DETAILED ACTION
This Office Action is in response to communications filed on April 28th, 2026 for Application No. 17/511,946, in which claims 1-20 are presented for examination. The amendments filed on April 28th, 2026 have been entered, where claims 1, 9, and 17 are amended.
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 .
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 1-20 are rejected under 35 U.S.C. 101 because the claimed inventions are directed to abstract ideas without significantly more.
Regarding Claim 1:
Step 1: Claim 1 is a process claim. Therefore, claims 1-8 are directed to a statutory category of eligible subject matter.
Step 2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the "Mental Processes" grouping of abstract ideas. Here, elements of the claimed subject matter are mental processes. Specifically, the claim recites
“A method comprising: generating, for each of a plurality of users, probabilities of object interactions, each probability of object interaction corresponding to a respective object attribute and representing a likelihood that the respective user will interact with an object having the respective object attribute during a forecasting window, wherein a given object recommendation ranking for a respective user comprises a ranked list of the object attributes ranked according to the corresponding probabilities of object interactions” (mental processes – amounts to exercising judgment to form an opinion on object interaction probabilities for users within a forecasting window, with reference to known or observed entities with known or observed attributes that correspond with the object interaction probabilities, which may be aided by pen and paper to create a corresponding ranked list of object attributes for each users object recommendation rankings);
“calculating per-user interaction probabilities for each user over a forecasting window, wherein each per-user interaction probability represents a likelihood that the respective user will make any interaction during the forecasting window” (mental processes – amounts to exercising judgement to form an opinion on per-user interaction probabilities over a specific time window, which may be aided by pen and paper to perform the calculations);
“calculating affinity group rankings based on the probabilities of object interactions and the per-user interaction probabilities for each user” (mental processes – amounts to exercising judgement to form an opinion on affinity group rankings, with reference to previously determined values, which may be aided by pen and paper to perform the calculations);
“grouping the plurality of users based on the affinity group rankings” (mental processes – amounts to exercising judgement to form an opinion on groupings of users, with reference to previously determined rankings, which may be aided by pen and paper to memorialize the groupings);
“transmitting content items to the grouped users based on their respective affinity group rankings” (mental processes – apart from the “transmitting” itself, amounts to exercising judgment to form an opinion on which known or observed content items are applicable to previously determined groups of users);
“maintains associations between user identifiers and corresponding affinity group rankings” (mental processes – amounts to exercising judgment to associate descriptive characteristics of known or observed users with their groupings, which may be aided by pen and paper to maintain and memorialize these associates);
“detecting observed user interactions with the transmitted content items” (mental processes – apart from the “detecting observed” itself, which may require a particular technological environment when applied to “transmitted” data or may be considered data gathering, amounts to exercising judgment to observe events);
“continuously updating the affinity group table based on the detected observed user interactions with the transmitted content items, wherein the updating modifies the . . . affinity group rankings based on the observed user interactions” (mental processes – apart from the “detected” itself, which as discussed above may require a particular technological environment when applied to “continuously” updated based on “transmitted” data or may be considered data gathering, amounts to exercising judgment to form opinions on updates to the previously determined table of user-ranking associates, based on known or observed interactions that alter previous opinions, which may be aided by pen and paper to maintain and memorialize the updates); and
“re-executing the generating and calculating steps for subsequent forecasting windows using the updated affinity group table to improve prediction accuracy of future user interactions through iterative refinement of the affinity group rankings” (mental processes – amounts to repeating the mental processes discussed above, with reference to known updates, in order to iteratively adjust the method and its outputs with the goal of improving accuracy of opinions).
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
The claim recites the additional elements:
“transmitting . . .” (transmitting content items amounts to insignificant extra-solution activity because transmission of data is incidental to the claimed subject matter);
“storing the affinity group rankings in an affinity group table comprising a data structure that . . . the stored” (storing affinity group rankings in a table amounts to insignificant extra-solution activity because the storing and retrieving of information in a memory data structure is incidental to the claimed subject matter);
“detecting observed . . . transmitted . . . continuously . . . based on the detected . . . transmitted” (in the event that the recitations of detecting events relating to transmitted data and continuously updating information based on the detected events relating to the transmitted data required a particular technological environment, merely generally linking the use of the judicial exception to a particular technological environment or field of use does not impose any meaningful limits on practicing the abstract idea); and
“detecting observed . . . based on the detected” (in the event that the recitations of detecting observed and based on the detected amounted to data gathering, the recitations would amount to insignificant extra-solution activity because mere data gathering is incidental to the claimed subject matter, see CyberSource v. Retail Decisions, Inc., 654 F.3d 1366, 1375, 99 USPQ2d 1690, 1694 (Fed. Cir. 2011))
Step 2B: The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
The claim recites the additional elements:
“transmitting . . .” (transmitting data is well‐understood, routine, and conventional, see Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362; see also buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014); therefore the limitation, which is recited with a high level of generality, remains insignificant extra-solution activity even upon reconsideration);
“storing the affinity group rankings in an affinity group table comprising a data structure that . . . the stored” (storing and retrieving information in memory is well‐understood, routine, and conventional, see Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015); OIP Techs., 788 F.3d at 1363, 115 USPQ2d at 1092-93; therefore the limitation, which is recited with a high level of generality, remains insignificant extra-solution activity even upon reconsideration);
“detecting observed . . . transmitted . . . continuously . . . based on the detected . . . transmitted” (merely generally linking the use of the judicial exception to a particular technological environment or field of use does not provide an inventive concept); and
“detecting observed . . . based on the detected” (mere data gathering is well‐understood, routine, and conventional, see OIP Techs., 788 F.3d at 1362-63, 115 USPQ2d at 1092-93; see also CyberSource v. Retail Decisions, Inc., 654 F.3d 1366, 1375, 99 USPQ2d 1690, 1694 (Fed. Cir. 2011); therefore the limitation, which is recited with a high level of generality, remains insignificant extra-solution activity even upon reconsideration).
For the reasons above, Claim 1 is rejected as being directed to an abstract idea without significantly more. This rejection applies equally to dependent claims 2-8. The additional limitations of the dependent claims are addressed below.
Regarding Claim 2:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 2 depends on. Here, the claim recites additional limitations that are mental processes. Specifically, the claim recites
“wherein generating the probability of object interactions for the respective user comprises classifying the respective user” (mental processes – amounts to exercising judgment to form an opinion on object interaction probabilities and classifications for users, with reference to known, observed, or imagined entities, which may be aided by pen and paper).
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
The claim recites the additional elements:
“using a classification model” (amounts to mere instructions to apply the judicial exception on generic and unspecialized computer components, which do not impose any meaningful limits on practicing the abstract idea).
Step 2B: The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
The claim recites the additional elements:
“using a classification model” (mere instructions to apply the exception using generic computer components does not provide an inventive concept).
Accordingly, Claim 2 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 3:
Step 2A Prong 1: See the rejection of Claim 2 above, which Claim 3 depends on. Here, the claim recites additional limitations that are mental processes. Specifically, the claim recites
“wherein classifying the respective user using the classification model comprises classifying the respective user” (mental processes – amounts to exercising judgment to form an opinion on classifications for users, with reference to known, observed, or imagined entities, which may be aided by pen and paper).
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
The claim recites the additional elements:
“using a multinomial random forest classifier” (amounts to mere instructions to apply the judicial exception on generic and unspecialized computer components, which do not impose any meaningful limits on practicing the abstract idea).
Step 2B: The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
The claim recites the additional elements:
“using a multinomial random forest classifier” (mere instructions to apply the exception using generic computer components does not provide an inventive concept).
Accordingly, Claim 3 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 4:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 4 depends on. Here, the claim recites additional limitations that are mental processes. Specifically, the claim recites
“wherein each attribute in the ranked list of object attributes is associated with a corresponding score, the corresponding score used to sort the ranked list of object attributes” (mental processes – amounts to exercising judgment to form an opinion scores associated with items in a list, which are evaluated to determine the ranking, which may be aided by pen and paper).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 4 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 5:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 5 depends on. Here, the claim recites additional limitations that are mental processes. Specifically, the claim recites
“wherein calculating the affinity group rankings comprises computing a predicted number of interactions using a lifetime value model” (mental processes – amounts to exercising judgement to form an opinion on affinity group rankings, with reference to previously determined values and a statistical model, which may be aided by pen and paper to perform the calculations in a manner consistent with a lifetime value model).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 5 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 6:
Step 2A Prong 1: See the rejection of Claim 5 above, which Claim 6 depends on. Here, the claim recites additional limitations that are mental processes. Specifically, the claim recites
“wherein computing the predicted number of interactions using the lifetime value model comprises computing a predicted number of interactions using a beta-geometric model” (mental processes – amounts to exercising judgement to form an opinion on affinity group rankings, with reference to previously determined values and a specific type of statistical model, which may be aided by pen and paper to perform the calculations in a manner consistent with a lifetime value model using a beta-geometric model).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 6 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 7:
Step 2A Prong 1: See the rejection of Claim 6 above, which Claim 7 depends on. As discussed above, if a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the "Mental Processes" grouping of abstract ideas. Additionally, if a claim limitation, under its broadest reasonable interpretation, covers mathematical relationships, mathematical formulas or equations, or mathematical calculations, then it falls within the “Mathematical Concepts” grouping of abstract ideas. Here, the claim recites additional limitations that are mental processes and could also be interpreted as mathematical concepts. Specifically, the claim recites
“wherein computing the predicted number of interactions comprises dividing an output of the beta-geometric model by a total number of expected interactions to obtain the predicted number of interactions for the respective user” (mental processes – amounts to exercising judgement to form an opinion on a predicted number of interactions, with reference to previously determined values, which may be aided by pen and paper to perform the calculations; mathematical concepts – dividing two inputs to determine a output amounts to a mathematical formula).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 7 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 8:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 8 depends on. As discussed above, if a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the "Mental Processes" grouping of abstract ideas. Additionally, if a claim limitation, under its broadest reasonable interpretation, covers mathematical relationships, mathematical formulas or equations, or mathematical calculations, then it falls within the “Mathematical Concepts” grouping of abstract ideas. Here, the claim recites additional limitations that are mental processes and could also be interpreted as mathematical concepts. Specifically, the claim recites
“wherein calculating affinity group rankings comprises multiplying the probability of object interactions by the per-user interaction probabilities for each user to obtain a likelihood for each user” (mental processes – amounts to exercising judgement to form an opinion on affinity group rankings, with reference to previously determined values, which may be aided by pen and paper to perform the calculations; mathematical concepts – multiplying two inputs to determine a product amounts to a mathematical formula).
Step 2A Prong 2 & Step 2B: There are no elements left for consideration of implementation within a practical application or for consideration of significantly more.
Accordingly, Claim 8 is rejected as being directed to an abstract idea without significantly more.
Regarding Claim 9:
Step 1: Claim 9 is a machine claim. Therefore, claims 9-16 are directed to a statutory category of eligible subject matter.
Step 2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the "Mental Processes" grouping of abstract ideas. Here, the claim recites elements that are substantially the same as the limitations of Claim 1. As a result, and as elaborated above, these limitations are abstract ideas because they are mental processes. Additionally, the claim recites additional limitations that are mental processes. Specifically, the claim recites
“wherein a given object for a respective user comprises a ranked list of the object attributes” (mental process – amounts to exercising judgement to form an opinion on a value comprised of a ranked list of information).
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
The claim recites the additional elements:
“A non-transitory computer-readable storage medium for tangibly storing computer program instructions capable of being executed by a computer processor, the computer program instructions defining steps of” (amounts to mere instructions to apply the judicial exception on generic and unspecialized computer components, which do not impose any meaningful limits on practicing the abstract idea);
“transmitting . . .” (transmitting content items amounts to insignificant extra-solution activity because transmission of data is incidental to the claimed subject matter); and
“continuously . . . based on observed . . . transmitted” (in the event that the recitations of continuously updating information based on observed events relating to the transmitted data required a particular technological environment, merely generally linking the use of the judicial exception to a particular technological environment or field of use does not impose any meaningful limits on practicing the abstract idea).
Step 2B: The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
The claim recites the additional elements:
“A non-transitory computer-readable storage medium for tangibly storing computer program instructions capable of being executed by a computer processor, the computer program instructions defining steps of” (mere instructions to apply the exception using generic computer components does not provide an inventive concept);
“transmitting . . .” (transmitting data is well‐understood, routine, and conventional, see Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362; see also buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014); therefore the limitation, which is recited with a high level of generality, remains insignificant extra-solution activity even upon reconsideration); and
“continuously . . . based on observed . . . transmitted” (merely generally linking the use of the judicial exception to a particular technological environment or field of use does not provide an inventive concept).
For the reasons above, Claim 9 is rejected as being directed to an abstract idea without significantly more. This rejection applies equally to dependent claims 10-16. The additional limitations of the dependent claims are addressed below.
Regarding Claim 10, the claim recites limitations that are all substantially the same as limitations of Claim 2, in the form of a machine. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 10 is rejected under the same rationale.
Regarding Claim 11, the claim recites limitations that are all substantially the same as limitations of Claim 3, in the form of a machine. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 11 is rejected under the same rationale.
Regarding Claim 12, the claim recites limitations that are all substantially the same as limitations of Claim 4, in the form of a machine. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 12 is rejected under the same rationale.
Regarding Claim 13, the claim recites limitations that are all substantially the same as limitations of Claim 5, in the form of a machine. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 13 is rejected under the same rationale.
Regarding Claim 14, the claim recites limitations that are all substantially the same as limitations of Claim 6, in the form of a machine. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 14 is rejected under the same rationale.
Regarding Claim 15, the claim recites limitations that are all substantially the same as limitations of Claim 7, in the form of a machine. The claim is also directed to performing mental processes and/or mathematical concepts without integration into a practical component or significantly more.
Accordingly, Claim 15 is rejected under the same rationale.
Regarding Claim 16, the claim recites limitations that are all substantially the same as limitations of Claim 8, in the form of a machine. The claim is also directed to performing mental processes and/or mathematical concepts without integration into a practical component or significantly more.
Accordingly, Claim 16 is rejected under the same rationale.
Regarding Claim 17:
Step 1: Claim 17 is a machine claim. Therefore, claims 9-16 are directed to a statutory category of eligible subject matter.
Step 2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the "Mental Processes" grouping of abstract ideas. Here, the claim recites elements that are substantially the same as the limitations of Claim 1. As a result, and as elaborated above, these limitations are abstract ideas because they are mental processes.
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
The claim recites the additional elements:
“A device comprising: a processor configured to” (amounts to mere instructions to apply the judicial exception on generic and unspecialized computer components, which do not impose any meaningful limits on practicing the abstract idea);
“transmit . . .” (transmitting content items amounts to insignificant extra-solution activity because transmission of data is incidental to the claimed subject matter); and
“continuously . . . based on observed . . . transmitted” (in the event that the recitations of continuously updating information based on observed events relating to the transmitted data required a particular technological environment, merely generally linking the use of the judicial exception to a particular technological environment or field of use does not impose any meaningful limits on practicing the abstract idea).
Step 2B: The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
The claim recites the additional elements:
“A device comprising: a processor configured to” (mere instructions to apply the exception using generic computer components does not provide an inventive concept);
“transmit . . .” (transmitting data is well‐understood, routine, and conventional, see Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362; see also buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014); therefore the limitation, which is recited with a high level of generality, remains insignificant extra-solution activity even upon reconsideration); and
“continuously . . . based on observed . . . transmitted” (merely generally linking the use of the judicial exception to a particular technological environment or field of use does not provide an inventive concept).
For the reasons above, Claim 17 is rejected as being directed to an abstract idea without significantly more. This rejection applies equally to dependent claims 18-20. The additional limitations of the dependent claims are addressed below.
Regarding Claim 18, the claim recites limitations that are all substantially the same as limitations of Claim 3, which depends upon the additional elements of claim 2, in the form of a machine. The claim is also directed to performing mental processes without integration into a practical component or significantly more.
Accordingly, Claim 18 is rejected under the same rationale.
Regarding Claim 19, the claim recites limitations that are all substantially the same as limitations of Claim 6, which depends upon the additional elements of claim 5, in the form of a machine. The claim is also directed to performing mental processes and/or mathematical concepts without integration into a practical component or significantly more.
Accordingly, Claim 19 is rejected under the same rationale.
Regarding Claim 20, the claim recites limitations that are all substantially the same as limitations of Claim 8, in the form of a machine. The claim is also directed to performing mental processes and/or mathematical concepts without integration into a practical component or significantly more.
Accordingly, Claim 20 is rejected under the same rationale.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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 1-4, 8-12, 16-18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Duvvuri et al. (hereinafter Duvvuri) (Pat. Pub. No.: US 2021/0125221 A1) in view of Suzuki et al. (hereinafter Suzuki) (“A Logit Model of Brand Choice and Purchase Incidence: A Real Options Approach”) and Kaufman et al. (hereinafter Kaufman) (Pat. Pub. No. US 2011/0218850).
Regarding Claim 1, Duvvuri teaches a method comprising (Fig. 5-7; Para. [0009] – [0011], “FIG. 5 illustrates an example set of operations for determining a target group of users for engaging with a campaign . . . FIGS. 6A-B illustrate example sets of operations for determining an inclusion group of users and an exclusion group of users . . . FIG. 7 illustrates an example set of operations for using machine learning to generate models for generating and/or enhancing user profiles”, where the “set[s] of operations” collectively form a method, depicted across Fig. 5-7):
generating, for each of a plurality of users, probabilities of object interactions, each probability of object interaction corresponding to a respective object attribute and representing a likelihood that the respective user will interact with an object having the respective object attribute during a forecasting window (Para. [0024], “One or more embodiments include determining a target group for engaging with a campaign based on product-specific affinity attributes . . . For example, the target group of users may have a greater likelihood of interacting with the campaign content, or even making a purchase in response to the campaign content, than the other users. Hence, the campaign management application is able to take a product-specific approach that considers user affinities when determining a target group”, where the probabilities of object interactions, “likelihood of interacting with the campaign content, or even making a purchase” is generated for each of a plurality of “users”, using the “user affinities”, and corresponding to a respective object attribute, “based on product-specific affinity attributes”, and representing a likelihood that the respective user will interact with an object having the respective object attribute during a forecasting window, such as “a campaign” and/or its associated window of time, see Para. [0025], “Users who, for example, live nearby the store and have positive affinity towards craft supplies may still be likely to engage with the back-to-school campaign”),
wherein a given object recommendation ranking for a respective user comprises a ranked list of the object attributes ranked according to the corresponding probabilities of object interactions (Para. [0025], “One or more embodiments include determining a target group for engaging with a campaign based on utilizing different weights for affinity attributes, confirmable user attributes, and/or inferred user attributes. Users are associated with . . . the affinity attributes described above. Different user attributes have different levels of significance and/or correlations towards the effectiveness of campaign content for a particular user . . . Users who, for example, live nearby the store and have positive affinity towards craft supplies may still be likely to engage with the back-to-school campaign. A user interface of a campaign management application is configured to accept tuning parameters specifying weights applicable to various user attributes. The campaign management application generates an inclusion score function using the specified weights. The campaign management application inputs affinity attributes, confirmable user attributes, and/or inferred user attributes of a user into the inclusion score function, applies the specified weights to the user attributes, and thereby determines an inclusion score”, wherein the above mentioned user affinities, “Users are associated with . . . the affinity attributes described above”, are used to determine the above mentioned probabilities of object interactions, likelihood of “engaging with a campaign”, by generating a given object recommendation ranking, “inclusion score”, comprising a ranked list of object attributes ranked according to the corresponding probabilities of object interactions, “weight[ed]” list of “affinity attributes”, which as discussed above includes product-specific attributes; see also Para. [0035], “An inclusion group determinator 204 includes an inclusion score function for determining an inclusion score for each user”, where the “inclusion score” is determined “for each user”; see generally Abstract, “A campaign profile specifies products and/or content items associated with a campaign. A target group selection engine applies an affinity attribute model to user information of a user. The affinity attribute model is used to determine the user's affinity towards (a) product attributes of the products associated with the campaign and/or (b) content attributes of the content items associated with the campaign . . . [a] user interface accepts target user tuning parameters that specify weights to be applied to the affinity attributes determined by the affinity attribute model. Based at least on applying the weights to the affinity attributes, an inclusion score”);
. . . ;
calculating affinity group rankings based on the probabilities of object interactions and . . . [another metric] (Fig. 6A and Fig. 6B; Para. [0137] – [0117], “One or more embodiments include determining whether the inclusion score is above a threshold value (Operation 616) . . . If the inclusion score is above the threshold value, then the user is added into the inclusion group (Operation 618). If the inclusion score is not above the threshold value, then the user is excluded from the inclusion group (Operation 620 . . . one or more embodiments include determining whether the exclusion score is above a threshold value (Operation 636) . . . If the exclusion score is above the threshold value, then the user is added into the exclusion group (Operation 638). If the exclusion score is not above the threshold value, then the user is excluded from the exclusion group (Operation 640)”, where the “inclusion group[s]” and “exclusion group[s]” are calculated using the probabilities of object interactions, which, as discussed above, is represented by the “inclusion score[s]”, and another metric, “the exclusion score”; see also Para. [0138], “Referring to FIG. 6B, the operations of FIG. 6B are similar to the operations of FIG. 6A. The operations of FIG. 6B are applied with respect to exclusion ranges, exclusion weights, an exclusion score, and an exclusion group, whereas the operations of FIG. 6A are applied with respect to inclusion ranges, inclusion weights, an inclusion score, and an inclusion group”, where the “inclusion group[s]” and “exclusion group[s]” are collectively within the broadest reasonable interpretation of affinity group rankings because they provide a hierarchical rank based on associated “threshold[s]” and “ranges” of user affinities, see Fig. 9C; Para. [0219], “the exclusion groups 912-916 are superimposed on the inclusion groups 902-904 in a Venn diagram”; and Abstract, “Based at least on applying the weights to the affinity attributes, an inclusion score and/or exclusion score for the user is determined. The user is included in a target group, for engaging with the campaign, based on the inclusion score and/or exclusion score”);
grouping the plurality of users based on the affinity group rankings (Fig. 9C; Para. [0219], “Referring to FIG. 9C, the exclusion groups 912-916 are superimposed on the inclusion groups 902-904 in a Venn diagram. Target group 920 is represented by the dotted-pattern regions. Users who are in any of inclusion groups 902-904, without being in any of exclusion groups 912-916 are within target group 920”, where the plurality of “Users” are grouped “within [the] target group”, which is based on the affinity group rankings, “the inclusion groups” and “exclusion groups” collectively);
transmitting content items to the grouped users based on their respective affinity group rankings (Para. [0024], “The campaign management application may direct campaign content about the products of interest and/or sales campaign towards the target group of users, without sending the campaign content to other users”, where, as discussed above, “the target group of users” is determined based on their respective affinity group rankings, see Fig. 9C and Para. [0219], “Referring to FIG. 9C, the exclusion groups 912-916 are superimposed on the inclusion groups 902-904 in a Venn diagram. Target group 920 is represented by the dotted-pattern regions. Users who are in any of inclusion groups 902-904, without being in any of exclusion groups 912-916 are within target group 920”, where the plurality of “Users” are grouped “within [the] target group”, which is based on the affinity group rankings, “the inclusion groups” and “exclusion groups” collectively);
storing the affinity group rankings in an affinity group table comprising a data structure that maintains associations between user identifiers and corresponding affinity group rankings (Para. [0024], “The campaign management application may direct campaign content about the products of interest and/or sales campaign towards the target group of users, without sending the campaign content to other users”, where user identifiers must be stored, either temporarily or permanently in a data structure, in a manner that maintains its associations with the “Target group”, in order to “direct campaign content” to “the target group of users, without sending the campaign content to other users”; see also Fig. 9C and Para. [0219], “Referring to FIG. 9C, the exclusion groups 912-916 are superimposed on the inclusion groups 902-904 in a Venn diagram. Target group 920 is represented by the dotted-pattern regions. Users who are in any of inclusion groups 902-904, without being in any of exclusion groups 912-916 are within target group 920”, where, as discussed above, the “Target group” is determined using the affinity group rankings, “inclusion groups” and “exclusion groups”, therefore, maintenance of the associations between the user identifiers and the target group requires maintenance of the associations between the user identifiers and the affinity group rankings, either temporarily or permanently in a data structure, either of which is within the broadest reasonable interpretation of an affinity group table; see generally Para. [0040], “In one or more embodiments, a data repository 230 is any type of storage unit and/or device (e.g., a file system, database, collection of tables, or any other storage mechanism) for storing data”);
. . . ;
continuously updating the affinity group table based on [user input] . . . wherein the updating modifies the stored affinity group rankings based on the [user input] . . . (Fig. 2; Para. [0058], “a user interface 200 accepts user input specifying target user tuning parameters 212. Target user tuning parameters 212 are parameters used to tune an inclusion score function for determining an inclusion group 222 and/or an exclusion score function for determining an exclusion group 224. As described above, an inclusion score function includes one or more inclusion ranges 214 and/or inclusion weights 216. An exclusion score function includes one or more exclusion ranges 218 and/or exclusion weights 220”, where the “user interface 200” and “user tuning parameters 212” allows the user to continuously update the affinity group rankings, “inclusion” and “exclusion” groups, by modifying the “tuning parameters” to adjust the “inclusion score[s]” and “exclusion score[s]”; which, as discussed above, must be stored in either a temporary or permanent data structure that is within the broadest reasonable interpretation of an affinity group table);
re-executing the generating and calculating steps . . . using the updated affinity group table . . . through iterative refinement of the affinity group rankings . . . (Fig. 8A-8B and Para. [0205] – [0207], “An administrator enters, via user interface 800, user attributes relevant for determining the inclusion group under confirmable user attributes 804, inferred user attributes 806, and/or affinity attributes 808 . . . The administrator may use a variety of input methods for entering the relevant user attributes . . . Additionally, the administrator enters, via user interface 800, (a) whether each relevant user attribute is to be evaluated using an inclusion range or an inclusion weight, and (b) a corresponding value. If an inclusion range is indicated, the corresponding value indicates a required range of values for the user attribute for including the user into the inclusion group. If an inclusion weight is indicated, the corresponding value indicates a weight to be applied to a value for the user attribute for determining inclusion of the user into the inclusion group”, where “enter[ing], via user interface 800” “input[s]” for “user attributes relevant for determining” requires the re-executing of the generating, see Para. [0024], “One or more embodiments include determining a target group for engaging with a campaign based on product-specific affinity attributes . . . For example, the target group of users may have a greater likelihood of interacting with the campaign content, or even making a purchase in response to the campaign content, than the other users. Hence, the campaign management application is able to take a product-specific approach that considers user affinities when determining a target group”; Fig. 2; Para. [0058], “a user interface 200 accepts user input specifying target user tuning parameters 212. Target user tuning parameters 212 are parameters used to tune an inclusion score function for determining an inclusion group 222 and/or an exclusion score function for determining an exclusion group 224. As described above, an inclusion score function includes one or more inclusion ranges 214 and/or inclusion weights 216. An exclusion score function includes one or more exclusion ranges 218 and/or exclusion weights 220”, where the use of “user interface 200” allows for iterative refinement through “user tuning parameters 212”, which requires re-exciting the calculating, based on the new “inclusion score[s]” and “exclusion score[s]”; where, as discussed above, the executing of the generating and calculating requires storage of user identifiers and affinity group rankings in either a temporary or permanent data structure, which is within the broadest reasonable interpretation of a updated affinity group table).
Duvvuri does not explicitly disclose . . . calculating per-user interaction probabilities for each user over a forecasting window, wherein each per-user interaction probability represents a likelihood that the respective user will make any interaction during the forecasting window . . . the per-user interaction probabilities for each user . . . detecting observed user interactions with the transmitted content items . . . the detected observed user interactions with the transmitted content items . . . the observed user interactions . . . for subsequent forecasting windows . . . to improve prediction accuracy of future user interactions . . . .
However, Suzuki teaches [generating probabilities of object interactions] (Pg. 4, Para. 4, “The probability of brand choice, conditional on purchase incidence is modeled by the Multinomial Logit Model as follows:
PNG
media_image1.png
98
562
media_image1.png
Greyscale
[including the] Deterministic component of consumer h’s utility for purchasing brand i at time τ”, where “Phτ(i)” is within the broadest reasonable interpretation of a probability of object interaction because it represents a per-user affinity for an object, “consumer h’s utility for purchasing brand i”)
. . . calculating per-user interaction probabilities for each user over a forecasting window, wherein each per-user interaction probability represents a likelihood that the respective user will make any interaction during the forecasting window (Pg. 5, Para. 2, “The probability of purchase incidence is modeled by Binomial Logit Model as follows:
PNG
media_image2.png
102
618
media_image2.png
Greyscale
W˜h1,τ is a utility including random component of consumer h’s purchase incidence at time τ , and W˜h0,τ is a no-purchase of it . . . given by log of the denominator of the brand choice probability”, where the per-user interaction probabilities, “Pht(inc)”, are calculated for each user over a forecasting window, “consumer h’s purchase incidence at time τ”, which represents a likelihood, “purchase incidence”, that the respective user, “consumer h”, will make any interaction, binary model of an interaction “purchase” or no interaction “no-purchase”, during the forecasting window, “time τ”)
[calculating a value based on the probabilities of object interactions and] . . . the per-user interaction probabilities for each user (Pg. 4, Para. 3, “The probability of consumer h purchases brand i at time τ is modeled as follows:
PNG
media_image3.png
66
566
media_image3.png
Greyscale
First of the right side is the probability of purchase incidence, and second of the right side is the probability of brand choice, conditional on purchase incidence”, where a value, “Phτ(i)”, is calculated based on the probabilities of object interactions, “Phτ(i|inc)”, and the per-user interaction probabilities, “Pht(inc)”).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the calculating affinity group rankings based on the probabilities of object interactions and another metric of Duvvuri with the generating per-user probabilities of object interactions, calculating per-user interaction probabilities for each user over a forecasting window, and calculating a value based on the probabilities of object interactions and the per-user interaction probabilities of Suzuki in order to increase the effectiveness of advertising campaigns by generating affinity group rankings that represent the probability a customer will purchase an object within a forecasting window (compare Duvvuri, Abstract, “A campaign profile specifies products and/or content items associated with a campaign. A target group selection engine applies an affinity attribute model to user information of a user. The affinity attribute model is used to determine the user's affinity towards (a) product attributes of the products associated with the campaign and/or (b) content attributes of the content items associated with the campaign”, where only a “user's affinity” is considered for advertising campaigns, with Suzuki, Pg. 1-2, Para. 2-1, “Therefore, revealing the consumers’ purchase behavior, when and what they purchase, is very effective to retailers conducting price promotions, on fixing the products and the discounted price”, where considering both a user’s affinity “what they purchase” and a user’s interaction probability over a forecasting window “when . . . they purchase” “is very effective to retailers conducting . . . promotions”), which maximizes utility through consideration of per-user object affinities and purchasing propensities (Suzuki, Pg. 3, Para. 4, “We model the probabilities of brand choice and purchase incidence as a Nested Logit Model, in the model the probability of purchase incidence is a function of the expected maximum utility of the brand choice outcome”) using a simple method with improved representations of purchasing behavior (Suzuki, Pg. 3, Para. 3, “we develop a consumers’ purchase incidence and brand choice model by Nested Logit Model considering postpone option by using real options approach. By using real options approach, we can represent the value to postpone purchase, and develop more elaborated consumers’ purchase behavior model than previous studies. In addition, our model’s method to estimate the parameters is simple, and easy to treat”).
Additionally, Kaufman teaches . . . [a method, comprising storing the affinity group rankings in an affinity group table comprising a data structure that maintains associations between user identifiers and corresponding affinity group rankings] (Para. [0058] – [0064], “The goal of segmenting viewers is to create segments of viewers, based on their scientific attributes, which can be used for one or more subsequent methods. The method of segmenting viewers is preferably performed on a computer and includes . . . segmenting the viewers into one or more groups via one or more computer based algorithmic processes, which use the intermediary or raw scientific data . . . [and] storing the resulting groupings in a database for later retrieval”, where the affinity group rankings, “groupings” based on “viewer” “scientific attributes”, are stored in a table, “groupings in a database”, which must be a data structure that maintains associations between user identifiers and the corresponding affinity group rankings in order for “the resulting groupings” of the “segments of viewers” to be “stor[ed] . . . for later retrieval”);
detecting observed user interactions with the transmitted content items (Para. [0072] – [0074], “The method of optimizing an advertisement or piece of content is a computer aided process that preferably includes: 1. Selecting and displaying a specific advertisement or piece of content to a multitude of viewers. 2. Recording the effectiveness or performance and display data for each specific display of the specific advertisement or piece of content to each specific viewer”, where the method includes detection and observation, “Recording the effectiveness or performance and display data”, of the transmitted content items, “Selecting and displaying a specific advertisement or piece of content”; see also Para. [0082], “Performance and display data in Step 2 may include: . . . whether or not the specific viewer clicked on or otherwise interacted with the advertisement or content”, where the detection and observation is of user interactions, “clicked on or otherwise interacted”);
[continuously updating data based on] . . . the detected observed user interactions with the transmitted content items [, wherein the updating modifies data based on] . . . the observed user interactions . . . (Para. [0014] – [0015], “retrieving said effectiveness data of said specific advertisement or piece of content . . . displaying said specific advertisement or piece of content to said specific viewer . . . observing an effectiveness of said specific advertisement or piece of content on said specific viewer . . . updating on said computer said effectiveness data of said specific advertisement or piece of content”, where updates are performed to modify data, “updating . . . said effectiveness data”, based on the detected observed user interactions with the transmitted content, “observing an effectiveness of said specific advertisement or piece of content on said specific viewer”)
[and re-executing algorithmic steps] for subsequent forecasting windows . . . to improve prediction accuracy of future user interactions . . . (Para. [0014] – [0015], “updating on said computer said effectiveness data of said specific advertisement or piece of content for said specific scientific segment in which said specific viewer is grouped with said observed effectiveness data, such that said determining or predicting of whether said specific advertisement or piece of content will be effective with respect to said specific viewer is improved . . . updating one or more algorithms for determining with said effectiveness data recorded”, where the updates, “updating . . . said effectiveness data”, results in “updating one or more algorithms for determining with said effectiveness” in order to improve prediction accuracy for future interactions, “such that said determining or predicting of whether said specific advertisement or piece of content will be effective with respect to said specific viewer is improved”; see also Para. [0098], “The recursive feedback loop . . . continuously incorporate[s] new performance data in order improve the overall optimization process . . . . Because the basis of the Optimization of Advertisement or Piece of Content to Display sub-method is preferably empirical observation of performance across scientific segments, the prediction process may also be improved via further empirical observation of the results of the prediction process”, where “the recursive feedback loop” that “continuously incorporate[s] new performance data” from “empirical observation” requires the re-executing for subsequent forecasting windows).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the required affinity group table for storing the affinity group rankings, wherein a necessarily required data structure maintains associations between user identifiers and corresponding affinity group rankings of Duvvuri in view of Suzuki with the explicitly described affinity group table for storing the affinity group rankings, where an explicit database data structure maintains associations between user identifiers and corresponding affinity group rankings of Kaufman in order to store information in a persistent database for future use (Kaufman, Para. [0058], “storing the resulting groupings in a database for later retrieval”; Kaufman, Para. [0069], “The purpose of the algorithmic process in step 4 is to place viewers in meaningful scientific segments which can then later be used for targeting and creation of advertisements and content”; and Kaufman, Para. [0072], “Storing the above optimized data for later retrieval”), which allows for reduction of repetitive operations and optimization of processes (Kaufman, Para. [0020], “it can be observed or determined which specific advertisements or pieces of content performs best with which scientific segments, and then advertisements and content can be shown solely or disproportionally more often to the one or more top-performing scientific segments of viewers. This in turn reduces wasted displays of a particular advertisement”; Kaufman, Para. [0098], “The recursive feedback loop . . . to continuously incorporate new performance data in order improve the overall optimization process”).
Additionally, before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the transmission of content items to grouped users based on their affinity group rankings, the continuously updating the affinity group table based on modifications through user input, and the re-executing of the generating and calculating steps using the updated affinity group table and through iterative refinement of the affinity group rankings of Duvvuri in view of Suzuki and Kaufman with the detection of observed user interactions with transmitted content items, the continuously updating data based on the detected observed user interactions, and the re-executing of algorithmic steps of the method for subsequent forecasting windows to improve prediction accuracy of future user interactions in further view of Kaufman in order to allow users to utilize interaction feedback for iterative fine tuning of affinity group rankings, which will increase the personalization of advertising campaign content (Kaufman, Para. [0003], “Advertisers and media companies have struggled to adapt and many online advertisements are not reaching the desired viewer demographic, as mass content being created by traditional media companies is no longer connecting well with the interests of viewers seeking to consume more personalized and relevant content in a wide variety of online forms”), increase the monetary outputs for advertising campaigns (Kaufman, Para. [0007], “A second method for pricing online advertisements is called Cost Per Click ("CPC"). In the CPC method, the advertiser only pays when and if the viewer engages with the advertisement by clicking on the advertisement and linking to the advertisers' desired redirected location. The CPC method is usually priced per individual click. As such, "$0.50 CPC" advertising campaign or agreement would cost the advertiser 50 cents for each and every click that the advertisement receives. With the CPC method, no payment is made for mere passive viewing of the advertisement” and Kaufman, Para. [0008], “Because an enormous amount money is spent in the advertising and content creation and distribution space, small optimizations have the power of scale to generate and/or conserve large sums of money”), and result in increased predictive accuracy for subsequent forecasting windows (Kaufman, Para. [0098], “The recursive feedback loop . . . continuously incorporate[s] new performance data in order improve the overall optimization process . . . . Because the basis of the Optimization of Advertisement or Piece of Content to Display sub-method is preferably empirical observation of performance across scientific segments, the prediction process may also be improved via further empirical observation of the results of the prediction process”).
Regarding Claim 2, Duvvuri in view of Suzuki and Kaufman teach the method of claim 1, wherein generating the probability of object interactions for the respective user comprises classifying the respective user using a classification model (Duvvuri, Para. [0073] – [0075], “A machine learning algorithm 434 may include supervised components and/or unsupervised components. Various types of algorithms may be used, such as . . . classification . . . and random forest . . . the inferred user attribute model 332 and the affinity attribute model 432 include different data models for determining inferred user attributes and affinity attributes, respectively . . . [a] machine learning algorithm is used to train the inferred user attribute model 332 and the affinity attribute model 432”, where either the “inferred attribute model 332” or “the affinity attribute model 432” can be a “classification” model, which are used to classify respective users, see Duvvuri, Para. [0070], “A training set for training an affinity attribute model 432 includes historical user information of various users collected from one or more data sources. The historical user information may also include inferred user attributes. Each set of historical user information, associated with a respective user, is labeled with one or more affinity attributes of the user. The labeled affinity attributes may be determined via user input and/or another application”, where the “affinity attribute model 432” is trained to “label” “the user” with “infinity attributes”, including attributes from the inferred user attribute model, “inferred user attributes”; see also Duvvuri, Abstract, “A target group selection engine applies an affinity attribute model to user information of a user. The affinity attribute model is used to determine the user's affinity . . . Based at least on . . . the affinity attributes” and Duvvuri, Para. [0024], “determining a target group for engaging with a campaign based on product-specific affinity attributes. An affinity attribute model is configured to determine an affinity of a user towards a product attribute . . . the target group of users may have a greater likelihood of interacting with the campaign content, or even making a purchase in response to the campaign content, than the other users. Hence, the campaign management application . . . considers user affinities when determining a target group”, where generating the probability of object interactions for users, “likelihood of interacting with the campaign content”, comprises use of the “affinity attributes” from the “affinity attribute model”).
Regarding Claim 3, Duvvuri in view of Suzuki and Kaufman teach the method of claim 2, wherein classifying the respective user using the classification model comprises classifying the respective user using a multinomial random forest classifier (Duvvuri, Para. [0073] – [0075], “A machine learning algorithm 434 may include supervised components and/or unsupervised components. Various types of algorithms may be used, such as . . . classification . . . and random forest . . . the inferred user attribute model 332 and the affinity attribute model 432 include different data models for determining inferred user attributes and affinity attributes, respectively . . . [a] machine learning algorithm is used to train the inferred user attribute model 332 and the affinity attribute model 432”, where either the “inferred attribute model 332” or “the affinity attribute model 432” can be a “random forest” “classification” model, which are used to classify respective users, see Duvvuri, Para. [0070], “A training set for training an affinity attribute model 432 includes historical user information of various users collected from one or more data sources. The historical user information may also include inferred user attributes. Each set of historical user information, associated with a respective user, is labeled with one or more affinity attributes of the user. The labeled affinity attributes may be determined via user input and/or another application”, where the “affinity attribute model 432” is trained to “label” “the user” with “infinity attributes”, including attributes from the inferred user attribute model, “inferred user attributes”, which, in view of Suzuki, is a multinomial model, see Suzuki, Pg. 4, Para. 3, “The probability of brand choice, conditional on purchase incidence is modeled by the Multinomial Logit Model”).
The reasons for obviousness were discussed in regard to the rejection of Claim 1 above and remain applicable here.
Regarding Claim 4, Duvvuri in view of Suzuki and Kaufman teach the method of claim 1, wherein each attribute in the ranked list of object attributes is associated with a corresponding score (Duvvuri, Para. [0035] – [0036], “an inclusion score function includes a set of inclusion ranges 214 respectively applied to a set of user attributes . . .of a user . . . an inclusion score function includes a set of inclusion weights 214 respectively applied to a set of user attributes . . . of a user. The inclusion score function computes a weighted sum of the user attributes. An example inclusion score function is as follows: Inclusion Score=w.sub.1a.sub.1+w.sub.2a.sub.2+w.sub.3a.sub.3+ . . . . wherein w.sub.x represent the inclusion weights 214, and a.sub.x represent the user attributes”, where each attribute in the ranked list of object attributes, “user attributes”, is associated with a “weight” that is multiplied by the strength or affinity value of each affinity attribute, see Duvvuri, Para. [0069], “An affinity attribute score 438 may be expressed within any type of data range. As an example, an affinity attribute score may be an integer within the range 0 to 100. A higher score may represent a higher confidence level in the determination of the affinity attribute”, to determine a corresponding Score, see Duvvuri, Fig. 6B, reference “632”),
the corresponding score used to sort the ranked list of object attributes (Duvvuri, Para. [0080]-[0081], “One or more embodiments include presenting, on a user interface, the product attributes, content attributes, and/or campaign attributes . . . One or more embodiments include obtaining, via the user interface, user input specifying target user tuning parameters . . . The target group selection system obtains user input specifying respective target user tuning parameters for confirmable user attributes, inferred user attributes, and/or affinity attributes”, where “tunning parameters” can be inputted in order to adjust the weights associated with each attribute, which altering the attributes impact on the inclusion score, see Duvvuri, Para. [0058], “a user interface 200 accepts user input specifying target user tuning parameters 212. Target user tuning parameters 212 are parameters used to tune an inclusion score function for determining an inclusion group 222 and/or an exclusion score function for determining an exclusion group 224”; see also Duvvuri, Para. [0083], “a campaign may promote products with striped patterns. An administrator may determine that users who favor striped patterns should be included in a target group. The administrator may hence specify an inclusion weight applicable to a positive affinity attribute towards striped patterns. The administrator may determine that users who disfavor striped patterns should be excluded from the target group. The administrator may hence specify an exclusion weight applicable to a negative affinity attribute towards striped patterns. Hence, a target group selection system obtains target user tuning parameters for product-specific affinity attributes, content-specific affinity attributes, and/or campaign-specific affinity attributes via a user interface”, where adjusting the weights so that the corresponding scores may be used to sort the ranked list of object attributes by significance, which, in its most basic form, the weights could be adjusted so that the corresponding scores were either zero or not zero, with the scores then used to sort the object attributes into categories of relevant or not relevant. Also, weights may be adjusted so that scores can be used to sort attributes based on whether, or the degree with which, it is dispositive toward user inclusion or exclusion within the target group, see Duvvuri, Para. [0086] – [0087], “An administrator may determine that users who favor the “patterned” style should be included in a target group. The administrator may hence specify an inclusion weight applicable to a positive affinity attribute towards “patterned.” The administrator may determine that users who disfavor the “striped pattern” should be excluded from the target group. The administrator may hence specify an exclusion weight applicable to a negative affinity attribute towards “striped pattern””).
Regarding Claim 8, Duvvuri in view of Suzuki and Kaufman teach the method of claim 1, wherein calculating affinity group rankings (Duvvuri, Fig. 6A-6B; Duvvuri, Para. [0138], “Referring to FIG. 6B, the operations of FIG. 6B are similar to the operations of FIG. 6A. The operations of FIG. 6B are applied with respect to exclusion ranges, exclusion weights, an exclusion score, and an exclusion group, whereas the operations of FIG. 6A are applied with respect to inclusion ranges, inclusion weights, an inclusion score, and an inclusion group”, where, as discussed above, the “inclusion group[s]” and “exclusion group[s]” are collectively within the broadest reasonable interpretation of affinity group rankings because they provide a hierarchical rank based on associated “threshold[s]” and “ranges” of user affinities, see Duvvuri, Fig. 9C; Duvvuri, Para. [0219], “the exclusion groups 912-916 are superimposed on the inclusion groups 902-904 in a Venn diagram”; and Duvvuri, Abstract, “Based at least on applying the weights to the affinity attributes, an inclusion score and/or exclusion score for the user is determined. The user is included in a target group, for engaging with the campaign, based on the inclusion score and/or exclusion score”)
comprises multiplying the probability of object interactions by the per-user interaction probabilities for each user to obtain a likelihood for each user (Suzuki, Pg. 4, Para. 3, “The probability of consumer h purchases brand i at time τ is modeled as follows:
PNG
media_image3.png
66
566
media_image3.png
Greyscale
First of the right side is the probability of purchase incidence, and second of the right side is the probability of brand choice, conditional on purchase incidence”, where, in view of Suzuki, the affinity group rankings are a likelihood for each user, “Phτ(i)”, which is calculated by multiplying the probabilities of object interactions, “Phτ(i|inc)”, and the per-user interaction probabilities for each user, “Pht(inc)”).
The reasons for obviousness were discussed in regard to the rejection of Claim 1 above and remain applicable here.
Regarding Claim 9, Duvvuri teaches a non-transitory computer-readable storage medium for tangibly storing computer program instructions capable of being executed by a computer processor, the computer program instructions defining steps of (Para. [0234], “a non-transitory computer readable storage medium comprises instructions which, when executed by one or more hardware processors, causes performance of any of the operations described herein”, where the “instructions” are computer program instructions, see Fig. 10 and Para. [0225] –[0228], “Computer system 1000 may implement the techniques described herein program logic which in combination with the computer system causes or programs computer system 1000 to be a special-purpose machine . . . The term “storage media” as used herein refers to any non-transitory media that store data and/or instructions that cause a machine to operate in a specific fashion . . .such as storage device 1010 . . . Various forms of media may be involved in carrying one or more sequences of one or more instructions to processor 1004 for execution”):
. . . wherein a given object for a respective user comprises a ranked list of the object attributes ranked according to the corresponding probabilities of object interactions . . . (Para. [0025], “One or more embodiments include determining a target group for engaging with a campaign based on utilizing different weights for affinity attributes, confirmable user attributes, and/or inferred user attributes. Users are associated with . . . the affinity attributes described above. Different user attributes have different levels of significance and/or correlations towards the effectiveness of campaign content for a particular user . . . Users who, for example, live nearby the store and have positive affinity towards craft supplies may still be likely to engage with the back-to-school campaign. A user interface of a campaign management application is configured to accept tuning parameters specifying weights applicable to various user attributes. The campaign management application generates an inclusion score function using the specified weights. The campaign management application inputs affinity attributes, confirmable user attributes, and/or inferred user attributes of a user into the inclusion score function, applies the specified weights to the user attributes, and thereby determines an inclusion score”, wherein the above mentioned user affinities, “Users are associated with . . . the affinity attributes described above”, are used to determine the above mentioned probabilities of object interactions, likelihood of “engaging with a campaign”, by generating a given object recommendation ranking, “inclusion score”, comprising a ranked list of object attributes ranked according to the corresponding probabilities of object interactions, “weight[ed]” list of “affinity attributes”, which as discussed above includes product-specific attributes, which a given object for a respective user comprises; see also Para. [0035], “An inclusion group determinator 204 includes an inclusion score function for determining an inclusion score for each user”, where the “inclusion score” is determined “for each user”; see generally Abstract, “A campaign profile specifies products and/or content items associated with a campaign. A target group selection engine applies an affinity attribute model to user information of a user. The affinity attribute model is used to determine the user's affinity towards (a) product attributes of the products associated with the campaign and/or (b) content attributes of the content items associated with the campaign . . . [a] user interface accepts target user tuning parameters that specify weights to be applied to the affinity attributes determined by the affinity attribute model. Based at least on applying the weights to the affinity attributes, an inclusion score”).
The remaining limitations are substantially the same as limitations of Claim 1, therefore it is rejected under the same rationale.
Regarding Claim 10, the additional elements of the dependent claim are substantially the same as limitations of Claim 2, therefore it is rejected under the same rationale.
Regarding Claim 11, the additional elements of the dependent claim are substantially the same as limitations of Claim 3, therefore it is rejected under the same rationale.
Regarding Claim 12, the additional elements of the dependent claim are substantially the same as limitations of Claim 4, therefore it is rejected under the same rationale.
Regarding Claim 16, the additional elements of the dependent claim are substantially the same as limitations of Claim 8, therefore it is rejected under the same rationale.
Regarding Claim 17, Duvvuri teaches a device comprising: a processor configured to . . . (Fig. 10; Para. [0225], “According to one embodiment, the techniques herein are performed by computer system 1000 in response to processor 1004 executing one or more sequences of one or more instructions contained in main memory 1006”, where the “computer system 1000” is a device, which comprises a “processor 1004”).
The remaining limitations are substantially the same as limitations of Claim 1, therefore it is rejected under the same rationale.
Regarding Claim 18, Duvvuri in view of Suzuki and Kaufman teach the device of claim 17, wherein generating the probabilities of object interactions for the respective user comprises (Duvvuri, Abstract, “A target group selection engine applies an affinity attribute model to user information of a user. The affinity attribute model is used to determine the user's affinity . . . Based at least on . . . the affinity attributes” and Duvvuri, Para. [0024], “determining a target group for engaging with a campaign based on product-specific affinity attributes. An affinity attribute model is configured to determine an affinity of a user towards a product attribute . . . the target group of users may have a greater likelihood of interacting with the campaign content, or even making a purchase in response to the campaign content, than the other users. Hence, the campaign management application . . . considers user affinities when determining a target group”, where generating the probability of object interactions for users, “likelihood of interacting with the campaign content”, comprises use of the “affinity attributes” from the “affinity attribute model”)
classifying the respective user using a multinomial random forest classifier (Duvvuri, Para. [0073] – [0075], “A machine learning algorithm 434 may include supervised components and/or unsupervised components. Various types of algorithms may be used, such as . . . classification . . . and random forest . . . the inferred user attribute model 332 and the affinity attribute model 432 include different data models for determining inferred user attributes and affinity attributes, respectively . . . [a] machine learning algorithm is used to train the inferred user attribute model 332 and the affinity attribute model 432”, where either the “inferred attribute model 332” or “the affinity attribute model 432” can be a “random forest” “classification” model, which are used to classify respective users, see Duvvuri, Para. [0070], “A training set for training an affinity attribute model 432 includes historical user information of various users collected from one or more data sources. The historical user information may also include inferred user attributes. Each set of historical user information, associated with a respective user, is labeled with one or more affinity attributes of the user. The labeled affinity attributes may be determined via user input and/or another application”, where the “affinity attribute model 432” is trained to “label” “the user” with “infinity attributes”, including attributes from the inferred user attribute model, “inferred user attributes”, which, in view of Suzuki, is a multinomial model, see Suzuki, Pg. 4, Para. 3, “The probability of brand choice, conditional on purchase incidence is modeled by the Multinomial Logit Model”).
The reasons for obviousness were discussed in regard to the rejection of Claim 1 above and remain applicable here.
Regarding Claim 20, the additional elements of the dependent claim are substantially the same as limitations of Claim 8, therefore it is rejected under the same rationale.
Claims 5-6, 13-14, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Duvvuri in view of Suzuki, Kaufman, and Castéran et al. (hereinafter Castéran) (“Modeling Customer Lifetime Value, Retention, and Churn”).
Regarding Claim 5, Duvvuri in view of Suzuki and Kaufman teach the method of claim 1, wherein calculating the affinity group rankings comprises computing a predicted . . . (Duvvuri, Fig. 6A-6B; Duvvuri, Para. [0138], “Referring to FIG. 6B, the operations of FIG. 6B are similar to the operations of FIG. 6A. The operations of FIG. 6B are applied with respect to exclusion ranges, exclusion weights, an exclusion score, and an exclusion group, whereas the operations of FIG. 6A are applied with respect to inclusion ranges, inclusion weights, an inclusion score, and an inclusion group”, where, as discussed above, the “inclusion group[s]” and “exclusion group[s]” are collectively within the broadest reasonable interpretation of affinity group rankings because they provide a hierarchical rank based on associated “threshold[s]” and “ranges” of user affinities, see Duvvuri, Fig. 9C; Duvvuri, Para. [0219], “the exclusion groups 912-916 are superimposed on the inclusion groups 902-904 in a Venn diagram”; and Duvvuri, Abstract, “Based at least on applying the weights to the affinity attributes, an inclusion score and/or exclusion score for the user is determined. The user is included in a target group, for engaging with the campaign, based on the inclusion score and/or exclusion score”, which, in view of Suzuki, comprises computed predicted “probability” values, see Suzuki, Pg. 4, Para. 3, “The probability of consumer h purchases brand i at time τ is modeled as follows:
PNG
media_image3.png
66
566
media_image3.png
Greyscale
First of the right side is the probability of purchase incidence, and second of the right side is the probability of brand choice, conditional on purchase incidence”).
The reasons for obviousness were discussed in regard to the rejection of Claim 1 above and remain applicable here.
Duvvuri in view of Suzuki and Kaufman do not explicitly disclose . . . number of interactions using a lifetime value model.
However, Castéran teaches . . . [calculating customer purchase behavior, comprising predicting a] number of interactions using a lifetime value model (Pg. 1, Para. 1, “Customer lifetime value (CLV) allows assessing their current and future value in a customer base . . . to predict . . . the purchase behavior of their customers” and Pg. 3, Para. 1, “CLV analyses involve distinguishing active customers from defectors and then predicting their lifetime and future levels of transactions according to their observed past purchase behavior”, where a number of interactions are predicted, “predicting their lifetime and future levels of transactions”, using a lifetime value model, “Customer lifetime value (CLV)”, in order to calculate “the purchase behavior of their customers”; for additional details see Pg. 12, Para. 6 and Pg. 17, Para. 3).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the calculation of affinity group rankings comprising computing predicted probabilities of Duvvuri in view of Suzuki and Kaufman with the calculating customer purchasing behavior comprising predicting a number of interactions using a lifetime value model of Castéran in order to maximize return on investments and guide budget allocations for advertising campaigns (Castéran, Pg. 2, Para. 2, “CLV is an important concept on which the customer relationship management strategy; marketing resource allocation (to profitable customers), such as promotions; and the assessment of the marketing efficiency are based on . . . Thus, CLV models offer a powerful means to maximize the return on marketing investments and guide allocations of the marketing budget”), with improved purchase prediction validity in instances with high buying rates (Castéran, Pg. 25, Para. 3, “
If one considers the mean absolute percent error (MAPE) as an empirical criterion, the explanatory Pareto/NBD model has slightly worse results than either the standard or the BG/NBD model (15.5% vs. 12% and 10.5%; the basic NBD achieves the worst results at 38.1%). This result makes sense. According to Fader et al. (2005a), the BG/NBD forecasts are better when purchase frequency is high, as in the grocery retailing context, because of the differences among the model structures. Under the Pareto/NBD model, dropout occurs at any time – even before a customer has made a first purchase. However, under the BG/NBD, a customer cannot become inactive before making his or her first purchase. If buying rates are fairly high, BG/NBD and Pareto/NBD perform similarly well”).
Regarding Claim 6, Duvvuri in view of Suzuki, Kaufman, and Castéran teach the method of claim 5, wherein computing the predicted number of interactions using the lifetime value model comprises computing a predicted number of interactions (Castéran, Pg. 1, Para. 1, “Customer lifetime value (CLV) allows assessing their current and future value in a customer base . . . to predict . . . the purchase behavior of their customers” and Castéran, Pg. 3, Para. 1, “CLV analyses involve distinguishing active customers from defectors and then predicting their lifetime and future levels of transactions according to their observed past purchase behavior”, where a number of interactions are predicted, “predicting their lifetime and future levels of transactions”, using a lifetime value model, “Customer lifetime value (CLV)”, in order to calculate “the purchase behavior of their customers”; for additional details see Castéran, Pg. 12, Para. 6 and Castéran, Pg. 17, Para. 3)
using a beta-geometric model (Castéran, Pg. 1, Para. 1, “Customer lifetime value (CLV) allows assessing their current and future value in a customer base . . . to predict . . . the purchase behavior of their customers . . . an empirical application of the stochastic “standard” Pareto/NBD, and the BG/NBD models, as well as an explanatory Pareto/NBD model with covariates to grocery retailing store loyalty program scanner data, is done”, where “BG” is “beta-geometric”, see Castéran, Pg. 12, Para. 6, “beta-geometric/NBD model (BG/NBD by Fader et al. 2005a)”).
The reasons for obviousness were discussed in regard to the rejection of Claim 5 above and remain applicable here.
Regarding Claim 13, the additional elements of the dependent claim are substantially the same as limitations of Claim 5, therefore it is rejected under the same rationale.
Regarding Claim 14, the additional elements of the dependent claim are substantially the same as limitations of Claim 6, therefore it is rejected under the same rationale.
Regarding Claim 19, Duvvuri in view of Suzuki, Kaufman, and Castéran teach the device of claim 17, wherein calculating the affinity group rankings comprises (Duvvuri, Fig. 6A-6B; Duvvuri, Para. [0138], “Referring to FIG. 6B, the operations of FIG. 6B are similar to the operations of FIG. 6A. The operations of FIG. 6B are applied with respect to exclusion ranges, exclusion weights, an exclusion score, and an exclusion group, whereas the operations of FIG. 6A are applied with respect to inclusion ranges, inclusion weights, an inclusion score, and an inclusion group”, where, as discussed above, the “inclusion group[s]” and “exclusion group[s]” are collectively within the broadest reasonable interpretation of affinity group rankings because they provide a hierarchical rank based on associated “threshold[s]” and “ranges” of user affinities, see Duvvuri, Fig. 9C; Duvvuri, Para. [0219], “the exclusion groups 912-916 are superimposed on the inclusion groups 902-904 in a Venn diagram”; and Duvvuri, Abstract, “Based at least on applying the weights to the affinity attributes, an inclusion score and/or exclusion score for the user is determined. The user is included in a target group, for engaging with the campaign, based on the inclusion score and/or exclusion score”, which, in view of Suzuki, comprises computed predicted “probability” values, see Suzuki, Pg. 4, Para. 3, “The probability of consumer h purchases brand i at time τ is modeled as follows:
PNG
media_image3.png
66
566
media_image3.png
Greyscale
First of the right side is the probability of purchase incidence, and second of the right side is the probability of brand choice, conditional on purchase incidence”)
computing a predicted number of interactions (Castéran, Pg. 1, Para. 1, “Customer lifetime value (CLV) allows assessing their current and future value in a customer base . . . to predict . . . the purchase behavior of their customers” and Castéran, Pg. 3, Para. 1, “CLV analyses involve distinguishing active customers from defectors and then predicting their lifetime and future levels of transactions according to their observed past purchase behavior”, where a number of interactions are predicted, “predicting their lifetime and future levels of transactions”, using a lifetime value model, “Customer lifetime value (CLV)”, in order to calculate “the purchase behavior of their customers”; for additional details see Castéran, Pg. 12, Para. 6 and Castéran, Pg. 17, Para. 3)
using a beta-geometric model (Castéran, Pg. 1, Para. 1, “Customer lifetime value (CLV) allows assessing their current and future value in a customer base . . . to predict . . . the purchase behavior of their customers . . . an empirical application of the stochastic “standard” Pareto/NBD, and the BG/NBD models, as well as an explanatory Pareto/NBD model with covariates to grocery retailing store loyalty program scanner data, is done”, where “BG” is “beta-geometric”, see Castéran, Pg. 12, Para. 6, “beta-geometric/NBD model (BG/NBD by Fader et al. 2005a)”).
The reasons for obviousness were discussed in regard to the rejection of Claim 1 above, for the combination of Duvvuri with Suzuki, and in regard to the rejection of Claim 5 above, for the combination of Duvvuri in view of Suzuki and Kaufman with Castéran, and remain applicable here.
Claims 7 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Duvvuri in view of Suzuki, Kaufman, Castéran, and Tyagi (“Naïve Bayes Classification”).
Regarding Claim 7, Duvvuri in view of Suzuki, Kaufman, and Castéran teach the method of claim 6, wherein computing the predicted number of interactions comprises . . . an output of the beta-geometric model . . . to obtain the predicted number of interactions for the respective user (Castéran, Pg. 1, Para. 1, “Customer lifetime value (CLV) allows assessing their current and future value in a customer base . . . to predict . . . the purchase behavior of their customers . . . an empirical application of the stochastic “standard” Pareto/NBD, and the BG/NBD models, as well as an explanatory Pareto/NBD model with covariates to grocery retailing store loyalty program scanner data, is done”, where “BG” is “beta-geometric”, see Castéran, Pg. 12, Para. 6, “beta-geometric/NBD model (BG/NBD by Fader et al. 2005a)”; and Castéran, Pg. 3, Para. 1, “CLV analyses involve distinguishing active customers from defectors and then predicting their lifetime and future levels of transactions according to their observed past purchase behavior”, where a number of interactions are predicted, “predicting their lifetime and future levels of transactions”, using a lifetime value model, “Customer lifetime value (CLV)”, in order to calculate “the purchase behavior of their customers”; for additional details see Castéran, Pg. 12, Para. 6 and Castéran, Pg. 17, Para. 3).
The reasons for obviousness were discussed in regard to the rejection of Claim 5 above and remain applicable here.
Duvvuri in view of Suzuki, Kaufman, and Castéran do not explicitly disclose . . . dividing. . . by a total number of expected interactions . . . .
However, Tyagi teaches . . . dividing [an output of a number of interactions] . . . by a total number of expected interactions . . . (Pg. 5, Table 8.1 and Pg. 5, Para. 1-4, “the prior probability of each class, can be computed based on the training tuples: . . . P(C1)=P(buys_computer =yes) =9/14 =0.643 (since total 9 rows of yes) P(C2)=P(buys_computer =no) =5/14 =0.357”).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the computing of a predicted number of interactions for respective users, comprising an output of a beta-geometric model of Duvvuri in view of Suzuki, Kaufman, and Castéran with the dividing an output of a number of interactions by a total number of expected interactions of Tyagi in order to compute interaction probability values with reduced computational complexity (Tyagi, Pg. 4, No. 6, “For a given data set with many attributes it would be computationally expensive to calculate P(X|Ci) because we need to consider each attribute separately so we. “To reduce computation in evaluating P(X|Ci), the naive assumption of class-conditional independence is made. This presumes that the attributes’ values are conditionally independent of one another””).
Regarding Claim 15, the additional elements of the dependent claim are substantially the same as limitations of Claim 7, therefore it is rejected under the same rationale.
Response to Arguments
Applicant's arguments filed on April 28th, 2026 have been fully considered. Each argument is addressed in detail below.
I. Applicant argues the objections to the claims should be withdrawn (Applicant’s Remarks, 04/28/2026, Pg. 9, Section “1 Objections to the Claims”).
Applicant’s amendments have overcome each and every objection to the claims previously set forth in the January 28th, 2026 Office Action. As a result, the objections to the claims have been withdrawn.
II. Applicant argues the rejections of the claims under 35 U.S.C. 101 should be withdrawn (Applicant’s Remarks, 04/28/2026, Pg. 9-13, Section “2 Rejections under § 101”).
Specifically, Applicant asserts the claims are subject matter eligible because the abstract ideas are integrated into a practical application (Step 2A, Prong Two). In support of this assertion, Applicant makes several arguments.
1) First, Applicant argues that the previously communicated analysis “mischaracterizes the claims by dissecting each recited step in isolation, labeling it a mental process or mathematical concept, and then declaring the remaining additional elements to be either insignificant extra-solution activity or mere instructions to apply it on a generic computer”, whereas “when the claim is considered as a whole, the recited steps cooperate to define a particular closed-loop recommendation” (Applicant’s Remarks, Pg. 9, Para. 4). However, this argument mischaracterizes the basis for rejection of the claims under 35 U.S.C. 101, which considers the additional elements individually and as part of the claim as a whole. Specifically, as discussed in detail above in regard to the rejections of the claims under 35 U.S.C. 101 and as further discussed below in regard to Applicant’s specific arguments, the additional elements of the claim, even considered as a whole, are not sufficient to integrate the judicial exception into a practical application.
Therefore, the argument is not persuasive.
2) Second, Applicant argues the “the claims recite a specific technical architecture in which affinity group rankings are not merely calculated and discarded but are persistently stored in an affinity group table that maintains associations between user identifiers and corresponding affinity group rankings, content items are transmitted to grouped users based on those stored rankings, the system detects observed user interactions with the specific transmitted content items, the stored rankings in the table are modified based on those detected interactions, and the generating and calculating steps are then re-executed using the updated affinity group table for subsequent forecasting windows” which “is a particular way of achieving improved prediction accuracy through iterative, empirically driven refinement of a stored data structure; it is not the abstract idea of prediction itself . . . [such that] the amended claims here recite precisely such an improvement to the data-structure-based operation of a content recommendation system” (Applicant’s Remarks, Pg. 9-10, Para. 5-6 and Pg. 11-12, Para. 3-1). Additionally, Applicant argues the claims “further specify the manner in which the probabilities of object interactions and the object recommendation ranking are produced-namely, as per-object- attribute probabilities and as a ranking of those object attributes by their corresponding probabilities” which reflect the particular implementation disclosed in the specification (Applicant’s Remarks, Pg. 9-10, Para. 5-6 and Pg. 12-13, Para. 3-1). However, while the specification sets forth the alleged improvements to object recommendation and affinity group generation and modification, these improvements are asserted in a conclusory manner and without the detail necessary to be apparent to a person of ordinary skill in the art as to how the disclosure results in the asserted improvements (see MPEP 2106.04(d)(1)). Additionally, the claims recite limitations, such as persistently storing entity associations in a table, transmission of data and observation of its performance, and iterative calculation and grouping based on observed results, that have broad applicability across many fields of endeavor (see MPEP 2106.05(f)), such as to fail to reflect the alleged improvements (see MPEP 2106.04(d)(1)).
Therefore, the argument is not persuasive.
3) Third, Applicant argues “Examiner's contrary assertion that the "maintaining of associations between user identifiers and group rankings within a table structure" and the "re-executing of the generating and calculating steps using the updated table" can be performed in the mind with pen and paper does not withstand scrutiny” because “[t]hese steps operate on a persistent data structure that is repeatedly written to, read from, and modified across successive forecasting windows in response to empirically detected interactions with transmitted digital content. A pen-and-paper analogue would require a human to contemporaneously detect and record every user's interactions with specific transmitted content items across a population, to update stored rankings for every user in response, and to regenerate probabilities based on the updated rankings for every subsequent forecasting window. That is not a mental process; it is a description of a computerized feedback system that could not practically be performed by a human” (Applicant’s Remarks, Pg. 10-11, Para. 2-1). However, the claims as currently formulated merely recite some degree of re-executing of generation and calculation in response to some degree of observed behavior and with reference to a table with an unspecified degree of complexity. As a result, and as discussed in detail above, the claims recite observations, evaluations, judgments, and opinions that are precisely the kinds of limitations that can practically be performed in the mind (see MPEP 2106.04(a)(2)(III)(A)). Additionally, claims that require a computer may still recite a mental process (see MPEP 2106.04(a)(2)(III)(C)).
Therefore, the argument is not persuasive.
4) Fourth, Applicant argues the “transmitting content items” step is not “insignificant extra- solution activity” because “[t]he transmitting step is not mere output; it is the mechanism that causes the observed user interactions that in turn drive modification of the stored affinity group rankings used as input to the next iteration. Without the transmitting step, there would be no detected interactions with transmitted content items, and without those detected interactions, there would be no modification of the stored rankings and no iterative refinement. The transmitting step is therefore an integral part of the technical feedback loop, not data output that is tangential to the claim” (Applicant’s Remarks, Pg. 11, Para. 2). Similarly, Applicant argues the “detecting observed user interactions” step is not “mere data gathering” because “[t]he claim does not recite detection of arbitrary user activity; it recites detection of interactions with the specific content items that were transmitted based on the stored affinity group rankings, and the detected interactions are used to modify the stored rankings for use in subsequent iterations. The detection step is therefore structurally coupled to both the transmitting step that precedes it and the updating step that follows it, and it provides meaningful limits on the claim”, which applicant asserts is comparable to the subject matter at issue in Ex parte Desjardins, Appeal No. 2021-004615 (PTAB Mar. 16, 2023) and Ex parte Carmody, Appeal No. 2023- 002124 (PTAB Nov. 27, 2023) (Applicant’s Remarks, Pg. 11-12, Para. 3-1).
However, a showing that a limitation recites a necessary action, such as “data gathering or outputting”, does not preclude a finding that the limitation is insignificant extra-solution activity (see MPEP 2106.05(g)). Instead, as discussed in detail above, the steps of “transmitting content items” and “detecting observed user interactions” were found to be incidental to the claimed subject matter, such as to fail to sufficiently limit the judicial exceptions of observations, evaluations, judgments, and opinions (see MPEP 2106.04(a)(2)(III)(A)). Specifically, the transmission of data is comparable to presenting offers and gathering statistics or the insignificant application of an abstract idea and the detection of object interactions is mere data gathering, all of which are established as insignificant extra-solution activity (see MPEP 2106.05(g)).
Therefore, the argument is not persuasive.
5) Fifth, Applicant argues “The amended claims here do not merely instruct a computer to "predict user interests better." They specify the particular way that improved prediction accuracy is achieved: through a persistent affinity group table that stores user-identifier-to-ranking associations; through detection of interactions with specifically transmitted content items; through modification of the stored rankings based on those detected interactions; and through re-execution of the generating and calculating steps using the updated table as input to subsequent forecasting windows”, which conforms with patent eligible subject matter discussed in the August 4, 2025 Memorandum on subject matter eligibility (Applicant’s Remarks, Pg. 12, Para. 2). However, as discussed above, the claims do not recite a particular way of improvement, and instead recite limitations, such as persistently storing entity associations in a table, transmission of data and observation of its performance, and iterative calculation and grouping based on observed results, that have broad applicability across many fields of endeavor (see MPEP 2106.05(f)), such as to fail to reflect the alleged improvements (see MPEP 2106.04(d)(1)). Additionally, the claims recite generic computer components that amount to mere instructions to apply the judicial exceptions (see MPEP 2106.05(f)).
Therefore, the argument is not persuasive.
6) Sixth , Applicant argues the dependent claims “recite additional technical features (e.g., multinomial random forest classifier, beta-geometric lifetime value model) that further confirm their eligibility” (Applicant’s Remarks, Pg. 13, Para. 2). However, as discussed in detail above, each of the dependent claims are directed to an abstract idea without significantly more. Additionally, Applicant has not “specifically point[ed] out the supposed errors” with this interpretation (see 37 C.F.R. 1.111(b)).
Therefore, the argument is not persuasive.
III. Applicant argues the rejections of the claims under 35 U.S.C. 103 should be withdrawn (Applicant’s Remarks, 04/28/2026, Pg. 13-17, Section “3 Rejections under § 103”).
Specifically, Applicant asserts the prior art of record, alone and in combination, fails to describe or suggest all elements of the amended independent claims (Applicant’s Remarks, Pg. 13-14, Para. 4-1). In support of this assertion, Applicant makes several arguments.
1) First, Applicant argues “Duvvuri's inclusion and exclusion groups are binary membership buckets determined by threshold comparison; they are not rankings of users” because “The Office Action's characterization of Duvvuri's inclusion and exclusion groups as collectively providing "plural rankings" because they are "hierarchical" conflates threshold-based group membership with a ranking of users. A claim element that recites affinity group rankings . . . is not taught by a reference that simply divides users into two threshold-defined sets and then intersects them” (Applicant’s Remarks, Pg. 14, Para. 2). However, as discussed in detail above, Duvvuri’s inclusion groups and exclusion groups are collectively within the “broadest reasonable interpretation consistent with the specification” of affinity group rankings because they provide a hierarchical rank based on associated thresholds and ranges of user affinities (see MPEP 2111; see also Phillips v. AWH Corp., 415 F.3d 1303, 1316, 75 USPQ2d 1321, 1329 (Fed. Cir. 2005)). Additionally, the specification does not define ranking in a manner that precludes the ranking from being a binary ranking or from being based on thresholds (id.).
Therefore, the argument is not persuasive.
2) Second, Applicant argues Duvvuri fails to disclose “a ranking of users within or across groups that is persistently stored and then modified through observed feedback” such that each of “storing those rankings as associations between user identifiers and affinity group rankings in a table that is modified based on observed user interactions”, “modifying stored rankings based on observed interactions with transmitted content” and “continuously updating the affinity group table based on the detected observed user interactions with the transmitted content items” are disclosed (Applicant’s Remarks, Pg. 14-15, Para. 2-1). However, as discussed in detail above, Duvvuri is relied upon as part of a combination of references to teach these elements of the amended independent claims. Specifically, Duvvuri is relied upon to teach the continuous updating of the affinity group table that modifies the stored affinity group rankings. Whereas Kaufman is relied upon to teach the detection of observed user interactions with transmitted content items. As a result, arguments that Duvvuri fails to fully disclose these elements amounts to “attacking references individually where the rejections are based on combinations of references” (see MPEP 2145; see also In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981)).
Therefore, the argument is not persuasive.
3) Third, Applicant argues Suzuki fails to cure the alleged deficiencies discussed in 1) and 2) (Applicant’s Remarks, Pg. 15, Para. 2). However, as discussed in detail above, the arguments in support of the allegations of deficiency discussed in 1) and 2) were not persuasive. Additionally, Suzuki is not relied upon to teach these aspects of the independent claims.
Therefore, the argument is not persuasive.
4) Fourth, Applicant argues Kaufman “fails to supply what Duvvuri and Suzuki lack” because its “stored groupings are segment memberships-not rankings of users-and Kaufman's updating step modifies effectiveness data associated with advertisements, not stored affinity group rankings of users . . . it does not store, and it does not modify, associations between user identifiers and affinity group rankings. The Office Action's assertion that Kaufman's "recursive feedback loop" satisfies the claimed continuous updating of the affinity group table therefore conflates updating advertisement effectiveness data with modifying stored rankings of users, which is a distinct technical operation” (Applicant’s Remarks, Pg. 15, Para. 3). However, for the same reasons as discussed in regard to 2), this amounts to “attacking references individually where the rejections are based on combinations of references” (see MPEP 2145; see also In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981)).
Therefore, the argument is not persuasive.
5) Fifth, Applicant argues “[n]one of these references-individually or in combination describes a persistent affinity group table storing user-identifier-to-ranking associations that is modified based on detected interactions with transmitted content items and then used as input to regenerate probabilities in subsequent forecasting windows” because “Duvvuri contributes threshold-based inclusion/exclusion group membership with administrator-tuned weights. Suzuki contributes an econometric per-user purchase incidence probability. Kaufman contributes storage of segment membership and updating of advertisement effectiveness data based on observed viewer behavior” (Applicant’s Remarks, Pg. 16, Para. 1). However, as discussed above, the prior art of record collectively discloses elements that are within the “broadest reasonable interpretation consistent with the specification” of each and every element of the independent claims (see MPEP 2111; see also Phillips v. AWH Corp., 415 F.3d 1303, 1316, 75 USPQ2d 1321, 1329 (Fed. Cir. 2005)). Additionally, Applicant’s argument amounts to “general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references” (see 37 C.F.R. 1.111(b)).
Therefore, the argument is not persuasive.
6) Sixth, Applicant argues Applicant argues “[t]he rejection relies on hindsight reconstruction of the claimed combination” because “[t]he Office Action's motivation to combine focuses on the general goal of improving advertising campaign personalization and increasing monetary outputs, but generalized motivations at that level of abstraction do not supply the missing technical architecture; they at most suggest that skilled artisans were interested in better ad targeting, which does not equate to a suggestion to build the particular feedback loop recited in the claims” (Applicant’s Remark’s, Pg. 16, Para. 1). However, as discussed in detail above, the Applicant-cited improvements to “advertising campaign personalization and increasing monetary outputs” are tied to specific technical implementations disclosed in Suzuki and Kaufman. As a result, one of ordinary skill in the art would have been motivated to combine the teachings of Duvvuri with these specific technical implementations to arrive at the claimed invention to utilize these improvements (see MPEP 2143(I)(G)).
Therefore, the argument is not persuasive.
7) Seventh, Applicant argues “the cited combination does not describe the predicate updated affinity group table, and it follows that the cited combination cannot describe re-executing the generating and calculating steps using that updated table” (Applicant’s Remarks, Pg. 16, Para. 2). However, the arguments in support of the allegations of deficiency discussed in 1) through 6) were not persuasive.
Therefore, the argument is not persuasive.
8) Eighth, Applicant argues that neither Castéran nor Tyagi cure the asserted deficiencies with the rejections of the independent claims as rendered obvious over Duvvuri in view of Suzuki and Kaufman (Applicant’s Remarks, Pg. 16-17, Para. 3-1). However, the arguments in support of the allegations of deficiency discussed in 1) through 7) were not persuasive. Additionally, neither Castéran nor Tyagi are relied upon to teach the independent claims.
Therefore, the argument is not persuasive.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MATTHEW BRYCE GOLAN whose telephone number is (571)272-5159. The examiner can normally be reached Monday through Friday, 8:00 AM to 5:00 PM ET.
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, Alexey Shmatov can be reached at (571) 270-3428. 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.
/MATTHEW BRYCE GOLAN/Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights