There has always been considerable doubt on whether machine translation is suitable for translating life sciences content. Many translators and LSPs maintain that machine translation is not suitable for this purpose because of the mission-critical nature of content that demands extensive human skills and expert domain knowledge. Mission critical is a term they apply to materials that, if mistranslated, could compromise patient safety. Translation errors in this domain can literally be a matter of life and death.
Thus, this is a domain that has always been much more focused on quality control and has always sought out subject matter expert linguists to do a lot of the critical translation work. Often the translators are doctors, nurses or technicians who used to work with the medical equipment and medical specialty that they are translating for. The extensive use of subject matter expert linguists at various phases in the translation process is more extensive in this domain than most others.
However, Omniscien Technologies contends that many of the very reasons that are provided as objections against the use of machine translation are actually the points that make the Life Sciences domain ideally suited to expert customized machine translation. These include:
- High levels of quality control on historical human translations produce high quality translation memories that are ideally suited for machine translation training data.
- Terminology has historically been managed and clearly defined. Once this terminology data is applied correctly to a custom machine translation engine, terminology will be corrected every time.
- When the Clean Data SMT model is leveraged, the focus is around quality rather than quantity of data. This methodology for customizing machine translation requires lower volumes of data than other approaches and delivers very high consistency in terminology and writing style.
- High levels of quality control are still employed whether the initial translation is performed by human or machine. The resulting post edited machine translations will improve the quality of the custom engine more rapidly than other approaches due to the quality and consistency of the final deliverable translation.
Read more by downloading our white paper “Why Life Sciences Is an Ideal Domain for Machine Translation”