News / Svarog capital

4.10.2010

PROMT announces groundbreaking PROMT DeepHybrid MT Solution

Press release

PROMT, a leading provider of automated translation solutions, announces its new groundbreaking approach to enterprise-scale machine translation, PROMT DeepHybrid. By utilizing the full spectrum of clients’ assets in each step of the translation process, PROMT DeepHybrid provides high-quality automated translation output which leads to significantly higher post-editing efficiency and easier translation process automation for content that is not post-edited.

Each pure-based approach has its limitations: a strict rules-based approach may not yield the best fluency and terminology relevance, whereas a strict statistics-based approach is prone to making syntactic errors and often strips off valuable metadata information. The DeepHybrid solution combines the control and predictability of RBMT with SMT fluency, yielding the best translation choice while preserving source content formatting, tone of voice and accurate terminology.

Svetlana Sokolova, CEO and founder of PROMT comments: "PROMT DeepHybrid represents a major milestone in the development of enterprise automated translation solutions. Our unique technology goes significantly beyond the concept of statistical post-editing, which defines many shallow hybrid approaches. PROMT DeepHybrid solution is comprised of a non-deterministic generation of multiple translation hypotheses at the transfer phase, followed by tree pruning and statistical post-editing. Thus, every source sentence will generate multiple translation candidates; the best-formed and most relevant translation is selected automatically by the engine."

Olga Beregovaya, President/CEO of PROMT Americas, the company’s enterprise solutions division, adds: “The majority of enterprise users have accumulated significant TM volumes and multi-thousand term glossaries. When you add corporate style guides and brand consistency requirements into the equation, you get a truly challenging task for translation. By allowing clients’ assets to influence the engine’s choices on every step of translation process, PROMT DeepHybrid addresses this challenge. We believe our new product to be a genuine breakthrough for improving machine translation quality."


		4.10.2010 PROMT announces groundbreaking PROMT DeepHybrid MT Solution Press release PROMT, a leading provider of automated translation solutions, announces its new groundbreaking approach to enterprise-scale machine translation, PROMT DeepHybrid. By utilizing the full spectrum of clients’ assets in each step of the translation process, PROMT DeepHybrid provides high-quality automated translation output which leads to significantly higher post-editing efficiency and easier translation process automation for content that is not post-edited. Each pure-based approach has its limitations: a strict rules-based approach may not yield the best fluency and terminology relevance, whereas a strict statistics-based approach is prone to making syntactic errors and often strips off valuable metadata information. The DeepHybrid solution combines the control and predictability of RBMT with SMT fluency, yielding the best translation choice while preserving source content formatting, tone of voice and accurate terminology. Svetlana Sokolova, CEO and founder of PROMT comments: "PROMT DeepHybrid represents a major milestone in the development of enterprise automated translation solutions. Our unique technology goes significantly beyond the concept of statistical post-editing, which defines many shallow hybrid approaches. PROMT DeepHybrid solution is comprised of a non-deterministic generation of multiple translation hypotheses at the transfer phase, followed by tree pruning and statistical post-editing. Thus, every source sentence will generate multiple translation candidates; the best-formed and most relevant translation is selected automatically by the engine." Olga Beregovaya, President/CEO of PROMT Americas, the company’s enterprise solutions division, adds: “The majority of enterprise users have accumulated significant TM volumes and multi-thousand term glossaries. When you add corporate style guides and brand consistency requirements into the equation, you get a truly challenging task for translation. By allowing clients’ assets to influence the engine’s choices on every step of translation process, PROMT DeepHybrid addresses this challenge. We believe our new product to be a genuine breakthrough for improving machine translation quality." ← \| → © 2008 Svarog Capital Advisors Limited Web “25 Kadr”