The Semiox workflow is tailored to the customer’s data and requirements. It is engineered to optimize the equilibrium between cost, time, and quality:
First, previously translated documents are reused. Nobody would use machine translation or translation memories for a document with only a few changed words. Instead, Content Recycling leverages safely larger chunks of translated text. Only the rest is passed to machine translation.
Next, the Multilingual Knowledge System annotates the translation and recognizes semantic discrepancies. Omitted text is identified. Style rules are applied. The outcome of these steps decide whether the translation is fit for purpose.
If not fit, translation and warnings are passed to expert revision. The job is offered to revisers with matching domain expertise. They claim the job and open the revision editor in the browser. The revision editor leads them through the document, issue by issue. Ideally no editing is required. Instead the reviser clicks on one of the proposed options and jumps to the next warning until all are processed.
After quality assurance the translation is delivered. The linguistic assets are updated thus further improving the performance of Semiox for the next job.