Assessing machine translation quality on uterine cancer content: MQM-based comparative study

This study evaluates the effectiveness of machine translation in conveying medical content on endometrial cancer from English to Arabic. Machine translation is essential for medical awareness, yet few studies evaluate its error-based performance in translating English medical articles on women's tumors into Arabic. This study assesses machine translation efficiency to enable readers and specialists to access medical knowledge. By comparing three translation engines—Yandex, Systran, and Microsoft—the study aims to determine which engine performs best in delivering accurate and reliable medical content. Additionally, the study examines the extent to which target readers can depend on machine translation to understand medical information accurately. By employing the Multidimensional Quality Metrics (MQM) framework, the study identifies accuracy and terminology errors. The CAT tool, SDL Trados, is used in the assessment process. Both qualitative and quantitative approaches are adopted: the qualitative approach identifies errors and assesses their impact on translation quality, while the quantitative approach calculates the frequency of each error type, assigns penalty points, and generates quality scores. The data, sourced from WebMD, covers content on endometrial cancer, a prevalent cancer among women. Although there has been great advancement in machine translation efficiency, machine translation is still inadequate in conveying precise medical content, particularly regarding terminology and accuracy. These two aspects are crucial in translating medical content. This study, in addition to other previous studies, highlights the inaccuracy of machine translation, needing further refinements in translation technologies when dealing with specialized domains like medicine