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.
Claim 1-20 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
MPEP 2106 (III) sets out steps for evaluating whether a claim is drawn to patent-eligible subject matter. The analysis of claims 1-20, in accordance with these steps, follows.
Step 1 Analysis:
Claims 1-16 are directed to methods (processes). Claims 17-20 are directed to an apparatus (machine). Therefore, claims 1-20 fall into one of four statutory categories (i.e., process, machine, article of manufacture).
As to claim 1,
Step 2A Prong 1: this claim recites the following abstract ideas:
computing, ..., channel contribution data based on the individual-level user interaction data; (the limitation describes computing contribution data from user interaction data, which is a mental process implemented using a pen and paper.)
generating, ..., an individual channel contribution value for the digital content channel based on the channel contribution data and the aggregate attribution model. (the limitation describes calculating a contribution value based on data and a model, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
obtaining, by a multi-touch attribution model, individual-level user interaction data from a digital content channel; (this limitation describes data collection/receiving, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
using the multi-touch attribution model (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
training, using a training component, an aggregate attribution model based on the channel contribution data; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
using a calibration component (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 2,
Step 2A Prong 1: this claim recites the following abstract ideas:
wherein the individual channel contribution value indicates a contribution of the digital content channel to an event of the individual-level user interaction data. (the limitation describes characterizing the meaning of a computed value, which is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
Step 2A Prong 2 and 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B. Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2. The claim is ineligible.
As to claim 3,
Step 2A Prong 1: this claim recites the following abstract ideas:
computing a preliminary individual channel contribution value for the digital content channel ...; (the limitation describes computing a contribution value, which is a mental process implemented using a pen and paper.)
computing an aggregate channel contribution value for the digital content channel ..., wherein the individual channel contribution value is generated based on the preliminary individual channel contribution value and the aggregate channel contribution value. (the limitation describes computing a value and combining two computed values to generate another value, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using the multi-touch attribution model (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
using the aggregate attribution model (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 4,
Step 2A Prong 1: This claim does not recite an additional abstract idea, but the claim depends on claim 1,
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
wherein the aggregate attribution model is trained using the channel contribution data and experimental testing data. (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 5,
Step 2A Prong 1: this claim recites the following abstract ideas:
generating, ..., an attribution prior based on the channel contribution data; (the limitation describes determining a prior from data, which is a mental process implemented using a pen and paper.)
computing, ..., an objective function for the aggregate attribution model based on the attribution prior; (the limitation describes computing an objective function, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using the training component (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
updating, using the training component, parameters of the aggregate attribution model based on the objective function. (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 6,
Step 2A Prong 1: this claim recites the following abstract ideas:
normalizing, ..., the individual channel contribution value based on a plurality of individual channel contribution values corresponding to a plurality of content channels. (the limitation describes normalizing a value against a set of values, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using the calibration component (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 7,
Step 2A Prong 1: this claim recites the following abstract ideas:
computing, ..., the channel contribution data for an interaction path, wherein the channel contribution data corresponds to a plurality of channels, respectively; (the limitation describes computing contribution data for a path across channels, which is a mental process implemented using a pen and paper.)
computing, ..., a plurality of aggregate channel contribution values corresponding to the plurality of channels, respectively; (the limitation describes computing a plurality of values, which is a mental process implemented using a pen and paper.)
generating, ..., a plurality of individual channel contribution values based on the plurality of aggregate channel contribution values. (the limitation describes generating values from other computed values, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using the multi-touch attribution model (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
using the aggregate attribution model (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
using the calibration component (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 8,
Step 2A Prong 1: This claim does not recite an additional abstract idea, but the claim depends on claim 1,
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
providing, using a content component, content to a user via the digital content channel based on the individual channel contribution value. (this limitation describes data transmission/outputting of content, which is insignificant extra-solution activity and a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 9,
Step 2A Prong 1: this claim recites the following abstract ideas:
generating, ..., a content distribution campaign based on the individual channel contribution value. (the limitation describes creating a content distribution campaign, which falls within certain methods of organizing human activity, i.e., advertising, marketing or sales activities, see MPEP 2106.04(a)(2)(II).)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using a campaign component (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 10,
Step 2A Prong 1: this claim recites the following abstract ideas:
computing, ..., an individual channel contribution value based on the individual-level user interaction data; (the limitation describes computing a contribution value from user interaction data, which is a mental process implemented using a pen and paper.)
updating, ..., the individual channel contribution value based on an aggregate attribution model to obtain an updated channel contribution value; (the limitation describes updating a computed value based on a model, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
obtaining, by a multi-touch attribution model, individual-level user interaction data for a user from a digital content channel; (this limitation describes data collection/receiving, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
using the multi-touch attribution model (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
using a calibration component (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
providing, using a content component, customized content to the user via the digital content channel based on the updated channel contribution value. (this limitation describes data transmission/outputting of content, which is insignificant extra-solution activity and a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 11,
Step 2A Prong 1: this claim recites the following abstract ideas:
wherein the updated individual channel contribution value indicates a contribution of the digital content channel to an event of the individual-level user interaction data. (the limitation describes characterizing the meaning of a computed value, which is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
Step 2A Prong 2 and 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B. Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2. The claim is ineligible.
As to claim 12,
Step 2A Prong 1: this claim recites the following abstract ideas:
computing an aggregate channel contribution value ..., wherein the individual channel contribution value is updated based on the aggregate channel contribution value. (the limitation describes computing a value and updating another value based on it, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using the aggregate attribution model (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 13,
Step 2A Prong 1: this claim recites the following abstract ideas:
computing, ..., a plurality of individual channel contribution values for an interaction path, wherein the plurality of individual channel contribution values corresponds to a plurality of channels, respectively; (the limitation describes computing a plurality of contribution values, which is a mental process implemented using a pen and paper.)
computing, ..., a plurality of aggregate channel contribution values corresponding to the plurality of channels, respectively; (the limitation describes computing a plurality of values, which is a mental process implemented using a pen and paper.)
updating, ..., each of the plurality of individual channel contribution values based on the plurality of aggregate channel contribution values. (the limitation describes updating computed values based on other computed values, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using the multi-touch attribution model (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
using the aggregate attribution model (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
using the calibration component (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 14,
Step 2A Prong 1: This claim does not recite an additional abstract idea, but the claim depends on claim 10,
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
training, using a training component, the aggregate attribution model using the individual channel contribution value and experimental testing data. (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 15,
Step 2A Prong 1: this claim recites the following abstract ideas:
generating, ..., an attribution prior based on the individual channel contribution value; (the limitation describes determining a prior from a computed value, which is a mental process implemented using a pen and paper.)
computing, ..., an objective function for the aggregate attribution model based on the attribution prior; (the limitation describes computing an objective function, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using a training component (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
updating, using the training component, parameters of the aggregate attribution model based on the objective function. (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 16,
Step 2A Prong 1: this claim recites the following abstract ideas:
generating, ..., a content distribution campaign based on the updated individual channel contribution value. (the limitation describes creating a content distribution campaign, which falls within certain methods of organizing human activity, i.e., advertising, marketing or sales activities, see MPEP 2106.04(a)(2)(II).)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using a campaign component (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 17,
Step 2A Prong 1: this claim recites the following abstract ideas:
trained to compute channel contribution data based on individual-level user interaction data from a digital content channel; (the limitation describes computing contribution data from user interaction data, which is a mental process implemented using a pen and paper.)
trained to compute an aggregate channel contribution value for the digital content channel; (the limitation describes computing an aggregate contribution value, which is a mental process implemented using a pen and paper.)
configured to generate an individual channel contribution value for the digital content channel based on the channel contribution data and the aggregate channel contribution value. (the limitation describes calculating a contribution value based on data and another computed value, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
at least one memory; (This limitation is directed to mere instruction to store the abstract idea on a generic memory, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(iv) Storing and retrieving information in memory.)
at least one processor executing instructions stored in the at least one memory; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
a multi-touch attribution model comprising multi-touch attribution parameters stored in the at least one memory (This limitation is directed to mere instruction to store and apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
an aggregate attribution model comprising aggregate attribution parameters stored in the at least one memory (This limitation is directed to mere instruction to store and apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
a calibration component configured (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 18,
Step 2A Prong 1: This claim does not recite an additional abstract idea, but the claim depends on claim 17,
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
a content component configured to provide content to a user via the digital content channel based on the individual channel contribution value. (this limitation describes data transmission/outputting of content, which is insignificant extra-solution activity and a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 19,
Step 2A Prong 1: this claim recites the following abstract ideas:
generate a content distribution campaign based on the individual channel contribution value. (the limitation describes creating a content distribution campaign, which falls within certain methods of organizing human activity, i.e., advertising, marketing or sales activities, see MPEP 2106.04(a)(2)(II).)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
a campaign component configured (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 20,
Step 2A Prong 1: This claim does not recite an additional abstract idea, but the claim depends on claim 17,
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
a training component configured to update parameters of the aggregate attribution model based on the channel contribution data. (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1-3,6-13, and 16-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Yan et al. (US 20190278378 A1) in view of Chittilappilly et al. (US 20160210661 A1).
As to claim 1, Yan teaches a method of training a machine learning model, comprising: (see Yan paragraph [0203] "the deep learning attribution system 104 uses the training touchpoint paths mentioned above to train the touchpoint attribution attention neural network 714 to determine attention weights that reflect the relative conversion significance of each target touchpoint in a touchpoint sequence.")
obtaining, by a multi-touch attribution model, individual-level user interaction data from a digital content channel; (see Yan paragraph [0065] "The content provider or advertisement service can display the digital content to the user via one or more digital media channels, such as serving or downloading the digital content to the user via web page, mobile application, a streaming service, and/or other digital media channels.", and see Yan paragraph [0076] "the deep learning attribution system 104 obtains touchpoint data from a database that maintains touchpoint information related to an entity and/or product, where each touchpoint includes a touchpoint identifier, a user identifier, and an interaction time (e.g., timestamp).")
computing, using the multi-touch attribution model, channel contribution data based on the individual-level user interaction data; (see Yan paragraph [0156] "the deep learning attribution system 104 utilizes the attention layer to determine touchpoint attention weights (e.g., a.sub.t) for each touchpoint in the target touchpoint sequence 502.", and see Yan paragraph [0192] "the incremental touchpoint attribution scores provide an estimate for a channel's impact. As shown, the results are aggregated at the channel level.")
Yan does not explicitly teach "training, using a training component, an aggregate attribution model based on the channel contribution data; and", and "generating, using a calibration component, an individual channel contribution value for the digital content channel based on the channel contribution data and the aggregate attribution model."
However, Chittilappilly teaches:
training, using a training component, an aggregate attribution model based on the channel contribution data; and (see Chittilappilly paragraph [0049] "the channel response predictive model 162 can use weekly summaries of the actual channel stimulus 182 and the channel response measurements 172 over, for example, the last six months to predict the temporal contributions of each instance (e.g., channel) of the media channels 155.sub.1 to the channel-level conversions", and see Chittilappilly paragraph [0052] "the channel response feedback module 164 uses the touchpoint response predictive model 168 to generate a set of predicted channel response parameters 184 that can be used to further train the channel response predictive model 162. Specifically, the predicted channel response parameters 184 can comprise the aggregated digital channel contribution values derived from the most recent touchpoint data received in real time from the Internet, providing a dynamic feedback loop to continually improve the accuracy of the channel response predictive model 162")
generating, using a calibration component, an individual channel contribution value for the digital content channel based on the channel contribution data and the aggregate attribution model. (see Chittilappilly paragraph [0064] "The mapping of touchpoints to channels generated by the map generator 187 can be used by an attribution calibration engine 186 to reconcile certain inconsistencies exhibited by the channel-level attribution parameters 175 and the user-level attribution parameters 179.", and see Chittilappilly paragraph [0095] "an adjustment factor for each channel can be determined from the ratio of the respective channel-level attribution and the respective user-level aggregate attribution (e.g., "Display" adjustment factor=4.2/8.2=0.51). The attribution adjustments for each channel can then be applied to touchpoints mapped to each channel to determine a set of calibrated attribution parameters 173")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Yan, which teaches a multi-touch attribution model computing channel contribution data from individual-level user interaction data, to include Chittilappilly's aggregate attribution model trained on contribution values fed back from the user-level model and Chittilappilly's attribution calibration engine, because Chittilappilly teaches that "a discrepancy can exist between the digital channel attribution predicted by the top-down attribution model, and the digital channel attribution predicted by the bottom-up attribution model" (paragraph [0035]) and that reconciling them provides "a dynamic feedback loop to continually improve the accuracy of the channel response predictive model" (paragraph [0052]), yielding the predictable result of more accurate, reconciled contribution values.
As to claim 2, Yan as modified by Chittilappilly teaches the method of claim 1,
Wherein: the individual channel contribution value indicates a contribution of the digital content channel to an event of the individual-level user interaction data. (see Yan paragraph [0157] "the touchpoint attributions 506 indicate the relative significance and conversion importance of each touchpoint in the target touchpoint sequence 502.")
As to claim 3, Yan as modified by Chittilappilly teaches the method of claim 1,
further comprising: computing a preliminary individual channel contribution value for the digital content channel using the multi-touch attribution model; and (see Yan paragraph [0156] "the deep learning attribution system 104 utilizes the attention layer to determine touchpoint attention weights (e.g., a.sub.t) for each touchpoint in the target touchpoint sequence 502.")
Examiner note: The user-level attribution computed before calibration serves as the preliminary value that Chittilappilly's calibration engine subsequently adjusts.
Yan does not explicitly teach "computing an aggregate channel contribution value for the digital content channel using the aggregate attribution model", and "wherein the individual channel contribution value is generated based on the preliminary individual channel contribution value and the aggregate channel contribution value."
However, Chittilappilly teaches:
computing an aggregate channel contribution value for the digital content channel using the aggregate attribution model, (see Chittilappilly paragraph [0071] "the attributed channel responses 273 and associated set of channel contribution values 275 can be generated by the channel response predictive model 162.")
wherein the individual channel contribution value is generated based on the preliminary individual channel contribution value and the aggregate channel contribution value. (see Chittilappilly paragraph [0095] "an adjustment factor for each channel can be determined from the ratio of the respective channel-level attribution and the respective user-level aggregate attribution (e.g., "Display" adjustment factor=4.2/8.2=0.51). The attribution adjustments for each channel can then be applied to touchpoints mapped to each channel to determine a set of calibrated attribution parameters 173")
As to claim 6, Yan-Chittilappilly as modified by Sun teaches the method of claim 1,
further comprising: normalizing, using the calibration component, the individual channel contribution value based on a plurality of individual channel contribution values corresponding to a plurality of content channels. (see Yan paragraph [0193] "Table 3 shows normalized values of Table 2 such that the total touchpoint attribution scores sum to one (i.e., 1).", and see Yan paragraph [0194] "the fractional touchpoint attribution scores normalize the incremental scores of each channel for each touchpoint sequence and aggregate the incremental touchpoint attribution contributions at the channel level.")
As to claim 7, Yan-Chittilappilly as modified by Sun teaches the method of claim 1,
further comprising: computing, using the multi-touch attribution model, the channel contribution data for an interaction path, wherein the channel contribution data corresponds to a plurality of channels, respectively; (see Yan paragraph [0156] "the deep learning attribution system 104 utilizes the attention layer to determine touchpoint attention weights (e.g., a.sub.t) for each touchpoint in the target touchpoint sequence 502.", and see Yan paragraph [0184] "the researchers ran the experiments on a large event dataset of a marketing organization having three primary media channels (display, email, and paid search) with six different touchpoints")
Yan does not explicitly teaches "computing, using the aggregate attribution model, a plurality of aggregate channel contribution values corresponding to the plurality of channels, respectively; and", "generating, using the calibration component, a plurality of individual channel contribution values based on the plurality of aggregate channel contribution values."
However, Chittilappilly teaches:
computing, using the aggregate attribution model, a plurality of aggregate channel contribution values corresponding to the plurality of channels, respectively; and (see Chittilappilly paragraph [0094] "channel-level attribution parameters 175 might describe a set of channel-level attributions 475 for a "Display" channel, a paid search channel (e.g., "Search (P)"), and an organic search channel (e.g., "Search (O)") for a certain time period.")
generating, using the calibration component, a plurality of individual channel contribution values based on the plurality of aggregate channel contribution values. (see Chittilappilly paragraph [0095] "The attribution adjustments for each channel can then be applied to touchpoints mapped to each channel to determine a set of calibrated attribution parameters 173")
As to claim 8, Yan-Chittilappilly as modified by Sun teaches the method of claim 1,
further comprising: providing, using a content component, content to a user via the digital content channel based on the individual channel contribution value. (see Yan paragraph [0063] "the deep learning attribution system 104, via the server device 101, can provide digital content to a client device (e.g., the client devices 112a) through the selected media channel.", and see Yan paragraph [0166] "the deep learning attribution system 104 can automatically send content to the target user via the one or more recommended media channels.")
As to claim 9, Yan-Chittilappilly as modified by Sun teaches the method of claim 1,
further comprising: generating, using a campaign component, a content distribution campaign based on the individual channel contribution value. (see Yan paragraph [0226] "generating a digital conversion prediction corresponding to a digital media channel for the target user based on the attention weights for the digital touchpoints in the digital target touchpoint sequence, generating a digital content campaign that includes the target user and the digital media channel based on the digital conversion prediction, and executing the digital content campaign by providing digital content via the digital media channel to a client device of the target user.")
As to claim 10, Yan taches a method for data processing, comprising:
obtaining, by a multi-touch attribution model, individual-level user interaction data for a user from a digital content channel; (see Yan paragraph [0076] "the deep learning attribution system 104 filters each of the touchpoints by user (e.g., user identifier) to identify touchpoints between an individual user and the entity/product.")
computing, using the multi-touch attribution model, an individual channel contribution value based on the individual-level user interaction data; (see Yan paragraph [0157] "the touchpoint attributions 506 indicate the relative significance and conversion importance of each touchpoint in the target touchpoint sequence 502.")
providing, using a content component, customized content to the user via the digital content channel based on the updated channel contribution value. (see Yan paragraph [0166] "the deep learning attribution system 104 can automatically send content to the target user via the one or more recommended media channels.", and see Yan paragraph [0169] "the deep learning attribution system 104 can provide a conversion prediction 514 that accounts for characteristics and attributes of the target user")
Yan does not explicitly teaches "updating, using a calibration component, the individual channel contribution value based on an aggregate attribution model to obtain an updated channel contribution value; and."
However, Chittilappilly teaches:
updating, using a calibration component, the individual channel contribution value based on an aggregate attribution model to obtain an updated channel contribution value; and (see Chittilappilly paragraph [0064] "The attribution adjustment can be used to calibrate the attribution of each touchpoint mapped to the "Display" channel such that the user-level attribution (e.g., bottom-up attribution) is consistent (e.g., reconciled) with the channel-level attribution (e.g., top-down attribution).", and see Chittilappilly paragraph [0095] "The attribution adjustments for each channel can then be applied to touchpoints mapped to each channel to determine a set of calibrated attribution parameters 173")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Yan, which teaches computing an individual channel contribution value and providing customized content to the user based on that value, to include Chittilappilly's calibration step updating the value based on an aggregate attribution model, because Chittilappilly teaches calibrating so that "the user-level attribution (e.g., bottom-up attribution) is consistent (e.g., reconciled) with the channel-level attribution (e.g., top-down attribution)" (paragraph [0064]), and a value corrected for aggregate-level effects such as seasonality more accurately identifies the effective channel, yielding the predictable result of better-targeted content delivery.
As to claim 11, Yan as modified by Chittilappilly teaches the method of claim 10,
wherein: the updated individual channel contribution value indicates a contribution of the digital content channel to an event of the individual-level user interaction data. (see Yan paragraph [0157] "the touchpoint attributions 506 indicate the relative significance and conversion importance of each touchpoint in the target touchpoint sequence 502.")
Yan does not explicitly teach "updated individual channel contribution value indicates a contribution"
However, Chittilappilly teaches:
wherein: the updated individual channel contribution value indicates a contribution of the digital content channel to an event of the individual-level user interaction data. (see Chittilappilly paragraph [0064] "the user-level attribution (e.g., bottom-up attribution) is consistent (e.g., reconciled) with the channel-level attribution (e.g., top-down attribution)."
As to claim 12, Yan as modified by Chittilappilly teaches the method of claim 10,
further comprising: computing an aggregate channel contribution value using the aggregate attribution model, (see Chittilappilly paragraph [0071] "the attributed channel responses 273 and associated set of channel contribution values 275 can be generated by the channel response predictive model 162.")
wherein the individual channel contribution value is updated based on the aggregate channel contribution value. (see Chittilappilly paragraph [0095] "an adjustment factor for each channel can be determined from the ratio of the respective channel-level attribution and the respective user-level aggregate attribution (e.g., "Display" adjustment factor=4.2/8.2=0.51).")
As to claim 13, Yan as modified by Chittilappilly teaches the method of claim 10,
further comprising: computing, using the multi-touch attribution model, a plurality of individual channel contribution values for an interaction path, wherein the plurality of individual channel contribution values corresponds to a plurality of channels, respectively; (see Yan paragraph [0156] "the deep learning attribution system 104 utilizes the attention layer to determine touchpoint attention weights (e.g., a.sub.t) for each touchpoint in the target touchpoint sequence 502.", and see Yan paragraph [0184] "the researchers ran the experiments on a large event dataset of a marketing organization having three primary media channels (display, email, and paid search) with six different touchpoints")
Yan does not explicitly teach "computing, using the aggregate attribution model, a plurality of aggregate channel contribution values corresponding to the plurality of channels, respectively; and", and "updating, using the calibration component, each of the plurality of individual channel contribution values based on the plurality of aggregate channel contribution values."
However, Chittilappilly teaches:
computing, using the aggregate attribution model, a plurality of aggregate channel contribution values corresponding to the plurality of channels, respectively; and (see Chittilappilly paragraph [0094] "channel-level attribution parameters 175 might describe a set of channel-level attributions 475 for a "Display" channel, a paid search channel (e.g., "Search (P)"), and an organic search channel (e.g., "Search (O)") for a certain time period.")
updating, using the calibration component, each of the plurality of individual channel contribution values based on the plurality of aggregate channel contribution values. (see Chittilappilly paragraph [0064] "Such adjustments can be applied to other channels of interest (e.g., channels comprising a given marketing campaign).", and see Chittilappilly paragraph [0095] "The attribution adjustments for each channel can then be applied to touchpoints mapped to each channel to determine a set of calibrated attribution parameters 173")
As to claim 16, Yan-Chittilappilly as modified by Sun teaches the method of claim 10,
further comprising: generating, using a campaign component, a content distribution campaign (see Yan paragraph [0226] "generating a digital content campaign that includes the target user and the digital media channel based on the digital conversion prediction")
Yan does not explicitly teach "based on the updated individual channel contribution value"
However, Chittilappilly teaches:
based on the updated individual channel contribution value. (see Chittilappilly paragraph [0120] "the campaign planning server 934 can be used to generate marketing campaign plans and associated marketing spend apportionments")
As to claim 17, Yan teaches an apparatus for data processing, comprising:
at least one memory; (see Yan paragraph [0240] "The computing device 1000 includes memory 1004, which is coupled to the processor(s) 1002. The memory 1004 may be used for storing data, metadata, and programs for execution by the processor(s).")
at least one processor executing instructions stored in the at least one memory; (see Yan paragraph [0239] "the processor(s) 1002 includes hardware for executing instructions, such as those making up a computer program. As an example, and not by way of limitation, to execute instructions, the processor(s) 1002 may retrieve (or fetch) the instructions from an internal register, an internal cache, memory 1004, or a storage device 1006 and decode and execute them.")
a multi-touch attribution model comprising multi-touch attribution parameters stored in the at least one memory, the multi-touch attribution model trained to compute channel contribution data based on individual-level user interaction data from a digital content channel; (see Yan paragraph [0100] "the deep learning attribution system 104 randomly initializes the embedding matrix and tunes its parameters through supervised training.", and see Yan paragraph [0203] "the deep learning attribution system 104 uses the training touchpoint paths mentioned above to train the touchpoint attribution attention neural network 714 to determine attention weights that reflect the relative conversion significance of each target touchpoint in a touchpoint sequence.", and see Yan paragraph [0208] "the components 710-742 can include one or more instructions stored on a computer-readable storage medium and executable by processors of one or more computing devices, such as a client device or server device.")
Yan does not explicitly teach "an aggregate attribution model comprising aggregate attribution parameters stored in the at least one memory, the aggregate attribution model trained to compute an aggregate channel contribution value for the digital content channel; and", and "a calibration component configured to generate an individual channel contribution value for the digital content channel based on the channel contribution data and the aggregate channel contribution value."
However, Chittilappilly teaches:
an aggregate attribution model comprising aggregate attribution parameters stored in the at least one memory, the aggregate attribution model trained to compute an aggregate channel contribution value for the digital content channel; and (see Chittilappilly paragraph [0066] "the parameters (e.g., channel response predictive model parameters 163) describing the learning model (e.g., channel response predictive model 162) can be stored in a measurement data store 526 for access by various computing devices (e.g., measurement server 110, management interface device 114, apportionment server 111, etc.).", and see Chittilappilly paragraph [0071] "the attributed channel responses 273 and associated set of channel contribution values 275 can be generated by the channel response predictive model 162.")
a calibration component configured to generate an individual channel contribution value for the digital content channel based on the channel contribution data and the aggregate channel contribution value. (see Chittilappilly paragraph [0064] "The mapping of touchpoints to channels generated by the map generator 187 can be used by an attribution calibration engine 186 to reconcile certain inconsistencies exhibited by the channel-level attribution parameters 175 and the user-level attribution parameters 179.", and see Chittilappilly paragraph [0095] "an adjustment factor for each channel can be determined from the ratio of the respective channel-level attribution and the respective user-level aggregate attribution (e.g., "Display" adjustment factor=4.2/8.2=0.51). The attribution adjustments for each channel can then be applied to touchpoints mapped to each channel to determine a set of calibrated attribution parameters 173")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Yan, which teaches an apparatus with memory, a processor, and a stored multi-touch attribution model, to include Chittilappilly's aggregate attribution model with parameters stored in a data store (paragraph [0066]) and its attribution calibration engine (paragraph [0064]), because Chittilappilly identifies the discrepancy between top-down and bottom-up attribution (paragraph [0035]) and teaches that the reconciled architecture provides "real-time, reconciled updates to multiple computer-generated predictive models" (paragraph [0045]), yielding the predictable result of an apparatus producing contribution values accurate at both the user and aggregate levels.
As to claim 18, Yan as modified by Chittilappilly teaches the apparatus of claim 17,
further comprising: a content component configured to provide content to a user via the digital content channel based on the individual channel contribution value. (see Yan paragraph [0063] "the deep learning attribution system 104, via the server device 101, can provide digital content to a client device (e.g., the client devices 112a) through the selected media channel.", and see Yan paragraph [0166] "the deep learning attribution system 104 can automatically send content to the target user via the one or more recommended media channels.")
As to claim 19, Yan as modified by Chittilappilly teaches the apparatus of claim 17,
further comprising: a campaign component configured to generate a content distribution campaign based on the individual channel contribution value.
(see Yan paragraph [0226] "generating a digital content campaign that includes the target user and the digital media channel based on the digital conversion prediction")
Yan does not explicitly teach "based on the individual channel contribution value."
However, Chittilappilly teaches:
based on the individual channel contribution value. (see Chittilappilly paragraph [0120] "the campaign planning server 934 can be used to generate marketing campaign plans and associated marketing spend apportionments, which information can be stored in the campaign plan data store 957 and/or used to generate various instances of the campaign specification data records 974.")
As to claim 20, Yan as modified by Chittilappilly teaches the apparatus of claim 17,
further comprising: a training component configured to update parameters of the aggregate attribution model based on the channel contribution data. (see Chittilappilly paragraph [0052] "the predicted channel response parameters 184 can comprise the aggregated digital channel contribution values derived from the most recent touchpoint data received in real time from the Internet, providing a dynamic feedback loop to continually improve the accuracy of the channel response predictive model 162", and see Chittilappilly paragraph [0100] "The channel response feedback module 164 can also use the touchpoint response predictive model 168 to generate a set of predicted channel response parameters (see operation 516) for use by the model generator 506 to improve the accuracy of the channel response predictive model 162.")
Claim(s) 4-5, and 14-15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Yan et al. (US 20190278378 A1) in view of Chittilappilly et al. (US 20160210661 A1) and Sun et al. (US 20190065638 A1).
As to claim 4, Yan as modified by Chittilappilly teaches the method of claim 1,
wherein: the aggregate attribution model is trained using the channel contribution data and (see Chittilappilly paragraph [0052] "the predicted channel response parameters 184 can comprise the aggregated digital channel contribution values derived from the most recent touchpoint data received in real time from the Internet, providing a dynamic feedback loop to continually improve the accuracy of the channel response predictive model 162")
Yan-Chittilappilly do not explicitly teach "experimental testing data."
However, Sun teaches:
experimental testing data. (see Sun paragraph [0060] "For example, past experiments or industry knowledge may define the distribution for a particular parameter. For example, there may be a known behavior of how television advertising behaves, thus an informative prior can be received and/or selected by the modeler 124 based on this knowledge.", and see Sun paragraph [0062] "Thus, both the geo-level data 136 and priors for the parameters may influence the generation of the posterior distribution for the GBHMMM 138.", and see Sun paragraph [0074] "Geo experiments are often used to measure the ad effectiveness by modifying the media spend in randomly picked treatment geos and estimating the response relative to the control geos. Independent variation in marketing spend across geos from the experiments offers the possibility to improve MMM results by eliminating or reducing advertisement bias, that occurs from the result of directed marketing efforts, and increasing the effective sample size.")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Yan as modified by Chittilappilly, which teaches training the aggregate attribution model based on channel contribution data, to include Sun's experimental testing data as an additional training input, because Sun teaches that "Independent variation in marketing spend across geos from the experiments offers the possibility to improve MMM results by eliminating or reducing advertisement bias... and increasing the effective sample size" (paragraph [0074]), yielding the predictable result of a less biased, more accurate aggregate attribution model.
As to claim 5, Yan-Chittilappilly as modified by Sun teaches the method of claim 1,
further comprising: generating, using the training component, an attribution prior based on the channel contribution data; (see Chittilappilly paragraph [0052] "the predicted channel response parameters 184 can comprise the aggregated digital channel contribution values derived from the most recent touchpoint data received in real time from the Internet, providing a dynamic feedback loop to continually improve the accuracy of the channel response predictive model 162")
Chittilappilly does not explicitly teach "an attribution prior based on the channel contribution data;", "computing, using the training component, an objective function for the aggregate attribution model based on the attribution prior; and", and "updating, using the training component, parameters of the aggregate attribution model based on the objective function."
an attribution prior based on the channel contribution data (see Sun paragraph [0029] "One improvement to the model may be to pool data from one or more brands into the same product category and pass the knowledge via informative priors to a single brand within the same category.", and see Sun paragraph [0033] "A Bayesian hierarchical model can be generated based on pooled information from multiple geographic regions and prior knowledge. The prior knowledge may be based on industry experience.", and see Sun paragraph [0059] "Generating the GBHMMM 138 may include generating the GBHMMM 138 based on one or more priors for one or more parameters of the GBHMMM 138 and the geo-level data 136. Modeler 124 can be configured to receive informative and/or non-informative priors (e.g., weak or strong priors). The priors may be distributions for one or more parameters of the GBHMMM 138. An informative prior (e.g., a strong prior) may be a pre-determined probability distribution for a parameter.", and see Sun paragraph [0065] "The priors 320-324 may be informative, i.e., they may be predetermined by a user based on industry knowledge or information learned from other data sources.")
computing, using the training component, an objective function for the aggregate attribution model based on the attribution prior; and (see Sun paragraph [0027] "Bayesian hierarchical models are used to generate posterior distributions for parameters of the model based on collected data for said parameters and a prior distribution for said parameters.", and see Sun paragraph [0061] "Fitting the GBHMMM 138 may include fitting the posterior of each of the one or more parameters based on the geo-level data 136, i.e., posterior of each of the one or more parameters may be the prior for each of the one or more parameters conditioned on the geo-level data 136.", and see Sun paragraph [0088] "If the data is strong enough, the posterior will be pulled towards the true parameters, otherwise, the posteriors will be close to the priors.")
updating, using the training component, parameters of the aggregate attribution model based on the objective function. (see Sun paragraph [0051] "Fitting the parameters of the GBHMMM 138 can include generating a posterior distribution for each of the parameters of the GBHMMM 138 by the modeler 124 based on the geo-level data 136 and the one or more priors.", and see Sun paragraph [0062] "Using one or more fitting algorithms (e.g., Gibbs-Sampling, Markov Chain Monte Carlo (MCMC), etc.) the modeler 124 can be configured to generate the GBHMMM 138. Generating the GBHMMM 138 may include determining posterior distributions for the one or more parameters of the GBHMMM 138. Thus, both the geo-level data 136 and priors for the parameters may influence the generation of the posterior distribution for the GBHMMM 138.", and see Sun paragraph [0089] "Gibbs sampling can be used to fit the geo-level model by alternating between sampling the transformation parameters and the hierarchical linear model parameters. When conditioning on the parameters of the hierarchical linear model, slice sampling can be used to draw samples from the full conditional distribution of the transformation parameters.")
As to claim 14, Yan as modified by Chittilappilly teaches the method of claim 10,
further comprising: training, using a training component, the aggregate attribution model using the individual channel contribution value and experimental testing data. (see Chittilappilly paragraph [0052] "the channel response feedback module 164 uses the touchpoint response predictive model 168 to generate a set of predicted channel response parameters 184 that can be used to further train the channel response predictive model 162.", and see Chittilappilly paragraph [0100] "The channel response feedback module 164 can also use the touchpoint response predictive model 168 to generate a set of predicted channel response parameters (see operation 516) for use by the model generator 506 to improve the accuracy of the channel response predictive model 162.")
Yan-Chittilappilly does not explicitly teaches "experimental testing data."
However, Sun teaches:
experimental testing data. (see Sun paragraph [0060] "past experiments or industry knowledge may define the distribution for a particular parameter. For example, there may be a known behavior of how television advertising behaves, thus an informative prior can be received and/or selected by the modeler 124 based on this knowledge.", and see Sun paragraph [0074] "Geo experiments are often used to measure the ad effectiveness by modifying the media spend in randomly picked treatment geos and estimating the response relative to the control geos. Independent variation in marketing spend across geos from the experiments offers the possibility to improve MMM results")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the invention of Yan as modified by Chittilappilly, which teaches training the aggregate attribution model using contribution values fed back from the user-level model, to include Sun's experimental testing data from treatment-versus-control geo experiments, because Sun teaches such data improves MMM results "by eliminating or reducing advertisement bias... and increasing the effective sample size" (paragraph [0074]) the same accuracy-improvement objective Chittilappilly states for its feedback training (paragraph [0052]) yielding the predictable result of a more accurate aggregate attribution model.
As to claim 15, Yan-Chittilappilly as modified by Sun teaches the method of claim 10,
further comprising: generating, using a training component, an attribution prior based on the individual channel contribution value; (see Chittilappilly paragraph [0052] "the predicted channel response parameters 184 can comprise the aggregated digital channel contribution values derived from the most recent touchpoint data received in real time from the Internet")
Yan-Chittilappilly does not explicitly teaches "an attribution prior", "computing, using the training component, an objective function for the aggregate attribution model based on the attribution prior; and", and "updating, using the training component, parameters of the aggregate attribution model based on the objective function."
However, Sun teaches:
an attribution prior (see Sun paragraph [0029] "One improvement to the model may be to pool data from one or more brands into the same product category and pass the knowledge via informative priors to a single brand within the same category.", and see Sun paragraph [0065] "The priors 320-324 may be informative, i.e., they may be predetermined by a user based on industry knowledge or information learned from other data sources.", and see Sun paragraph [0088] "Alternatively, category level priors can be used as informative priors for the GBHMMM 138.")
computing, using the training component, an objective function for the aggregate attribution model based on the attribution prior; and (see Sun paragraph [0061] "Fitting the GBHMMM 138 may include fitting the posterior of each of the one or more parameters based on the geo-level data 136, i.e., posterior of each of the one or more parameters may be the prior for each of the one or more parameters conditioned on the geo-level data 136.", and see Sun paragraph [0088] "If the data is strong enough, the posterior will be pulled towards the true parameters, otherwise, the posteriors will be close to the priors.")
updating, using the training component, parameters of the aggregate attribution model based on the objective function. (see Sun paragraph [0062] "Using one or more fitting algorithms (e.g., Gibbs-Sampling, Markov Chain Monte Carlo (MCMC), etc.) the modeler 124 can be configured to generate the GBHMMM 138. Generating the GBHMMM 138 may include determining posterior distributions for the one or more parameters of the GBHMMM 138.", and see Sun paragraph [0089] "Gibbs sampling can be used to fit the geo-level model by alternating between sampling the transformation parameters and the hierarchical linear model parameters.")
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ABDULLAH K ABOUD whose telephone number is (571)272-0025. The examiner can normally be reached Mon-Fri 8am-5pm.
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, Li B Zhen, can be reached at (571) 272-3768. 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.
/ABDULLAH KHALED ABOUD/ Examiner, Art Unit 2121
/Li B. Zhen/ Supervisory Patent Examiner, Art Unit 2121
Prosecution Insights