How can one manage his own bibliography? (using biber/biblatex)

Question

I used to work with Citavi and recently tried out Jabref.

Due to the nature of the best, i.e. I always had re-touch the .bib-file, I recently exported consolidated my sources I had stored in Citavi and Jabref and worked them into a single .bib-file which I am going to use with Biblatex and biber.

I intend to use either Notepad++, Texstudio or Geany for editing the .bib-files. So now I am thinking about simplifying / improving the workflow when I need to edit it.

First of I thought I'd split it up into 27 files, which are A-Z and numerals. Yet sometimes in the past when I added the sources, I didn't consider that in a case of a collaboration, I didn't specifically account for the corresponding author if there was one, which are of course often named first. That might prove tedious if I'd like to look at the specific code for an entry. In terms of functionality this is irrelevant since I assume one is not going to omit any letters.

What do you do for staying sane when working with large .bib files? Or is it not that much of an hassle for you?

Answer 1

I use one file and notepad++ when I'm on my windows box (jEdit on linux most of the time). My keys are (normally*) in the form Surname_Key_Words. I strip any keywords fields (manually or using a regex find/replace), because I just find thy slow down searching. A find is pretty simple: searching for Nakamura_ would only find me the following entry.

@article{Nakamura_GaN_LED,
    title = {High-Power {GaN} {P}-{N} Junction Blue-Light-Emitting Diodes},
    author = {Shuji Nakamura and Takashi Mukai and Masayuki Senoh},
    journal = {Japanese Journal of Applied Physics},
    volume = {30},
    number = {Part 2, No. 12A},
    pages = {L1998-L2001},
    numpages = {4},
    year = {1991},
    url = {http://jjap.jsap.jp/link?JJAP/30/L1998/},
    doi = {10.7567/JJAP.30.L1998},
    publisher = {The Japan Society of Applied Physics}
}

I also keep copies of the listed papers in a flat folder with the file name equal to the bibtex key. This works on bibtex and biblatex (which I use now but I have to maintain back-compatibility without too much hassle). It seems to work well for ~200 entries and counting.

I just found jabref led to too much click this then click that then type this nonsense, but switched to using quality editors when I had many entries with similar title elements that needed {}, easily solved by a regex find/replace.

I use the same editors for all my .tex files with appropriate command macros for compiling etc.

*other keys:

• colleagues -- Firstame_key_words
• me -- Me_keyword
• others -- company_Key_Words
• etc.

Answer 2

I don't think there is a definitive answer to this question at all. For one thing, I think what suits one person will not suit another. Moreover, people differ in how many entries they have to handle, where they obtain new data from and what they need to be able to do with that data later on.

However, I realised that trying to explain this in comments was not working very well. Maybe this will be a little clearer.

Overview

I currently use different files for different kinds of abbreviations. One for author, editor and organisation names. One for journals and series. One for places. One for publishers. One for everything else. Then I have bib files by category e.g. one for general reference (dictionaries etc., regular non-specialist non-fiction), one for literature etc. This makes it easy to cross-reference because all entries for papers in a given anthology, for example, are in the same file as the entry for the anthology. (Note that not all of the required data is in the same file. But neither bibtex nor biber care about that.) Then I have some scripts for transforming downloaded references into my preferred format.

Example

Hopefully this will make the above a little clearer...

Suppose I have an anthology of papers on aardvarks edited by Camel B. Jones which contains a paper on their diet by Maria N. Davies.

Camel B. Jones (ed.). 1992. The Life and Times of Aardvarks. Oxford: Oxford University Press.

Maria N. Davies. 1992. 'What Aardvarks Like To Eat'. In Camel B. Jones (ed.), The Life and Times of Aardvarks. Oxford: Oxford University Press. Pages 141-156.

Then in authors.bib, I might have:

@string{davies-maria-n = {Davies, Maria N.}}
@string{jones-camel-b = {Jones, Camel B.}} 

In pub.bib, I might have:

@string{oup = {Oxford University Press}}

places.bib might contain:

@string{oxon = {Oxford}}

Then zoo.bib might contain all entries to do with zoology. For new keys, I try to use something like lastname1-lastname2-...-initiallettersoftitle so my entries would look something like this:

@incollection{davies-wal2e,
  author = davies-maria-n,
  crossref = {jones-lta},
  title = {What Aardvarks Like To Eat},
  pages = {141--156}}
@collection{jones-lta,
  editor = jones-camel-b,
  publisher = oup,
  address = oxon,
  title = {The Life and Times of Aardvarks},
  booktitle = {The Life and Times of Aardvarks},
  year = 1992}

The idea is to ensure consistency by using strings for things which are used multiple times (or might be used multiple times). Titles are unique - or, rather, they do not vary in consistent, patterned ways. So they go in in the normal way as there is no point in bothering to define strings for them. But anything where consistency is an issue gets a string defined in the appropriate file of abbreviations.

Now, if I need to cite 'What Aardvarks Like To Eat', I of course need to load all of the relevant bib files in the appropriate order. That is, I need to load the files containing the strings before loading the .bib files which use them. For example:

\bibliography{authors,pub,places,zoo}

But that's just code you copy and paste into a new document, put in a template or load using \input.