“As a result, the collections of the two largest languages in patents are now united as full text documents on the same website through the EPO’s global patent database, Espacenet, and linguistically accessible for innovators from both regions using a single tool, Patent Translate,” said the EPO in a press release.
“About 80% of all patents worldwide will now be translateable with Patent Translate,” said EPO President Benoît Battistelli while SIPO Commissioner Tian Lipu described it as “another stellar example of our [EPO-SIPO] cooperation”.
“It is estimated that, by the year 2013, the total volume of Chinese patent documentation will be over 10 million,” said Tian Lipu, adding: “It [the machine translation service] will further promote science and technology communications between China and Europe and eventually serve global innovation development.”
Patent Translate has expanded to cover 14 European languages since it was launched 10 months ago. It aims to cover all 28 languages of the 38 countries who are members of the EPO plus Chinese, Japanese, Russian and Korean by 2014.
An agreement has been signed with Japan and two agreements have been signed relating to Russian [with the Russian patent office and a regional patent office of former Soviet Union countries], said Battistelli, who estimated that this would cover “95% of the patent documentation available in the world”. Discussions about adding Arabic are also underway.
The machine translation service brings together Google, which provides the software, technology and applications, and the EPO, which has spent around €10 million providing the linguistic corpora (patent terminology).
As to the issue of the quality of the translations, Battistelli said that “I’m not saying that the grammar and syntax of the sentences is perfect” but that “scientists and engineers competent in a specialist field will understand the content of the document”.
He stressed that the machine translation had “no legal value” and was “for information purposes”, in other words helping users select pertinent documents. Once they have done that, the idea is that they would then have the document translated by a qualified person if they wanted to use it legally.
An EPO official described the service as providing a “fit-for-purpose translation giving a patent searcher an indication of relevance or non-relevance” and a “good basis to decide if they want a manual translation or not”.
“A machine tool is the only practical solution for the amount of data concerned. If this was manually translated, it would take 16,000 man years (based on a translator translating one patent per day),” he said.
On the subject of quality, he said that there was functionality whereby users could compare the translation of a Chinese text in English, French and German. “You never reach perfection,” he said “but what we can say today is that there is no better tool available in the market.”
“Quality will change. The more it is used, the more information there is in the database, the better the quality. I think it will change almost every day,” said Batistelli.
Battistelli also said the EPO ensured “the confidentiality of searches [on the EPO website]” and that there were “on average 20,000 connections per day”.
