Machine Translation Quality Estimation: Advances and Emerging Challenges

Dear Colleagues,

As machine translation (MT) systems continue to evolve and gain widespread use across various domains, the challenge of assessing translation quality without reference translations has become increasingly important. Quality estimation (QE) for machine translation allows the assessment of MT output in real-time, without human intervention. This is vital in many real-world applications, such as post-editing MT output, improving MT systems, or deploying MT systems, in contexts where reference translations or human assessors are scarce.

Despite considerable progress, numerous challenges persist in this expanding field, and new approaches are needed to improve QE models. As MT applications broaden to areas like healthcare, legal systems, education, and global communication, the demand for accurate, explainable, and robust quality estimation methods continues to grow. This demand is especially pronounced for low-resource languages, where MT systems face challenges due to scarcity of training data. Moreover, QE models are being increasingly integrated into MT workflows, where they not only guide systems toward improved translations but also aid in automated post-editing (APE) of detected errors and filter out low-quality translation candidates. Effectively utilising QE models in such integrated MT or APE approaches, particularly alongside large language models (LLMs), remains an open question. This Special Issue seeks to address both methodological innovations and practical applications, offering a comprehensive view of the current research and future directions in QE.

Topics of interest include, but are not limited to, the following:

• Novel architectures for QE of MT output;
• QE for low-resource languages and specialised domains;
• Integrating QE systems into MT and APE workflows/approaches;
• Transfer learning, domain adaptation, and multitask learning for QE;
• QE for specialised neural MT systems and large language models (LLMs);
• Integrating any type of external information to improve QE systems;
• Data efficiency: semi-supervised and unsupervised QE techniques;
• Building compact and computationally efficient QE models;
• Explainability and interpretability in QE models;
• Real-world applications in sectors such as healthcare, legal, education, and global communication;
• Benchmark datasets and resources for QE of MT output.

Dr. Arda Tezcan
Dr. Frédéric Blain
Dr. Miquel Esplà-Gomis
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the Special Issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 250 words) can be sent to the Editorial Office for assessment.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-anonymized peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Information is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1800 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.