4

Recently, I've noticed many top conference papers in the CS field have extended versions submitted to and published in reputable journals. (At least in mobile computing)

While such extended versions generally retain the original paper's title, they sometimes have a slightly different title. Anyway, when I've read both the original and the extended versions, the main contribution of the latter seems to typically lie in additional experiments and analyses.

So, I am wondering that despite the underlying idea and methodology remaining largely consistent between both versions, are these journal-published extended versions of conference papers considered separate achievements?

If yes, extending conference papers seems like there would be significant advantages when seeking positions in both academia and industry. Nevertheless, why don't some people choose that way?

2 Answers 2

2

My answer is probably very opinionated.

If it came down to just bean-counting, a published paper is a published paper is a published paper. But "curriculum evaluation" in academia, and in industry, is (should be?) more than bean counting, and good hiring committees/managers should go beyond the number of papers, and should consider both the set of papers holistically, and each paper individually.

In my opinion, consistently publishing extended journal versions signals an attitude: one's desire to go deep about one own's results (by doing the extended analysis, experiments, extensions, ...), and one's desire to communicate in the best way possible (by adding examples, figures, intuition, roadmaps, ...), the best result they can achieve. This is an attitude that I personally find highly desirable in a colleague.

Thus, while an extended journal version could certainly be a separate contribution, it can also play the above additional role, which is not to be discounted: don't forget that departments do not necessarily try to hire the best researcher they can, they try to hire the best colleague they can.

0

Please note that different countries and different disciplines use different rules. EE departments sometimes only count journal publications and so do certain universities in certain countries. Publishing in conferences under these circumstances thus only serves to establish priority, a reputation, and derive benefits from attending a conference. In this case, only the journal version counts towards advancement and salary increases.

This of course does not apply to most Computer Scientists. Journal versions are (usually) extended. Typically, 30% of the journal version or more has to be new. So, an adversarial bean-counter that uses a conference publication as 10 points should give the journal version 15 points. The adversarial bean counter is of course acting ridiculously.

For longer time evaluation, citation counts are being used. Journal publications might be easier accessible and therefore higher visible, gathering additional citations. For a really good, important paper, the total number of citations is probably almost the same, whether the researcher published only in a conference or gave an extended journal version. However, researchers who cite the contribution will tend to cite the journal version as the more final word by the author. Thus, for evaluation by citation count, journal papers might only give marginal benefits.

Hopefully, evaluation is done by actual looking at the record and the individual contributions. In this case, as Matteo's answer shows, taking the trouble to generate a more definite version that takes the discussion at and after the conference into account is evidence for a much better attitude to obtaining and publicizing results.

To answer your question: The journal version is of course not an independent contribution. Writing it is difficult and tedious, and many people do not receive enough benefits. There is also a certain amount of excitement gained by solving a completely new problem, so spending time on writing a journal version is maybe not the best choice for the individual.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .