Hierarchical Density Order Embeddings
This improves lexical entailment modeling for natural language processing applications, though it is an incremental advance over existing probabilistic word embeddings.
The paper tackles the problem of representing hierarchical semantic relationships between words by introducing density order embeddings, which use probability densities instead of point vectors to capture uncertainty and entailment. The approach achieves state-of-the-art performance on WordNet hypernym prediction and the HyperLex dataset.
By representing words with probability densities rather than point vectors, probabilistic word embeddings can capture rich and interpretable semantic information and uncertainty. The uncertainty information can be particularly meaningful in capturing entailment relationships -- whereby general words such as "entity" correspond to broad distributions that encompass more specific words such as "animal" or "instrument". We introduce density order embeddings, which learn hierarchical representations through encapsulation of probability densities. In particular, we propose simple yet effective loss functions and distance metrics, as well as graph-based schemes to select negative samples to better learn hierarchical density representations. Our approach provides state-of-the-art performance on the WordNet hypernym relationship prediction task and the challenging HyperLex lexical entailment dataset -- while retaining a rich and interpretable density representation.