<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; ">==============================================================<br> Final Call for Paper Submissions<br><br> NAACL-HLT 2010 Workshop on<br>Active Learning for Natural Language Processing (ALNLP)<br><br> June 5 or 6, 2010, Los Angeles, CA<br><br> <a href="http://www.active-learning.net/alnlp2010">http://www.active-learning.net/alnlp2010</a><br><br> Submission Deadline: March 8, 2010<div><span class="Apple-tab-span" style="white-space: pre; "> </span>(NOTE: extended from March 1)<br>==============================================================<br><br>Labeled training data is required to achieve state-of-the-art performance for<br>many machine learning solutions to NLP tasks. While traditional supervised<br>methods rely exclusively on existing labeled data to induce a model, active<br>learning allows the learner to select unlabeled data for labeling in an effort to<br>reduce annotation costs without sacrificing performance. Thus, active learning<br>appears promising for NLP applications where unlabeled data is readily<br>available (e.g., web pages, audio recordings, minority language data), but<br>obtaining labels is cost-prohibitive.<br><br>Ample recent work has demonstrated the effectiveness of active learning over<br>a diverse range of applications. Despite these findings, active learning has<br>not yet been widely adopted for many ongoing large-scale corpora annotation<br>efforts -- resulting in a dearth of real-world case studies and copious research<br>questions. Machine learning literature has primarily focused on active learning<br>in the context of classification, devoting less attention to issues specific to NLP<br>including annotation user studies, incorporation of semantic information, and<br>more complex prediction tasks (e.g. parsing, machine translation).<br><br><br>TOPICS<br><br>The aim of this workshop is to foster innovation and discussion that advances<br>our understanding in these and other practical issues for active learning in NLP.<br>Topics of particular interest include:<br><br> -- Alternative query types: labeling features rather than instances,<br> mixed-resolution queries for structured instances, etc.<br> -- Creative ways for obtaining data via active learning (e.g., online games,<br> Mechanical Turk)<br> -- Managing multiple, possibly non-expert annotators (e.g., "crowdsourcing"<br> environments)<br> -- Reusability: using data acquired with one active learner to train other<br> model classes<br> -- Domain adaptation and active learning<br> -- Multi-task active learning<br> -- Criteria for stopping and monitoring active learning progress<br> -- Active learning in coordination with semi-supervised or unsupervised<br> learning approaches<br> -- Interactive active learning interfaces and other HCI issues<br> -- Parallelization of active learning and its computational challenges<br> -- Software engineering considerations for active learning and NLP<br> -- Theoretical analysis of active learning<br><br>We also welcome case-study papers describing the application of active<br>learning in real-world annotation projects and lessons learned thereby.<br>Additionally, we would consider papers with insights applicable to NLP from<br>other machine learning communities (e.g., computer vision, bioinformatics,<br>and data mining), where annotation costs are also high.<br><br><br>SUBMISSIONS<br><br>We invite submissions of two kinds: 1. original and unpublished work as<br>full papers, limited to 8 pages (+1 extra page for references); 2. position or<br>work-in-progress papers, limited to 4 pages (including references). Both kinds<br>of papers will appear in the proceedings and presented orally. As reviewing<br>will be double-blind, author information should not be included in the papers<br>and self-reference should be avoided.<br><br>All submissions must be made in PDF format using the START paper<br>submission website:<br><a href="https://www.softconf.com/naaclhlt2010/alnlp/">https://www.softconf.com/naaclhlt2010/alnlp/</a><br>Submissions must follow the NAACL HLT 2010 formatting requirements:<br><a href="http://naaclhlt2010.isi.edu/authors.html">http://naaclhlt2010.isi.edu/authors.html</a><br>Authors are strongly encouraged to use the LaTeX or Microsoft Word style<br>files available there. Papers not conforming to these requirements are<br>subject to rejection without review.<br><br><br>IMPORTANT DATES<br><br>March 8, 2010: Paper Submission Deadline (extended from March 1)<br>March 30, 2010: Notification of acceptance<br>June 5 or 6, 2010: Workshop held in conjunction with NAACL-HLT<br><br><br>ORGANIZERS AND CONTACT<br><br>- Burr Settles, Carnegie Mellon University, USA<br>- Kevin Small, Tufts University, USA<br>- Katrin Tomanek, University of Jena, Germany<br><br>Please address any queries regarding the workshop to:<br> <a href="mailto:alnlp2010@gmail.com">alnlp2010@gmail.com</a><br><br><br>PROGRAM COMMITTEE<br><br>- Markus Becker (SPSS, UK)<br>- Claire Cardie (Cornell University, USA)<br>- Hal Daume III (University of Utah, USA)<br>- Ben Hachey (Macquarie University, Australia)<br>- Robbie Haertel (Brigham Young University, USA)<br>- Udo Hahn (University of Jena, Germany)<br>- Eric Horvitz (Microsoft Research, USA)<br>- Rebecca Hwa (University of Pittsburgh, USA)<br>- Ashish Kapoor (Microsoft Research, USA)<br>- Prem Melville (IBM T.J. Watson Research Center, USA)<br>- Ray Mooney (University of Texas at Austin, USA)<br>- Fredrik Olsson (SICS, Sweden)<br>- Foster Provost (New York University, USA)<br>- Eric Ringger (Brigham Young University, USA)<br>- Dan Roth (University of Illinois at Urbana-Champaign, USA)<br>- Burr Settles (Carnegie Mellon University, USA)<br>- Kevin Small (Tufts University, USA)<br>- Katrin Tomanek (University of Jena, Germany</div></body></html>