Difference between revisions of "Suggestions"

From ACL Wiki
Jump to: navigation, search
(Suggestions)
(Suggestions)
Line 2: Line 2:
  
 
* The capacity for individuals to pursue research in NLP often depends on access to annotated corpora, but this is difficult for isolated researchers, especially those at small institutions or based in countries with weak currencies.  Could ACL mediate access to a selection of standard non-free corpora, perhaps via a special membership status?  (Steven Bird, May 23, 2008)
 
* The capacity for individuals to pursue research in NLP often depends on access to annotated corpora, but this is difficult for isolated researchers, especially those at small institutions or based in countries with weak currencies.  Could ACL mediate access to a selection of standard non-free corpora, perhaps via a special membership status?  (Steven Bird, May 23, 2008)
 +
 +
* Effective teaching of CL is difficult given diverse student backgrounds and the complex resource requirements.  Evidence for what methods work in which contexts is largely anecdotal.  The ACL could commission a working group to make a more systematic study of CL pedagogy, and prepare a report comparable to the one that exists for Computer Science education http://www.sigcse.org/cc2001/  (Steven Bird, May 23, 2008)
  
 
* I would suggest that ACL conferences adopt an author response phase during reviewing where authors can see initial reviews and post a short response to correct errors and misunderstandings in the reviews before a final decision is made.  Author response is now widely used in many related conferences such as AAAI, ICML, and NIPS, where as an author and a reviewer I have found it to be a valuable addition to improving the reviewing process. Many members of the SIGDAT board have recently suggested it be established for EMNLP-08. (Ray Mooney, May 19, 2008)
 
* I would suggest that ACL conferences adopt an author response phase during reviewing where authors can see initial reviews and post a short response to correct errors and misunderstandings in the reviews before a final decision is made.  Author response is now widely used in many related conferences such as AAAI, ICML, and NIPS, where as an author and a reviewer I have found it to be a valuable addition to improving the reviewing process. Many members of the SIGDAT board have recently suggested it be established for EMNLP-08. (Ray Mooney, May 19, 2008)

Revision as of 21:44, 22 May 2008

Suggestions

  • The capacity for individuals to pursue research in NLP often depends on access to annotated corpora, but this is difficult for isolated researchers, especially those at small institutions or based in countries with weak currencies. Could ACL mediate access to a selection of standard non-free corpora, perhaps via a special membership status? (Steven Bird, May 23, 2008)
  • Effective teaching of CL is difficult given diverse student backgrounds and the complex resource requirements. Evidence for what methods work in which contexts is largely anecdotal. The ACL could commission a working group to make a more systematic study of CL pedagogy, and prepare a report comparable to the one that exists for Computer Science education http://www.sigcse.org/cc2001/ (Steven Bird, May 23, 2008)
  • I would suggest that ACL conferences adopt an author response phase during reviewing where authors can see initial reviews and post a short response to correct errors and misunderstandings in the reviews before a final decision is made. Author response is now widely used in many related conferences such as AAAI, ICML, and NIPS, where as an author and a reviewer I have found it to be a valuable addition to improving the reviewing process. Many members of the SIGDAT board have recently suggested it be established for EMNLP-08. (Ray Mooney, May 19, 2008)
  • ACL should start giving more space (back?) to models that incorporate linguistic insights, and to work in areas such as discourse and dialogue, where the issue of evaluation can only rarely be answered by reporting results on available corpora. Linguistic insights are sometimes valuable even if they don't improve performance. In discourse/ dialogue / NLG, meaningful available corpora are rare, even if one wanted to use them for evaluation. Evaluation of such work can be carried out from many different points of view, foremost among those, IMO, user evaluations. (Barbara Di Eugenio, May 19, 2008)
  • The ACL Anthology should be augmented to contain the complete back issues of "the Finite String" from Vol. 1, No. 1 (Jan., 1964). The issues contained in J79 (the microfiche issues, starting with Vol. 11) should be separately pointed to from a new "the Finite String" identification line. I've many of the paper copies and can scan them for ACL. (Robert A. Amsler, May 14, 2008)
  • In recognition of the amount of effort they have put in organizing their workshop, workshop chairs should not have to pay registration fees to attend their own workshop (Dragomir Radev, May 1, 2008)
  • Authors of papers should, wherever feasible, include in their papers a URL from which source code and/or data relevant to their papers can be downloaded (Dragomir Radev, May 1, 2008)
  • It would be nice if part of the waived "student volunteer" package at least included tutorials (which are of big interest to junior students), if not also workshops. (Hal Daume III, May 14, 2007)

Notes

  • Progress on the ideas proposed in 2007 (an open access journal, a video archive, and the extension of the ACL anthology) will be reported on at the 2008 ACL business meeting.
  • Suggestions entered before May 31, 2008, will be considered for discussion at the ACL business meeting at ACL 2008
  • Suggestions entered after May 31, 2008, will be considered for ACL 2009