This paper examines the effectiveness of various evaluation metrics for Kurdish machine translation systems in low-resource settings. The research compares automatic evaluation metrics with human assessments to determine the most reliable methods for evaluating Kurdish translation quality. The study provides insights into evaluation challenges specific to under-resourced languages.
Bnar, Polla Fattah. ‘Evaluation Metrics in Low-Resource MT: Insights from Using Automatic and Human Evaluation in Kurdish.’ Machine Translation Summit, 2024.