The objective of this report is to present an overview on the available machine translation engines and services. The report summarises machine translation methods and their key concepts, takes a quick overview on the machine translation service market and summarises some of the benefits and potential challenges of those services from the perspective of the Cross-Border Data Exchange (CBDE) Project.
Machine translation is the process of using an engine to automatically translate human input text from one language to another. There are both cost-free and commercial machine translation tools available on the market. The engines’ methodology varies: the translations are either rule-based, statistical-based or neural-based. The nature and characteristics of the needed translation defines what kind of machine translation engine suits the situation best.
This report views machine translation engines through the glasses of the above-mentioned CBDE project: using machine translation engines in cases like studying abroad, transferring health data and understanding legislation texts smoothly over the borders of Nordic and Baltic countries.
This overview on machine translation tools is based on literature and theory – the engines have not been tested for this report. The report takes a slightly closer view on three (3) machine translation services that provide a wide enough set of languages to be used across all Nordic and Baltic countries. Google Translate, Amazon Translate and EU eTranslation are capable of translating the whole set of Nordic and Baltic languages. These three translation engines can all also be integrated to core systems via API interface – but before taking any machine translation service into use, not only the translation method but also information security and privacy issues are to be considered.
The report also suggests that there are no machine translator services that offer 100% accurate translations in every context; there is always a need for post-translation quality inspection done by a human.
To be able to make a more concrete and detailed analysis on the usability of machine translation services in different real-life environments, there is a need for more detailed information gathering, field testing and discussions with the service providers. In addition, it would be interesting and reasonable to explore more on utilising existing specialised terminology and vocabularies together with machine translation. These specialised vocabularies have been created by different administrative branches for varying needs. It would seem highly reasonable to evaluate whether these vocabularies could be used as an external source for broadening machine translators’ vocabulary capabilities.