Unsupervised Multilingual Dense Retrieval via Generative Pseudo Labeling

TL;DR

This paper presents UMR, an unsupervised method for training multilingual dense retrieval models without paired data, using pseudo labeling and iterative improvement, outperforming supervised baselines on benchmark datasets.

Contribution

Introduces UMR, a novel unsupervised framework for multilingual dense retrieval that leverages language models for pseudo labeling without requiring paired data.

Findings

UMR outperforms supervised baselines on benchmark datasets.

The iterative framework improves retrieval performance.

The approach reduces dependence on costly paired data.

Abstract

Dense retrieval methods have demonstrated promising performance in multilingual information retrieval, where queries and documents can be in different languages. However, dense retrievers typically require a substantial amount of paired data, which poses even greater challenges in multilingual scenarios. This paper introduces UMR, an Unsupervised Multilingual dense Retriever trained without any paired data. Our approach leverages the sequence likelihood estimation capabilities of multilingual language models to acquire pseudo labels for training dense retrievers. We propose a two-stage framework which iteratively improves the performance of multilingual dense retrievers. Experimental results on two benchmark datasets show that UMR outperforms supervised baselines, showcasing the potential of training multilingual retrievers without paired data, thereby enhancing their practicality. Our source code, data, and models are publicly available at https://github.com/MiuLab/UMR