Hierarchical Embeddings for Hypernymy Detection and Directionality

TL;DR

HyperVec is a neural model that learns hierarchical embeddings to improve hypernymy detection and directionality, outperforming previous methods especially with limited training data and cross-lingual generalization.

Contribution

The paper introduces HyperVec, a novel unsupervised neural model that captures hierarchical relationships and generalizes to unseen hypernymy pairs and multiple languages.

Findings

HyperVec outperforms existing unsupervised measures in hypernymy detection.

HyperVec effectively captures hypernym-hyponym hierarchies.

The model generalizes well to unseen pairs and different languages.

Abstract

We present a novel neural model HyperVec to learn hierarchical embeddings for hypernymy detection and directionality. While previous embeddings have shown limitations on prototypical hypernyms, HyperVec represents an unsupervised measure where embeddings are learned in a specific order and capture the hypernym$-$hyponym distributional hierarchy. Moreover, our model is able to generalize over unseen hypernymy pairs, when using only small sets of training data, and by mapping to other languages. Results on benchmark datasets show that HyperVec outperforms both state$-$of$-$the$-$art unsupervised measures and embedding models on hypernymy detection and directionality, and on predicting graded lexical entailment.