I have been very dismissive of Mendeley for many years now. For one good reason: the data is always extracted from Google Scholar. I get the worst, most incomplete reference from Mendeley. Being an early adopter (staring from its beta stage; around 2008), I was left with frustrations with Mendeley. Now, it is time to appreciate one great quality of Mendeley that no other reference manager can emulate: its attempt to do the undo-able. That is, Mendeley tries to get the reference information by reading the PDF file directly. This technology is unique to Mendeley, so far as I can tell. While both Bookends and Zotero can extract some identifiers like DOI and ISBN, they never try to get the Title, the author and the date by directly reading the PDF file. Mendeley does that. As a result, it is a life saver when you have a lot of junk to clean up.
I recently downloaded more than 3400 pdf files from a linguistic archive. Importing them to any of the references gives not a single relevant reference data–both Zotero and Bookends gave me zero result. I also tried Papers3. Quite interestingly, Papers was able to pick some of them. But, the data it gets was less than 20% success rate.
Then, I dragged them to Mendeley, majority of them get their references filled. Most of them get junk reference, of course, as usual. But, hey, this is technology. We have to do a lot of trial and error. Cleaning the junk library was much better than inserting references, one by one, for 3400 item. For that, I am now grateful of Mendeley.
But, ultimately I cannot live with Mendeley because it gets data from Google Scholar only–always junk data. That is why I have to move back these partially filled references to either Bookends or Zotero.
Completing the incomplete references in Zotero is a nightmare, I learned by the hard way. Zotero excels at getting data from browser (internet) and the attach the PDF over the reference. Having PDFs with incomplete reference or no DOI, however, Zotero is a huge pain. I am a PDF guy. I rarely pick the reference from the web page. I always go to the pdf; and attempt to fill up the reference when I have some extra time latter. For that, Bookends is much better. BE has a feature called Autocomplete (similar to the Targeted Browsing feature in Sente) which helps me highlight the Title of the book (article) from the pdf and tell it to search it somewhere in the web engines (google scholar, World Cat, or my local library website). That way, I don’t have to write the reference manually.
For Zotero, if you have missed to get the data from the website first, or that the PDF contains no DOI, the only option you have is to manually write the reference. Jabref is even better on that because you can copy and paste references from Google scholar to the existing PDF.
But, I find Zotero better than Bookends on these two aspects.
- Direct syncing of the Bibtex file: using the Better BibTex plugin
- Automatically getting the ISBN of the books. Zotero picks the ISBN of the books almost always correctly. This feature is coming to Bookends as well. But, BE is not really effective at yet.
How about Jabre?
It excels at manipulating references in the bibtex format. Furthermore, it as one unique feature that no other reference manager yet implemented–embedding the XML metadata into the pdf files. There are two good reasons to write metadata into the PDF files.
- it improves searching: you can search Spotlight by the author or the Title of the book; or order the books by their date of publication. This is specially very useful if you use more advanced searching tools like FoxTrot, or (DTsearch in the windows)
- you can lose the reference, or share the pdf without losing the reference data about it. If your library is lost or corrupted, you don’t have to fill the reference data again. You can just drag the pdf and Jabref will populate the reference data for you.
The conclusion is: every reference manager has its own strengths and weakness. Each of them have their own niche users; and niche features. Bookends and Jabref are my all time favorite reference managers. I think I will keep all the 3 reference managers me for now. Some people own three cars, just for sheer fun of it, even if one car is usually enough. I don’t have to chose among these great softwares. I will use Zotero for the books, Bookends for everything else and Jabref for bibtex.