The machine translation solution MT4client™ that KERN developed itself is based on the latest developments in artificial intelligence (AI) as well as rule-based, statistical and neuronal approaches. The software is tuned to the individual requirements of the customers, meaning that their specific terminology can be taken into account during the translation process. The engines are trained for this - they don’t simply swap out terminology, but rather they consider the subject and customer-specific requirements and language style.
Is machine translation the right solution?
The fully automated translation of texts already happens in various areas - for example in social networking, the provision of support texts on the web or the acceleration of internal corporate communication: For short-lived texts such as emails, where a quick response is most important, the use of machine translations can be sufficient. In such cases, simply conveying the meaning is enough. Grammatical and orthographical correctness only plays a minor role.
Conversely, catalogues, product descriptions, instruction manuals and industry-specific texts and documents must have exactly the right content and precisely explain the way the product works. Therefore, such texts require a translation which is correct and exact in terms of both language and content. This is achieved via machine translation and an additional intensive post-editing process.
Which options are there to use MT4client™ in an effective manner?
1) Pure mechanical use: MT4client™ enables the rapid creation of high-quality translations. This process can be individualised through the use of further language resources (e.g. terminology databases). Since the system is based on neural networks, these can be trained with a range of implicit rules to precisely meet individual requirements. The AI experts of KERN regularly improve the machine translation solution and implement current developments in the area of MT. Furthermore, MT4client™ is adjusted for specific industries and text forms by our AI experts (known as specialised MT).
2) Integration of human translators into the process: The quality and effectiveness of machine translated texts can be further optimised through post-editing. A human translator with the relevant subject area expertise is responsible for this. The machine translation is seamlessly integrated into the working environment of the translator, so they can concentrate on the most important thing: optimising their texts in the foreign language.
Trainable engines - subject and customer-specific MT engines
KERN is constantly using specialised and customer-specific MT engines to expand MT4client™. With the help of proprietary language data corpora as well as intelligent data filters, the AI experts at KERN can optimise neural networks for the translation of specific text types.
With subject-specific MT, multilingual technical communication is given a modern and powerful technological basis. This allows you to transfer complex, industry-specific texts into the required language with consistency and the appropriate style, as well as with terminological and technical accuracy.
We offer a choice of neural engines in various language combinations for numerous industries and subject areas:
- Agricultural economics
- Mechanical engineering
- Life sciences
When does it make sense to use customer-specific MT (Custom MT)?
If you already have substantial translation datasets, and have a high need for translations of a specific text form or domain, specialised MTs are a good option. These are also advantageous with narrowly defined text types and subject areas.
The advantages of Custom MTs at a glance:
- The highest standard of machine translation
- Can be individually fine-tuned
- Can be enhanced with data from KERN if required
- Customers can benefit from them for years to come
- Constant optimisation of the engine
- Coverage of various sectors
- Comprehensive analyses of text characteristics by experts and optimisation of the engine based on these characteristics
Overview of the various MTs:
- e.g. for general-purpose texts or high error tolerance
- Large number of language combinations
- TMs or glossaries for even more accuracy
- Good for product names, addresses, customer-specific terms
- Directly available for translation
- In the finance, IT and hardware industries, among others
- For frequently demanded languages
KERN Custom MT
- Based on customer-specific data
- Can be enhanced by data from KERN if required