What do you mean, BERT? Assessing BERT as a Distributional Semantics Model

Timothee Mickus (1) , Mathieu Constant (1) , Denis Paperno (2) , Kees van Deemter (2)

1 ATILF - Analyse et Traitement Informatique de la Langue Française
2 Universiteit Utrecht / Utrecht University [Utrecht]

Contextualized word embeddings, i.e. vector representations for words in context, are naturally seen as an extension of previous non-contextual distributional semantic models. In this work, we focus on BERT, a deep neural network that produces contextualized embeddings and has set the state-of-the-art in several semantic tasks, and study the semantic coherence of its embedding space. While showing a tendency towards coherence, BERT does not fully live up to the natural expectations for a semantic vector space. In particular, we find that the position of the sentence in which a word occurs, while having no meaning correlates , leaves a noticeable trace on the word embeddings and disturbs similarity relationships.

Mots clés: distributional semantics; contextualized word embeddings; neural networks

Domaines: Informatique et langage [cs.CL]; Linguistique; Informatique

Timothee Mickus, Mathieu Constant, Denis Paperno, Kees van Deemter. What do you mean, BERT? Assessing BERT as a Distributional Semantics Model. Proceedings of the Society for Computation in Linguistics, 2020, 3, ⟨10.7275/t778-ja71⟩. ⟨hal-02484933⟩