2011Q3 Reports: Program Chairs

From Admin Wiki
Jump to: navigation, search

Program Chairs (Rada Mihalcea and Yuji Matsumoto)

ACL 2011 received a total of 1,146 papers, out of which 634 were submitted as long papers and 512 were submitted as short papers (these are the figures after removing all non-qualifying submissions; the original numbers were 672 submissions for long papers and 531 for short papers). 25.9% of the long paper submissions and 25.0% of the short paper submissions were accepted for presentation at the conference. 116 of the long papers and 56 of the short papers will have an oral presentation, and 48 of the long papers and 72 of the short papers will be presented as posters. The papers were selected by a program committee of 27 area chairs, from Asia, Europe, and North America, assisted by a panel of 648 reviewers.

We had almost the same number of long paper submissions as last year (ACL 2010), but a significantly larger number of short papers; this is probably explained by the fact that the deadline for short papers was set after the notification of long papers, and also by the smaller number of events (and thus opportunities to submit papers) compared to last year.

The following table shows the number of submissions in each area for long and short papers, as well as the number of papers accepted in each area.

Areas Long Received Long Accepted Short Received Short accepted
Discourse, Dialogue, and Pragmatics 31 9 22 5
Information Retrieval 26 5 14 1
Language Resources 32 9 26 7
Linguistic Creativity 20 4 18 6
Machine Translation: Methods, Applications and Evaluations 33 11 40 12
Machine Translation: Statistical Models 56 15 43 11
Multilinguality 24 6 20 5
Natural Language Processing Applications 30 7 32 10
NLP for Web 2.0 31 8 18 2
Phonology/Morphology, Tagging and Chunking, Word Segmentation 34 10 26 8
Question Answering 19 3 11 1
Semantics 48 13 40 6
Sentiment Analysis, Opinion Mining and Text Classification 45 12 36 10
Spoken Language Processing 20 5 17 5
Statistical and Machine Learning Methods 42 11 31 7
Summarization and Generation 36 7 37 9
Syntax and Parsing 56 15 47 15
Text Mining and Information Extraction 51 14 34 8
TOTAL 634 164 512 128

One problem we encountered was a fairly large number of papers that violated the anonymity requirements by including the names of the authors on the first page (about 15 papers were in this category), as well as several papers that violated the length requirements (about 15 more papers were in this category). We rejected all the papers that violated the anonymity requirement. We asked the authors of papers with less than half page of excess content to shorten their papers without changing the content; the papers that exceeded the required length by a page or more were rejected without a review.

To achieve the goal of a broad technical program, we followed the initiative from last year and solicited papers under four different categories: "theoretical computational linguistics," "empirical/data-driven approaches," "resources/evaluation," and "applications/tools." We also continued to accept other types of papers (e.g., surveys or challenge papers), although unlike the previous year, no separate category was created for these papers. The papers falling under one of the four categories were reviewed using specialized reviewed forms; we also had a general review form that was used to review the papers that did not fall under one of the four main categories. The set of review forms can be found at http://acl2011.org/call.shtml#submission.

A new initiative this year was to also accept papers accompanied by supplemental materials (software and/or data) described in the paper. In addition to the regular review of the research quality of the paper, the accompanied resources were also reviewed for their quality. Papers that are submitted with accompanying software/data received additional credit toward the overall evaluation score, and acceptance or rejection decision was made based on the quality of both the research and the software/data component. Among all the submissions, a total of 84 papers were accompanied by a software package and 117 papers were accompanied by a dataset. Among all the accepted papers, 30 papers are accompanied by software and 35 papers are accompanied by a dataset. These materials will be hosted on the ACL web site under http://www.aclweb.org/supplementals.

A problem that we encountered with papers that had accompanying resources was that some authors had difficulty in anonymizing the resources. Some reviewers pointed out that they could find out the identity of the authors or the group that the authors belonged to. We did not reject the papers unless the anonymity violation was apparent; in fact, no paper was rejected because of an anonymity violation of resources.

The program will consist of five parallel sessions for oral presentations, and one poster session during the evening of the second day of the conference. There will be a total of 172 oral presentations, which include 56 short papers and 116 long papers. The poster session will include presentations of both long and short papers, for a total of 128 posters.

Based on the recommendation received from the student session co-chairs (Miles Osborne and Thamar Solorio), it was decided that all the student session papers will also be presented during the poster session.

As in previous years, there will be three awards, one for the best long paper, one for the best long paper by a student, and one for the best short paper. The candidates for the best paper awards were nominated by the area chairs, who took into consideration the feedback they received from the reviewers on whether a paper might merit a best paper prize. From among the nominations we received, we selected the top five candidates for the long and short papers, and the final awards were then selected by the area chairs together with the program co-chairs. The recipients of the best paper awards will present their papers in a plenary session during the second day of the conference.

We are delighted to have two distinguished invited speakers: Dr. David Ferrucci (Principal Investigator, IBM Research), who will talk about his team's work on building "Watson" - a deep question answering system that achieved champion-level performance at Jeopardy!, and Lera Boroditsky (Assistant Professor, Stanford University), who will give a presentation on her research on how the languages we speak shape the way we think. In addition, the recipient of the ACL Lifetime Achievement Award will present a plenary lecture during the final day of the conference.