BibLaTeX preserve case of acronyms in title

Question

I have an extensive list of references(mostly journal articles) exported from Mendeley to a .bib file automatically.

The issue that I have currently is the IEEE style uses sentence case and thus I lose capitalisation of my acronyms in my title. Considering the following example,

@article{Paik2007,
author = {Paik, Bu Geun and Kim, Jin and Park, Young Ha and Kim, Ki Sup and Yu, Kwon Kyu},
doi = {10.1016/j.oceaneng.2005.11.022},
isbn = {00298018},
issn = {00298018},
journal = {Ocean Engineering},
keywords = {Convection velocity,Phase-averaging,Propeller wake,Swirling strength,Tip vortex,Trailing vorticity,Two-frame PIV},
mendeley-groups = {Msc Research},
pages = {594--604},
title = {Analysis of Wake Behind a Rotating Propeller using PIV Technique in a Cavitation Tunnel},
url = {http://202.114.89.60/resource/pdf/679.pdf},
volume = {34},
year = {2007}
}

I use the following settings for biblatex

\usepackage[url=false,        % Ignore url field
            backend=biber,    % Not bibtex
            sorting=none,     % Order of appearence in text
            bibencoding=utf8, % For those special characters
            maxbibnames=10,   % Use et all only for more than 10
            style  =numeric-comp]{biblatex}

With the biblatex style set to ieee, the title is produced as

Analysis of wake behind a rotating propeller using piv technique in a cavitation tunnel

I'd like the title to be represented as

Analysis of wake behind a rotating propeller using PIV technique in a cavitation tunnel

If I change the style to numeric-comp, I then get the title field as an exact representation of the the bib file, but the bib file isn't of a uniform standard. Some journals follow sentence case, while others don't.

I do know that I can manually edit the bib file and wrap all acronyms in {} to preserve the case, but is there a smarter way of automatically maintaining the case if there are two or more subsequent uppercase letters in the title?

Answer 1

I wholeheartedly agree with jon that a bibliography file is something a lot of thoughts and manual labour should go into to make sure the information is as accurate as possible; there are so many subtleties and corner cases that a script or other automation cannot catch. I also agree fully with his advice

Invest the time (even if not for this task) to cultivate a good and compliant bibliography. Most data sucked up from other sources is defective in subtle and non-so-subtle ways....

That said, it is possible to protect acronyms with curly braces using Biber's sourcemapping and Regex

\DeclareSourcemap{
  \maps[datatype=bibtex]{
    \map{
       \step[fieldsource=title, match=\regexp{\b([A-Z]{2,})\b}, replace={{}{$1}}]
    }
  }
}

This will match all groups of two or more uppercase letters surrounded by word breaks (spaces, newlines and the like) and wrap them in curly braces (for some reason the empty group is needed for this to work, heaven knows why).

MWE

\documentclass[american]{article}
\usepackage{babel}
\usepackage[style=ieee, backend=biber]{biblatex}
\usepackage{filecontents}
\begin{filecontents*}{\jobname.bib}
@article{Paik2007,
author = {Paik, Bu Geun and Kim, Jin and Park, Young Ha and Kim, Ki Sup and Yu, Kwon Kyu},
doi = {10.1016/j.oceaneng.2005.11.022},
journal = {Ocean Engineering},
pages = {594--604},
title = {Analysis of Wake Behind a Rotating Propeller using PIV and CIV Technique in a Cavitation Tunnel},
url = {http://202.114.89.60/resource/pdf/679.pdf},
volume = {34},
year = {2007}
}
@article{un,
author = {Ban Ki-moon},
journal = {Ocean Engineering},
title = {I Quite like the UN},
year = {2007}
}
@article{faq,
author = {Anne Uthor},
journal = {Ocean Engineering},
title = {FAQ and Other Questions Answered},
year = {2007}
}
\end{filecontents*}
\addbibresource{\jobname.bib}

\DeclareSourcemap{
  \maps[datatype=bibtex]{
    \map{
       \step[fieldsource=title, match=\regexp{\b([A-Z]{2,})\b}, replace={{}{$1}}]
    }
  }
}

\begin{document}
\nocite{*}


\printbibliography
\end{document}

---

Bonus,

sed "s/\([A-Z]\{2,\}\)/{\1}/g" <file>

should do the same thing as sed script.