Publication at Faculty of Mathematics and Physics |
2020
Abstract
In this paper, we present a new dataset for testing geometric properties of sentence embeddings spaces. In particular, we concentrate on examining how well sentence embeddings capture complex phenomena such paraphrases, tense or generalization.
The dataset is a direct expansion of Costra 1.0, which we extended with more sentences and sentence comparisons. We show that available off-the-shelf embeddings do not possess essential attributes such as having synonymous sentences embedded closer to each other than similar sentences with a significantly different meaning.
On the other hand, some embeddings appear to capture the linear order of sentence aspects such as style (formality and simplicity of the language) or time (past to future).