SDL International, the world’s leading provider of translation services and technology solutions, announced today the immediate availability of its innovative new SDL Knowledge-based Translation System, which delivers a breakthrough in translation throughput and significantly reduces the total cost of translation.

 For the first time, translation memory (TM) technology is tightly integrated with terminology extraction and adaptive machine translation to automate the majority of the translation process. With humans to post-edit the material, high quality translations are achievable faster and at lower total cost than from a traditional TM-based process.

Knowledge-based Translation (KbT) is the intelligent combination of translation automation software with human translation skills, delivering up to five times the translation throughput of traditional human translation. SDL is unique in its ability to blend technology with linguistic skills to provide its customers with faster translations, reduced total cost and high quality output. SDL KbT System leverages experience gained from over a decade serving global customers in many different industry sectors.

SDL’s integrated software manages and automates the complex end-to-end processes involved in localizing business content. For many localization requirements the majority of the translation can now be automated, significantly speeding up the process. In addition, its self-learning capability continually updates the knowledge base and speeds up the process for subsequent translations.

The SDL services team has the skills and experience to process different types and styles of content through all the different language combinations available within SDL KbT System, ensuring that customers achieve optimal translations for their business requirement.

“For many years, the choice has been simple: high quality translations using experienced translators or machine translation,” said Mark Lancaster, chairman and CEO of SDL International. “The SDL KbT System delivers quality much more cost-effectively in a robust, commercial solution that uses new software and a new approach to finally realize the potential of machine translation.”

SDL KbT System combines, for the first time in a single, integrated system, the following new or updated translation automation components:

• SDL Enterprise Translation Server: At the heart of the SDL KbT System is the adaptive machine translation engine of SDL Enterprise Translation Server. Its sophisticated rules-based technology rapidly translates between many common language pairs, dramatically reducing translation time.
  
• SDLPhrase Finder: This new and unique software component uses the intelligence of the adaptive machine translation engine to automatically identify words, phrases and terms that are suitable for once-only translation. Unlike statistical approaches, SDLPhrase Finder uses knowledge of language structure to correctly identify common terms in context and generate the required terminology database and dictionary entries. This substantially reduces the time taken to analyze source texts and improves consistency and quality of translations.
  
• SDLWorkFlow: Controls multilingual assets and the processes associated with translation and localization. Its role-based web portal provides all parties involved across the end-to-end content value chain with the appropriate access to their tasks and activities. Multiple workflows can be defined and routing from one step to the next is automatic, ensuring no unnecessary delays are incurred.
  
• SDL Translation Memory Server: Content is pre-processed using the SDL Translation Memory Server to identify sentences for which translated material already exists. Intelligence is built in to the system to understand how to process text where a fuzzy match occurs, calling on the adaptive machine translation engine where appropriate.
  
• Knowledge base repository: Previously translated content is stored in the knowledge base, which holds translation memories, terminology databases and the adaptive machine translation dictionaries. These are updated as new text flows through the system, ensuring that the amount of text to be translated is always kept to a minimum.
  

New opportunities for localization of content will open up due to the easing of the time and costs constraints that previously applied. Mark Lancaster continues, “We are very excited about the potential for the SDL KbT System. In addition to improving performance for traditional translation needs, an e-Support system suddenly becomes commercially viable in multiple languages, and the translation of volatile content such as news feeds becomes fast enough to meet customer demands. This unique combination of advanced technologies and quality service delivery is unmatched in the market today.”

The SDL KbT System is built on the open standards XML and TMX* to enable it to integrate easily with existing systems and ensure that knowledge built up in the translation process is not locked away in proprietary formats.