Lexical Semantic Change (LSC) detection, i.e. the task of identifying words that change meaning over time, is a very active research area, with applications in NLP, lexicography, and linguistics. Evaluation is currently the most pressing problem in LSC detection, as no gold standards are available to the community, which hinders progress. We organize a shared task that addresses this gap by providing researchers with an evaluation framework and high-quality manually annotated datasets for English, German, Latin, and Swedish.
Participants are asked to solve two related tasks for computational LSC detection, aiming to identify change in senses of words over time using corpus data.
* Sub-Tasks *
We provide two time-specific (t1, t2) corpora for each language. Both tasks will be evaluated using ground truth as annotated by humans, native speakers (except for Latin, for which scholars of Latin were used).
Sub-Task 1 - Binary classification: given a set of words, determine which gained or lost senses between t1 and t2 and which did not.
Sub-Task 2 - Ranking: given a set of words, rank them by the degree of LSC they underwent between t1 and t2.
Important Dates
Test data ready: February 19, 2020 (*NOW*)
Evaluation start: February 19, 2020 (*NOW*)
Evaluation end: March 11, 2020
Paper submission due: April 17, 2020
Notification to authors: June 10, 2020
Camera ready due: July 1, 2020
The complete list of dates is available on the task website.
Task Organizers
Barbara McGillivray, University of Cambridge and The Alan Turing Institute, UK
Dominik Schlechtweg, University of Stuttgart, DE
Simon Hengchen, University of Helsinki, FI
Haim Dubossarsky, University of Cambridge, UK
Nina Tahmasebi, University of Gothenburg, SE