Organizing references

by Willie Wong

Sort of on the opposite end of what I wrote about last week (on managing edits of my own papers), I recently came across a good way of managing citations and references.

Previously I’ve been compiling BibTex files by hand: I have a complicated scheme set-up with a central bib directory. Inside it is a master.bib file and a bunch of subdirectories. The subdirectories are categories: general relativity, PDE, differential geometry, etc. In each category are several smaller bibfiles. Each bibfile represent a subcategory, and the bibtex key I used are always of the form “[category]-[subcategory]-[authors’ initial]-[year]”. By keeping separate bibfiles for separate subcategories, editing is easier. And I have a script that recompiles all the bib entries from the subcategory files and dump them into the master.bib file. Furthermore, the script also compiles the key-author-title information and dump them into a dictionary file, which I can then import in Vim to get quick citations. (I’m actually quite proud of this abuse of the vim dictionary that I am using.)

I’ve heard many good things about Zotero and Mendeley, and I have seen Zotero in action. While I really like its direct integration to firefox which allows it to very simply import citations from the web, I am not so thrilled about using a cloud service for managing my citations.

By way of Slashdot, I learned about Jabref. It is Java based and automatically platform agnostic (so I can use it both on my primary Linux computers and also on my Windows computer if I have to). I like its ability to group/tag articles making them easier to find, and its BibTex based backend (making exports simpler). Furthermore, one can link entries with stored PDF files. And with the database stored locally, syncing the bib file and the “link” is as simple as copying the entire directory.

Gradually I plan to migrate not just papers that I cite, but also papers that I read to the database. It provides a good, searchable interface for papers in general. Right now finding a paper on my harddrive involves invoking the unix command “find” and requires remembering either the lastname of one of the authors, or some keyword in the title in proper abbreviation. This contributed in part to my having two or three copies of the same papers on my harddrive. Jabref will certainly help.

I also like the fact that it has a Review field in the bib entry, so you can jot down some thoughts you (or other people) have about the paper.