Unsupervised Word Mapping Using Structural Similarities in Monolingual Embeddings

TL;DR

This paper introduces an unsupervised method for aligning monolingual word embeddings to induce bilingual dictionaries without relying on parallel data or seed dictionaries, leveraging structural similarities in the embeddings.

Contribution

It presents a novel unsupervised approach that exploits local and global structural similarities in monolingual embeddings for bilingual dictionary induction.

Findings

Performance comparable to supervised methods

Effective without prior alignments or seed dictionaries

Applicable to language pairs lacking parallel corpora

Abstract

Most existing methods for automatic bilingual dictionary induction rely on prior alignments between the source and target languages, such as parallel corpora or seed dictionaries. For many language pairs, such supervised alignments are not readily available. We propose an unsupervised approach for learning a bilingual dictionary for a pair of languages given their independently-learned monolingual word embeddings. The proposed method exploits local and global structures in monolingual vector spaces to align them such that similar words are mapped to each other. We show empirically that the performance of bilingual correspondents learned using our proposed unsupervised method is comparable to that of using supervised bilingual correspondents from a seed dictionary.