Saved in:
Bibliographic Details
Main Authors: Haq, Sami Ul, Osuji, Chinonso Cynthia, Castilho, Sheila, Davis, Brian
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.13980
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915499327094784
author Haq, Sami Ul
Osuji, Chinonso Cynthia
Castilho, Sheila
Davis, Brian
author_facet Haq, Sami Ul
Osuji, Chinonso Cynthia
Castilho, Sheila
Davis, Brian
contents In this paper, we present our submission to the Tenth Conference on Machine Translation (WMT25) Shared Task on Automated Translation Quality Evaluation. Our systems are built upon the COMET framework and trained to predict segment-level Error Span Annotation (ESA) scores using augmented long-context data. To construct long-context training data, we concatenate in-domain, human-annotated sentences and compute a weighted average of their scores. We integrate multiple human judgment datasets (MQM, SQM, and DA) by normalising their scales and train multilingual regression models to predict quality scores from the source, hypothesis, and reference translations. Experimental results show that incorporating long-context information improves correlations with human judgments compared to models trained only on short segments.
format Preprint
id arxiv_https___arxiv_org_abs_2509_13980
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Long-context Reference-based MT Quality Estimation
Haq, Sami Ul
Osuji, Chinonso Cynthia
Castilho, Sheila
Davis, Brian
Computation and Language
Machine Learning
In this paper, we present our submission to the Tenth Conference on Machine Translation (WMT25) Shared Task on Automated Translation Quality Evaluation. Our systems are built upon the COMET framework and trained to predict segment-level Error Span Annotation (ESA) scores using augmented long-context data. To construct long-context training data, we concatenate in-domain, human-annotated sentences and compute a weighted average of their scores. We integrate multiple human judgment datasets (MQM, SQM, and DA) by normalising their scales and train multilingual regression models to predict quality scores from the source, hypothesis, and reference translations. Experimental results show that incorporating long-context information improves correlations with human judgments compared to models trained only on short segments.
title Long-context Reference-based MT Quality Estimation
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2509.13980