The workshop on Extrinsic Parse Improvement (EPI) will provide a forum for researchers who work on improving parsing as pre- or post-processing steps, that is, for adapting and improving parses without modifying the parser itself. The focus of the workshop is on methods for modifying treebank representations, which can then be used to train parsers; on methods for automatically extending the training set with new examples; and on methods for improving parser output by post-corrections, or even on detecting reliable or deviant parses.
All these methods are used regularly for adapting parsers to new types of text, with different goals: either to adapt the syntactic annotation scheme to improve parsing performance (Manning and Klein 2003), to improve parser output (Hall and Novak 2011), to adapt a parser to new domains (Candito et al. 2011), to parse the web (Le Roux et al. 2012), to find errors in treebanks or parses (Dickinson and Smith 2011), or to discover reliable parses (Kawahara and Uchimoto 2008). However, many of these methods are used only in specific contexts. This workshop will bring together researchers from different areas to foster cross-fertilization between different areas of parsing correction and adaptation. We are especially encouraging young researchers to submit completed work as well as work in progress, which may profit from discussions with other researchers.
The areas of interest for this joint workshop include, but are not limited to, the following topics:
Submissions will be in form of short papers: 6 pages, including references, in pdf, following the most recent ACL guidelines. Submission will be via the EasyChair system: submit here.
Each paper will be reviewed by at least two reviewers. Reviewing will be blind. Thus, papers have to be anonymous. Accepted papers will be published online at CEURS. After the workshop, we plan to have a special issue on the workshop topic in the Journal of Language Modelling. For this, a separate call will be issued.
The program for the workshop can be found here
The invited talk will be by Feiyu Xu, on "Domain-Adaptive Parse Reranking with Automatically Learned Relation Extraction Rules". Information can be found here
The workshop will be held as part of ESSLLI in August 2013 in Duesseldorf, Germany. More information about ESSLLI can be found at the ESSLLI web pages
Tim Baldwin (University of Melbourne, Australia)
Marie Candito (University of Paris 7, France)
Jinho Choi (University of Massachusetts, USA)
Jennifer Foster (Dublin City University, Ireland)
Kim Gerdes (University of Paris 3, France)
Gülsen Eryigit (Istanbul Technical University, Turkey)
Jonas Kuhn (IMS Stuttgart, Germany)
Joseph Le Roux (Université Paris-Nord, France)
David McClosky (Stanford University, USA)
Stephan Oepen (University of Oslo, Norway)
Ines Rehbein (University of Potsdam Germany)
Djamé Seddah (University of Paris 4, France)
Reut Tsarfaty (Uppsala University, Sweden)
Josef van Genabith (Dublin City University, Ireland)
Fei Xia (University of Washington, USA)
For questions, contact Sandra Kübler (skuebler) or Markus Dickinson (md7) (both addresses at indiana.edu)