Unseen Word Representation by Aligning Heterogeneous Lexical Semantic Spaces

Word embedding techniques heavily rely on the abundance of training data for individual words. Given the Zipfian distribution of words in natural language texts, a large number of words do not usually appear frequently or at all in the training data. In this paper we put forward a technique that exp...

Full description

Saved in:
Bibliographic Details
Published in:Proceedings of the ... AAAI Conference on Artificial Intelligence 2019-07, Vol.33 (1), p.6900-6907
Main Authors: Prokhorov, Victor, Pilehvar, Mohammad Taher, Kartsaklis, Dimitri, Lio, Pietro, Collier, Nigel
Format: Article
Language:English
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Word embedding techniques heavily rely on the abundance of training data for individual words. Given the Zipfian distribution of words in natural language texts, a large number of words do not usually appear frequently or at all in the training data. In this paper we put forward a technique that exploits the knowledge encoded in lexical resources, such as WordNet, to induce embeddings for unseen words. Our approach adapts graph embedding and cross-lingual vector space transformation techniques in order to merge lexical knowledge encoded in ontologies with that derived from corpus statistics. We show that the approach can provide consistent performance improvements across multiple evaluation benchmarks: in-vitro, on multiple rare word similarity datasets, and invivo, in two downstream text classification tasks.
ISSN:2159-5399
2374-3468
DOI:10.1609/aaai.v33i01.33016900