ACL-2000 PC CO-CHAIRS (K Vijayashanker and Chang-Ning Huang) =========================================================== ACL-2000 PC Co-Chair report A. The Program -------------- For this year's ACL, 267 submissions were received. Of these 201 were submitted to the general session and 65 to the theme sessions. A total of 70 papers were accepted, 53 from the main session and 17 from the thematic sessions. Thus the acceptance rate for the entire conference was 26.2 percent with the rate being 26.4 and 26.2 percent for the general and thematic sessions. In addition to the two keynote speakers (at the welcome reception), we have three invited talks during the conference. A1. Thematic Sessions --------------------- We liked some of the reasons behind special thematic sessions introduced at last year's conference. We discussed the thematic sessions with last year's PC co-chairs particularly about the problems they encountered in implementing the themes. One problem they had was with the number of submissions for some of the themes. Therefore, we decided to ask that the theme proposers offer reasons why they would expect that the proposed theme area would be viable and likely to generate a good number of submissions. We accepted four of the six theme proposals we received. We thought these would generate sufficient interest and met some of the reasons underlying the thematic sessions. We believe that the number of submissions did mostly meet our expectations. The following gives the numbers of submissions and acceptances in each theme. Theme Submissions #ACC % of papers accepted Question-Answering 6 2 33% ML in Dialogue 8 2 25% Summarization 13 4 30.8% Proc. of Asian Languages 38 9 23.7% Total 65 17 26.2% A2. Breakdown by Research Areas (General Session Only) ------------------------------------------------------ It was hard to decide how to breakdown the numbers by areas. Finally, to simplify our task, we decided to break it down according to the area chairs. Area Submissions #accepted %accepted Syntax, Grammars 15 4 26.7% and morphology Machine Translation 22 5 22.7% and Multilinguality Machine Learning and 31 11 35.5% Statistical Methods (Syntax) Speech, Systems, and 20 3 15.0% Evaluation Discourse, Dialogue, 35 10 28.6% and Generation Lexicon and Semantics 21 5 23.8% Corpus-based and Statistical 36 8 22.2% NLP (not syntactic) Parsing Algorithms and Models 23 8 34.8% Total 201 53 26.4% A2. Geographical Breakdown -------------------------- We are pleased to report that the time of internationalization efforts of ACL, we received papers from around the world. The numbers below reflect the host countries of the contact authors. Continent Countries Conference Count Submission Acceptance % Submissions Acceptance % Americas USA 68 32.3 total 73 32.9 Australasia Japan 44 22.7 China/HK 26 26.9 Korea 14 35.7 total 97 23.7 Europe Germany 25 28.0 UK 19 21.1 France 15 13.3 Spain 7 42.9 total 95 22.1 The substantial increase in number of submissions from Asia can be attributed partly to the location of the conference, and the inclusion of a theme focussing on NLP techniques especially for Asian Languages. However, it must be noted that nearly all papers from Australasia were from China, Japan, and Korea. We were surprised by the low number of submissions from some Australasian countries such as Australia, Singapore, Taiwan. B. The Process -------------- Reviewing was blind. The program committee comprised of eight area chairs who were assisted by a team of reviewers. The thematic session submissions were reviewed by individual theme chairs and their team of reviewers. All reviews and theme chair recommendations were available for the program committee deliberations. For each theme paper, an area chair was responsible for leading the discussion at the program committee meeting. We used hard copy for paper distribution and email for reviewing. Despite realizing that mailing incurs considerable expenses. We chose not to go for electronic submission because (1) we did not have sufficient support for this, and more importantly (2) we expected an increased number of papers that might use fonts which often create problems, and during this stage of the process, we don't have sufficient time to deal with so many submissions. In contrast for the submission of the final version of the selected papers, we felt we would have more time to deal with any problems with electronic versions (for the CDROM version of the proceedings). This was indeed the case. We encountered such problems while printing/examining the pdf files of accepted papers. We were able to handle these problems as the time pressure was not as intense at this stage and furthermore the number of papers was only around a quarter of the original number of submissions. Prior to the submission, all authors had to pre-register their papers using a web-based form. Greg Silber developed a software package for us that included this. Upon submission of information, each submitter was immediately informed of a paper-id which was confirmed in a subsequent email. In addition, various files (html, text and spreadsheets) recording different pieces of the entered information (either collectively for all submissions or individually for each submission) were produced. These were especially useful for us during the entire process. We used a program developed by David Yarowsky that produced a ranking of area chairs/reviewers for papers. This information was useful for us in making our decisions of assignment of papers to reviewers. C. Reviewing ------------ We believe that overall the area chairs, theme chairs and the reviewers were conceintcous and did a very good job. The chairs made sure that at least three reviews were received for each paper and on a few occasions requested the reviewers to provide additional comments/elaborate. The reviewers were also requested to keep in mind the high number of submissions from non-native-English-speaking authors and to be constructive in their remarks regarding language use. D. Proceedings -------------- We asked the authors of accepted papers to send us a hard-copy of the paper for the printed proceedings and a pdf file for the CDROM proceedings. The hard-copies were put together in order and assigned page numbers and the prefatory pages created. These were shipped to Hong Kong for final production. Despite very good support from Samuel Harp at Hong Kong, we wondered whether it might have been simpler to have continued using OmniPress given their knowledge and experience in producing the ACL proceedings. The time difference and distance were the main problems with the arrangement used this year. As mentioned above, there were a substantial number of pdf files where there were problems with inclusion of fonts. Also there were quite a few cases where all the pages wouldn't print etc. However, with cooperation from the authors, these were sorted in time. All of remaining items in the production of the CDROM version were handled by David Yarowsky. We are very thankful to David Yarowsky for managing this task and for doing a wonderful job. E. Other Items -------------- While we were not unsatisfied with the number of submissions to the theme sessions, we were a little concerned with the unevenness of the reviewing of the theme papers. It would be nice to have a clear demarcation of the responsibilities with the new conference committee organization. Likewise a repository of files used and instructions on actions to be taken by the various chairs would be useful.