A company in the aerospace engineering and future technologies industry recently contacted our US team regarding approximately 200 pages of financial data (text, graphs, tables and charts) for its M&A team to be reviewed in English. The required turnaround time meant that human translation was impossible and, since the documentation was intended only for internal review, the client opted for machine translation. Our translation technology team provided a customised MT solution that would satisfy the client’s deadline while producing a readable document.
Understanding different types of machine engines
We applied a generic stock engine (largely trained on data made publicly available by the European Union specifically for machine learning purposes) that had been further trained by our MT specialists with clean, relevant and anonymised data and in accordance with existing data protection laws. For interested clients, EVS Translations also offers custom topic-driven engines (e.g. for finance or law) based on the client’s existing terminology. Even further customisation and integration is possible for clients with already existing MT workflows.
Applying a ‘machine memory’ to speed up the translation process
Machine engines can be used to produce translations of a text but also for building a ‘memory’. Per client request, EVS Translations builds custom client translation memories but machine memories, also. The client translation memory is a database which is developed over time and that contains all the source texts and their translated equivalents which our translators have produced. A machine memory, however, is created purely from source texts and their equivalent MT output. In machine augmented translation workflows, the translator accesses both memories and chooses the most appropriate suggestion for a term or phrase and amends it if needed. Memories, are of course, used and stored in accordance with GDPR legislation and current data protection guidelines. This process can help optimise workflows and can enable clients to meet submission or publishing deadlines.
When do you choose an MT solution?
It’s important to consider the type of content that needs translating and weigh-up the pros and cons of applying MT. It’s also good to think about long-term demand for translation services within your department or business and build a translation strategy that will deliver long-term cost savings but at the quality levels your teams need. From chatbots and marketing copy to legal documents and corporate information for the public domain, all of these content types will carry a different set of demands which may or may not require the capabilities of machine translation.
Adding value for clients: effective document formatting to aid internal review
For this particular project, our desktop publishing team was able to reproduce the new English document so that the document layout was identical to the source PDF. This makes comparing and finding information in the source and target languages simple and, in this case, helped the client’s team to successfully navigate through the document during an internal meeting. The client particularly appreciated this detail.
If your department is considering MT as a solution for rapid turnaround times or very large volumes, speak with our team today. They can advise on customisation according to what your specific requirements are.
EVS Translations UK Ltd.
+44 (0)115 964 4288
EVS Translations USA Inc.
EVS Translations GmbH
+49 69/82 97 99-99