Can downloaded citations be used directly?

Question

I am using plain bibliography style and I found the citations that I downloaded from IEEE Xplore quite self-contradicting... In the block quote below is a really bad example that I have encountered:

   @INPROCEEDINGS{652309, 
    author={Bangun, R.A. and Beadle, H.W.P.}, 
    booktitle={Information, Communications and Signal Processing, 1997. ICICS., Proceedings of 1997 International Conference on}, 
    title={A network architecture for multiuser networked games on demand}, 
    year={1997}, 
    month={sep}, 
    volume={3}, 
    number={}, 
    pages={1815 -1819 vol.3}, 
    keywords={ADSL;cable-modem;computer games;high bandwidth networks;multiplayer networked  games;multiuser networked games on demand;network architecture;software architectures;computer games;computer networks;entertainment;interactive systems;multi-access systems;network topology;}, 
    doi={10.1109/ICICS.1997.652309}, 
    ISSN={},}

By self-contradicting I mean:

• The author field is often inconsistent, sometimes they have full names but sometimes they just have initials (such as in the example above).
• They included volume number in the pages number, which produces duplicate volume numbers after compilation.
• The booktitle field is simply not right.

The result is that I have to proofread and correct all downloaded citations, as I want my reference list to look consistent. I am not writing for any journal or conference so I don't have to stick to a particular style (though it would help to make my references consistent I guess), but I do want a time-saving method to help me get a nice, clean, uniform reference list.

Any suggestions?

Answer 1

Unfortunately, I know not one database that is well-kept. Bute some of the errors appear to be systematic. From my experience with the IEEE database (of course just an extrapolation of the samples I have seen), the following can be automated:

• In pages, “ -” can be replaced by “--”, almost all entries contain this error.
• In month, the month name should have 3 chars, lowercase, without braces, in the IEEE database you will often find one of these: “{mar}”, “{Mar}”, “{MAR}”, “{March}”, “{march}”, that should all translate to “mar”.
• In the author field, multiple name abbreviations like “R.A.” are sometimes with, sometimes without space inbetween, can be corrected by regular expression.
• Empty fields can be deleted, I also suggest to delete keywords and abstract, the first is not the keywords given by the authors but some automatically assigned ones, therefore they are not of good quality. The latter is usually taken from the PDF file of the paper and usually contains therefore errors, too.
• Uppercase abbreviations in the title should be covered by curly braces to keep them uppercase.
• I personally remove other fields like doi or url rather than correcting them, but that's up to your taste.

These could be automatically done with some regular expression hacking (sed or whatever you prefer), but you have to go through every entry manually and correct things like booktitle, so I usually do everything by hand.

Google scholar is also full of errors, but not necessarily the same ones, and this holds for all other databases I know.

Answer 2

Reasonably good bibtex entries are provided by the DBLP database. Most IEEE publications are indexed there.