Baidu Searches for Machine Translation

Chinese search giant Baidu is locked in a massive battle with Alibaba and Tencent over the future of over a billion consumers. With many traditional sectors of the Chinese economy sputtering, Baidu and other technology-based companies are looking to bring billion-dollar-plus R&D advances to help transform the retail experience.

Baidu President Zhang Yaqin was interviewed by Bloomberg Business from Davos, where he commented on the company’s direction in engaging in “online-to-offline” (O2O) opportunities.

To enable this, Mr. Yaqin emphasized the need for “voice recognition, picture recognition, machine translation, deep learning and neural networks and all those things will make us competitive in the long term, and in the short term, too.”

Machine Translation (MT) caught our attention. Baidu launched its first MT effort, Baidu Translate, in 2011. Slator, which has been covering Alibaba’s advances for years, took this opportunity to look back on other related developments the search giant has made.Baidu Translate’s overall 94% accuracy rating is usually “good enough” for many consumer uses. Yet that 6% leaves a significant scattering of gaps in understanding, especially around key technical terms and other domain-specific language. For your business, that 6% of specialized language the machine translation misses, or messes up on, may make or break your customer experience.

While we definitely appreciate MT tools, our own perspective and best practices suggest that for high-quality needs, such as your own web site or your product documentation, MT should serve as a basis for a professional translation — not as an exclusive substitution for it.

We’d love to hear your own experiences in terms of human versus machine translation. And if you have a project that you’re in need of translating, contact us at projects@e2f.com to see how we stack up against the machines!