In this blog's early days, you could throw a dart at the weekly Class 705 issuances and easily hit a patent that likely "failed" Bilski. Not so, anymore. As the contours of the machine-or-transformation test have (ostensibly) begun to take more definite form through examiner guidelines and BPAI decisions, and as the examining corps is more fully clued in, it's far less common now (in my weekly searching, at least) to find the true outlier.
Which brings us to the following patent that issued yesterday:
|
U.S. Patent No. 7,653,568 System and method of identifying individuals of influence Inventors: Edward B. Keller, et al. Issued: January 26, 2010 Assignee: GfK U.S. Holdings, Inc.
1. A method for generating a database scoring algorithm for use in identifying from demographic data those individuals in a population having a greater probability than other individuals in the population of influencing the choices made by others, the method comprising:
b. providing demographic data for each individual in the first population, wherein the demographic data corresponds to a set of demographic variables;
c. creating categorical intervals in each demographic variable such that each categorical interval is associated with one or more possible data values for the demographic variable;
d. reformatting the demographic data based on the categorical intervals;
e. identifying a plurality of predictive variables from the set of demographic variables, wherein identifying the plurality of predictive variables comprises: calculating for each categorical interval a response index value, and selecting as predictive variables only those demographic variables having a categorical interval with an index value exceeding a pre-determined threshold index value;
f. calculating a percent gain for each categorical interval in a predictive variable as a measure of the correlation between the categorical interval and whether a person is an influential;
g. creating a predictive model based upon the plurality of predictive variables, the categorical intervals, and the percent gains;
h. validating the predictive model; and
i. generating a database scoring algorithm based on the predictive model, wherein the database scoring algorithm is configured to receive demographic data for individuals in a second population and to output a ranking of individuals in the second population based on the probability that each individual is influential;
|
The inventor here, Edward B. Keller, calls these folks "The Influentials", and wrote a book a few years back under that title. The premise, of course, is that advertisers waste too much money targeting the intended purchasers of a product, and instead should target those who can influence others to purchase via, e.g., word of mouth.
It's an interesting thought. Whether it is patentable, or how it can be claimed, is another story... The inclusion of the last element calling for the computer, especially in a "wherein" clause, smacks of a last-minute attempt at machine-tying. And not even particular machine-tying. Let's look under the hood:
- 101 Not Initially Raised. This claim was originally filed in a much shorter form. In particular, the last element of involving a computer was missing. In July 2006, however, the Examiner did not reject the claim as unpatentable. A 101 rejection first was raised in a Final Rejection in March 2007, citing the "useful, concrete and tangible" test. An amendment led to the withdrawal of the rejection in February 2008.
- 101 Returns After Comiskey. While the application was being appealed, the Examiner entered a new rejection in the Examiner's Answer to the appeal brief (a "rare" but permitted occurrence, per MPEP 1207.03), stating "Comiskey explains that mental processes per se are not statutory under 101." While it's nice to see an Examiner staying on top of the law, it would have made more sense to make this rejection prior to the appeal, in the Feb. 2008 office action -- Comiskey was decided in September 2007. Ironically, the Examiner's Answer was mailed October 30, 2008 -- the same day as the Federal Circuit's Bilski decision.
- Post-Bilski Activities. In reopening prosecution, the Applicant added several claim elements (what became steps f), g) and, most notably, k)), arguing that it was now "tied to a computer." The Examiner was unmoved, noting in a Final Rejection that the "identifying" step "may merely involve retrieving predictive variables from a database, which would amount to nothing more than insignificant post-solution activity." (GfK had also argued that the claim "transforms the underlying demographic data for a first population into a database scoring algorithm that can be applied to other sets of demographic data." The Examiner did not respond to that argument.)
- Mysterious Interview. Following the Final Rejection, the Applicants interviewed the Examiner. The substance of that interview, as summarized by the Examiner:
"The rejections under 101, 112, and 103 were discussed."
That's it. We know nothing more about what arguments were raised, apparently with success. There is no subsequent summary from the Applicant on record (MPEP 713.04). All that we know is that a Notice of Allowance, with no reasons for allowance, followed.The NOA did, however, contain an examiner's amendment in which certain terms were tweaked, and even two elements were added: d) and i). Could it be that the Examiner thought that "reformatting" data is a sufficient transformation under Bilski? We're left guessing.
I don't consider myself a Gladwellian Connector, and I doubt this invention would identify me as one of "The Influentials". But people probably don't need much influencing to start wondering about what was going on here.
Recent Comments