*apologies for cross-posting*
TIAD-5 shared task - 2nd call for participation - Translation Inference Across
Dictionaries @ LREC 2022
We are pleased to invite you to take part in the fifth shared task for
Translation Inference Across Dictionaries (TIAD
2022<https://tiad2022.unizar.es/>), that will be held in the framework of
GLOBALEX Workshop on Linked Lexicography (GWLL
2022<https://globalex2022.globalex.link/lrec2022/>) in conjunction with the
13th Language Resources and Evaluation Conference (LREC
2022<https://lrec2022.lrec-conf.org/en/workshops-and-tutorials/ws-tut-schedule/>)
in Marseille (France) on June 20, 2022.
TASK DEFINITION
The aim of the TIAD shared task is to explore and compare best methods and
techniques to infer translations indirectly between language pairs, in the
context of a coherent experiment that enables reliable validation and
comparison of the results and to stimulate and enhance further research on the
topic. The resources consist of datasets from the Apertium RDF graph
(v2.0<https://zenodo.org/search?page=1&size=20&q=%22Ontolex-lemon%20and%20TIAD%20versions%22>),
which includes 53 RDF bilingual dictionaries across 43 languages. Such
techniques should help in auto-generating new bilingual and multilingual
dictionaries based on existing ones.
The participating systems are asked to generate translations indirectly among
three languages - English, French, Portuguese - based on already known
translations contained in the Apertium RDF graph, where these languages are not
connected directly.
Participants will apply their methodologies to discover indirect translations
(mediated by any other language in the graph) between the pairs EN/FR, FR/PT
and PT/EN, and may also make use of other freely available sources of
background knowledge to improve performance, as long as no direct translations
among the final language pairs are used.
Evaluation of the results will be carried out against manually compiled pairs
from the Global Series<https://lexicala.com/resources#dictionaries> of K
Dictionaries - Lexicala.
PUBLICATION OF RESULTS
Participants will submit a system paper including a description of their
system, the way the data were processed, the applied algorithms, the obtained
results, as well as the conclusions and ideas for future improvements. The
papers will be peer reviewed prior to publication to confirm that all aspects
are well covered.
The workshop will accept also regular papers from participants who are not
taking part in the shared task but have worked in the area of translation
inference and want to publish novel results or ideas, possibly with different
datasets and experimental basis from those proposed here.
Please submit a 1,000-word abstract, which will be evaluated based on
scientific merit. Authors of accepted abstracts will be invited to submit a
full version of the paper.
Papers should have 4-8 pages, according to the LREC submission format. The
accepted papers will be published as part of the LREC workshops proceedings and
presented during the Globalex workshop. Paper submission will go through the
START system<https://lrec2022.lrec-conf.org/en/submission2022/authors-kit/>.
IMPORTANT DATES
10/01/2022 - Technical description of evaluation process and data provided by
the organisers
10/04/2022 - Submission of results by participating systems
29/04/2022 - Evaluation results communicated by organisers
27/05/2022 - Submission of system description papers
20/06/2022 - Globalex Workshop on Linked Lexicography day
NOTE: for regular papers, see the 'submission and timetable' section of GWLL
2022<https://globalex2022.globalex.link/lrec2022/>.
ORGANISERS
* Jorge Gracia, University of Zaragoza, Spain
* Besim Kabashi. Friedrich-Alexander-University of Erlangen-Nuremberg,
Germany
In conjunction with the GWLL 2022 organisers.
REVIEW COMMITTEE
See https://globalex2022.globalex.link/lrec2022/.
WEBSITES AND CONTACT
A full description of TIAD-2022 is available at https://tiad2022.unizar.es/.
The GWLL 2022 website is at https://globalex2022.globalex.link/lrec2022/.
For queries, contact Jorge Gracia jogracia@xxxxxxxxx<mailto:jogracia@xxxxxxxxx>.