Computing distributed representations of words using the CoRoLa corpus

摘要

We investigate the usability of the CoRoLa corpus for generating high quality vectorrepresentations of words for Romanian language. Different model parameters are tested and modelquality is compared in three test cases: two word analogies data sets and a word similarity correlationwith human judgment. Furthermore, we prove that CoRoLa provides superior word representationscompared to other known Romanian corpora, such as the Wikipedia corpus.