Word Embeddings, Analogies, and Machine Learning: Beyond king - man + woman = queen

By Aleksandr Drozd, Anna Rogers, Satoshi Matsuoka. 2016

174 citations recorded by the source corpus.

Subfields: Natural Language Processing, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

In Representation & Self-Supervised Learning

Citations in the corpus

Fair Is Better than Sensational: Man Is to Doctor as Woman Is to Doctor (2019, methodology): “Also, Drozd, Gladkova, and Matsuoka (2016) have shown that the original and classically used 3 COSADD method (Mikolov et al. 2013) is not able to capture all linguistic regularities present in the embeddings.”
Inducing Relational Knowledge from BERT (2019, methodology): “The extent to which vector differences and other linear transformations between two words capture their relationship was also analyzed by subsequent works (Vylomova et al. 2016; Drozd, Gladkova, and Matsuoka 2016; Bouraoui, Jameel, and Schockaert 2018).”
Measuring Social Biases in Grounded Vision and Language Embeddings (2020, background): “Such relative distances between word vectors indicate how related two concepts are and are directly used in many natural language processing tasks, e.g., analogy completion (Drozd et al., 2016).”
The (too Many) Problems of Analogical Reasoning with Word Vectors (2017, methodology): “We have already mentioned two problems with the task: the problem of the Google test scores being ﬂattering to the VSMs (Gladkova et al., 2016), and also 3CosAdd disad-vantaging them, because the required semantic information may be encoded in more complex ways (Drozd et al., 2016).”
The Linear Representation Hypothesis and the Geometry of Large Language Models (2023, methodology): “We consider 22 concepts defined in the Big Analogy Test Set (BATS 3.0) (Gladkova et al., 2016), which provides such counterfactual pairs.”
Bias in word embeddings (2020, methodology): “Caliscan et al. [17] introduce a general methodology to trace bias in word embeddings, while Drozd [30] et al. automatize the process.”
Probing Pretrained Language Models for Lexical Semantics (2020, methodology): “We evaluate our models on the Bigger Analogy Test Set (BATS) (Drozd et al., 2016) with 99,200 analogy questions.”