Low quality patents exist in part because it can be part to find relevant "prior art." A new search tool from Google could change that.
Google on Thursday announced an overhaul to its patent search tools as part of a bid to help the public and examiners spot reference material that can prevent low quality patents from being granted in the first place.
“Good patents support innovation while bad patents hinder it. Bad patents drive up costs for innovative companies that must choose between paying undeserved license fees or staggering litigation costs,” said the company in a blog post, citing a rise in patent litigation and the problem of “trolls,” which use suspect patents to shake down companies.
The most significant feature of the new version of Google Patents is a search function that combines the existing patent database with so-called “prior art” sources. Prior art, which includes everything from scholarly journals to books to blueprints, is used by examiners to decide if an invention is both new and non-obvious, which are the criteria required for receiving a patent in the first place.
By combining the patent and prior art searches, it will be easier to flush out materials that can show that a product or concept is ineligible for patent protection – or that it is indeed new and innovative. In theory, the new system will funnel all relevant material information into one spot, and make it easier to assess the merits of an invention. In its blog post, Google cited the example of a controversial “shopping cart” patent that a troll used to demand money from a wide number of online retailers; the patent was eventually invalidated, but only after the discovery of prior art in two books the examiner who issued the patent had overlooked.
Laura Sheridan, senior patent counsel for Google GOOG , explained in a phone interview that the search tool is based on machine classification, and relies on the CPC codes, which patent offices around the world use to sort inventions into different categories. Here is a screenshot of a search for “autonomous vehicles”
“The way we’ve trained our model is to classify based on earlier patents,” said Sheridan. The service is also capable of integrating prior art searches in other languages by drawing on Google Translate.
The new tool could also help patent examiners who are under time pressures. While they spend 19 hours on a typical application, this is often not enough to review all prior art. And while defendants can still use prior art to challenge a patent after it is issued, the process is difficult and expensive – in part because of a legal rule that patents are presumed to be valid.
The new tool comes at a time when Congress is in the midst of debating a reform bill intended to rein in the ongoing problem of patent trolls.