Patent information providers take on Asian challenge
It is a familiar story in the business press: the emerging presence of Chinese, Korean and other Asia Pacific companies cannot be ignored.
For potential collaborators – as well as for competitors – this presents a significant challenge: how do you find out what is going on if you don’t speak the language?
One way to find out about technology being developed in a particular country is to look at the patents being filed with that country’s patent authority. Increasingly, this can be done by looking at that patent authority’s website. However, discussion groups on this topic at the recent IPI ConfEx meeting in Seville, Spain revealed that searching by keyword or inventor name can be problematic on local patent office websites because different patent authorities index things in different ways. Delegates noted that the Derwent World Patent Index (DWPI) classification codes are a good way to search because they are the same around the world.
For the chemical and pharmaceutical industries there is also sometimes the option to search by internationally-recognised chemical structures and sequences. However, not all patent offices and tools enable these searches.
Although going to the Korean Patent Office website to find information about Korean patents, for example, seems like a good place to start, searching many different sites is not a popular approach to finding such information. ‘It’s too time-consuming to go to all the patent offices,’ noted one delegate. For this reason, users of patent information often rely on tools that search multiple databases of patents.
But the providers of this information face similar challenges to those of the individual searchers. One of the biggest of these is language. It’s not a new issue, of course: companies in English-speaking countries are used to keeping abreast with the patents filed by their German-or French-speaking rivals, for example. But such situations are relatively simple. Both free and commercial patent search tools index patents from large bodies like the European Patent Office. And even if the person interested in a patent doesn’t speak the language of the patent, they are likely to be able to find an English abstract, if not a translation of the full text. They could also get an idea of a patent’s contents through an internet translation tool or even some level of guesswork.
There has been a need for English translations for non-Latin-based languages for some time too. As Pierre Buffet, chief executive officer delegate of Questel commented, ‘People were first talking about Asian data in the 1960s but then it was information from the USSR.’ Western companies and individuals have also been interested in Japanese patent information for a long time.
Machines versus people
The quickest and cheapest way to translate patent information is to use machine translation. However, errors can creep in that make huge differences to the meaning of a patent. For example, one delegate at IPI ConfEx found that the word ‘cycle’ in a technical patent had been translated as ‘bicycle’. Beyond simple machine translation, some information providers employ people to check and edit the translations, particularly of patent abstracts. Even this might not be enough, though. As another delegate pointed out, ‘human translation really should be by someone who understands the patent.’ There was also concern about situations where the patent office might not represent an invention clearly enough.
To pursue litigation, patent professionals still need to obtain professional translations of a particular patent.
The tools available
So what do the commercial information providers offer in this area?
STN International currently has five databases covering patents from Asia. The value-added databases CAplus and DWPI both provide editor-written abstracts in English. There are also three first-level databases, INPADOCDB, JAPIO and KOREAPAT.
Chemical Abstracts Plus (CAplus) is a chemistry bibliographic database from Chemical Abstracts Service (CAS). It has information on patent literature from Japan, China, Korea, Taiwan, Singapore, India and Hong Kong. The other value-added database on STN is DWPI, which is produced by Thomson Reuters. This provides information on patent publications from the 41 most important patent issuing authorities of the world, including those covered by CAplus and also the Philippines.
The first-level database INPADOCDB has the most comprehensive country coverage, with the bibliographic and family data of patent documents and utility models of 81 patentissuing organisations. In the Asia Pacific region, INPADOCDB covers patents for Japan, China, Korea Taiwan, Singapore, India, Indonesia, Vietnam, Hong Kong and the Philippines. It also has utility models for Japan, China, Korea and Taiwan. JAPIO, from Japan’s patent office, contains bibliographic information, abstract and drawings of Japanese unexamined patent applications that are machine-translated into English. KOREAPAT, produced by Korean Institute of Patent Information (KIPI), contains bibliographic information, abstracts and drawings of Korean (un)examined patent applications in English. The translations are done by specialists.
Translating full text
Many of the current tools provide access to English patent abstracts. The full text in English, however, is more of a challenge but it is something that companies are aiming for. Thomson Innovation includes English translations of the full text of Japanese patents.
‘The Japanese full text available through Thomson Innovation is translated using a machine-assisted translation process,’ explained Cindy Poulos, vice president of product management for the Scientific business of Thomson Reuters. ‘The first pass of the translation is machine-based. Then editors review the output and make corrections. Corrections are also fed back into the thesaurus used by the machine translation engine, so the quality improves over time.’
There is also a trend towards enabling the patents in databases to be searched in their native languages. ‘Currently, we do not offer searching in non-Western characters but a Japanese user interface will be available for Thomson Innovation in a future release,’ said Poulos.
Minesoft and its partner RWS also see patents in their original languages as important. The companies have added millions of Asian language patent documents to Minesoft’s PatBase Database Service over the past few months, according to Ann Chapman, director of Minesoft. ‘In each case, we supply English language data. The Asian text has been added for the convenience of customers who read those languages,’ she explained. She added that whatever language the full patent is in, chemists and engineers can benefit from seeing the images and technical drawings.
There are still plenty of challenges ahead with making Asian-language patents available though. ‘Japan is known for the high numbers of patents being published. And throughout Asia economies are developing fast and patenting activity is becoming important. It is not easy to set up the production runs for machine translating such large amounts of data and the development of technical dictionaries is a challenge,’ observed Chapman.