Online reference managers: not quite there yet

For my Paper Writing Dream Machine I obviously need a reference manager. My list of required features includes:

  • Easy addition of references by integration with Pubmed, Google Scholar and other online databases. A special bookmarklet would be a bonus.
  • Commenting on references, e.g. custom tags or free-form text
  • Creation of reference lists, e.g. by tags
  • Sharing of reference lists to user-defined groups or to everybody
  • Integration of the references into your word processor of choice. An export function into a common format such as .ris is a minimum. A better solution would be an open programming interface (API).

For obvious reasons (sharing references with your coauthors) an online reference manager is the easiest way to accomplish these feature requests.

Connotea
You import references into Connotea via bookmarklet or by entering the DOI. A direct query of Pubmed is not possible. References can be tagged and commented on. Unfortunately it is not possible to create a reference list for a particular paper that is only seen by your coauthors (Connotea groups are all or nothing). References can be exported into common formats, but no direct word processor integration. A Connotea Web API is available.

CiteULike
CiteULike is very similar in features to Connotea. The same limitations (no private group for reference list, no word processor integration) also apply.

2collab
A newer service created by Elsevier. Again similar in features to CiteULike and Connotea. Private groups for your coauthors to share a reference list are supported. No direct word processor integration. 2collab has a public API.

RefWorks
In contrast to Connotea and CiteULike this is a commercial product (our university has a subscription). Again no direct integration with Pubmed (only copy and paste), but a bookmarklet. And again difficulties to create a list of references that is visible just to your coauthors. Write-n-Cite is a RefWorks plugin for Microsoft Word. No API.

EndNote Web
EndNote Web is promoted as a sort of Endnote Lite, i.e. a web-based tool with just the basic features of the commercial desktop application. Again no direct Pubmed Search, but integration with Web of Science (just like Endnote a product from Thomson Scientific). Sharing of reference lists to just your coauthors is possible. Endnote Web comes with a Microsoft Word plugin for ease paper writing.

With the exeption of Connotea and 2collab, all the systems mentioned above allow you to store the fulltext PDF. This is a handy feature, but online storing of creates copyright issues for papers that are not public access.

Zotero, a Firefox extension, is not on this list, because references are stored locally and cannot easily be shared with others.

Surprisingly, none of the tools supports all my requirements. Endnote Web comes closest, but is far from perfect (e.g. sometimes slow, also needs Web of Science). Is it so difficult to build a Pubmed query function and group reference list into the tools?

Related Posts Plugin for WordPress, Blogger...
This entry was posted in Snippets. Bookmark the permalink.

15 Responses to Online reference managers: not quite there yet

  1. Pedro Beltrao says:

    My wish would be for a combination of Zotero and Connotea. I use connotea for the social tagging to get suggestions on topics of interest but Zotero is much nicer to use as a library. I tried to look around some time ago to have a way to submit to both at the sime time but I could not find a pluging to do it when I searched.

  2. Neil Saunders says:

    It’s not difficult to do any of these things. It’s just that almost nobody seems to be working on them, presumably because bibliography/citation management is rather boring and software developers prefer “sexier” projects.

  3. Martin Fenner says:

    Pedro, my favorite application is “Papers”:http://mekentosj.com/papers/ (Macintosh only). But it is a desktop application and – just like Zotero – currently can’t easily share references with other people (e.g. your coauthors). Maybe what we need is a common data format and synchronisation mechanism (just like with calendar software), so that everybody can use his favorite desktop application but at the same time can share his references.
    Neil, it would be fairly easy for Pubmed or Google Scholar to add these features.

  4. Bob O'Hara says:

    OpenOffice is developing a “bibliographic tool”:http://bibliographic.openoffice.org/. How close will that come to doing what you want, particularly if it was connected with (say) Connotea?

  5. Martin Fenner says:

    The OpenOffice tool right now doesn’t allow online sharing of references. But I understand that it can connect to online databases.

  6. Duncan Hull says:

    Nice summary, BTW Refworks (and Citeulike) do have a kind-of API, if you count the “RESTful Web Service Interfaces”:http://www.crummy.com/writing/RESTful-Web-Services/ (Just HTTP GET) they provide as a primitive (but useful) API of sorts.

  7. David Crotty says:

    The downfall of all of the online managers is the inefficiency of the tagging. If, like most established scientists, you’ve got a reference list that has thousands of entries, it could take you weeks of dedicated work to tag them all. And continuous effort to update your tags. And even after they’re tagged, it’s still not a good way to help you find what you’re looking for. I think that a high quality full-text search is vastly better and it requires no previous effort towards tagging.
    Until there’s full-text search of the papers themselves, the online repositories are going to be of secondary use. Sure, they’re probably good for sharing lists with collaborators, but you’re much better off using something like Papers, Yep or Yojimbo for your own writing and research.

  8. Martin Fenner says:

    My current workflow is built around Papers. I search for manuscripts from within Papers or import the PDFs right away. I do have about 1300 papers in the program and I don’t do any tagging. When writing a manuscript, I create an folder for that manuscript in Papers and export the references to Endnote (in RIS format).
    Papers has three disadvantages: it is Macintosh-only, it is a desktop application and it has no group features. A web-based application could solve all these shortcomings and would ideally sync with Papers, Zotero, etc.
    Maybe I should add storing of full-text PDFs to the list of my requirements. This creates additional copyright issues, but that problem can be solved in a variety of ways (e.g. the PDFs are not shared, linking to the journal website instead of storing a local copy, only storing of public access PDFs, etc.).

  9. William Gunn says:

    I agree that there is no good reference manager that easily integrates with whatever people use use for reading, sharing, and writing. My money’s on Zotero, personally, because I think they’re closest to being able to put it all together.
    I don’t think tagging is the failure here, though. Don’t get caught up in back-tagging. Just fetch keywords using eutils or something like that for your old ones and start tagging as you go forwards. I’ve never seen a need to update tags on a previously tagged article, so I think you might have gotten the wrong idea about what tags are good for. Look at how Flickr uses them in conjunction with date uploaded to help you find and organize pictures.
    Doesn’t Zotero store and index PDFs now, allowing full text search?
    2collab helpfully picks some keywords for you, but beware the Connotea import. It requires a .ris export and import, but the two services treat URL fields differently. They are working on it, “I hear”:http://synthesis.williamgunn.org/2008/06/16/connotea-youve-been-good-to-me/.
    I’ve made a Connotea -> Zotero -> Reference Manager situation work for me. I think the Zotero steo was necessary because the direct import didn’t work. maybe it works better now.

  10. Martin Fenner says:

    The following resourcs provide further information on the topic:
    * “A brief review of RefWorks”:http://duncan.hull.name/2008/06/20/a-brief-review-of-refworks/ (Duncan Hull, from last week)
    * “Finding the right bibliographic/reference tool”:http://www.fauskes.net/nb/bibtools/ (fauskes.net)
    * “Comparison of reference management software”:http://en.wikipedia.org/wiki/Comparison_of_reference_management_software (Wikipedia)

  11. David Crotty says:

    William, I guess it depends on what you want to do with your tags. If the whole purpose is sharing the material with others, then yes, a rough system that never gets updated or organized is probably sufficient. But if you’re trying to organize your entire library so you can quickly find what you need, then you either need extensive tagging, or full text search, the latter being superior (why do you think Apple, Microsoft and Google have put so much effort into desktop searching, rather than creating tagging systems for your files?). To me, finding what I need is the reason I’d invest time in tagging my references. If I want to use Connotea as a discovery tool, then I’ll look at other people’s tags, and not spend much time adding any of my own, as there’s no advantage in it for me, other than being a good citizen. But if I want to find that one paper from about 10 years ago, the one that used differential display, tagging is probably going to fall short for me, as it’s unlikely I’ll tag every single paper with every single method used therein.
    I do use Zotero, but as others have pointed out, it’s not a system that you can use for discovery or sharing with others, if that’s something that’s important to you. I also like Papers a lot, but that’s probably because I’m so used to the iPhoto/iTunes style interface.

  12. deathraypizza software says:

    For a better way of searching PubMed, try “PubSearch”:http://www.deathraypizza.com, it’s much better than using the PubMed web interface.

  13. Christina Pikas says:

    I’ve successfully exported as many as 200 references at a time from RefWorks – the limiting factor is how slow it is and server/proxy timeouts. We have RefShare, too, but it isn’t the same as sharing on connotea or citeulike

  14. Jyoti Shankar says:

    Hello, This is a very late comment but I have been searching for such a dream paper-writing system myself. I think CiteULike comes very close. It has four of the five features and executes all of them pretty efficiently.
    * Easy addition of references by integration with Pubmed, Google Scholar and other online databases. A special bookmarklet would be a bonus.
    _CiteULike has a bookmarklet that can be found “here”:http://citeulike.org/post. Via this bookmark, posting from Pubmed is a breeze and takes me around 2 seconds. There is also a “Greasemonkey script available for CiteULike-Pubmed integration if you work with Firefox”:http://userscripts.org/scripts/show/2478
    A similar script exists for “Google Scholar”:http://userscripts.org/scripts/show/11648 _
    * Commenting on references, e.g. custom tags or free-form text
    _CiteULike has an unlimited tag form-field. You can insert your personal notes (that can be either public or private).You can also store PDFs in your CiteULike Library along with the notes._
    _Another delightful complement to CiteULike is “SyncUThink”:http://www.andrewberman.org/projects/sync/ It automatically fetches and uploads PDFs available from the IP range you are working on. Most of the time, the correct PDFs are fetched and reduce my workload by a huge chunk. In the end, I just need to track down a few PDFs that “SyncUThink”:http://www.andrewberman.org/projects/sync/ hasn’t managed to find for some reason or the other._
    * Creation of reference lists, e.g. by tags
    _If you click any tag in your CiteULike library, all the references with that tag are loaded and you could export the tag-specific list as an RIS or BibTeX file._
    * Sharing of reference lists to user-defined groups or to everybody
    _You could create private/public groups on CiteULike and share all or part of your reference-lists with a single check-mark. The references are shared with everyone on CiteUlike by default._
    * Integration of the references into your word processor of choice. An export function into a common format such as .ris is a minimum. A better solution would be an open programming interface (API).
    _Though integration with word processors is not yet an option, “export to .ris and .bibTeX format do exist”:http://citeulike.org/profile/Zephyrus/export. Springer has recently extended sponsorship to CiteULike and I think some more good additions and integrations are on the way._
    _As has been mentioned earlier, “CiteULike does have an API”:http://svn.citeulike.org/faq/all.adp
    and a “very responsive bunch of administrators”:http://www.citeulike.org/groupfunc/3124. I don’t have any personal experiences but I have heard it mentioned several times on the discussion board that they are a helpful lot and encourage developers._
    _I reckon, after so much praise-heaping, a conflict-of-interest statement is in order. :) I am not related to the CiteUlike or make any profit from it but I just had to share my joy at having found this nifty reference management tool that is very near perfect._
    _I think what would make it completely perfect would be some sort of integration of citeulike with Zotero!_

  15. Martin Fenner says:

    Jyoti, thanks for mentioning *SyncUThink* which I didn’t know about. Since writing the blog post in June, a few things have happened. CiteULike is now sponsored by Springer, and a few new social networking sites for scientists with paper handling features have appeared, most importantly “Mendeley”:http://www.mendeley.com and “Labmeeting”:http://www.labmeeting.com. For me the most important issue right now is how to best integrate these different tools. The discussion about this topic has currently shifted to Ian Mulvany’s blog (“More feedback wanted, where do we go now?”:http://network.nature.com/blogs/user/ianmulvany/2008/08/19/more-feedback-wanted-where-do-we-go-now).