Google’s “Prior Art Finder”

Google’s “Prior Art Finder”

Google announced a major update to their Google Patents search tool yesterday. These changes include two main offerings:
1. The inclusion of EP documents
2. The introduction of the “Prior Art Finder” tool
As of the writing of this blog post, it appears that while EP documents are available within the Google Patents search system (such as this document), they do not come up normally in results lists for search queries. This is the case even when using verbatim title searching or document number searching, and the advanced search form does not include any options for searching EP documents, either. However, this seems like a glitch or gradual roll-out that Google is well known for and not a permanent issue. The inclusion of EP documents is a major boon to those who prefer Google Patents over Espacenet (in the free search tool domain). EP data is but the first step towards making Google Patents a more viable search tool for more than a quick starting point or “silver-bullet” type search.
As for the Prior Art Finder tool: I’ve got a lot to say about how that feature works, what it can do for the professional searcher, and what it means for the future of prior art search after the jump.

Google’s Prior Art Finder tool is accessible from any individual patent document within Google Patents’ system.
Prior Art Finder pros
1. Query generation is automatic – Starting a query with the Prior Art Finder couldn’t be easier. Users need only to click the blue “Find prior art” button at the top of any individual patent document within Google Patents. From there, search terms are automatically mined and generated from the full available text of the document. Accompanying the term generation is an automatic end date for search results that coincides with the extracted filing date from the patent document (putting the “prior” in prior art).

2. Federated search through various Google entities is helpful – Users accustomed to running separate strings in Google Scholar and Google Books will be assisted by the fact that both of these valuable search tools will be queried simultaneously through the Prior Art Finder. Switching between results sets is accomplished by selecting the category at the top of the results page, and the pre-loaded entries load instantaneously and conveniently.
3. Post search modification is user-friendly – Automatically generated search terms are clearly displayed (no black-box here!). Along with supplementary generated terms, can be toggled on-and-off with ease and search results will be instantly updated. Additionally, the custom date range fields can similarly be modified.

 Prior Art Finder cons
1. Results are limited to Google search engines – Prior Art Finder merely searches Google Scholar, Google Patents, Google Books, Google People, and Google Web. It will not search valuable prior art databases that Google does not have the access to index, ruling out a vast majority of the available “non-patent” prior art available to expert searchers. As a result, those seeking a complete professional prior art search will find that Prior Art Finder is merely a first drop in the bucket.

 2. Queries must be generated from a document present in Google Patents’ database – This means if it’s not a US or EP document in Google’s database, you’re out of luck for automatic query generation. There is no currently available easy way to create your own query, but a work-around can be accomplished by using any patent document in the Google system as a starting point, erasing the automatically generated terms, and entering your own. At this point, however, users may be better suited to generate their own queries in each separate Google search product. At the same time, it would be nice to have the interface of Prior Art Finder and the functionality of the federated nature of the tool available for brand new queries.

 3. Automatic query generation and results are hit-and-miss – Testing on a patent for a “Thread-measuring feeding device” came up with results for yarn properties, blood regulation, human motion analysis, fly fishing, and computer threading (and this was just in the “Top 10″). At this stage, Prior Art Finder is far from a panacea for those looking for the next evolution in automated prior art search. It just goes to show that the “neck-top computer” still can’t be beat when it comes to complex and semantic heavy problems such as patent and prior art searching.

 One further note: the Google Research blog post about this update leads me to believe that many further updates may be coming to Google Patents. One can only hope that the sleeping giant in the search field has been awakened to the possibilities and growing necessity of providing patent search tools!