Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 09/28/2023 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-15 are rejected under 35 U.S.C. 101 for reciting an abstract idea without significantly more.
Regarding claim 1
Claim 1 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 1 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“executing a first model within a multi-model system for estimating a race for each of the plurality of individuals included in the dataset;” – this limitation recites evaluating identification information and making a classification/judgment as to race. Under its broadest reasonable interpretation, this is a mental evaluation/classification that can be performed in the human mind, even if the claim states it is carried out by a “model”; the use of a computer as a tool does not remove such a limitation from the mental-process grouping. See MPEP § 2106.04(a)(2)(III) (mental processes).
“subsequent to executing the first model, executing a second model within the multi-model system for estimating a race for individuals included in the dataset that were unable to be estimated via the first model;” – this limitation likewise recites further evaluation and judgment on remaining records not yet classified, i.e., another stage of mentally classifying individuals based on available information. This falls within the mental-process grouping because it is still a concept performed in the human mind, such as observation, evaluation, judgment, or opinion. See MPEP § 2106.04(a)(2)(III).
“and subsequent to executing the second model, executing a third model within the multi-model system for estimating a race for individuals included in the dataset that were unable to be estimated via the first model and the second model, wherein the third model estimates a race for individuals with insufficient data fields.” – this limitation recites mental evaluation/judgment, namely making a final race estimate for remaining individuals, including those with incomplete information. The fact that the claim uses a third “model” for insufficient fields does not change the character of the limitation as a judgment/classification step that falls within MPEP § 2106.04(a)(2)(III).
Claim 1 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“performing, using a processor and a memory” – this limitation does not integrate the abstract idea into a practical application because it merely invokes generic computing components to perform the abstract classification process. This is no more than implementing the abstract idea on a computer or using a computer as a tool, which does not impose a meaningful limit on the judicial exception. See MPEP § 2106.05(f).
“acquiring, over a network, a dataset of identification information of a plurality of individuals;” – this limitation constitutes mere data gathering to obtain the information that will later be analyzed by the abstract classification steps. Such obtaining of data is insignificant extra-solution activity and does not integrate the abstract idea into a practical application. See MPEP § 2106.05(g).
Claim 1 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“performing, using a processor and a memory” – this limitation does not amount to significantly more because it recites only generic computer components performing their ordinary functions, i.e., using a processor and memory as tools to carry out the abstract idea. Such generic computer implementation is well-understood, routine, and conventional (WURC), and also reflects mere instructions to apply the exception on a computer. See MPEP § 2106.05(d) and § 2106.05(f).
“acquiring, over a network, a dataset of identification information of a plurality of individuals;” – this limitation is mere data gathering/insignificant extra-solution activity and therefore does not amount to an inventive concept. Obtaining data for use by the abstract idea, especially at a high level of generality over a generic network, is also well-understood, routine, and conventional (WURC). See MPEP § 2106.05(g) and § 2106.05(d).
Regarding claim 2
Claim 2 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 2 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“performing one or more pre- processing operations on the data set of identification information of the plurality of individuals.” – this limitation broadly recites preparing, organizing, or conditioning information before the later race-estimation steps are performed. Under its broadest reasonable interpretation, such preprocessing is an abstract information-evaluation/manipulation step that can performed mentally, such as reviewing, standardizing, sorting, or conditioning input information before making a judgment, and thus falls within mental processes; to the extent the preprocessing includes formatting or transforming values, it also reasonably falls within mathematical concepts. See MPEP § 2106.04(a)(2)(III) and § 2106.04(a)(2)(I).
Claim 2 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application.
Claim 2 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception.
Regarding claim 3
Claim 3 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 3 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein the first model is a BIFSG (new Bayesian Improved First Name Surname Geocoding) model.” – this limitation specifies that the first race-estimation step is performed using a Bayesian geocoding model based on first name, surname, and related information. Under its broadest reasonable interpretation, this step recites the abstract evaluation/classification of information to estimate race, and the Bayesian nature of the model also recite a mathematical relationship/calculation. See MPEP § 2106.04(a)(2)(III) and MPEP § 2106.04(a)(2)(I).
Claim 3 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application.
Claim 3 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception.
Regarding claim 4
Claim 4 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 4 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein the second model is a BISG (Bayesian Improved Surname Geocoding) model.” - this limitation specifies that the second race-estimation step is performed using a Bayesian geocoding model. Under its broadest reasonable interpretation, this step recites the abstract evaluation/classification of information to estimate race, and the Bayesian nature of the model also recite a mathematical relationship/calculation. See MPEP § 2106.04(a)(2)(III) and MPEP § 2106.04(a)(2)(I).
Claim 4 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application.
Claim 4 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception.
Regarding claim 5
Claim 5 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 5 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein the third model is a machine learning model.” – this limitation merely specifies that the third race-estimation step is performed using a machine learning model. Under its broadest reasonable interpretation, the limitation still recites the abstract classification of information to estimate race, i.e., a mental process, and additionally recites use of a mathematical model or algorithm to generate the estimate, i.e., a mathematical concept. See MPEP § 2106.04(a)(2)(III) and MPEP § 2106.04(a)(2)(I).
Claim 5 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application.
Claim 5 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception.
Regarding claim 6
Claim 6 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 6 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 6 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein the second model is being executed prior to completion of the first model.” – this limitation does not integrate the abstract idea into a practical application because it merely specifies that the abstract estimation/classification operations are overlapped in time on generic computing hardware. It amounts to using the computer as a tool to carry out the abstract idea without a particular execution order/timing, without any recited improvement to computer functioning, model architecture, or other technology. See MPEP § 2106.05(f).
Claim 6 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“wherein the second model is being executed prior to completion of the first model.” – this limitation does not amount to significantly more because it merely recites generic scheduling or ordering of execution of abstract analytical steps on conventional computer components. The claim does not recite any specific technical mechanism for parallelization, resource allocation, synchronization, or other implementation that would amount to an inventive concept; instead it reflects well-understood, routine, and conventional (WURC) use of a computer processor to execute tasks with overlapping timing. See MPEP § 2106.05(d).
Regarding claim 7
Claim 7 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 7 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 7 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein the third model is being executed prior to completion of the second model.” – this limitation merely specifies that the abstract estimation/classification operations are overlapped in time on generic computing hardware. It amounts to using the computer as a tool to carry out the abstract idea with a particular execution order or timing. See MPEP § 2106.05(f).
Claim 7 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“wherein the third model is being executed prior to completion of the second model.” – this limitation merely recites generic scheduling or ordering of execution of abstract analytical steps on conventional computer components. It reflects well-understood, routine, and conventional (WURC) use of a computer processor to execute tasks with overlapping timing. See MPEP § 2106.05(d).
Regarding claim 8
Claim 8 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 8 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 8 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein the second model is executed upon completion of the first model.” – merely specifying a sequential ordering for carrying out the abstract estimation/classification operation on generic computing hardware, amounts to using the computer as a tool to perform the abstract idea, without any recited improvement to computer functioning, processor operation, model architecture, or other technology. See MPEP § 2106.05(f).
Claim 8 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“wherein the second model is executed upon completion of the first model.” – this limitation does not amount to significantly more because it merely recites generic sequencing of execution of abstract analytical steps on conventional computer components. This reflects well-understood, routine, and conventional (WURC) use of a processor to execute tasks in sequence. See MPEP § 2106.05(d).
Regarding claim 9 (with assuming correction to “third model”)
Claim 9 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 9 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 9 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein the third mode is executed upon completion of the second model.” – this limitation merely specifies a sequential ordering for carrying out the abstract estimation/classification operations on generic computing hardware. It amounts to using the computer as a tool to perform the abstract idea in a particular order, without any recited improvement to computer functioning, processor operation, model architecture, or other technology. See MPEP § 2106.05(f).
Claim 9 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“wherein the third mode is executed upon completion of the second model.” – this limitation does not amount to significantly more because it merely recites generic sequencing of execution of abstract analytical steps on conventional computer components. This reflects well-understood, routine, and conventional (WURC) use of a processor to execute tasks in sequence. See MPEP § 2106.05(d).
Regarding claim 10
Claim 10 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 10 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 10 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein the first model is unable to estimate a race for individuals with insufficient data fields.” – this limitation merely further characterizes the first abstract race-estimation step by stating that the first model cannot perform the estimate when data fields are insufficient, i.e., it describes a condition under which the abstract evaluation/classification is not completed. This does not improve computer functionality or other technology, and does not impose a meaningful limit on the judicial exception. See MPEP § 2106.05(f).
Claim 10 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“wherein the first model is unable to estimate a race for individuals with insufficient data fields.” – this limitation merely states a result of the first abstract analytical step when insufficient information is present, while the remaining non-abstract elements are still the generic processor, memory, and network-based data acquisition of claim 1. Those extra elements remain well-understood, routine, and conventional (WURC). See MPEP § 2106.05(d).
Regarding claim 11
Claim 11 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 11 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
Claim 11 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application. The additional elements:
“wherein the second model is unable to estimate a race for individuals with insufficient data fields.” – this limitation merely describes a condition under which the abstract evaluation/classification is not performed. This does not improve computer functionality or any other technology, and does not impose a meaningful limit on the judicial exception. See MPEP § 2106.05(f).
Claim 11 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception. The additional elements are:
“wherein the second model is unable to estimate a race for individuals with insufficient data fields.” –this limitation merely states a result or condition of the second abstract analytical step when insufficient information is present, while the remaining non-abstract elements are still the generic processor, memory, and network-based data acquisition of claim 1. Those extra elements remain well-understood, routine, and conventional (WURC). See MPEP § 2106.05(d).
Regarding claim 12
Claim 12 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 12 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein the first model requires more data fields than the second model for performing the estimating.” – this limitation recites a comparison of informational requirements between two models used for the claimed race-estimation process. Under its broadest reasonable interpretation, this is still an evaluation or judgment regarding what information is needed to perform the abstract classification, and thus falls within mental processes under MPEP § 2106.04(a)(2)(III).
Claim 12 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application.
Claim 12 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception.
Regarding claim 13
Claim 13 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 13 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein accuracy and coverage of an estimate provided by the first model are improved upon execution of the second model.” – this limitation recites an evaluative result of the claimed race-estimation process, namely that the second model improves the quality and extent of the estimate relative to the first model. Under BRI, this is still part of the abstract evaluation/classification concept and can be characterized as a mental process because it reflects assessing and comparing estimation results. See MPEP § 2106.04(a)(2)(III).
Claim 13 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application.
Claim 13 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception.
Regarding claim 14
Claim 14 – Step 1 – Is the claim to a process, machine, manufacture or composition of matter?
Yes, the claim is to a process.
Claim 14 – Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites an abstract idea.
“wherein accuracy and coverage of an estimate provided by the second model are improved upon execution of the third model.” – this limitation recites an evaluative result of the claimed race-estimation process, namely that the third model improves the quality and extent of the estimate relative to the second model. Under BRI, this is still part of the abstract evaluation/classification concept and can be characterized as a mental process because it reflects assessing and comparing estimation results. See MPEP § 2106.04(a)(2)(III).
Claim 14 – Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No. There are no additional elements that integrate the judicial exception into a practical application.
Claim 14 – Step 2B – Does the claim recite additional elements that amount to significantly more than the judicial exception?
No. There are no additional elements that amount to significantly more than the judicial exception.
Regarding claim 15
Claim 15 is directed to the same judicial exception as claim 1 because it recites the same substantive limitations in system form, with the processor, memory, and communication circuit merely providing generic computer implementation of the abstract idea and not adding any meaningful limitation.
Regarding claim 16
Claim 16 is directed to the same judicial exception as claim 1 because it recites the same substantive limitations in non-transitory computer-readable-medium form, with the stored instructions merely causing processor/system to perform the same abstract idea and not adding significantly more.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-5 and 10-16 are rejected under 35 U.S.C. 103 as being unpatentable over Ioan Voicu (Using First Name Information to Improve Race and Ethnicity Classification) in view of Douglas C. Merrill et al. (US20190043070A1).
Regarding claim 1, Voicu in view of Merrill, teach a method for performing sequential multi-model race estimation, the method comprising: performing, using a processor and a memory:
“” – Voicu teaches this limitation in part. Voicu teaches Bayesian Improved Surname Geocoding (BIFSG) using identifying information including:
“calculates the probability that a person with surname (s), first name (f), and geographic area of residence (g)” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
“executing a first model within a multi-model system for estimating a race for each of the plurality of individuals included in the dataset;” – Voicu teaches this limitation. Voicu teaches:
“The new Bayesian Improved First Name Surname Geocoding (BIFSG) method” (Voicu, p. 1, § Abstract)
“the BIFSG formula … calculates the probability that a person with surname (s), first name (f), and geographic area of residence (g) belongs to racial or ethnic group (r)” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
Voicu makes clear that BIFSG is an improvement over BISG, i.e., part of a multi-modal estimation context:
“illustrating the accuracy and coverage improvements that BIFSG achieves compared with BISG.” (Voicu, p. 6, § 3. Evaluation Results)
“subsequent to executing the first model, ” – Voicu teaches this limitation in part. Voicu teaches:
“The BIFSG formula requires that all input probabilities are nonmissing.” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
And teaches when this is not the case, fallback may be performed:
“it may be desirable to enhance the algorithm so that it also creates proxies when one or more of the input probabilities are missing. This can be easily accomplished by computing proxies using a BISG-like formula if two attributes have nonmissing values” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
“and subsequent to executing the second model, executing a third model within the multi-model system for estimating a race for individuals included in the dataset that were unable to be estimated via the first model and the second model,” – Voicu teaches this limitation. Voicu teaches fallback to lower-information estimation:
“using surname-only (SO), first-name-only (FO), or geography-only (GO) probabilities if a single attribute bas nonmissing values.” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
That is a third estimation model used after the richer models cannot be applied.
“wherein the third model estimates a race for individuals with insufficient data fields.” – Voicu teaches this limitation. Voicu teaches that:
“The BIFSG formula requires that all input probabilities are nonmissing.” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
Voicu then teaches fallback when information is insufficient by using:
“a BISG-like formula if two attributes have nonmissing values” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
and:
“surname-only (SO), first-name-only (FO), or geography-only (GO) probabilities if a single attribute bas nonmissing values.” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
Voicu does not teach these limitations:
“acquiring, over a network … “
“… executing a second model … for estimating a race“
Merrill, however, teaches these limitations:
“… executing a second model … for estimating a race“ – Merrill that a second model may be used to generate a protected-class prediction, including race, for each data set/person, and that the model may specifically be BISG. Merrill teaches that:
“the protected class information for a data set indicates a probability of the person associated with the data set being a member of a protected class being evaluated.” (Merrill, p. 2, ¶[0027])
“the protected class being evaluated includes one of a class of race, ethnicity, national origin, religion, sex, sexual orientation, age, and disability status” (Merrill, p. 2, ¶[0027])
“the model evaluation system generates a protected class prediction for each data set by executing computer-executable instructions for a protected class model” (Merrill, p. 2, ¶[0028])
Merrill also expressly states in claim 19 that:
“the evaluation system generates the protected class membership information from the input rows by using a BISG process.” (Merrill, p. 29, claim 19)
“acquiring, over a network … “ – Merrill teaches this limitation. Merrill teaches:
“modeling system 110 runs on a separate device from the model evaluation system 120 and variables and scores are transmitted between the modeling system 110 and model evaluation system 120 via a computer network.” (Merrill, p. 13, ¶[0175])
“the evaluation system providing information identifying each variable of the predictor, the determined protected class model ranking value, and the determined decisioning system ranking value to an operator device via a network” (Merrill, p. 28, claim 2)
A POSITA would have been motivated to implement Voicu’s staged race/ethnicity estimation workflow in Merrill’s conventional processor/memory/network environments because doing so would have predictably enabled automated acquisition and processing of plural electronic records while preserving Voicu’s disclosed fallback estimation for incomplete data. Voicu already teaches successive estimation approaches depending on missing inputs; Merrill teaches BISG-based generation of protected-class membership information from input rows teaches transmitting over a computer network and processing it with processors and memory.
Regarding claim 2, Voicu in view of Merrill, teach the method according to claim 1, further comprising
“performing one or more pre- processing operations on the data set of identification information of the plurality of individuals.” – Voicu teaches this limitation. Voicu teaches:
“Following the initial data cleaning and ma1ching steps, the probability inputs to the BlFSG formula are computed.” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
Voicu further teaches preprocessing of identifying information by addressing unmatched or missing fields before estimation, including:
“for applications with missing surnames or surnames 1hat cannot be matched with the census list, the surname-based probabilities are considered missing.” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
A POSITA would have been motivated to use such initial data cleaning and matching operations before performing the staged race-estimation process, because doing so predictably prepares the identification information for subsequent estimation.
Regarding claim 3, Voicu in view of Merrill, teach the method according to claim 1, wherein
“the first model is a BIFSG (new Bayesian Improved First Name Surname Geocoding) model.” – Voicu teaches this limitation. Voicu teaches:
“The new Bayesian 1rnproved First Name Surname Geocoding (BIFSG) method” (Voicu, p. 1, § Abstract)
And further teaches:
“Implementation of the BIFSG Algorithm … Constructing the BIFSG proxies … applying the BIFSG formula to compute the proxies.” (Voicu, p. 3, § 2.3. Implementation of the BIFSG Algorithm)
Voicu also teaches that:
“the BISFG formula … calculates the probability that a person with surname (s), first name (f), and geographic area of residence (g) belongs to racial or ethnic group (r)” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
A POSITA would have been motivated to use Voicu’s expressly disclosed BIFSG model as the first model in the claim 1 workflow because BIFSG uses the richest available set of identifying inputs and predictably provides the most informative initial race/ethnicity estimate before fallback to lower-information models.
Regarding claim 4, Voicu in view of Merrill, teach the method according to claim 1, wherein
“the second model is a BISG (Bayesian Improved Surname Geocoding) model.” – Voicu teaches this limitation in part. Voicu teaches that the second model is “a BISG-like formula” used in fallback estimation when the richer BIFSG inputs are unavailable. Voicu teaches:
“This can be easily accomplished by computing proxies using a BISG-like formula if two attributes have nonmissing values” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
Voicu does not expressly teach:
that the second model is expressly a BISG model
Merrill, however, teaches this remaining aspect:
“the second model is a BISG … model” – Merrill teaches that:
“the evaluation system generates the protected class membership information from the input rows by using a BISG process.” (Merrill, p. 27, ¶[0329])
A POSITA would have been motivated to use Merrill’s expressly identified BISG process as the second model in Voicu’s staged estimation workflow because Voicu already teaches fallback to a BISG-like formula when fuller BIFSG inputs are unavailable, and using the known BISG process would have been a predictable way to implement that fallback.
Regarding claim 5, Voicu in view of Merrill, teach the method according to claim 1, wherein
“the third model ” – Voicu teaches this limitation in part. As already set forth in the claim 1 combination, Voicu teaches the third fallback estimation stage:
“using surname-only (SO), first-name-only (FO), or geography-only (GO) probabilities if a single attribute bas nonmissing values.” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
Voicu does not teach:
“… is a machine learning model.”
Merrill, however, teaches this limitation:
“… is a machine learning model.” – Merrill teaches:
“In some embodiments the ensembler is built via stacking or blending or both, and other combinations of ensembling methods, without limitation, where model weights are determined based on the execution of a machine learning model” (Merrill, p. 15, ¶[0189])
A POSITA would have been motivated to use Merrill’s machine-learning-based model techniques as the third model in Voicu’s staged estimation workflow because Voicu already teaches a third fallback model for lower-information records, and Merrill teaches machine-learning-model-based ensembling, with the predictable benefit of implementing the later-stage fallback using a known machine-learning model.
Claim 5 depends from claim 1; therefore, the same motivation to combine applies to claim 5.
Regarding claim 10, Voicu in view of Merrill teach the method according to claim 1, wherein:
“the first model is unable to estimate a race for individuals with sufficient data fields.” – Voicu teaches this limitation. Voicu teaches that:
“The BIFSG formula requires that all input probabilities are nonmissing.” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm)
Voicu teaches the insufficient-data-fields context, namely:
“For applications with missing first names or first names that cannot be matched with the first name list, the first-name-based probabilities are considered missing” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm (b))
“For applications with missing geo-demographic information, the geography-based probabilities are considered missing.” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm (c))
A POSITA would have understood from Voicu that the first model – BIFSG, is unable to estimate race when the required input probabilities are missing, i.e., when individuals have insufficient data fields.
Claim 10 depends from claim 1; therefore, the same motivation to combine applies to claim 10.
Regarding claim 11, Voicu in view of Merrill, teach the method according to claim 1, wherein
“the second model is unable to estimate a race for individuals with insufficient data fields.” – Voicu teaches this limitation. Voicu teaches fallback to the second model only:
“if two attributes have nonmissing values” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm)
And further teaches the lower-information fallback:
“using surname-only (SO), first-name-only (FO), or geography-only (GO) probabilities if a single attribute bas nonmissing values.” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
Voicu also teaches the insufficient-data fields context, namely:
“For applications with missing first names … the first-name-based probabilities are considered missing” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm (b))
“For applications with missing geo-demographic information, the geography-based probabilities are considered missing.” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm (c))
A POSITA would have understood from Voicu that the second model, i.e., the BISG-like fallback requiring two non-missing attributes, is unable to estimate race when the necessary input fields are insufficiently available.
Claim 11 depends from claim 1; therefore, the same motivation to combine applies to claim 11.
Regarding claim 12, Voicu in view of Merrill, teach the method according to claim 1, wherein:
“the first model requires more data fields than the second model for performing the estimating.” – Voicu teaches this limitation. Voicu teaches that:
“The BIFSG formula requires that all input probabilities are nonmissing.” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm)
Voicu further teaches that the second model uses fewer inputs (whereas BIFG uses surname, first name, and geographic information):
“This can be easily accomplished by computing proxies using a BISG-like formula if two attributes have nonmissing values” (Voicu, p. 4, 2.3. Implementation of the BIFSG Algorithm)
A POSITA would have understood from Voicu that the first model, BIFSG, requires more data fields than the second model, i.e., the BISG-like fallback, because BIFSG uses surname, first name, and geography, while the BISG-like fallback operates when only when two attributes have non-missing values.
Claim 12 depends from claim 1; therefore, the same motivation to combine applies to claim 12.
Regarding claim 13, Voicu in view of Merrill, teach the method according to claim 1, wherein:
“accuracy and coverage of an estimate provided by the first model are improved upon execution of the second model.” – Voicu teaches this limitation. Voicu teaches that adding the second-model fallback to the first-model workflow improves coverage, while maintaining/improving accuracy of the resulting estimate set:
“it may be desirable to enhance the algorithm so that it also creates proxies when one or more of the input probabilities are missing. This can be easily accomplished by computing proxies using a BISG-like formula if two attributes have nonmissing values” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm)
And further teaches that:
“the difference in accuracy between BISG and the BIFSG ·'extended’ with B!SG narrows as the missing rate for first names increases, since the extended BIFSG becomes more similar to BISG as more observations have race/ethnicity imputed based on BISG.” (Voicu, p. 5, footnote 16)
Voicu also teaches that the first-model approach retains superior accuracy/coverage characteristics in comparison to BISG, including that:
“the new approach offers improvements, in terms of accuracy and coverage, over BJSG” (Voicu, p. 2, § 1. Introduction)
And that:
“The improvements of Bi FSG also extend to the accuracy and coverage of various classification schemes.” (Voicu, p. 7, § 3.2.2. Classification Accuracy)
A POSITA would have understood from Voicu that executing the second-model fallback improves the resulting estimate set by increasing coverage for records with missing inputs while maintaining the accuracy characteristics discussed by Voicu.
Claim 13 depends from claim 1; therefore, the same motivation to combine applies to claim 13.
Regarding claim 14, Voicu in view of Merrill, teach the method according to claim 1, wherein:
“accuracy and coverage of an estimate provided by the second model are improved upon execution of the third model.” – Voicu teaches this limitation. Voicu teaches that, when fuller-input estimation cannot be performed, fallback may proceed by:
“computing proxies using a BISG-like formula if two attributes have nonmissing values, and using surname-only (SO), first-name-only (FO), or geography-only (GO) probabilities if a single attribute has nonmissing values.” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm)
Voicu further teaches the coverage effect of the further fallback, namely that extending the estimation process to cases with missing inputs allows estimation to continue for records that otherwise would not be estimated, explaining that:
“it may be desirable to enhance the algorithm so that it also creates proxies when one or more of the input probabilities are missing.” (Voicu, p. 4, § 2.3. Implementation of the BIFSG Algorithm)
A POSITA would have understood from Voicu that executing the third-model fallback after the second-model stage improves the resulting estimate set by increasing coverage for records with still fewer inputs, while preserving the staged estimation approach disclosed by Voicu.
Claim 14 depends from claim 1; therefore, the same motivation to combine applies to claim 14.
Regarding claim 15
Claim 15 is unpatentable over Voicu in view of Merrill for the same reasons set forth for claim 1, because claim 15 recites the same substantive multi-modal race-estimation limitations in system form, with the processor, memory, and communication circuit functionality for acquiring identification information and executing the staged estimation models. Therefore, claim 15 is unpatentable over Voicu in view of Merrill for the reasons discussed above with respect to claim 1.
Regarding claim 16
Claim 16 is unpatentable over Voicu in view of Merrill for the same reasons set forth for claim 1, because claim 16 recites the same substantive multi-modal race-estimation limitations in non-transitory computer-readable storage medium form, with the stored instructions causing a processor/system to perform the same method steps already taught by the combination of Voicu and Merrill.
Claims 6-9 are rejected under 35 U.S.C. 103 as being unpatentable over Voicu in view of Merrill et al. and further in view of Parangi et al. (US20210232920A1).
Regarding claim 6, Voicu in view of Merrill and further view of Parangi, teach the method according to claim 1, wherein
“the second model is being executed prior to completion of the first model.” – Voicu does not teach this limitation. Parangi, however, teaches this limitation. Parangi teaches:
“generating by the machine learning engine, a second machine learning model based upon the at least one characteristic of the user-specified data set and at least one characteristic of the user-specified task, responsive to receiving the user-specified data set and the user-specified task, during execution of the first machine learning model (214).” (Parangi, p. 4, ¶[0037])
A POSITA would have been motivated to execute the second model prior to execution of the first model in the Voicu/Merrill workflow in view of Parangi because Parangi teaches overlapping execution of first and second machine-learning models, and applying that known timing arrangement to the established multi-model estimation workflow would have predictably reduced latency and improved throughput while still generating the respective model outputs.
Regarding claim 7, Voicu in view of Merrill and further in view of Parangi, teach the method according to claim 1, wherein
“the third model is being executed prior to completion of the second model.” – Voicu does not teach this limitation. Parangi, however, teaches this limitation. Parangi teaches the execution of a later model prior to completion of an earlier model. Parangi teaches:
“the method 200 includes generating, by the machine learning engine, a second machine learning model … during execution of the first machine learning model (214).” (Parangi, p. 4, ¶[0037])
A POSITA would have found it obvious to apply Parangi’s known overlapping execution timing to the later handoff between the second and third models in the established Voicu/Merrill workflow, because selecting whether a later-stage model begins during execution of an earlier-stage model would have been a predictable scheduling variation in a staged model pipeline, with the predictable benefit of reducing latency and increasing throughput.
Regarding claim 8, Voicu in view of Merrill and further view of Parangi, teach the method according to claim 1, wherein
“the second model is executed upon completion of the first model.” – Voicu does not teach this limitation. Parangi, however, teaches this limitation. Parangi teaches that, in some embodiments, the method includes:
“generating, by the machine learning engine, a second machine learning model based upon the at least one characteristic of the user-specified data set and at least one characteristic of the user-specified task, responsive to receiving the user-specified data set and the user-specified task, subsequent to execution of the first machine learning model.” (Parangi, p. 4, ¶[0037])
A POSITA would have been motivated to execute the second model upon completion of the first model in the Voicu/Merrill workflow in view of Parangi because Parangi expressly teaches sequential handoff from a first machine-learning model to a second machine-learning model, and applying that known timing arrangement to the established multi-modal estimation workflow would have predictably supported orderly staged processing of records from the first model to the second model.
Regarding claim 9, Voicu in view of Merrill and further in view of Parangi, teach the method according to claim 1, wherein
“the third mode is executed upon completion of the second model.” – Voicu does not teach this limitation. Parangi, however, teaches this limitation. Parangi teaches execution of a later model after completion of an earlier model. Parangi teaches:
“subsequent to execution of the first machine learning model.” (Parangi, p. 4, ¶[0037])
A POSITA would have found it obvious to apply Parangi’s known completion-based sequential execution timing to the later handoff between the second and third models in the established Voicu/Merrill workflow, because selecting whether a later-stage model begins after completion of an earlier-stage would have been a predictable scheduling choice in a staged model pipeline, with the predictable benefit of orderly staged processing and simplified execution control.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Paul Coleman whose telephone number is (571)272-4687. The examiner can normally be reached Mon-Fri.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, David Yi can be reached at (571) 270-7519. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/PAUL COLEMAN/ Examiner, Art Unit 2126
/DAVID YI/ Supervisory Patent Examiner, Art Unit 2126