USCORE: An Effective Approach to Fully Unsupervised Evaluation Metrics for Machine Translation
arXiv:2202.10062 · doi:10.18653/v1/2023.eacl-main.27
Abstract
The vast majority of evaluation metrics for machine translation are supervised, i.e., (i) are trained on human scores, (ii) assume the existence of reference translations, or (iii) leverage parallel data. This hinders their applicability to cases where such supervision signals are not available. In this work, we develop fully unsupervised evaluation metrics. To do so, we leverage similarities and synergies between evaluation metric induction, parallel corpus mining, and MT systems. In particular, we use an unsupervised evaluation metric to mine pseudo-parallel data, which we use to remap deficient underlying vector spaces (in an iterative manner) and to induce an unsupervised MT system, which then provides pseudo-references as an additional component in the metric. Finally, we also induce unsupervised multilingual sentence embeddings from pseudo-parallel data. We show that our fully unsupervised metrics are effective, i.e., they beat supervised competitors on 4 out of our 5 evaluation datasets. We make our code publicly available.
Accepted at EACL 2023 (main track)
References in corpus (8)
- BARTScore: Evaluating Generated Text as Text Generation
- Experts, Errors, and Context: A Large-Scale Study of Human Evaluation for Machine Translation
- Multilingual Alignment of Contextual Word Representations
- Attenuating Bias in Word Vectors
- Theoretical Analysis of Self-Training with Deep Networks on Unlabeled Data
- UniTE: Unified Translation Evaluation
- Automatic Text Evaluation through the Lens of Wasserstein Barycenters
- EffEval: A Comprehensive Evaluation of Efficiency for MT Evaluation Metrics