Pharma’s Cutting Edge

Google has launched its Google Patents search beta.  I use Google quite a bit.  Google itself is my customized home page, and I frequently use Google News, Google Ads, Google Analytics, Google Scholar, Google Print, Google Maps, etc in my daily work and play.  It’s well known that Google releases new tools to the public early after their development, while in beta testing, so I expect some deficiencies and bugs at first.  Sometimes these deficiencies are addressed; sometimes not.

I’ve been awaiting a free patent search tool that improves upon the existing U.S. and international patent office tools and was gratified to learn that Google was taking on the task of developing one.  I put it through a quick test.  I wasn’t impressed.

Sure the tool allows you to search for and within documents using the ultra-quick Google engine, but other than that I can’t find much to recommend its early adoption for people who count on comprehensive, accurate search results or who want results sorted in a usable way.

First off, Google’s service is far less comprehensive than some other free internet-based tools.  Google has indexed the USPTO patent data, but none of the international patent databases have been indexed.  The USPTO makes for a good starting place for pharma/biotech patent searches, but most researchers will want to look at international filings.  They’ll need to use another search engine like espacenet to do that.  Also, only granted patents have been indexed.  A comprehensive service must also index patent applications.

As far as accuracy, I ran identical searches for three major pharmas on Google Patents and nearly simultaneously on the USPTO site.  In each case, the Google search yielded far fewer “hits”.  Eli Lilly, for instance, yields 1,084 patents on Google and 3,468 patents on USPTO.  It would have been nice to compare the two search engine results directly, but I couldn’t do that easily, because Google returns its findings in a non-sortable Googley way, the same way it returns results for its other services…according to Google’s search algorithm’s idea of relevance.  USPTO returns its search results sorted by date, most recent first.  The USPTO data are also not easily sortable, but at least the output makes some sense to the end-user who needs to find a specific patent.  The Google results display contains a bit more information for each patent than the USPTO display but not enough to give it a practical advantage.

Finally, Google scans the full USPTO patent text into its database as images over text, allowing within-document search.  I found the within-document search to work quite well, finding chemical terms nested within longer chemical names that included numbers and symbols.  The USPTO doesn’t offer within-document searching, but it displays text as HTML, which may be searched by a browser.  The USPTO output can also easily be saved as text.  Google’s output is an image of okay quality (figure quality is sub-par though) that may be saved and printed, but getting that image onto your hard drive as text requires additional manipulation with likely loss of information.

I’ll check back with Google Patents again in the future and will let readers know if my opinions change.

December 14th, 2006 | Category: Technology | Subscribe to comments | Leave a comment | Trackback URL