5 (!) years ago, I discussed my article pdf file naming scheme with Toshi Aoyama, and he agreed it would be a nice idea to add support for it. I just remembered iPapers and the discussion we had, and confirmed that one can construct a PubMed search (with a priori a unique result) from the naming scheme:
When obtaining the pdf for an article, I rename the file to something that I can recognise easily when browsing through my directories:
JournalName-Year-Volume-Issue-pNumber.pdf
here I use some abbreviation for the journal's name, and then the numbers for year of publication, volume, issue, and the 1st page number (prepended by a p), for instance:
JOV-2003-3-11-p831.pdf
As with the X11 font naming scheme, a missing issue number would give JOV-2003-3--p831.pdf rather than JOV-2003-3-p831.pdf
Since not all journals have an "official" compact abbreviation (like JOV for Journal of Vision or EBR for Experimental Brain Research), that particular field would be a bit tricky to use in PubMed searches to retrieve the corresponding reference, but the remaining information should in most cases lead to a unique (and correct) reference result:
Year[pdat] AND Volume[volume] AND Issue[issue] and Number[page]
(2003[pdat] AND 3[volume] AND 11[issue] AND 381[page] for the JOV reference above)
parsing the filename to retrieve the relevant data shouldn't be too hard: a call (or series of calls) to sscanf with the appropriate format string(s) should do the trick.
Would there be interest in pursuing this thought, and possibly adding support for this file naming scheme (through a plugin or otherwise)?