The future of machine translation with artificial intelligence
In a globalized world where organizations expand into multiple countries across the world, communication becomes an increasingly important topic. Therefore, the demand for translations is increasing resulting in organizations trying to find ways to translate in a quick and cost-effective manner.
Franciska Jurendic, 2021
Bachelor Thesis, Belimo Automation AG
Betreuende Dozierende: Stephan Jüngling
Keywords: machine translation, artificial intelligence, translation
Views: 12 - Downloads: 2
Belimo Automation AG is present in over 80 countries and has a great need for fast and efficient translations. This is where a translation process using specific tools and machine translation comes in place. In the current process, the inability to integrate their own terminology database is an issue that needs solving. Additionally, Belimo wonders how machine translation will develop through the influence of artificial intelligence and if keeping German over English as a source language for certain language pairs would produce a better machine-translation output with less post-editing effort.
The development of the thesis was split into four phases. Phase 1 focused on researching machine translation, how it developed, and how it is expected to develop in the future with the influence of artificial intelligence. In phase 2 interviews were conducted to analyses different translation processes in other organizations. In phase 3 different translation tools were explored to find a solution for specific objectives. In the last phase, a source language experiment was conducted, and based on the thesis different actions were explored and suggested.
Machine translation went through four different stages: rule-based, example-based, statistical, and neural machine translation. The last two stages were greatly improved by developments in artificial intelligence through deep learning and machine learning. The translation industry is expected to ex-perience a strong shift towards human post-editing. More translations will be handed to machines, while humans edit the output in the last process step. Out of the five companies, only two take advantage of machine translation. The other three use computer-assisted translation, translation management systems, and from scratch translations. From the different tools that were explored, only two presented potential for an implementation of an internal terminology database: Watson Language Translator and TextShuttle AI. The source language experiment showed no clear favour of German sources for Dutch and Danish, but French did. It was suggested that Belimo Automation AG considers switching to a different machine translator. Additionally, it was recommended that they prepare for the post-editing shift. Lastly, a suggestion was made on how to further expand the source language experiment.
Studiengang: Business Information Technology (Bachelor)
Fachbereich der Arbeit: Business Information System & IT-Management