ACL 2001 PC Co-Chair Report Norbert Reithinger and Giorgio Satta 1. Submissions and Reviewing ---------------------------- The web pages for submission notification at acl2001.dfki.de were up and running at the beginning of December 2000. At the end of the registration deadline we had about 320 notifications with 260 papers being finally submitted. Due to the not so positive experience with last year's thematic session and the short time frame we had for this year, we didn't include those type of sessions this year. As a surprise to some authors we really kept very close to the submission deadline. Since our overall schedule was pretty short, we had no buffer for late submissions. We had approximately 10 papers rejected for this reason. The area chairs, Jennifer Chu-Carroll, Joshua Goodman, Pierre Isabelle, Adam Kilgarriff, Lillian Lee, Adwait Ratnaparkhi, Ehud Reiter, James Rogers and David Weir, selected 211 reviewers from 25 countries. Reviewing was blind to all of the reviewers. As in the last years, we started without disclosing author identities to the area chairs as well. Due to unexpected changes of paper distributions in the 9 areas, as compared with previous years, several area chairs rushed to get more reviewers after papers were already routed. It was then impossible for us to manage with the conflicts of interest that newly arose without slowing down considerably the reviewing process. So we decided to disclose author identities to the area chairs. After a posteriori discussion with all of the area chairs, we now believe that our decision is the one that minimizes the chances of incurring into conflicts of interest, since the area chairs have a much more detailed knowledge of possible relations between their own reviewers and the paper authors, as compared with PC co-chairs. To keep costs down, the PC meeting was held on April 1st-2nd, hosted by ITRI, Univ. of Brighton (many thanks to Donia for arrangements). All the 9 area chairs plus the two chairs attended. The area chairs and the reviewers all did a great job. We spent the week after April 2nd solving the double submission conflicts. Papers accepted at ACL-2001 and with double submission were divided into two categories: A. multiple submissions of same paper; and B. submissions to different meetings of papers with apparently close matchings. All cases of A. resulted in the confirmation of the ACL-2001 submission and the withdrawal of the other submission, with the exception of two papers. One such paper was already in the NAACL-2001 program and had to be rejected by us; the other paper got accepted in the meanwhile at ICML, and was withdrawn by the authors from ACL-2001. The B cases (three papers) were more difficult: we gave to these authors the opportunity of arguing their point, and then discussed each case with all of the area chairs. At the end, all of the B. papers were confirmed to be in the ACL-2001 program. The problem of Minimal Publishable Units, as we called them, should be given a second thought for later conferences. At the end of the whole process, we accepted 68 papers. The acceptance rate is thus 26.5%. 2. Breakdown of the Papers by Research Areas --------------------------------------------- The following table follows the break down according to area chairs. area accept reject submissions Lexicon and Semantics 4 (16%) 20 24 Machine Learning and Statistical Methods for Syntax 8 (26%) 22 30 Parsing Algorithms 10 (27%) 27 37 Generation and Multi-Modality 8 (30%) 18 26 Discourse and Dialogue 7 (18%) 30 37 Speech, Systems and Evaluation 6 (46%) 7 13 Syntax, Grammars, Morphology and Phonology 10 (30%) 20 30 Corpus-based Natural Language Processing 8 (21%) 30 38 Machine Translation, Multilinguality 7 (28%) 18 25 result 68 (26%) 192 260 3. Geographical breakdown ------------------------- This year, papers from 33 countries were submitted for the main program. Among the accepted papers 31 come from the European-African Zone (45.6%), 30 come from the American Zone (44%), and 7 come from the Asian-Pacific Zone (10%). The accepted papers were contributed by authors from 14 different countries. The details according to zone are as follows. We also include the top three submission countries for each zone, cont. country reject accept total ------------------------------------- America US 42 29 71 Canada 10 1 11 Mexico 2 2 total 55 30 85 Asia-Pacific Japan 15 5 20 China 4 4 HK 3 3 Korea 3 3 total 27 7 34 Europe-Africa Germany 23 6 29 UK 21 8 29 France 19 6 25 total 110 31 141 Overall 192 68 260 Compared to last year's ACL, the submissions from the Asia-Pacific region for the main program were reduced by two-thirds, while the submissions from Europe-Africa increased by almost 50%. The American submissions increased slightly. 4. Paper processing ------------------- For the first time, the submission format was close to the final one (two columns etc) to ease the transition from the submission version to the final one. The authors were also requested to submit an electronic version as backup. Only in cases where an assigned reviewer couldn't review a paper properly we used the e-version as backup. The e-version were however used to generate proposals for the areas chairs and the reviewers. About 75% of our area selections coincided with the first or second ranked area proposal from David's and Radu's program. Our thanks go to David Yarowsky and Radu Florian who contributed this service. Some people asked us to switch reviewing from sending around hardcopies to electronic distribution. We relied on hardcopies for this year. The reason maily was that we didn't have the infrastructure (scripts for automatic submission, unpacking etc) to handle an all electronic conference and were unable to set them up (or collect them from other sources) in the short time frame. ACL should think about building up standard tools that can be handed to PC chairs (of ACL, NAACL, EACL) to switch to electronic handling of submissions and final versions. For the final version we put LaTeX styles and Word .doc and .dot templates on the PC web site. This year we also worked closely with the publications chair Roberto Zamparelli to write very precise formatting instructions that authors should follow in the preparation of their finals, with the main goal of avoiding non-uniformity of paper format in the final proceedings. We have found that authors are not very sensible to this problem, and tend to ignore formatting instructions details. Several authors squeezed paper lines beyond our indications, or used smaller font sizes. Some authors even changed the .sty we distributed. We spent quite a lot of time in arguing with these authors, and due to the extremely small time interval between final submission and publisher dead-line, we had to give up on some cases. We think that further thought to this problem should be given for later conferences, especially in consideration of the fact that some authors pay extra money to have additional pages in the proceedings (as opposed to squeezing). For the production of the CD-ROM the authors were also requested to submit electronic versions that were also used as back ups.