OpenLID-v3: Improving the Precision of Closely Related Language Identification -- An Experience Report

TL;DR

OpenLID-v3 enhances language identification accuracy for closely related languages by expanding training data, merging variants, and adding noise labels, outperforming previous tools on multiple benchmarks.

Contribution

The paper introduces OpenLID-v3, a significantly improved language identification system with new datasets, merging strategies, and noise handling, addressing limitations of existing tools.

Findings

Ensemble methods improve precision but reduce coverage for low-resource languages.

OpenLID-v3 outperforms GlotLID on multiple benchmarks.

New evaluation datasets for closely related languages were created.

Abstract

Language identification (LID) is an essential step in building high-quality multilingual datasets from web data. Existing LID tools (such as OpenLID or GlotLID) often struggle to identify closely related languages and to distinguish valid natural language from noise, which contaminates language-specific subsets, especially for low-resource languages. In this work we extend the OpenLID classifier by adding more training data, merging problematic language variant clusters, and introducing a special label for marking noise. We call this extended system OpenLID-v3 and evaluate it against GlotLID on multiple benchmarks. During development, we focus on three groups of closely related languages (Bosnian, Croatian, and Serbian; Romance varieties of Northern Italy and Southern France; and Scandinavian languages) and contribute new evaluation datasets where existing ones are inadequate. We find that ensemble approaches improve precision but also substantially reduce coverage for low-resource languages. OpenLID-v3 is available on https://huggingface.co/HPLT/OpenLID-v3.