[epe-users] instructions for the EPE 2017 proceedings volume

Stephan Oepen oe at ifi.uio.no
Mon Aug 28 22:22:57 CEST 2017


dear colleagues,

earlier today, we sent out acceptance notices for the application and
system descriptions from the EPE 2017 shared task.

we expect to gather a total of ten papers in the proceedings: a
high-level task summary, three application descriptions (events,
negation, and opinions), and six system descriptions.  two teams did
not submit a system description, and another two teams decided to
document their submissions in a joint paper.  we much look forward to
a strong collection of papers!

to increase cohesion in the presentation of EPE 2017 results, we would
like to kindly make a number of suggestions for all papers:

+ please always use the BibTeX references from the following file to
refer to, for example, the task summary, application descriptions, or
other participating systems:

  http://svn.nlpl.eu/epe/2017/public/epe.bib

+ as you will see, we suggest a uniform naming scheme for system
descriptions and, hence, would ask that you update the titles of your
manuscripts accordingly; in case you would prefer a different title,
please get in touch!

+ we will continue to update the EPE 2017 BibTeX entries, e.g. adding
page numbers once we see the breakdown of pages in the proceedings
volume; please use our ‘epe.bib’ without making local changes to the
file.

+ each system description should make sense ‘by itself’, i.e. it
should be possible to read individual papers outside the context of
the full proceedings volume.  please provide at least a paragraph or
two on the EPE task set-up, downstream applications, and metrics.

+ for the output representation(s) of your parser(s), please provide a
visual example, i.e. a dependency graph for one sentence (of suitable
complexity); where there are only minor differences between the
representations of different runs, it may be sufficient to give one
example graph and indicate minor variation in the running text.

+ as much as possible, please compare EPE 2017 results to intrinsic
evaluation metrics, both for your own parser(s) but maybe also for
others.  for participating systems where there are published
(intrinsic) results available, please reference these.

+ our reviewers have often given detailed and constructive comments.
please make sure you read their feedback carefully and try to address
their suggestions for clarification, extension, or revision.

+ please include the canonical web address for the EPE 2017 task in
the paper, where all submissions, results, and the complete evaluation
infrastructure will be made available for public download:

  http://epe.nlpl.eu

—so much for tonight :-).  we will need everything in START by monday,
september 5.

with thanks in advance, oe



More information about the epe-users mailing list