Why Microsoft’s Machine Translation System Is a Big Deal

07th Apr 2020

Computerized language translation services are nothing new. The most well known of them all, Google Translate, has been around since 2006. In stark contrast, Microsoft stepped into the consumer automatic translation services industry a whole five years later (which is a lot in the fast-paced online world).

With their early entry into the market, Google snagged the largest piece of the pie for automated translation services. It still enjoys a very extensive user base to this day, and is the automated translation service provider of choice of many people.

Microsoft, on the other hand, is not too far behind. Even if it cannot beat Google yet, it is pulling its own weight in the automated translation segment.

Why You Need to Pay Attention to Microsoft’s Machine Translation System

As popular as it is, Google Translate is infamous for making many mistakes in translating long phrases or sentences. Google Translate takes an extremely literal approach to translation, translating each sentence one word at a time. Thus, the translation can be hit or miss. As it does not take into account the complete context of the sentence, the translated phrase can often be complete gibberish. This tends to happen more frequently with longer phrases and documents.

Microsoft’s machine translation system, on the other hand, seems to take the context of the whole sentence or phrase into account. Thus, it is able to provide much higher quality and more accurate translations.

Recently, Microsoft has been taking steps to improve its automated translation system. In 2018, the company claimed to create the first machine translation system that could rival humans in quality and accuracy of Chinese to English translations.

Microsoft Translator – How Did They Do It

Even a few years ago, critics deemed the act of achieving human parity with computer-generated translations as a pipe dream. When Microsoft announced the creation of an extremely accurate machine translation system, it took the world by storm. They had effectively broken through a major milestone in the professional translation services industry. However, how did they do it?

The answer – Artificial Intelligence. Using powerful computer hardware and complex algorithms for translations, Microsoft’s Artificial Intelligence created accurate translations.

The AI used a host of different methods, such as dual learning, deliberation networks , joint training, and agreement regularization, among a lot more advanced techniques to reach this milestone. Processing ginormous amounts of language data every second, and cycling them through the AI to get more and more accurate results, Microsoft was able to achieve significant improvements in their automated translations. These processes would usually take lots of time on a traditional computer to work, but with the massive computing power of their dedicated supercomputers, the AI was able to reach these results in a reasonable amount of time.

However, the competition for accuracy in translations between man and machine is far from over. The machine translation is only a proof of concept as of now. The technology is not yet ready for widespread public use.

The leader of a natural language processing group that worked on Microsoft’s project, Ming Zhou, admitted that they still had some challenges to overcome. For one, they had not tested the system outside of the lab and on real-time news stories. Guaranteeing the accuracy of the translation all of the time proved to be another difficult problem.

Final Word

Machine Translation System is looking to pave the translations of tomorrow. Until we get there, however, there is no substitute for professional translation services that depend on humans for translation. The technology definitely has the potential to automatically generate accurate and contextually coherent translations. However, the technology has not matured enough to overtake human translators.

As of now, machine translation can be an incredibly useful tool for translation companies to lessen their workload. The machine translation software can attempt to accurately translate documents by itself. However, we are not at that point yet where we can blindly trust the translations generated by machine learning to be 100% accurate and contextually correct. A certified human translator is still required to guarantee the accuracy and quality of the translations produced.

 If you require accurate translations today, you will have to go with professional language translation services. To guarantee the quality and accuracy of your translations, choose a reputable translation agency such as Waterstone Translations.