Learning Word Sense Embeddings from Word Sense Definitions
This addresses the challenge of word sense ambiguity in NLP systems, but it is incremental as it builds on existing word sense embedding methods by using definitions instead of corpus data.
The paper tackles the problem of representing polysemous and homonymous words in NLP by proposing to learn word sense embeddings from definitions, rather than from a corpus, and shows that this approach yields high-quality embeddings on similarity and disambiguation tasks.
Word embeddings play a significant role in many modern NLP systems. Since learning one representation per word is problematic for polysemous words and homonymous words, researchers propose to use one embedding per word sense. Their approaches mainly train word sense embeddings on a corpus. In this paper, we propose to use word sense definitions to learn one embedding per word sense. Experimental results on word similarity tasks and a word sense disambiguation task show that word sense embeddings produced by our approach are of high quality.