WIPO Translate: Cutting-Edge Translation Tool For Patent Documents Extends Language Coverage

Geneva, September 25, 2017
PR/2017/809

WIPO’s ground-breaking "artificial intelligence"-based translation tool for patent documents is now available in ten languages, marking an important expansion of the highest-quality service yet available for accessing information on new technologies.

WIPO Translate uses cutting-edge neural machine translation technology to render highly technical patent documents into a second language in a style and syntax that more closely mirrors common usage, out-performing patent-translation tools built on previous technologies as well as other web-based products also using artificial intelligence.

WIPO Translate powers PATENTSCOPE, a 65 million record strong database used for research by inventors before filing international patent applications around the world via WIPO’s Patent Cooperation Treaty (PCT).

WIPO has "trained" the new technology to translate all patent documents in one of the official languages of the PCT (Arabic, German, Spanish, French, Korean, Japanese, Portuguese, Russian and Chinese) into English and vice-versa. A version of the tool launched in 2016 was initially available in English and Chinese.

“Extending WIPO Translate to more language pairs is a welcome development for innovators around the world. It ensures accessibility to the state-of-the-art knowledge created in the main languages of the production of technology,” said WIPO Director General Francis Gurry.  “The speed and accuracy of translations through WIPO Translate is unique because this tool is trained by and focused solely on patent documents, instead of a more-disparate array of texts, thereby producing higher-quality translations,” he added.

WIPO Translate is trained exclusively with huge amounts of patent texts and includes a “domain-aware-technique” that translates according to the specificity of the invention. The tool internally integrates 32 technical domains taken from the International Patent Classification. This allows the system to eliminate ambiguity in the translation process.  The technology takes into consideration the specific domain when translating a particular sentence thereby yielding more accurate translations.  This is a unique facility in the world of patent translations.

WIPO’s PATENTSCOPE database now fully integrates this new technology (previous statistical-based translation technology remains available for comparison purpose), making translations of patent documents retrieved through PATENTSCOPE more readily accessible.
 
Neural machine translation is an emerging technology. It is based on huge neural network models that “learn” from previously translated sentences. The specificity of neural machine translation (compared to previous “phrase based” statistical methods) is that it produces more natural word order, with particular improvements seen in so-called distant language pairs, like Japanese-English or Chinese-English. Even languages that are considered closer (like English-French) benefit from this new technology which results in better quality translations.

WIPO created its own software, based on open-source software and libraries and capitalized on in-house expertise in handling large datasets.

WIPO Translate is being offered gratis for use by other United Nations bodies and international organizations, which frequently see large and costly translation needs.

Background

The PATENTSCOPE database provides access to international Patent Cooperation Treaty (PCT) applications in full text format on the day of publication, as well as to patent documents of participating national and regional patent offices.  The information may be searched by entering keywords, names of applicants, international patent classification and many other search criteria in multiple languages. The database contains some 65 million records.