Today we're looking at an article on Path splicing that appears in Proc. ACM SIGCOMM [more on that later] in August of '08. This is one of the articles that is available through Nick Feamster's website.
This article was written with Mutaza Motiwala, Megan Elmore, and Santosh Vempala. Motiwala and Elmore are students of his while Santosh Vempala is a fellow professor at Georgia Tech's College of Computing/School of Computer Science. Vempala is the Director of the Algorithms and Randomness Center and Think tank. The article is on the design and evaluation of path splicing and on where this can be applied.
Looking at the References for the article they are predominantly conference and symposium works. There are a few Internet drafts [Internet drafts are exactly that- drafts of articles that appear on the Internet. They are not necessarily in final form and have not been formerly published, work may still be ongoing]. Also used are a few white papers [a marketing tool used to explain a new product. They could contain how the product/technology was developed, evaluated, and where the product can be best used. White papers may include or accompany FAQs and Spec sheets].
I was surprised by the presence of a few works from early in the decade. In the rapidly changing technology field I would think that anything over a year old would be outdated. This would explain the heavy use of conference papers and the use of Internet drafts and white papers. These would be the sources for the latest work and developments. The older works were mentioned in the Related work section, where they surveyed "related work in three areas- multihoming and mulitpath routing, fast recovery schemes, and overlay networks" (Section 3). So the older works were used to evaluate ideas that had already come up.
2 years ago
I really liked that about the outdated material still being posted. It reminds me of astronomy. How each new theory was based upon or an attempt to replace the older material. Science is all about building on past ideas and theories. It's nice to see that they keep even the "outdated" material.
ReplyDeleteMaybe "outdated" is a bit too strong - 'older' yes ... but not obsolete. :-)
ReplyDeleteI am finding similiar results with my scientist, whose research focuses on designing algorithms. The resources he uses from fields other than mathematics stem mainly from conferences and symposium proceedings. His mathematic resources come mostly from journal articles and are a few years older than conference papers. It seems helpful to his research to use older articles because in math, researchers are always building on past findings and mathematical formulas that have probably been around for several years. Perhaps your scientist has similiar needs.
ReplyDelete1) Note that conferences in computer science are the rough equivalent of journals in other fields in terms of the maturity of research published in them and the degree of peer-review to which they're subjected. It's a peculiarity of CS.
ReplyDelete2) A year is close to the minimum time you'd expect to see for citations -- the publication cycle for, say, SIGCOMM is submissions in January, paper publication in August.
3) Fundamental, important things change a little more slowly than you might think, even in computer science. :)