Design of a Machine Translation Service for Variable Data
The paper addresses the problem of designing a machine translation service for variable data that operates at the level of document segments rather than monolithic text. In contemporary publishing, marketing and educational workflows, documents are increasingly represented as sets of structured segments associated with content keys, language attributes and, in many cases, personalization markers inherited from CRM or content management systems.