Basic Rules

Tips and tricks for writing a bibliography

What is the purpose of a bibliography?

A bibliography actually serves three different purposes:
  1. for the readers of your paper, to understand what is the paper you're talking about:
    contrary to what they did in [BSD03]...,
  2. for someone who tries to reproduce your results, to get back to what you used as a basis:
    We start with a classical photon splitting algorithm [BTF78]...
  3. for agencies and recruitment committees, to estimate the quality of your work:
    He has 6 publications in international journals, 10 publications in reviewed international conferences...
    or the impact of a journal:
    Papers published in the CGRT journal are cited 56 % of the time...
Given that we live in the Google age, for the first two purposes a basic amount of information (such as author and title) could be sufficient.

But some papers are not available in digital format (sad, but true). For these papers, the bibliography must give the information required to look them up in the library: title of the journal, volume, number, year, or name of the conference and year. Page numbers are a bonus, of course (especially if the conference proceedings are more than 800 pages...)

For the third purpose, it is important to cite things properly: if half the people cite a certain journal as "Real-Time Computer Graphics" and the other half as "Computer Graphics in Real-Time", even though it is the same journal, each of them will get half the credit.

Basic rules

Be accurate:

The information given in your bibliography must be sufficient to retrieve the paper easily and without hassles.

When you cite a paper in a journal, make sure the year, volume and issue numbers appear in the bibliography. It's not enough to ensure that they are in the bibtex file, as bibtex can decide not to use them (for various reasons, the most common being that a journal paper was entered as an @inproceedings instead of an @article).

[Doe01] Doe J.: The proper use of paradigms, AEM Transactions on Paradigms, vol 17, n° 2 (2001), 128-149.

Similarly, if you reference a technical report, you must include the institution and the number of the tech report. Your readers must be able to find the work you referenced.

[Doe01] Doe J.: Paradigms in Practice, Technical report n° CSD-01-753, Enormous State University, 2001.
(note that here the number contains the department (CSD for Computer Science Department). If that is not the case, you should state the name of the department as well.)

Page numbers are also useful. Some conference proceedings are more than 800 pages long, your readers will appreciate if you tell them to look at page 365 directly.

Be consistent:

If you cite several papers from the same conference, make sure they all appear with the same typesetting (same conference name, for example...) Here is an example of 4 papers, published at the same conference in 2003 and 2004, using bibtex entries available on the internet:
[AL04] Timo Aila and Samuli Laine. Alias-free shadow maps. In Proceedings of Eurographics Symposium on Rendering 2004, pages 161-166. Eurographics Association, 2004.
[MT04] Tobias Martin and Tiow-Seng Tan. Anti-aliasing and continuity with trapezoidal shadow maps. In Alexander Keller and Henrik Wann Jensen, editors, Proceedings of the 15th Eurographics Workshop on Rendering Techniques, Norköping, Sweden, June 21-23, 2004, pages 153-160. Eurographics Association, 2004.

[SYGM03] J. Stewart, J. Yu, S. J. Gortler, and L. McMillan. A new reconstruction filter for undersampled light fields. In EGRW '03: Proceedings of the 14th Eurographics workshop on Rendering, pages 150-156, Aire-la-Ville, Switzerland, Switzerland, 2003. Eurographics Association.

[WSP04] Michael Wimmer, Daniel Scherzer, and Werner Purgathofer. Light space perspective shadow maps. In Alexander Keller and Henrik W. Jensen, editors, Rendering Techniques 2004 (Proceedings of the Eurographics Symposium on Rendering 2004), pages 143-151. Eurographics, Eurographics Association, June 2004.
Similarly, if you cite several papers from the same author, make sure he always appears with the same name:

[Doe99] John Doe. Searching for the lost whale, volume 1. Large Publisher, 1899.
[Doe00] J. Doe. Searching for the lost whale, volume 2. Large Publisher, 1900.
[Doe01] John H. Doe. Searching for the lost whale, volume 3. Large Publisher, 1901.
[Doe02] J. H. Doe. Searching for the lost whale, volume 4. Large Publisher, 1902.
[Doe03] JH Doe. Searching for the lost whale, volume 5. Large Publisher, 1903.

Use the internet wisely:

It's not because you see something on the internet that it becomes automatically true, right? Well, the same applies to bibliographic references. It's not because you found it on the author's web site, or in a Digital Library, that it becomes automatically correct. Read the bibliographic references that you download, and use your wisdom. When in doubt, go back to the paper version (of the journal or the proceedings).

Most bibtex entries you find on the internet contain extra information (name of the publisher, address of the publisher, ISBN...). This information is not necessarily useful in a scientific paper. Don't hesitate to remove it (for example, by renaming the field address to alt_address).

On the other hand, the internet can be a great tool to find the original paper from John Doe in 1967, or to know the page numbers for this paper you're citing.

Know your history:

Some conferences change name, e.g. going from a Workshop to a Symposium. The change served a purpose, usually it's about the impact of the conference. In your paper, you must acknowledge it, by using the proper name (Symposium after the change, Workshop before).

Don't assume that things were always as they are. Conferences change name, they also change the way they are published. For example, SIGGRAPH is published as issue 3 of the current volume of ACM Transactions on Graphics. So SIGGRAPH 2006 will be jointly published as "ACM Transactions on Graphics, vol. 25, n° 3". But this is a recent agreement. SIGGRAPH 2000, for example, was not published that way.

I'll give you some historic information below.

Give all relevant information:

If a conference proceedings is also published as a journal, it is in order to enhance (or recognize) the status of the conference. It benefits to all the authors who publish at this conference. But it is important to acknowledge this in your bibliography. Use, for example:
[Doe04] Doe J.: The use of paradigms in Computer Graphics. ACM Transactions on Graphics (Proceedings of SIGGRAPH 2004) vol. 23, n° 3 (2004), 234-245.
This version includes both the fact that it was published as an article in the journal ACM Transactions on Graphics, and that this issue of the journal contains the proceedings of the SIGGRAPH 2004 conference. Both informations are important: someone looking for this paper will know he can find it either in the "ACM Transactions on Graphics" shelf or on the "SIGGRAPH Proceedings" shelf.

Be concise:

Give the conference the shortest name by which it can be identified without confusion. Concision is especially important when writing a conference paper, where you have a limit on the number of pages, and the intended audience is from the same research field.

For example:

[Doe93] Doe J.: A new paradigm shifting algorithm. In SIGGRAPH (1993), pp. 606-612.
is better than:
[Doe93] Doe J.: A new paradigm shifting algorithm. In Proceedings of the 18th annual conference on Computer graphics and interactive techniques (1993), pp. 606-612.
as most people will know what the first is, and may not know that the second version is its official name.


[Doe02] Doe J.: A paradigm shifting algorithm implemented on the GPU. In Graphics Hardware (2002), pp. 96-105.
is better than:
[Doe02] Doe J.: A paradigm shifting algorithm implemented on the GPU. In Proceedings of the 2002 ACM SIGGRAPH/EUROGRAPHICS conference on Graphics Hardware (2002), pp. 96-105.
This one has the other problem that the useful information (the name of the conference) appears only at the very end of the entry, after the name of the associations organizing it. I've seen people confuse the associations and the conference...

On the other hand, it is better not to use acronyms for the conference names. GH '03 does not give enough information about the conference name.

Corollary: it is usually not necessary to use the words "Proceedings of..." or "... Conference". The bibliographic styles treat journals and conferences differently, so from the typesetting it is obvious whether this is a conference or a journal.

Corollary 2: Bibtex has a field address, meaning the address of the publisher or of the association organizing the conference. When it is used, readers tend to confuse it with the address of the conference. It is preferable not to use it, as it is not helpful in locating the paper.

Know BibTeX:

I'm not asking you to be on a first name basis with Oren, but a basic understanding of how bibtex works will help (This section has grown so big that it now has its separate page).

The history

There are things to know about the various conferences and the way they are published. I putting this information together mostly for historical/cultural purposes.