Bored of asking your embeddings to predict word similarity?
Looking for exciting challenges for your smartest model?
If you believe you deserve something new and cool, join the SemEval-2018 Task 10: Capturing Discriminative Attributes
Perceiving similarities is surely an important part of understanding word meaning, but spotting differences between similar items is equally crucial and often much more difficult.
Cappuccino and expresso are very similar because they contain coffee, but only cappuccino has milk and foam.
Strawberries and bananas are fruits, but only the former are red.
Lions and tigers are both dangerouns, wild animals, but only tigers have stripes.
Can your model discriminate the attributes that are shared by two concepts (e.g., coffee for Expresso and Cappuccino) from those that are typical just for one of them (e.g. foam for Cappuccino)?
If you really want to see whether your computational model is able to tackle these key aspects of meaning, make a change!
Join SemEval 2018 Task 10: Capturing Discriminative Attributes: https://competitions.codalab.org/competitions/17326
Be different by spotting differences!
Important dates:
Mon 08 Jan 2018: Evaluation start*
Mon 29 Jan 2018: Evaluation end*
Mon 05 Feb 2018: Results posted
Mon 26 Feb 2018: System description paper submissions due by 23:59 GMT -12:00
Mon 02 Apr 2018: Author notifications
Mon 16 Apr 2018: Camera ready submissions due
Organizers:
Alicia Krebs (Textkernel BV, Amsterdam, Netherlands)
Alessandro Lenci (Department of Philology, Literature, and Linguistics of the University of Pisa, Italy)
Denis Paperno (Lorraine Laboratory of Computer Science and its Applications (Loria, UMR 7503), National Center for Scientific Research (CNRS), France)