送交者: BerkeleyWolf 于 2006-3-21, 21:22:57:
I feel there are many misunderstandings in computer
science publication in previous discussions. Maybe
previous posters don't have enough time to elaborate.
Nevertheless, I outline few things as follows:
(1) Conference papers are very important in
computer science. The statement that “don’t count
conference papers” is simply not true in computer
science. If we don’t count conference papers, I am
sure most assistant professors in the U.S. can not get
tenure.
(2) Journal papers are also very important in
computer science. ACM and IEEE have a series of
journals you can extend and publish your significant
results. However, the journal publication time
duration is long and some significant work will never
appear in a journal version.
(3) A first-class conference paper is more
important than second-class journal papers. To support
this statement, I have to define “second-class”
journals. As a rough approximation, any non
ACM/IEEE/USENIX transactions, major magazines are
second class. Please be noted that it is only an
approximation, there are always exceptions.
(4) A first-class journal paper is better than
second class conference papers.
(5) Not all IEEE conferences are good. Some are
really bad.
(6) In general, ACM conferences have a higher
quality than IEEE conferences. Especially most SIG
conferences are first class conference. E.g. SIGCOMM,
SIGGRAPH, SIGMOD, SIGCHI, blab la…
(7) Acceptance rate is only an approximation in
evaluating the importance of the publication. We
should consider the self-selection (few people are
willing to waste time to submit if the hope is zero)
in a first class conference and significant watering
in a second class conference. Sometimes, a first class
conference has a higher acceptance rate than a second-
class.
(8) Impact factor is not an great indicator in
computer science. We can never compare the impact
factor to magazines like Science and Nature.
(9) Citations is a good approach, however, we
should carefully distinguish original research papers
v.s. review articles.
(10) Nature and Science magazine are not the venue
for computer science publication, letting along CELL
etc. I read few papers in CS in Nature and Science. I
should admit that they are good research. However,
they are not the most important ones in CS.
(11) Judge the ranking of publication is very
subjective. It depends on years of training in this
field. I outline my thinking in networking field, and
don’t hope everybody will agree.
General interest: JACM (very rare Networking papers),
CACM (greatest 10+ years before, but towards common in
recent years, still great).
1 class Journal /conf: ACM/IEEE TON == SIGCOMM, ACM
TOSN, IEEE TOC, IEEE JSAC
1.5 class conf: INFOCOM, ICNP, IPCN, Sensys, MobiSys,
MobiHoc, Mobicom, ICDCS, PPoPP, ICC, SECON, Milcom,
USENIX, IMC, SIGMetrics, middleware
blab la….
(12) For a compressive story of measure research,
please read Douglas Comer’s article “Ways To Measure
Research” at
http://www.cs.purdue.edu/homes/dec/essay.research.measure.html