Making sense of patent information

Author:Catherine Jewell - Edward Harris - Steven Kelly
Position:Communications Division, WIPO

The patent system is designed both to recognize and reward inventors and to make technological information available to the general public with a view to spurring new innovations. Patent documents contain a huge amount of technological information and are a valuable source of business intelligence.


WIPO has developed a variety of new services and tools to enable innovators and companies to mine this information in support of their own research or business goals. In autumn 2016, it launched two new tools that will make it even easier to search and make sense of the huge volume of patent information generated every year in diverse languages across the globe.

Leading the way in machine translation

WIPO Translate is a ground-breaking translation tool for patent documents based on artificial intelligence which promises to unlock a wealth of previously inaccessible technological information.

Unveiled in late October 2016, the latest version of WIPO Translate employs neural machine translation (NMT) technology to offer innovators the highest-quality service yet available for accessing information about new technologies.

“One of the aims of the patent system is to make technology available,” explains WIPO Director General Francis Gurry. “Language is often a barrier to achieving that aim universally.”

Mr. Gurry heralded the development of WIPO Translate as “another great step forward” which means that “a vast and ever-increasing trove of patent documents will soon be more easily accessible to innovators who search these records for inspiration or technical know-how" (video on Youtube).

WIPO Translate outperforms existing machine translation tools. It incorporates cutting-edge neural machine translation technology that enables it to convert highly technical patent documents into a second language in a style and syntax that closely mirrors common usage.

Neural machine translation is an emerging technology inspired by the structure and function of brain’s biological neural networks – a radical departure from the phrase-based statistical translation around which most established tools have been developed. With this technology, a large neural network “learns” from previously translated sentences, enabling it to generate highly accurate translations. In the case of WIPO Translate, some 60 million sentences from Chinese patent documents provided by the State Intellectual Property Office of the People’s Republic of China through WIPO’s PATENTSCOPE database were compared with their equivalent official English translation as filed at the United States Patent and Trademark Office.

Emphasis on East Asian languages

The neural version of WIPO Translate complements WIPO’s existing statistical machine-based translation tools, which are available for 16...

To continue reading