There is an ever-growing glut of patent and related information swilling around global databases and libraries. But what makes the search task of professional patent researchers even harder is that that information is not always input in a standardised way and is increasingly likely to be filed in an East Asian language.

No wonder, then, that tracking down a patent can be horribly reminiscent of looking for a needle in a haystack. And to do their job properly, professional researchers have to find that needle ­ every single time.

Despite these problems, an undercurrent of optimism was evident among delegates at the recent Information Retrieval Facility symposium held in Vienna.

The Information Retrieval Facility (IRF) is a platform designed to encourage dialogue and co-operation between patent and information retrieval professionals, who can compare experiences in their respective fields.

Admittedly, there was concern in Vienna, and warnings of problems to come were traded across the symposium floor between the experts. But the tone was buoyant and a crowd of ideas politely jostled to get a hearing.

Henk Tomas, from consultancy IP Search Services, began his keynote speech with the line: “Fifty million patents later and we still search the way we did 28 years ago.”

Efficiency tool
Tomas explained how patent information could be a useful tool for monitoring the activities of suppliers as well as competitors. Tracking patents in this way helps to avoid the unnecessary duplication of effort as well as protecting intellectual property. “Patents are part of a globally accepted, harmonic legal system,” Tomas pointed out.

As an information specialist, he then summed up the other side of the patent search problem: “We look for one patent that takes away the novelty of an idea that may be out there, but we don’t know if that patent even exists.”

Patent search and knowledge management issues are complex and apply on a global scale. There has been a massive growth in patent information over the last 30 years, and a significant and growing amount of that is East Asian in origin. Patents filed in languages heavily influenced by classical Chinese, such as modern Chinese, Japanese and Korean, present translation problems for Western organisations, while different data input methods raise further challenges.

Another key issue is how the structure of patent databases can vary significantly. That can come down to how organisations prioritise data ­ scientific versus legal information ­ or how coverage is considered, such as by country or by timeframe.

The current setup is not a particularly efficient model, especially when researchers need to cross-check patent information between industries.

The patent industry’s information systems also lack a decent set of coherent and joined-up standards (or at least ones that are used properly and consistently).
Mistakes, inconsistencies and omissions are all too frequent. Then there are the whispers of intentional errors and ambiguity made in filings in order to gain a competitive advantage.

The consensus at the IRFS event was that the situation would not get better until these challenges had been addressed and overcome.

The world of patent information is clearly facing some fundamental challenges. Could the solution be to change its very nature?

Lost in translation
The attendees turned their attention to how both humans and optical character recognition software make mistakes, and to the effects on the patent retrieval process.

The language gap is caused by an intricate and sometimes subtle difference of structure between East Asian and Indo-European languages. Automatic machine translation has been trialled for viability and effectiveness, and may help towards a solution, but there is some way to go towards the ideal of flawless automated translation.

The bigger discussions about how professionals actually conduct patent searches and which tools they use uncovered pain points. A lengthy evaluation of the old approaches of patent search concluded with general agreement that a new way was needed.

Wrapping up the symposium Stephen Adams, IRF board member and MD of information consultancy Magister, summarised how the two disparate communities of academic information retrieval and patent search had for the first time come together, how the patent system had evolved and why it now faced the challenges it did.

Most importantly, Adams asked if information retrieval research could solve the issues.

Progress, he said, depended on collaboration, patience and commitment between all parties.

Patent offices needed to impose standardisation, penalising applicants with the loss of priority filing dates if they didn’t comply.

Work also had to continue on technology to clean up existing patent data. This work should run in parallel with the release of new untainted patent information.
Finally, Adams called on his peers to “lend their efforts or at least moral support for the development of tools”.

He added: “Unless there is that spirit of generosity, we are not going to make progress here.” His suggested agenda for change could be a sign of what is to come.