EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Artificial Intelligence-Driven Translation Quality Assessment: A Computational Framework for Multilingual Discourse Analysis and Decision Support

Maria Nasr, Rana Raad

Computer and Decision Making: An International Journal · 03.11.2026 · 3 (6S) · 3171-3185

DOI: 10.1922/comdem.v3i6s.19061 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 03.11.2026. Verfasst von Maria Nasr und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Although Artificial Intelligence has transformed machine translation evaluation, most metrics have not been able to include aspects of discourse coherence, contextual meaning and interpretation that can be used to make decisions. This work introduces and formulates a multilingual translation quality assessment (TQA) framework that combines semantic evaluation based on neural models, discourse analysis, fine-grained error detection, and explanations to support decision making. The proposed architecture carries out translation processing, generates multi-lingual contextual representation, evaluates the semantic coverage, the linguistic fluency, the referential consistency, the coherence and the severity of errors between the aligned source text and the translation. The multi-dimensional quality indicators, uncertainty estimates and localized explanations in its decision-support component emphasize human review and correction. Linguistically different translation databases are defined, along with independent bilingual (translation) annotation, baseline lexical and neural evaluation levels. Planned analyses include correlations with human judgments, precision and recall of error detection, cross-language robustness assessment, and ablation experiments, and error detection precision and recall. This framework addresses a problem of having a score per sentence and correlates the document level evidence to clear review recommendations. At this stage there is no room to voice any conclusion about performance improvement since the empirical testing is not yet completed. The study provides a methodological blueprint for further research in the field of trustworthy, context-sensitive translation assessment and accountable translation work with human-AI collaboration in multilingual situations and in professional translation review contexts.

Zitieren

APA

Nasr, M., & Raad, R. (2026). Artificial Intelligence-Driven Translation Quality Assessment: A Computational Framework for Multilingual Discourse Analysis and Decision Support. Computer and Decision Making: An International Journal, 3(6S), 3171-3185. https://doi.org/10.1922/comdem.v3i6s.19061

Vancouver

Nasr M, Raad R. Artificial Intelligence-Driven Translation Quality Assessment: A Computational Framework for Multilingual Discourse Analysis and Decision Support. Computer and Decision Making: An International Journal. 2026;3(6S):3171-3185. doi:10.1922/comdem.v3i6s.19061

BibTeX
@article{Nasr2026,
  author = {Nasr, Maria and Raad, Rana},
  title = {Artificial Intelligence-Driven Translation Quality Assessment: A Computational Framework for Multilingual Discourse Analysis and Decision Support},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {3171--3185},
  doi = {10.1922/comdem.v3i6s.19061}
}
Kostenlos starten →