home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!edcastle!aisb!eagle!reiter
- From: reiter@eagle.ed.ac.uk (Ehud Reiter)
- Newsgroups: comp.ai
- Subject: Conf vs Journals (Was IJCAI Reviewing)
- Message-ID: <1993Jan24.134658.18841@aisb.ed.ac.uk>
- Date: 24 Jan 93 13:46:58 GMT
- References: <C10q84.390@cpsc.ucalgary.ca> <POLLACK.93Jan21181428@dendrite.cis.ohio-state.edu>
- Sender: news@aisb.ed.ac.uk (Network News Administrator)
- Reply-To: reiter@aisb.ed.ac.uk (Ehud Reiter)
- Organization: Dept AI, Edinburgh University, Scotland
- Lines: 33
-
- In article <POLLACK.93Jan21181428@dendrite.cis.ohio-state.edu> pollack@cis.ohio-state.edu writes:
- >The importance with which we take publication decisions for AI
- >conferences is phenomenal ... Because of an early dearth of
- >refereed journals in the field, many people have come to believe, and
- >stake their career on, the idea that rapid publication of notes in the
- >conferences should count as much as archival journal articles.
-
- This point has been made before (even on this newsgroup), and people in general
- seem to agree that `it would be nice' if finished work appeared in journals,
- while conferences were reserved for work-in-progress, speculative ideas, etc.
- Unfortunately, nothing seems to change. The situation even seems to be
- getting worse in some respects, e.g., in their justification for their
- policy against multiple submissions, the IJCAI committee have now specifically
- said that they regard IJCAI papers as `archival' publications of finished work,
- similar to journal articles.
-
- Within my own subfield of Natural Language Processing, the main journal,
- COMPUTATIONAL LINGUISTICS, has been encouraging people to submit short
- `squibs' of roughly conference-paper size and with conference-paper
- turnaround times - but I don't see many of these appearing. Of the two
- best-known conferences in the field, one, ACL, takes a selective approach
- (25% acceptance rate, 35 papers in all), while the other, COLING, takes a
- more `anything vaguely plausible is accepted' approach (50% acceptance rate,
- 200 papers in all); and although COLING is probably more in line with the
- general idea of what conferences should be like in scientific domains,
- ACL has become and remains a lot more prestigious.
-
- I suspect that in the end AI will evolve more towards the scientific norm, but
- it will be by a haphazard process pushed by external factors (e.g., standards
- of tenure review committees), not a carefully planned change.
- --
- Ehud Reiter
- (e.reiter@edinburgh.ac.uk)
-