcher@u.northwestern.edu author_url: '' date: !binary |- MjAxMC0xMi0wOCAxNDoyMDowMyArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0wOCAxNDoyMDowMyArMDEwMA== content: ! "From what I can see you're not quite right about Mendeley winning the hearts and minds - I think it's a geography thing. Zotero is hugely, hugely more popular in the US. It's not even close. And I assume you're right that Mendeley is increasingly more popular in the UK - which would just mean that it's easier to reach out to the scientific community of the country you're in. \r\n\r\nZotero and Mendeley will be one step closer to interoperability once they both use the citerproc.js citation generator for csl 1.0. But that still leaves a lot of issues, including how to match identifiers and, apparently, how to address the additional information Zotero allows you to add to citations - which is indeed critical. I think the former is actually much harder to do, the latter, if that's really the case, just seems to be a limitation of Mendeley that they need to fix to make the software useful for anyone in the social sciences and humanities." - id: 4055 author: Martin Paul Eve author_email: martin@martineve.com author_url: '' date: !binary |- MjAxMC0xMi0wOCAxNzowODozOSArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0wOCAxNzowODozOSArMDEwMA== content: ! "Hi Adam,\r\n\r\nThanks for that - as you say, I'm working from a UK perspective and, from what I hear, Mendeley gets huge uptake relative to Zotero although I personally use the latter.\r\n\r\nThe identifier matching shouldn't really be a problem. If both plugins exposed the correct APIs, then all that is needed is for Mendeley, on import, to store the Zotero identifier in an internal field of its own for that specific user prior to document conversion.\r\n\r\nI hadn't heard about the citerproc.js plan, but is there, technically, any reason why both sets couldn't use the same document plugin system? Admittedly, this would probably have to end up being LGPL licensed so as to fit with Mendeley's proprietary nature, but would be a huge step forward in terms of interoperability. At the moment, reference systems seem to compete on unequal grounds. Once you've got far enough into a piece to know whether you like a system, you are unlikely to want to convert to another. If this lock-in was eliminated, they could compete on grounds of features and user interface.\r\n\r\n(P.S. I'm really pleased about US uptake of Zotero -- I'm a big fan, although, as I say, I'd still like to know of and see how Mendeley progresses)" - id: 4062 author: adam.smith author_email: karcher@u.northwestern.edu author_url: '' date: !binary |- MjAxMC0xMi0wOCAxODoyMToyNyArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0wOCAxODoyMToyNyArMDEwMA== content: ! "I think the problem with the plugin would be that it also has to work with the internal data model of the respective software - things like Zotero groups (and whatever Mendeley has in that respect) would probably be an additional challenge.\r\n\r\nI think a (very) longterm plan for Zotero is to include citation data in citations - that would make conversion easier, I guess.\r\n\r\nAs for the license of a joint plugin - I think there would be a reluctance among a good part of Zotero's non-staff contributors against LGPL. But I don't see why Mendeley couldn't use a general GPL component - actually I think it already does: csl is (I'm pretty sure) released under GPL, and almost all csl styles under cc by-sa\r\n\r\nI wonder, though - who would have the incentive to put in the significant work? I don't know much about Mendeley internals, but Zotero doesn't have a lot of manpower for development and I don't them making this a priority.\r\n\r\n(just to clarify my status - I'm just an active Zotero user and volunteer some time on the forum - I don't have any formal association with or inside knowledge of Zotero)." - id: 4228 author: Mr. Gunn author_email: william.gunn@mendeley.com author_url: http://mendeley.com/profiles/william-gunn/ date: !binary |- MjAxMC0xMi0wOSAxNjoxNDoxNCArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0wOSAxNjoxNDoxNCArMDEwMA== content: Well, at the risk of being unhelpful again, let me point out that Mendeley and Zotero do both use citeproc.js, so that's not an issue, it's just that how the particular word processor plugins were implemented are different and use different field codes. It would be great if we could agree on a uniform way to do markup in word processors, so that Zotero and Mendeley could be used interchangeably to format citations in a document, but we just aren't there yet. I'll bring this up for discussion. - id: 4229 author: Martin Paul Eve author_email: martin@martineve.com author_url: '' date: !binary |- MjAxMC0xMi0wOSAxNjoyMDo0MiArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0wOSAxNjoyMDo0MiArMDEwMA== content: ! "Thanks for that response, Mr. Gunn. I didn't find your response unhelpful, it was utterly courteous throughout and only \"unhelpful\" in the sense that, for my purposes, it was impractical -- I think Mendeley's approach towards the community, especially on Twitter, is exemplary.\r\n\r\nThis is another example of that. Nobody from Zotero (I appreciate that they are a community, albeit funded, effort, as opposed to a private entity) has initiated that discussion; it is Mendeley who are taking the lead.\r\n\r\nThat said, I would be extremely happy if all parties were to work together for a standardized citation field reference. At present, I am locked-in to Zotero. If this were interchangeable, the reference systems would be competing based on their actual merits, rather than a captive user-base, which I think we all agree would be a good thing!" - id: 4237 author: adam.smith author_email: karcher@u.northwestern.edu author_url: '' date: !binary |- MjAxMC0xMi0wOSAxNzowNzozMCArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0wOSAxNzowNzozMCArMDEwMA== content: ! "\"Nobody from Zotero (I appreciate that they are a community, albeit funded, effort, as opposed to a private entity) has initiated that discussion; it is Mendeley who are taking the lead.\"\r\nwell - document conversion has been discussed by Zotero and on the Zotero forums for quite a while - and (rightly imho) not limited to Mendeley, but more focused on older software like Endnote, which presumably has many more locked-in users.\r\n\r\nZotero is also apparently hiring a new community liasion/evangelist, which should professionalize that part - I don't know where they are on this, the announcement and call for applications was about 2 months ago and they sounded very pleased by the applicant pool." - id: 4251 author: Martin Paul Eve author_email: martin@martineve.com author_url: '' date: !binary |- MjAxMC0xMi0wOSAxOTozMjoyOSArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0wOSAxOTozMjoyOSArMDEwMA== content: I feel I've been hugely overcritical of Zotero now -- well, perhaps the community engagement aspect will change in time.... - id: 6029 author: Bruce author_email: bdarcus@gmail.com author_url: '' date: !binary |- MjAxMC0xMi0yOSAxNjoyNjoyOCArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0yOSAxNjoyNjoyOCArMDEwMA== content: ! 'I, the creator of the CSL language on which citation support in Mendeley and Zotero is based, have long been talking about this. There is no technical reason at all that we couldn''t have a single plug-in code-base and API that could interact with different services: mendeley, zotero, citeulike, etc. But someone needs to write it.' - id: 6030 author: Bruce author_email: bdarcus@gmail.com author_url: '' date: !binary |- MjAxMC0xMi0yOSAxNjozMDo0MSArMDEwMA== date_gmt: !binary |- MjAxMC0xMi0yOSAxNjozMDo0MSArMDEwMA== content: ! "Oh, and to offer some balance to this discussion, Zotero was the first major CSL implementation, and Mendeley benefited from that groundwork substantially. For example. the first Mendeley plugin code was pretty much the same open source code Zotero wrote. It just happens that they inherited some of what I consider to be the same design mistakes as Zotero ;-).\r\n\r\nSo I'd prefer if the respective user communities would refrain from the competitive language." ---

One of the best things about Mendeley is that, the second you mention their name on Twitter, a horde of helpful and informative community liaison team members descend upon you. This means that, if you have a query, you can be sure that someone knowledgeable is never far away.

That said, it seems that my current issue -- being able to convert either an OpenOffice or Word Zotero document to Mendeley -- is not possible. Having communicated with @mrgunn he suggested two approaches:

  1. Save the document without Zotero markup and re-insert the citations
  2. Run the document through a LaTeX converter (untested)

As the document in question has over 500 citations, I didn't find this to be of much use (which is not a criticism of @mrgunn, but rather the sad state of interoperability). However, I decided to do a little digging myself to work out what the actual issues are...

Now, in OpenOffice (at least), a Zotero field code looks like this: Reference: ZOTERO_ITEM {"citationItems":[{"locator":"17","uri":["http://zotero.org/users/64077/items/WKMCH3XK"]}]} RND5qVC9BQOk1. Upon visiting the URL in question, it is quite clearly a Zotero catalogue entry. A Mendeley citation, on the other hand, looks like: Reference: Mendeley Citation{f19dafe4-bd54-4b6c-9585-7acfd39428c9} RNDSDLGneoN3n. The task of a converter would be to match these two styles up.

However, it doesn't seem to be that simple. Zotero reference fields can also contain a plethora of other information that Mendeley seemingly doesn't support, the foremost (and to my mind a deal-breaker) being page numbers (!) This means that any conversion to Mendeley's format would be lossy; it would be impossible to preserve all the data from Zotero citations because Mendeley simply doesn't support including all this information.

While Mendeley seems, from my Twitter conversations, to be winning the hearts and minds of the research community, time and time again I query whether this is because they have a superior product to, say, Zotero or because they are able to market it better owing to the monetarization of their service. Things they win on: pretty interface; social aspect of sharing; online profiles generated automatically are great. Sadly, however, some basic stuff is missing. If they could implement this and then work with Zotero to allow document cross-compatibility (alongside not asking me to enter my email password to garner contacts -- this is incredibly bad practice -- OAuth?), they'd be on to a winner with a superior product.