CSH researcher Anna Di Natale will give a talk at the Natural Language Processing Group at University of Utrecht on December 3, 2020.
Details on how to join will be announced here.
Abstract:
Colexification is a linguistic concept that has recently been used to assess the semantic similarity of words. Unfortunately, such assumption lacks validation at scale. In this talk, I consider one form of meaning, affective meaning, and show that colexification networks follow such relationship in the creation of links between words. Moreover, I explore the possible applications of such properties, in particular the unsupervised expansion of pre-existing affective lexica and the interpolation of affective meaning of words. I then compare the performance of the network method with a state of the art machine learning corpus method.