=======================================================================
NAACL HLT 2009 Workshop on
Active Learning for Natural Language Processing
June 5, 2009, Boulder, Colorado, USA
Submission Deadline: March 6, 2009
Endorsed by the following ACL Special Interest Groups:
* Special Interest Group on Natural Language Learning (SIGNLL)
* Special Interest Group for Annotation (SIGANN)
=======================================================================
MOTIVATION
Labeled data is a prerequisite for many popular algorithms in natural
language processing and machine learning. While it is possible to
obtain large amounts of annotated data for well-studied languages in
well-studied domains and well-studied problems, labeled data are rarely
available for less common languages, domains, or problems.
Unfortunately, obtaining human annotations for linguistic data is
labor-intensive and typically the costliest part of the acquisition of
an annotated corpus.
It has been shown before that active learning can be employed to reduce
annotation costs but not at the expense of quality. While diverse work
over the past decade has demonstrated the possible advantages of active
learning for corpus annotation and NLP applications, active learning is
not widely used in many ongoing data annotation tasks. Much of the
machine learning literature on the topic has focused on active learning
for classification problems with less attention devoted to the kinds of
problems encountered in NLP.
TOPICS
We are interested in bringing together researchers to explore the
challenges and opportunities of active learning for NLP tasks, language
acquisition, and language learning. General work on active learning on
NLP classification tasks, sequence labeling, parsing, semantics, and
other more complex tasks will be welcome in the workshop. More specific
topics of interest include, but are not limited to:
- theoretical analysis of active learning in the context of NLP
applications
- novel active learning approaches to estimate the training utility
of individual selection units
- cost-sensitive active learning approaches incorporating data
acquisition costs
- approaches to model or predict annotation costs as well as studies
on factors that influence annotation time
- criteria for stopping or monitoring progress of active learning
- overfitting of data acquired with active learning: how much is the
data biased towards the learning scheme involved in the selection
and what are the limitations of re-use with other learning schemes
- Human-Computer Interaction aspects of annotation including
requirements, impact of interface design on annotation time, and
methods to deal with reliability of annotators
- approaches to multi-task active learning
- approaches to deal with or reduce computational complexity of
active learning approaches including parallelization, issues of
pool- or batch-size, varying degrees of look-ahead, etc.
- active learning and domain adaption
- active learning compared to or combined with other semi-supervised
or even unsupervised learning approaches
- application of active learning in real annotation projects and
experiences gained thereby
SUBMISSIONS
We invite submissions of two kinds:
1. original and unpublished work as full papers, limited to 8 pages of
text (up to one extra page may be used for references);
2. position papers or papers describing ongoing work as short papers,
limited to 4 pages in total (including references).
Both kinds of papers will appear in the proceedings and will be
presented orally. As reviewing will be double-blind, author information
should not be included in the papers and self-reference should be avoided.
All submissions must be made in PDF format using the START paper
submission website:
https://www.softconf.com/naacl-hlt09/ActiveLearningNLP2009/
Submissions must follow the NAACL HLT 2009 formatting requirements:
http://clear.colorado.edu/NAACLHLT2009/stylefiles.html
Authors are strongly encouraged to use the LaTeX or Microsoft Word style
files available there. Papers not conforming to these requirements are
subject to rejection without review.
IMPORTANT DATES
March 6, 2009: Submission Deadline
March 30, 2009: Notification of acceptance
April 12, 2009: Camera-ready copies due
June 5, 2009: Workshop held in conjunctions with NAACL HLT
ORGANIZERS AND CONTACT
- Eric Ringger, Brigham Young University, USA
- Robbie Haertel, Brigham Young University, USA
- Katrin Tomanek, University of Jena, Germany
Please address any queries regarding the workshop to:
al.nlp2009 [at] googlemail.com
PROGRAM COMMITTEE
- Shlomo Argamon (Illinois Institute of Technology, USA)
- Jason Baldridge (University of Texas at Austin, USA)
- Markus Becker (SPSS, UK)
- Ken Church (Microsoft Research, USA)
- Hal Daume (University of Utah, USA)
- Robbie Haertel (Brigham Young University, USA)
- Ben Hachey (University of Edinburgh, UK)
- Udo Hahn (University of Jena, Germany)
- Eric Horvitz (Microsoft Research, USA)
- Rebecca Hwa (University of Pittsburgh, USA)
- Ashish Kapoor (Microsoft Research, USA)
- Mark Liberman (University of Pennsylvania/LDC, USA)
- Prem Melville (IBM T.J. Watson Research Center, USA)
- Ray Mooney (University of Texas at Austin, USA)
- Miles Osborne (University of Edinburgh, UK)
- Eric Ringger (Brigham Young University, USA)
- Kevin Seppi (Brigham Young University, USA)
- Burr Settles (University of Wisconsin, USA)
- Victor Sheng (New York University, USA)
- Katrin Tomanek (University of Jena, Germany)
- Jingbo Zhu (Northeastern University, China)