Mendeley's automatic BiBTeX file generation is a godsend in terms of avoiding the mindless and potentially error-prone work of copying citation information from PDFs of research papers to to BiBTeX. However, its automatic bibliographic information extraction algorithm is a little error-prone, and sometimes, even with proper bibliographic information, the export to BiBTeX file process tends to mess up accented characters (such as å, ø, é, etc.).
Does anyone have any suggestions on a workflow for correcting the .bib files automatically generated by Mendeley? I'm particularly interested in workflows that back-propagate the corrections back to the Mendeley database so that I don't have to keep correcting .bib files repeatedly.
Right now, I work on a Mac, so I currently edit the automatically generated .bib file with BibDesk; something reasonably close to equivalent on Linux would be editing the .bib file with JabRef, or even going through the .bib file with a text editor and BiBTeX-specific package (like emacs and AUCTeX).