Back when I first heard about credit default swaps and mortgage backed securities, I applied my rule-of-thumb for non-traditional investments. That is, if I try to understand the mechanics and value of a process and can’t, maybe there’s less there than meets the eye. It’s an egocentric approach but, it’s kept me from following the crowd into some pretty awful investments. If I don’t get it, I don’t buy it.
Searching ESI is what I do during much of my waking hours, so I pretty well “get” searching ESI. I still regard myself as more a student than a teacher of technology assisted review, but I’m trying very, very hard to “get” all the variations of these advanced analytic technologies…with mixed success. I should mention: You can’t use the If I don’t get it, I don’t buy it credo unless you also bust your hump trying to get it.
I mention all of this by way of noting that Nicholas Economou, the Chairman and CEO of e-discovery service provider H5, kindly dropped me a note to announce that H5 had secured a patent for its ”high recall and high precision relevancy searching” process. Per H5′s press release, U.S. patent number 8,296,309 covers H5’s method for “generating and iteratively refining relevance rules used as complex search queries to find relevant documents. Based on direction from counsel and informed by review of example documents, relevance rules are populated with linguistic search terms and iteratively refined to increase accuracy.”
Holy smokes! That’s what I do! H5 has patented practical keyword search!
So, if your approach to e-discovery is an iterative process whereby, with the benefit of input from counsel, you tweak searches to secure better-and-better results using automated search tools to identify relevant documents, you might need to start paying big fat royalties to H5 for using their patented process.
I’m only partly joshing about that. Here’s how the H5 process is described in the patent. The bracketed simplifications are mine:
A computer-implemented method comprising: generating, by a processing device [your search tool of choice], a filter for identifying a relevant document based on an initial relevance rule related to a set of documents [your initial keywords]; applying, by the processing device, the filter to the set of documents thereby identifying a subset of relevant documents [extracting the documents with hits]; receiving, by the processing device from an assessor, the subset of relevant documents comprising an identification of key information [reviewing the documents with hits to see which searches worked and which didn't]; generating, by the processing device, an updated relevance rule based on the key information and the initial relevance rule, wherein the generating comprises identifying, by the computer, a conflict between the key information and the initial relevance rule, wherein one of the key information and the initial relevance rule identifies at least one document within the set of documents to be relevant and the other of the key information and the initial relevance rule identifies the at least one document within the set of documents to be non-relevant [flagging the search terms that get relevant stuff but still pick up junk]; generating, by the processing device, a query based on the updated relevance rule for identifying the relevant documents within the set of documents [changing the search up a bit with,say, a Boolean AND NOT connector to eliminate some of the junk]; and outputting, by the processing device, the set of documents within which the relevant documents have been identified [running the now-tweaked search again].
The balance of the patented process can be neatly summarized as ‘okay, now have your system do that again and again until you have divided the collection into various piles of really relevant stuff and the other stuff. Lather, rinse, repeat.’
So, if this is how you and your company or firm approach e-discovery, maybe you’d better just cut it out! This is now a patented process, and continuing to infringe the H5 patent makes you a patent ogre (not a patent troll, that’s a different beast). Or you could write fat checks to H5.
More seriously, congratulations to Nicholas Economou and the whole H5 team for securing this patent that I clearly don’t get; and please accept my apologies for all of my apparent infringing on same these last 20 years or so.
Hey, do you suppose I could get a process patent on “lather, rinse, repeat?”
Nicholas clarifies: “[the] process includes a number of key elements, including: user modeling, assessment, query creation, measurement, and iteration. What the patent recognizes as unique includes both certain aspects of the elements individually and the way in which those elements are integrated into a single coherent, replicable, reliable technological process.”
As I am neither single nor reliably coherent, I guess I’m not infringing after all.