Bilingual Lexicon Induction for Low-Resource Languages using Graph Matching via Optimal Transport

TL;DR

This paper introduces a graph-matching approach based on optimal transport to improve bilingual lexicon induction, especially for low-resource languages, enhancing cross-lingual NLP tasks.

Contribution

It presents a novel graph-matching method utilizing optimal transport that significantly boosts bilingual lexicon induction for low-resource language pairs.

Findings

Improved performance across 40 language pairs.

Effective with minimal supervision.

Enhances cross-lingual NLP applications.

Abstract

Bilingual lexicons form a critical component of various natural language processing applications, including unsupervised and semisupervised machine translation and crosslingual information retrieval. We improve bilingual lexicon induction performance across 40 language pairs with a graph-matching method based on optimal transport. The method is especially strong with low amounts of supervision.