Do Supervised Distributional Methods Really Learn Lexical Inference Relations?

Omer Levy, Steffen Remus, Chris Biemann, Ido Dagan

Abstract: Distributional representations of words have been recently used in supervised settings for recognizing lexical inference relations between word pairs, such as hypernymy and entailment. We investigate a collection of these state-of-the-art methods, and show that they do not actually learn a relation between two words. Instead, they learn an independent property of a single word in the pair: whether that word is a “prototypical hypernym”.