Primerjava slovenskih besednih vektorskih vložitev z vidika spola na analogijah poklicev

In recent years, the use of deep neural networks and dense vector embeddings for text representation have led to excellent results in the field of computational understanding of natural language. It has also been shown that word embeddings often capture gender, racial and other types of bias. One of the methods for assessing the quality of word embeddings are analogies calculations. The article focuses on the evaluation of Slovene word embeddings in terms of gender. We compiled a list of male and female equivalents of occupations and evaluated the gender bias of fastText, word2vec and ELMo embeddings