PROMT DeepHybrid technology combines the flexibility and predictability of rule-based machine translation (RBMT) with the fluency and lexical acumen of statistical machine translation (SMT).

The PROMT DeepHybrid engine can be trained on client's parallel texts (for example, translation memories) for a specific subject domain. The trained MT engine produces high-quality translation, which is sufficient for comprehension and requires minimal post-editing.

The PROMT DeepHybrid Translation Server is a web-enabled server-based product. It is a reliable, robust, and scalable enterprise solution which can handle huge production volumes and has both Web interface (for users) and API (for integration with external applications).

This product was initially designed for the Localization Industry and Language Service Providers (LSPs) but also works for any client wishing to translate large text volumes on a regular basis and that has amassed enough content for engine training.

Key Features

• Translation of texts in multiple formats (XML, HTML, MS Office documents, etc.)
• Retention of the document's original formatting, layout, and metadata in MT output
• Full support of industry standards (XLIFF, TMX, OLIF, etc.)
• Integration with GMS (SDL WorldServer, SDL TMS, GlobalSight, etc.) and CAT tools (Déjà Vu, memoQ, etc.)

Return on Investment (ROI)

Purchasing an MT engine offers an excellent return on investment due to the following benefits:

• Decreased Time-to-Market. The translation cycle with MT+Post-Editing takes considerably less time than translating from scratch.
• Cost Saving. The trained MT engine produces raw translation of such quality that LSPs are willing to offer post-editing discounts of up to twenty-five percent (25%).