Keeping track of your PDF papers with Papers

by Jane on July 31, 2009

The New York Times (of all places) wrote about the chore of keeping track of all the scientific papers on your desktop:  all those PDF files cryptically named 1998A+A__338_781M.pdf and fulltext.pdf .  Or worse, 0903(4).3037v1, because it’s faster to re-download the PDF each time you need it, than to find it in your download folder.   Mac’s Searchlight tool is useless for finding PDF papers, because it doesn’t understand that titles and author lists of PDFs are most important.  So a search for “Arnett” will find papers citing Arnett, not the actual paper by Arnett.

Astronomers describe elaborate schemes for keeping track of papers:  renaming each file with an author name and year; sorting PDFs in descriptive directories; printing out hundreds of papers per year and filing them in a cabinet.

I’ve been trying Papers for OS X ($43, free 30-day trial), an application that searches for scientific papers (through ADS), then organizes those papers for you.

The best part of Papers are “Collections”, which are like iTunes playlists.  Say I want to write a new proposal, so I’m reviewing the literature on the topic of…  the hard X-ray luminosities of ultraluminous IR galaxies.  I make a new collection, “ULIRGs in the X-ray”, then drag in the papers I’ve found.  Some papers I’ll drag to several collections.  Now, a year later, when I start writing a paper on that subject,  I simply open the Papers collection (screenshot below), review my old literature search, add new papers, and quickly grab citations.  Pretty cool.


If you search ADS from within Papers, the program keeps track of the metadata (author, ADS link, bibcode) and associates them with the PDF file.  However, this requires using Papers’ ADS interface.  Searching for Rigby et al. 2008 would be Rigby[1AU] 2008[yr], or maybe Rigby[AU] Papovich[AU] 2008[yr].  The syntax isn’t hard to learn, but you have to learn a new syntax.


You can also import papers that are cluttering your desktop.   However, most PDFs lack metadata, so Papers doesn’t know what they are.  (ADS, CAN THIS BE FIXED?)  As a result, for every “Unknown” paper, you have to type in the author name and date, have Papers search ADS, find the right paper, and hit “Match” to link the metadata to the paper.  This is a pain, and is why I’ve been slow to switch to Papers.

Also, I’ve heard good things about Zotero — maybe an AstroBetter reader could review it, and/or compare it to Papers?

{ 7 comments… read them below or add one }

1 Kelle July 31, 2009 at 2:17 pm

Thank you so much for writing this up! Multiple people have bugged me about how AstroBetter is lacking a post about Papers!

My take so far is that Papers looks great and has enormous potential, but for the time being, I’m sticking with BibDesk. I’m working on a post right now describing why.

BTW, there’s a 40% discount (taking it down to $25) for undergrads and grad students. Details here:


2 Marshall July 31, 2009 at 11:13 pm

You ask “ADS, Can this be fixed?” about the lack of metadata – but it’s not ADS that’s producing the PDFs, it’s the journals. A&A has for several years already included the DOI codes which allow automatic metadata identification, and IOP Publishing is doing so on all new ApJ and AJ PDFs. (Yet another way in which they’ve vastly improved on U Chicago Press!) No one’s yet tried to update all the old PDFs out there, though…

It does take a while to move over your old library, but in my opinion it’s more than worth it. I can’t imagine working without Papers these days.

Jane, if you prefer using the real ADS interface instead of the Papers one (which I do), there’s a nice solution in the Papers bookmarklet, which gives you a web browser button “Open in Papers”. Do a search in ADS, click on the paper you want and press that button to have it show up in Papers, usually with all the metadata in place.


3 Adam August 2, 2009 at 11:14 pm

I just started using Papers and I’m rather liking it. Way better than loose PDFs! I’ve got my own suggestions list but it’s definitely got potential. So I’m looking forward to see what you have to say about BibDesk, Kelle, before I go ahead and take the plunge into that student discount 🙂


4 Jessica August 5, 2009 at 4:34 pm

FYI, mekentosj (the makers of Papers) have been very responsive about feature requests and such. One request I have seen is to improve the matching for astro-ph papers when using the “Open in Papers” feature from a browser. I use this feature all the time. My typical papers usage is:

1) surf astro-ph in the morning
2) identify any papers I want to keep permanently in my library
3) from the PDF in my browser, click Open in Papers
4) in Papers, use the match button to get all the metadata
5) by default it tries to match on the DOI, but it returns nothing… so right now I have to manually type in author names to get the match…. this is the problem that I hope is fixed in future Papers releases.


5 Marcos August 6, 2009 at 10:15 am

Agreed with Marshall – you can use the bookmarklet or otherwise use the Papers built-in browser to find articles in a web interface – rather than use the built-in ADS search.

Also, you don’t need to memorize the weird syntax – once you type in a search term and hit return, you can then click on the blue term and pick what it is first author, what have you – and it’ll fill in the [1AU] for you.


6 Jane August 6, 2009 at 12:34 pm

Great discussion, everyone! I sent quick emails to Papers and ADS, raising the metadata issue and linking to this discussion. Here’s a responose from the Papers developers:

From: mekentosj
Subject: Re: Website feedback
Date: August 6, 2009 9:24:39 AM PDT

Hi Jane,
Great blog post, many thanks! Indeed like one of the posters mentions the main problem is that publishers put very little metadata in their PDFs, second there’s the issue of metadata in the webpages (what is what the bookmarklet uses). ADS could actually do a much better job at incorporating a lot of metadata in the html headers of their webpages, that would already picked up by Papers today (like it does from the webpages of Nature or Science for example).

We do plan to add better detection of ADS identifiers in their webpages but this will take more work from our end. We’ll do our best!

Again, many thanks for the feedback. Enjoy Papers!
Best wishes, Alex


7 John August 12, 2009 at 9:12 am

That’s great — I played with papers a while ago, but found it frustrating that it was unable to pull metadata out of ADS (which other systems, like CiteULike, have been doing for years). If they really can fix that, I’ll be sold…


Leave a Comment

Previous post:

Next post: