Hierarchical Embeddings for Hypernymy Detection and Directionality
This addresses a key challenge in NLP for tasks like semantic analysis, though it appears incremental as it builds on existing embedding approaches.
The paper tackles the problem of detecting hypernymy relationships and their directionality in natural language processing by introducing HyperVec, a neural model that learns hierarchical embeddings. Results show HyperVec outperforms state-of-the-art unsupervised measures and embedding models on benchmark datasets for hypernymy detection, directionality, and graded lexical entailment.
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.