PATENT ANALYSIS SOFTWARE

Visualising patterns of meaning in patents

Three years ago intellectual property - a fancy name for patents - became the common currency of everyday conversations just like house values and share prices. Companies really began to take notice of their patent portfolios and some even talked of generating the major part of their income by managing this 'intellectual capital'.

That hyperbole has fallen out of favour but, even after the bursting of the stock market technology bubble, patents are still big business and sometimes big news. In August this year, a US judge ordered Microsoft to pay $521 million to a tiny start-up, Eolas Technologies, and the University of California for using their patented technology in its Web browser. Earlier in the month, online auction firm eBay was ordered to pay nearly $30 million to MercExchange, a small private company owned by inventor Thomas Woolston, for infringing its patents. Such judgments are not new but they are a sign that the never-ending patent wars are still in a 'hot' phase, especially in the software sector.

Patents have always been used for checking up on (and of course sometimes for suing) competitors. But they have also been essential to research and development staff. So much of the information in the tens of millions of patents in the world is never published anywhere else. Fifteen years ago, as a bench chemist in the pharmaceutical industry, I remember puzzling through the archaic legal language of a patent to extract an experimental method that might be useful in the synthesis of a new drug. Then, each week, I'd share the job of flicking through printed patent abstracts, spotting interesting new chemical structures. Having identified something worth following up, there was sometimes a long wait for the printed patent to arrive. Searching for related patents was a job that called for the expertise of the company's librarian.

This scenario is now a fading memory for most users. Derwent's vice-president of industrial property markets, Rob Willows, confirms that his company still provides paper-based products but '97 per cent of our business is electronic,' he says. And it's increasingly web-based too. This is because, in the past 10 years, patents have moved beyond their traditional strongholds in the R&D and legal departments. Staff in every division and at every level are now using patent information for business planning, says Willows. That can be anyone from the finance director assessing the potential value of the patent portfolio of a takeover candidate during due diligence, to a business-development manager of a pharmaceutical company searching for drugs to license-in to expand the product portfolio. Users like these need search tools that are easy to learn and that produce a straightforward - often graphical - output. Patent professionals like the new tools too, says Willows, because they can produce a search report that includes charts or spreadsheets that are easier for clients to understand.

The result of this changing landscape is that web-based searching is a fast-moving area, and the past year has seen all the main players in the sector extending their databases, adding analytical tools and improving user-interfaces. But the big news has been sector consolidation, with information giant Thomson (Derwent's parent) snapping up web-search newcomer Delphion for $22m at the very end of 2002. Delphion had seen itself as something of a pioneer of patent searching on the web. The firm, which was spun out of IBM, gained a reputation not only for storing and providing web access to full-text patent information from US, European, Japanese, and World Intellectual Property Organization (WIPO) patent authorities (and commercial databases such as Derwent's), but also for implementing tools to analyse and track market activities, and to view and download patent images.

Delphion claimed to have the most complete patent record available anywhere with access to more than 40 million records (compared to Derwent's 20 million-plus). Its software tools: let subscribers extract data from selected patent fields in mutually compatible formats for spreadsheet analysis; build charts to help spot trends; include citation linking; enable clustering of similar patents, and so on. All this proved too tempting a target for Thomson. 'We gained a very nice web platform with simple but effective graphing and charting capabilities,' says Willows. And the company hasn't wasted any time in capitalising on the technology. Delphion's software is already being used as a front-end for other Thomson products. It could even end up being a portal for most of the company's products, says Willows.

Another trend has been towards direct delivery of patents. In March, Thomson launched its 'Patent Store', which enables users to view and download a patent document from the same software - making downloading part of the searching process. The same technology is used to provide patent document-delivery facilities for other services, such as Derwent's World Patents Index and the Patents Citation Index. This improved access to patent images has been extended to Delphion, which is now integrated with the 'Patent Store'. This extends access to full-text patent specifications for over four million additional patents from national collections worldwide, bringing the number of patent images available to over 10 million, the company claims. Images can be downloaded in bulk or saved one at a time, says Willows. Some patents, however, are still not immediately available for download, and these can be ordered and delivered via e-mail.

Thomson has also been combining other services in the past 12 months, and has rolled out an ambitious programme of web-based tools that draw on research information and expertise from ISI for example. The focus has been on fast-moving or business-critical areas. ISI's Web of Software search engine trawls journals, magazines, and selected websites for examples of prior art - just the sort of service that can save a company from getting entangled in a multi-million pound lawsuit.

In January this year, the same technology was extended to cover the nanotechnology sector, with content culled from Derwent's patent databases, ISI's journals, and specially-vetted web sites. This sort of combined searching is necessary because nanotechnology information is primarily available in journals; in all some 35,000 patent records and more than 185,000 journal records are covered. But a dramatic rise in the number of nanotechnology patents is currently under way. Despite the increasing focus on the web, Thomson still supports Dialog, and has made improvements in the past year that allow faster, more sophisticated searching, says Willows. The trend toward direct patent delivery, and the inclusion of visualisation tools to analyse citation patterns, is evident in product improvements from the other major companies in the sector - Questel-Orbit, Patent Cafe, and Micropatent.

Questel-Orbit has also introduced new databases and enhanced existing ones, including an improved SDI automatic alerting service, and has unveiled a new version of its Internet patent search tool, QPAT. The improvements include more extensive patent delivery, easier navigation, patent citation analysis, and data visualisation. The last two features come courtesy of a partnership forged by Questel with a specialist visualisation software firm, anacubis, in February this year. The company's name - taken from the Cubist art movement pioneered by Braque and Picasso - means to depict different viewpoints simultaneously.

The technology behind the viewer has been developed over the past 12 years by anacubis' parent company, i2, which is a provider of data visualisation and link analysis software to intelligence services and police forces around the world.

The enhanced QPAT was launched in June this year and Questel is also planning to include the anacubis viewer into its QWEB services. The viewer automatically creates a visual representation of data in real-time, which web searchers can then manipulate and make use of. Search hits are displayed as icons and the relationships between them are shown as different types of link. These views can subsequently be dragged into anacubis' desktop tools for further analysis and comparison with other commercial information. (Those who want to see what the anacubis viewer looks like in action can try out an enhanced version of the popular Google search engine on the company's web site at www.anacubis.com/googledemo/googledemo.html).

This shift in emphasis from patent searching to patent analysis is typified by some - off-line - software now being distributed by Questel in partnership with the Danish Patent Office. This is the patent-valuing software, IPscore, which was developed by the Danish Patent Office in co-operation with the Copenhagen Business School, as a tool for managers evaluating patents and development projects. The software bases its results on 40 assessment factors drawn from a company's strategy, legal, technology, market, and financial standing. It then produces qualitative and financial estimates of that company's patent portfolio. A sign of web developments to come, perhaps.

PatentCafe's newly launched Global Patent Search sticks with traditional Boolean-based searching and enhances it with concept-based searching, dubbed Semetric. Global Patent Search also has a built-in document manager that improves documentation and communication of search results.

The aim of the Semetric software is to improve the accuracy of the first round of searching, which nearly always misses some important patents.

Like the anacubis visualisation software, Semetric is a product of the intelligence community. It is based on a statistical technique called 'latent semantic analysis' - which is also being employed in spam-killing e-mail software - and works by mapping relationships between each word and every other word in large sets of documents. It also uses a neural net, so the software is capable of 'learning'. PatentCafe says the results from a query are conceptual and, because of this, the software can retrieve relevant documents even when those documents do not share any words with the query.

Micropatent has improved its Aureka web-based patent software several times, adding visualisation software and extending links to non-patent literature. Its ThemeScape software includes a number of visualisation tools, one of which is text mining. This analyses the text of documents, identifies key themes, and uses computer algorithms to categorise patents by subject matter and plot them on a map. Users can choose to pull full-text, claims, titles, or abstracts into Themescape and search using keywords. Results are grouped into topic areas and displayed on a three-dimensional (contoured) map for easier viewing and comparison of data. Despite these advances in visualisation and other so-called 'intelligent' searching, these techniques are a long way from taking over from Boolean logic searching, sceptics warn.

This is particularly true when the search is carried out by a patent professional who has day-by-day familiarity with a particular business sector. Software that produces excellent results on specially selected libraries may not work so well in the real world. Anyone who has read a patent knows how difficult it can be to determine exactly what it means. And that's without considering the question of stealth patents, which have been deliberately designed to evade searches so that the owner can turn up to claim a priority over an existing technology and, with it, a healthy income. Scientific patents, with their chemical formulae, may be a particularly tough nut for visualisation methods to crack, chemical industry experts warn. And it may be getting harder. The Chemical Abstract Service (CAS) has observed three significant trends in patent information: the main patent organisations are granting more patents than ever; more new organic and inorganic substances are being first disclosed in patents; and patents covering the biosciences are becoming more complex and dense with information.

While users are getting to grips with these changes, the companies providing patent searching services are facing more radical changes in the way they do business. The greater awareness of the importance of patents has stimulated some unexpected developments in the sector, says Derwent's Willows. Large pharmaceutical companies now want to use Derwent's software to link patent databases with 'legacy information' held in internal databases. 'The trend is for major organisations to try to take all resources in-house,' he says. In the future, database suppliers could end up increasingly as systems integrators - supplying software and programming expertise as patent-searching becomes an integral part of doing business.

back to main features page
home