For anyone who attended the PIUG conference in Denver earlier this month, you may have left Denver with the impression that patent analytics and patent data mining are well on their way to supplanting more traditional approaches to searching and reviewing patent information. It's apparent that some businesses have certainly succumbed to the hype. And it's easy to see why. It seems these days that a variety of analytic programs are popping up with the promise of reducing, if not eliminating, the labour intensive task of searching patent information and having to review each patent document one by one.
The promise of patent analytics certainly looks appealing. The ability to cluster the data, understand the relationship between the nodes in citation maps or create patent family trees without having to review hundreds if not thousands of records can be a time saver when looking for trends. It can provide a useful overview before embarking on a particular R&D direction, but it will not tell you whether the particular outcome of R&D (i.e. the invention) is novel.
When one attempts to work with these patent analytic programs, one discovers that the data needs to be cleaned and categorized and that this endeavour takes time. And it takes time to understand and interpet the data output and then transform that understanding into something the CEO will understand. Not to mention convincing the CEO to pay the hefty price tag attached to some of the analytic tools out there.
Even if a contour map shows you "white space", how can you be certain that there isn't an unbeknownst patent lurking in that supposed white space that could pose a real patent infringement risk? It's foolish to assume that just because the map says there is nothing that there really is nothing. It all boils down to the strategy used to collect the data in the first place. As it currently stands, the patent analytic tools can handle the structured data of patents, such as patent assignee, but what about contextually understanding the unstructured data such as the description, where the patent applicant can make up or use whatever phrases it so chooses. You need a human to mentally draw the connections and understand whether the phrase "fastener" in the patent claim can mean a staple, VELCRO or a safety pin.
So until a computer can think for itself like a human, it appears that macro analysis of patent information will not supplant the micro analysis of patent information. There will still be a need for those patient souls who dutifully look through patent records one by one for that killer art or the "oh, oh, that could pose a serious problem" patent.
Don't you agree?