1st Workshop on Learning to Generate Natural Language

LGNL 2017


Artificial Intelligence



The 1st Workshop on Learning to Generate Natural Language (LGNL) invites
papers of a theoretical and experimental nature on language generation and
their conditional variants. The event will be co-located with ICML in Sydney,
Australia. Relevant topics include (but are not limited to):
Content planning
Decoding strategies
Dialogue response generation
Discourse coherence
Evaluation of language generation
Image caption generation
Instructions generation
Language modelling
Machine translation
Multimodal language generation
Poetry/Lyric generation
Question answering
Semantic coherence
Story generation
Summarisation
Video description generation
Submissions
We welcome three categories of paper: regular workshop papers,
cross-submissions and extended abstracts. Only regular papers will be
included in the archival workshop proceedings.
Regular Workshop Papers
Authors should submit a paper of up to 4-6 pages long, with unlimited
reference and supplementary material pages. The submission should follow the
ICML 2017 style and formatting guidelines
(https://2017.icml.cc/Conferences/2017/StyleAuthorInstructions [1]). The
papers should present novel research. Reviewing will be double-blind, so
please do not include any self-identifying information in the submission.
Best Paper Prizes and Student Travel Grants
We are aiming to provide best paper prizes and student travel grants with the
exact number and amount of support to be decided.
Extended Abstracts
Preliminary but interesting ideas that have not been published before may be
submitted as extended abstracts. These should be up to 2 pages long. They
will not be included in the archival proceedings, and so are an ideal for
work which would benefit from additional exposure and discussion but is not
ready for publication. Reviewing will be double-blind, so please do not
include any self-identifying information in the submission.
Cross-submissions
In addition to previously unpublished work, we invite papers on relevant
topics which have appeared in alternative venues (such as NLP or speech
conferences). Accepted cross-submissions will be presented as posters, with
an indication of the original venue, but will not be included in the archival
proceedings. Selection of cross-submissions will be determined solely by the
organising committee.
Read more:
https://www.aclweb.org/portal/content/1st-workshop-learning-generate-natural-language-0
[1] https://2017.icml.cc/Conferences/2017/StyleAuthorInstructions