Machine Translation, or MT for short, is the use of intelligent automated software capable of translating large amounts of source data into multiple languages. As a rough guide, there are three main categories of machine translation:
“Generic” Machine Translation describes a ‘one size-fits-all’ solution and refers to engines such as Google-type engines that translate written text from one language to another. Used by consumers or individual business users for ad hoc translations of short texts, “Generic MT” is more prone to grammar and syntax mistakes and is less accurate than “customized” machine translation.
“Customized” Machine Translation involves “training” or adaptation of the translation software to recognize language belonging to a specific domain , industry or organization. Using basic statistical or rule-based translation technologies, “Customized” Machine Translation offers business a higher level of accuracy with high volume. However, even when it comes bundled with data pre-processing functionality and a user friendly interface (UI) translation quality can be compromised due to the inherent limitations of the translation technology selected.
“Enterprise” Machine Translation represents the next generation of “augmented” machine translation engines that employ sophisticated technology to reproduce style, format and terminology more faithfully than alternative solutions. “Enterprise” Machine Translation answers the need of today’s global business for high volume and high velocity content localization and real-time multi-lingual communication.
“Generic” vs. “Customized” vs. “Enterprise” Machine Translation. While all three categories described above may use statistical-based engines – there are some important performance differences. “Generic” machine translation engines throw massive amounts of data at its engines in hope for them to become better with time; they also lack the advanced technologies needed to deal with nuanced branded translation, syntax and formatting. “Customized” solutions that ‘train’ their engines offer improved translation quality but are ultimately only as good as the data provided and are exposed to the “data dilution effect” and other statistical MT technology limitations.
By comparison, as a result of its augmented technologies “Enterprise” Machine Translation is designed to overcome some of the inherent limitations of statistical-engines such as the challenges of formatting, styling and the adherence to corporate language fidelity and target language conventions.