DETAILED ACTION
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 .
Status of Claims
Applicant’s communications filed on 1/20/2022 have been considered.
Claims 1, 8, 11 and 17 have been amended.
Claim 20 has been canceled.
Claim 21 is newly added.
Claims 1-19 and 21 are currently pending and have been examined.
Response to Arguments
Applicant’s arguments filed with respect to the rejection of claims under 35 USC 101 have been fully considered but they are not persuasive.
Applicant argues that “amended claim 1 recites… non-human functions that cannot be reasonably considered sales or marketing activities” (Remarks Page 10). This argument has been fully considered but is not persuasive. The MPEP discloses, “certain activity between a person and a computer… may fall within the ‘certain methods of organizing human activity’ grouping… the determination should be based on whether the activity itself falls within one of the sub-groupings” MPEP 2106.04. Accordingly, merely reciting functions performed by a computer (non-human functions) does not automatically disqualify the claim from reciting certain methods of organizing human activity. Furthermore, the independent claims recite additional including a system, comprising: a memory storing instructions and a trained machine learning model trained, a processor operatively connected to the memory and configured to execute the instructions to perform operations including: initiating a widget on a device, the widget including a first trained machine learning model locally operating on the user device, the first trained machine learning model having been trained; an active user session; a remote server; a computer-implemented method; and the first trained machine learning model having been trained on the user device based on a training dataset local to the user device, which are not analyzed under Step 2A, Prong 1 of 101 analysis. The claims further recite limitations such as learning associations between an item offering and one or more categories, outputting an item label category in response to an input data set, predicting a score of an item based on training data sets, retrieving a first data set from the user associated with a user session, scoring a first item associated with the user session, labeling a category for the first item based on the result of the scoring, sending a result of said labeling, and applying, based on the result of said labeling, the category to a transaction record that is associated with the user session without accessing the first data set and that lacks item-level data. These limitations represent Certain Methods of Organizing Human Activity, in that they recite commercial/legal interactions. For example, Applicant’s specification discusses finding associated between item and/or merchant data and classifications or categories (see at least Spec [0043]), wherein categorization data includes transaction receipts ([0055]). Accordingly, the recited limitations claim collection and analysis of data for the purpose of categorizing said data.
Applicant further argues that amended claim 1 “[does] more than just link the claim to a technical environment… such features are also technical functions that provide more than mere instructions to ‘apply it’… [and] result in improvement to the technology for data privacy and security,” with reference to Applicant’s specification ([0081]) (Remarks Pages 10 and 11). This argument has been fully considered but is not persuasive. While Applicant’s specification (see at least [0081]) states that the claimed features can reduce the amount of data transmitted between the user device and the remote server, potentially improving data privacy and efficiency, this merely represents a conclusory statement. The statements made in Appellant’s specification, specifically at paragraph [0081], do not provide any detail regarding how the claimed invention is providing any improvement to the functioning of the computer/other technology, thereby making the statements merely conclusory. For example, reducing the amount of data transmitted and improving data privacy and efficiency do not represent how a change or improvement has been made to the claimed technology. Although the claims include technology such as a system, comprising: a memory storing instructions and a trained machine learning model trained, a processor operatively connected to the memory and configured to execute the instructions to perform operations including: initiating a widget on a device, the widget including a first trained machine learning model locally operating on the user device, the first trained machine learning model having been trained; an active user session; a remote server; a computer-implemented method; and the first trained machine learning model having been trained on the user device based on a training dataset local to the user device, such elements are merely peripherally incorporated in order to implement the abstract idea. The claimed process, while arguably resulting in more efficient categorization, is not providing any improvement to another technology or technical field. While the additional elements are utilized, it is the categorization of items that is an improvement. Accordingly, the rejection has been maintained.
Applicant’s arguments filed with respect to the rejection of claims under 35 USC 103 have been fully considered but are rendered moot under new grounds of rejection.
Applicant argues that the amended claims overcome the currently cited prior art because “Rigney is silent to [the amended limitations]” (Remarks Pages 11 and 12). This argument has been considered but is rendered moot under new grounds of rejection. Independent claim 1 currently stands rejected in view of the newly cited combination of Rigney, Chhibber, and Ovick. Rigney has been further relied upon to teach a first trained machine learning model, the first trained machine learning model having been trained, at (Rigney, [0048-0049][0051]), disclosing the machine learning model trained by users and based on a database of common business expenses. Rigney further teaches applying, by the remote server, based on the result of said labeling, the category to a transaction record, at (Rigney, [0057]), disclosing that the system saves categorization information to an expense log. Chhibber, on the other hand, has been relied upon to teach the widget including a machine learning model locally operating on the user device, retrieving one or more datum of an active user session, scoring, by the model operating on the user device, items associated with the active user session, sending, by the widget, information to a remote server, and processing, by the remote server, the information associated with the active user session without accessing the first data set from the user device, as discussed below. Ovick has additionally been relied upon to teach applying the category to a transaction record that lacks item-level data, as discussed below. Regarding independent claim 11, it is further noted that Rigney has been further relied upon to teach the first trained machine learning model having been trained to predict a score of an item based on training data sets, at (Rigney, [0048-0049][0051]), disclosing that a recurrent neural network seeded with a database of common business expenses tailored to the individual’s occupation or industry, where parameters include occupation and receipt item descriptions is trained by the users, and other similar users, and can determine a probability that particular receipt items fall in a particular category. Accordingly, independent claims 1 and 11 stand rejected in view of a new grounds of rejection, and Applicant’s arguments are moot.
With regards to Applicant’s arguments that new independent claim 21 is allowable for the reasons discussed with regards to independent claims 1 and 11 (Remarks Page 12), this argument has been considered but is rendered moot under new grounds of rejection. As discussed above, independent claims 1 and 11 stand rejected in view of the newly cited combination of Rigney/Chhibber/Ovick. Independent claim 21 recites substantially similar subject matter to that recited in independent claim 11, and therefore has similarly been rejected in view of the combination of Rigney/Chhibber/Ovick.
It is further noted that the amended limitations of dependent claim 8 are taught by newly cited Chhibber, as discussed below. Furthermore, amended dependent claim 17 additionally stands rejected under new grounds of rejection, in view of the combination of Rigney/Chhibber/Ovick/Walter.
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 1-19 and 21 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claims recite an abstract idea. The judicial exception is not integrated into a practical application. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Under Step 1 of the Subject Matter Eligibility Test for Products and Processes, the claims must be directed to one of the four statutory categories. See MPEP 2106.03. Claims 1-10 and 21 are directed towards a process. Claims 11-19 are directed towards a machine. Therefore, claims 1-19 and 21 are directed to one of the four statutory categories (Step 1: YES, regarding claims 1-19 and 21).
Under Step 2A of the MPEP, it is determined whether the claims are directed to a judicially recognized exception. See MPEP 2106.04. Step 2A is a two-prong inquiry.
Under Prong 1, it is determined whether the claim recites a judicial exception. In determining whether the claims are directed to a judicial exception, the claims are analyzed to evaluate whether the claims recite a judicial exception.
Taking Claim 11 as representative, claim 11 recites limitations that fall within the certain methods of organizing human activity groupings of abstract ideas, including:
A method for categorization, comprising:
a model, based on item information data and ground truth, to learn associations between an item offering and one or more categories and to output an item label category in response to an input data set; and
a user;
predict a score of an item based on training data sets;
retrieving a first data set from the user, the first data set associated with one or more datum of a user session;
scoring a first item associated with the user session based at least in part on the first data set;
labeling a category for the first item based on a result of the scoring;
sending a result of said labeling; and
applying, based on the result of said labeling, the category to a transaction record that is associated with the user session without accessing the first data set and that lacks item-level data.
Claims 1 and 21 recite the same limitations believed to be abstract as recited in claim 11.
Claim 11, as exemplary, recites the abstract idea of product categorization. These recited limitations fall within the "Certain Methods of Organizing Human Activities" Grouping of abstract ideas as it relates to commercial interactions of sales activities or behaviors. Accordingly, the claim recites an abstract idea. See MPEP 2106.04.
Accordingly, under Prong One of Step 2A of the Alice/Mayo test, claims 1, 11 and 21 recite an abstract idea (Step 2A, Prong One: YES).
Under Prong 2, it is determined whether the claim recites additional elements that integrate the exception into a practical application of the exception.
Claim 11 recites additional elements beyond the judicial exception(s), including a system, comprising: a memory storing instructions and a trained machine learning model trained, a processor operatively connected to the memory and configured to execute the instructions to perform operations including: initiating a widget on a device, the widget including a first trained machine learning model locally operating on the user device, the first trained machine learning model having been trained; an active user session; and a remote server. Claim 1 recites the same additional elements as recited in claim 11, and additionally recites a computer-implemented method; and the first trained machine learning model having been trained on the user device based on a training dataset local to the user device. Claim 21 recites the same additional elements as recited in claims 1 and 11.
These additional elements are described at a high level in Applicant’s specification without any meaningful detail about their structure or configuration. As such, these computer-related limitations are not found to be sufficient to integrate the abstract idea into a practical application. Claims 1, 11 and 21 specifying that the abstract idea of product categorization is executed in a computer environment merely indicates a field of use in which to apply the abstract idea because this requirement merely limits the claims to the computer field, i.e., to execution on a generic computer. As such, under Prong Two of Step 2A of the Alice/Mayo test, when considered both individually and as a whole, the limitations of claims 1, 11 and 21 are not indicative of integration into a practical application (Step 2A, Prong Two: NO).
Since claims 1, 11 and 21 recite an abstract idea and fail to integrate the abstract idea into a practical application, claims 1, 11 and 21 are “directed to” an abstract idea (Step 2A: YES). Accordingly, the judicial exception is not integrated into a practical application.
Next, under Step 2B, the instant claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as discussed above, the additional elements of a system, comprising: a memory storing instructions and a trained machine learning model trained, a processor operatively connected to the memory and configured to execute the instructions to perform operations including: initiating a widget on a device, the widget including a first trained machine learning model locally operating on the user device, the first trained machine learning model having been trained; an active user session; a remote server; a computer-implemented method; and the first trained machine learning model having been trained on the user device based on a training dataset local to the user device method amount to no more than mere instructions to apply the exception using generic computer components. For the same reason these elements are not sufficient to provide an inventive concept. Therefore when considering the additional elements alone, and in combination, there is no inventive concept in the claim, and thus the claim is not patent eligible (Step 2B: NO).
Dependent claims 2-10 and 12-19, when analyzed as a whole, are held to be patent ineligible under 35 U.S.C. 101 because they do not add “significantly more” to the abstract idea. As for dependent claims 3-7, 9-10, 13-16 and 18-19 these claims recite limitations that further define the same abstract idea noted in independent claims 1, 11 and 21. Therefore, claims 3-7, 9-10, 13-16 and 18-19 are considered patent ineligible for the reasons given above.
As for dependent claims 2, 8, 12 and 17, these claims recite limitations that further define the abstract idea noted in independent claims 1, 11 and 21. Additionally, they recite the following additional limitations:
wherein the one or more datum is continuously received throughout the active user session and includes one or more of HyperText Markup Language (HTML) code, metadata, or an item tag;
wherein the first trained machine learning model is part of a federated learning system across a plurality of user devices; and
injecting code into a browser associated with the first item, wherein the code includes… an adjustment to a layout of a webpage associated with the first item.
The additional elements of HyperText Markup Language (HTML) code, a federated learning system across a plurality of user devices, and injecting code into a browser associated with the first item, wherein the code includes… an adjustment to a layout of a webpage associated with the first item are all recited at a high level of generality such that they amount to no more than instructions to apply the judicial exception in a generic technological environment. Even in combination, these additional elements do not integrate the abstract idea into a practical application and do not amount to significantly more than the abstract idea itself. Accordingly, under the Alice/Mayo test, claims 1-19 and 21 are ineligible.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
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 1-3, 8-9, 11-13, 17-18 and 21 are rejected under 35 U.S.C. 103 as being unpatentable over previously cited Rigney (US 2018/0308179 A1), in view of newly cited U.S Patent Application No. 2022/0414529 A1 to Chhibber et al., hereinafter Chhibber, and further in view of U.S Patent Application No. 2014/0040051 A1 to Ovick et al., hereinafter Ovick.
Regarding Claim 1, Rigney discloses A computer-implemented method for categorization, comprising ([0047]):
initiating a widget on a user device ([0023] embodiments of the invention may take the form of a smart phone application (app) configured to be installed on a user smart phone),
a first trained machine learning model ([0048-0049] Item descriptions are input into the machine learning model which cross references the line item with the user’s occupation and items in the common business expense database, then the model determines the probability that the receipt item is a business expense; [0051] a machine learning model will consult a database of business expenses organized by occupation and vendor to determine the probability of whether a particular receipt item is a business expense. Business expenses organized by occupation can initially be procured by the system… see further [0049] a recurrent neural network trained by the users, and other similar users),
the first trained machine learning model having been trained ([0049] a recurrent neural network seeded with a database of common business expenses tailored to the individual’s occupation or industry, where parameters include occupation and receipt item descriptions… trained by the users, and other similar users);
retrieving, by the widget, a first data set from the user device, the first data set associated with one or more datum of a user ([0023] The smart phone app may therefore be able to access receipts and other purchase records across these various platforms so as to be more likely to capture relevant information; [0042] In Step 200, the item description and other available information is gathered from the receipt or other purchase record, from the user device and sent to a remote server for categorization);
scoring, by the first trained machine learning model, a first item based at least in part on the first data set ([0051] a machine learning model will consult a database of business expenses organized by occupation and vendor to determine the probability of whether a particular receipt item is a business expense; [0052] In Step 210, the system determines a probability that the line item actually falls within the predicted category. The probability may be based upon a degree of certainty that the item description matches a corresponding item description in the database of common expenses);
labeling a category for the first item based on a result of the scoring ([0053] In Step 212, the system compares this probability to a threshold… if the probability is above the threshold, the system may query the user to confirm line items; [0054] In Step 214, the system requests verification from the user that the line item was categorized correctly. The verification may present the line item and the determined categorization);
sending, by the widget, a result of said labeling to a remote store ([0057] In Step 216, the system receives the verification from the user. If the user approves of the categorization, in Step 218 the system saves the information to an expense log or other data store (such as discussed above in FIG. 1). The system may additionally save the information to the data store of common items… see [0031] data stores can be remotely accessible over internet 442); and
applying, by the remote server, based on the result of said labeling, the category to a transaction record ([0057] in Step 218 the system saves the information to an expense log or other data store (such as discussed above in FIG. 1). The system may additionally save the information to the data store of common items);
But does not explicitly disclose the widget including a machine learning model locally operating on the user device; the machine learning model trained on the user device based on a training data set local to the user device; retrieving data of an active user session; a first item associated with the active user session; sending information to a remote server; and applying the category to a transaction record associated with the active user session without accessing the first data set from the user device and that lacks item-level data.
Chhibber, on the other hand, teaches the widget including a machine learning model locally operating on the user device ([0015] User computing device 110A, in the illustrated embodiment, includes baseline model 120 and device-trained model 130A; [0031] user computing device 110A includes secure storage 212 and application 240, which in turn includes training module 250 and updated model 130);
the machine learning model trained on the user device based on a training data set local to the user device ([0016] User computing device 110A, in the illustrated embodiment, receives a stream 104 of user data. The stream 104 of user data is a continuous flow of information into the user computing device 110A; [0018] User computing device 110A trains a baseline model 120 using one or more sets 114 of user data from the stream 104 of user data to generate device-trained model 130A. User computing device 110A trains baseline model 120 using one or more machine learning techniques);
retrieving one or more datum of an active user session ([0032] Application 240 receives user request 102 from user 120 and stream 104 of user data; [0040] real-time module 310 pre-processes adjusted user data 222 as it is received. For example, as new user requests 102 are received at user computing device 110A and as new data comes in from the stream 104 of user data, real-time module 310 performs pre-processing techniques… see [0016] The stream 104 of user data is a continuous flow of information into the user computing device 110A. This stream of data may be continuous and includes device characteristics, characteristics associated with user 120, characteristics associated with user request 102);
scoring, by the model operating on the user device, items associated with the active user session ([0019] Device-trained model 130A outputs risk score 132 for the user request 102 based on set 106 of characteristics and user computing device 110A transmits the risk score 132 to decisioning module 160. Risk score 132 indicates an amount of risk associated with user request 102 based on the set 106 of characteristics… see [0023] a first user submitting a transaction request for a wrench from a hardware store, a second user submitting a transaction request for a diamond ring from a pawn shop);
sending, by the widget, information to a remote server ([0020] In response to sending risk score 132 to system 150, user computing device 110A receives a decision 162; [0022] Server computer system 150 receives risk score 132 for user request 102 from device-trained model 130A of user computing device 110A); and
processing, by the remote server, the information associated with the active user session without accessing the first data set from the user device ([0022] Server computer system 150, in the illustrated embodiment, receives risk score 132 for user request 102 from device-trained model 130A of user computing device 110A. System 150 executes decisioning module 160 to generate a decision 162 for request 102 based on risk score 132… Decisioning module 160 may receive information specifying the type of request 102 from user computing device 110A in addition to the obfuscated user data 116 (which includes information about the user and the user's device). Decisioning module 160 selects a set of rules and heuristics for request 102 based on one or more characteristics indicated in obfuscated user data 116; [0047] Rule selection module 470 receives obfuscated user data 116 from user computing device 110A and selects a set 464 of rules from a plurality of security rules 462 (e.g., for evaluating user request 102) based on the obfuscated user data 116; [0063] the computing device obfuscates, using one or more privacy techniques, a portion of the user data).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, the widget including a machine learning model locally operating on the user device, the machine learning model trained on the user device based on a training data set local to the user device, retrieving one or more datum of an active user session, scoring, by the model operating on the user device, items associated with the active user session, sending, by the widget, information to a remote server, and processing, by the remote server, the information associated with the active user session without accessing the first data set from the user device, as taught by Chhibber, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. It further would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Rigney, to include the teachings of Chhibber, in order to improve transaction security while maintaining the integrity of private user information stored at edge devices (Chhibber, [0011]).
Ovick, on the other hand, teaches applying the category to a transaction record that lacks item-level data ([0347-0349] a transaction handler processes transactions to receive transaction records (301) providing information such as merchant ID and merchant category… transaction records (301) further include details about the products and/or services involved in the purchase; [0355] The transaction records (301) are aggregated to generate aggregated measurements (e.g., variable values (321)) that are not specific to a particular transaction, such as frequencies of purchases made with different merchants or different groups of merchants, the amounts spent with different merchants or different groups of merchants, and the number of unique purchases across different merchants or different groups of merchants; [0372] In FIG. 3, transaction data (109) are summarized (371) using the factor solutions and cluster solutions to generate the aggregated spending profile (341)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney and Chhibber, applying the category to a transaction record that lacks item-level data, as taught by Ovick, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. It further would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Rigney and Chhibber, to include the teachings of Ovick, in order to provide intelligence information about the spending patterns, preferences and/or trends of a user based on transaction patterns (Ovick, [0088][0346-0348]).
Regarding Claim 2, Rigney, Chhibber and Ovick teach the limitations of claim 1.
Rigney further discloses wherein the one or more datum is continuously received and includes one or more of HyperText Markup Language (HTML) code, metadata, or an item tag ([0021] This access to the repository may be an ongoing permission, such that embodiments of the invention will periodically or continuously monitor for new information; [0027] in Step 106, line items identified are analyzed to determine information about the related good or service that is associated with that line item… this step may be performed by acquiring more information about the line item by searching for the product identification number, the description, and other information on the receipt) (Examiner notes that, according to the limitation “one or more of…”, at least one of the subsequent forms of data must be present. Accordingly, Rigney ([0021][0027]) has been relied upon herein to teach wherein the data includes one or more of metadata or an item tag);
But does not explicitly disclose wherein data is received throughout the active user session.
Chhibber, on the other hand, discloses wherein data is received throughout the active user session ([0032] Application 240 receives user request 102 from user 120 and stream 104 of user data; [0040] real-time module 310 pre-processes adjusted user data 222 as it is received. For example, as new user requests 102 are received at user computing device 110A and as new data comes in from the stream 104 of user data, real-time module 310 performs pre-processing techniques).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, wherein data is received throughout the active user session, as taught by Chhibber, for the same reasons discussed above with respect to claim 1.
Regarding Claim 3, Rigney, Chhibber and Ovick teach the limitations of claim 1.
Rigney further discloses wherein the scoring is further based on one or more user interactions via the active user session that are indicative of one or more user behavior ([0051] a machine learning model will consult a database of business expenses organized by occupation and vendor to determine the probability of whether a particular receipt item is a business expense; [0052] In Step 210, the system determines a probability that the line item actually falls within the predicted category… see [0019] A “line item” is a good, service, donation, or other purpose for which money is sent);
But does not explicitly disclose interactions via the active user session.
Chhibber, on the other hand, discloses interactions via the active user session ([0030] existing user data can be updated in real-time as the user data is collected or otherwise becomes available; [0046] the recommendation engine 124 can identify user-content interactions (e.g., application usage history data, transaction history data, content viewing history data, etc.)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, interactions via the active user session, as taught by Chhibber, for the same reasons discussed above with respect to claim 1.
Regarding Claim 8, Rigney, Chhibber and Ovick teach the limitations of claim 1.
Rigney does not explicitly disclose wherein the first trained machine learning model is part of a federated learning system across a plurality of user devices.
Chhibber, on the other hand, teaches wherein the first trained machine learning model is part of a federated learning system across a plurality of user devices ([0018] User computing device 110A trains a baseline model 120 using one or more sets 114 of user data from the stream 104 of user data to generate device-trained model 130A. User computing device 110A trains baseline model 120 using one or more machine learning techniques; [0066] a server computer system receives from a plurality of user computing devices, a plurality of device-trained models and obfuscated sets of user data stored at the plurality of user computing devices, where the device-trained models are trained at respective ones of the plurality of user computing devices using respective sets of user data prior to obfuscation… see [0011] the disclosed techniques perform all or a portion of model training on edge devices rather than performing training at a central system. Performance of such training at edge devices instead of on a central server may be referred to herein as “federated learning.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, wherein the first trained machine learning model is part of a federated learning system across a plurality of user devices, as taught by Chhibber, for the same reasons discussed above with respect to claim 1.
Regarding Claim 9, Rigney, Chhibber and Ovick teach the limitations of claim 1.
Rigney further discloses wherein the method further includes adjusting a user experience based on a result of the labeling ([0053] In Step 212, the system compares this probability to a threshold… if the probability is below the threshold, the system may end the process as to that line item. For example, if the probability that the line item is a business expense is below a certain low threshold, the system may presume that it is not a business expense and not query the user for a verification).
Regarding Claim 11, Rigney discloses A system for categorization, comprising ([0047]):
a memory storing instructions ([0060]) and
a trained machine learning model trained, based on item information data and ground truth, to learn associations between an item offering and one or more categories and to output an item label category in response to an input data set ([0048-0049] Item descriptions are input into the machine learning model which cross references the line item with the user’s occupation and items in the common business expense database, then the model determines the probability that the receipt item is a business expense… this step can be accomplished by using unique parameters in a recurrent neural network seeded with a database of common business expenses tailored to the individual's occupation or industry. Parameters include user's occupation and receipt item descriptions, among other information. The network will be trained by the user, and other similar users, as business expenses identified by the network are either verified or denied by the user); and
a processor operatively connected to the memory and configured to execute the instructions to perform operations including ([0065]):
initiating a widget on a user device ([0023] embodiments of the invention may take the form of a smart phone application (app) configured to be installed on a user smart phone);
a first trained machine learning model ([0048-0049] Item descriptions are input into the machine learning model which cross references the line item with the user’s occupation and items in the common business expense database, then the model determines the probability that the receipt item is a business expense; [0051] a machine learning model will consult a database of business expenses organized by occupation and vendor to determine the probability of whether a particular receipt item is a business expense. Business expenses organized by occupation can initially be procured by the system… see further [0049] a recurrent neural network trained by the users, and other similar users);
the first trained machine learning model having been trained to predict a score of an item based on training datasets ([0049] a recurrent neural network seeded with a database of common business expenses tailored to the individual’s occupation or industry, where parameters include occupation and receipt item descriptions… trained by the users, and other similar users; [0051] a machine learning model will consult a database of business expenses organized by occupation and vendor to determine the probability of whether a particular receipt item is a business expense… see [0048] The model then determines the probability that the receipt item is a business expense);
retrieving, by the widget, a first data set from the user device, the first data set associated with one or more datum of a user ([0023] The smart phone app may therefore be able to access receipts and other purchase records across these various platforms so as to be more likely to capture relevant information; [0042] In Step 200, the item description and other available information is gathered from the receipt or other purchase record, from the user device and sent to a remote server for categorization);
scoring, by the first trained machine learning model, a first item based at least in part on the first data set ([0051] a machine learning model will consult a database of business expenses organized by occupation and vendor to determine the probability of whether a particular receipt item is a business expense; [0052] In Step 210, the system determines a probability that the line item actually falls within the predicted category. The probability may be based upon a degree of certainty that the item description matches a corresponding item description in the database of common expenses);
labeling a category for the first item based on a result of the scoring ([0053] In Step 212, the system compares this probability to a threshold… if the probability is above the threshold, the system may query the user to confirm line items; [0054] In Step 214, the system requests verification from the user that the line item was categorized correctly. The verification may present the line item and the determined categorization);
sending, by the widget, a result of said labeling to a remote store ([0057] In Step 216, the system receives the verification from the user. If the user approves of the categorization, in Step 218 the system saves the information to an expense log or other data store (such as discussed above in FIG. 1). The system may additionally save the information to the data store of common items… see [0031] data stores can be remotely accessible over internet 442); and
applying, by the remote server, based on the result of said labeling, the category to a transaction record ([0057] in Step 218 the system saves the information to an expense log or other data store (such as discussed above in FIG. 1). The system may additionally save the information to the data store of common items);
But does not explicitly disclose the widget including a machine learning model locally operating on the user device; retrieving data of an active user session; a first item associated with the active user session; sending information to a remote server; and applying the category to a transaction record associated with the active user session without accessing the first data set from the user device and that lacks item-level data.
Chhibber, on the other hand, teaches the widget including a machine learning model locally operating on the user device ([0015] User computing device 110A, in the illustrated embodiment, includes baseline model 120 and device-trained model 130A; [0031] user computing device 110A includes secure storage 212 and application 240, which in turn includes training module 250 and updated model 130);
retrieving one or more datum of an active user session ([0032] Application 240 receives user request 102 from user 120 and stream 104 of user data; [0040] real-time module 310 pre-processes adjusted user data 222 as it is received. For example, as new user requests 102 are received at user computing device 110A and as new data comes in from the stream 104 of user data, real-time module 310 performs pre-processing techniques… see [0016] The stream 104 of user data is a continuous flow of information into the user computing device 110A. This stream of data may be continuous and includes device characteristics, characteristics associated with user 120, characteristics associated with user request 102);
scoring, by the model operating on the user device, items associated with the active user session ([0019] Device-trained model 130A outputs risk score 132 for the user request 102 based on set 106 of characteristics and user computing device 110A transmits the risk score 132 to decisioning module 160. Risk score 132 indicates an amount of risk associated with user request 102 based on the set 106 of characteristics… see [0023] a first user submitting a transaction request for a wrench from a hardware store, a second user submitting a transaction request for a diamond ring from a pawn shop);
sending, by the widget, information to a remote server ([0020] In response to sending risk score 132 to system 150, user computing device 110A receives a decision 162; [0022] Server computer system 150 receives risk score 132 for user request 102 from device-trained model 130A of user computing device 110A); and
processing, by the remote server, the information associated with the active user session without accessing the first data set from the user device ([0022] Server computer system 150, in the illustrated embodiment, receives risk score 132 for user request 102 from device-trained model 130A of user computing device 110A. System 150 executes decisioning module 160 to generate a decision 162 for request 102 based on risk score 132… Decisioning module 160 may receive information specifying the type of request 102 from user computing device 110A in addition to the obfuscated user data 116 (which includes information about the user and the user's device). Decisioning module 160 selects a set of rules and heuristics for request 102 based on one or more characteristics indicated in obfuscated user data 116; [0047] Rule selection module 470 receives obfuscated user data 116 from user computing device 110A and selects a set 464 of rules from a plurality of security rules 462 (e.g., for evaluating user request 102) based on the obfuscated user data 116; [0063] the computing device obfuscates, using one or more privacy techniques, a portion of the user data).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, the widget including a machine learning model locally operating on the user device, retrieving one or more datum of an active user session, scoring, by the model operating on the user device, items associated with the active user session, sending, by the widget, information to a remote server, and processing, by the remote server, the information associated with the active user session without accessing the first data set from the user device, as taught by Chhibber, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. It further would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Rigney, to include the teachings of Chhibber, in order to improve transaction security while maintaining the integrity of private user information stored at edge devices (Chhibber, [0011]).
Ovick, on the other hand, teaches applying the category to a transaction record that lacks item-level data ([0347-0349] a transaction handler processes transactions to receive transaction records (301) providing information such as merchant ID and merchant category… transaction records (301) further include details about the products and/or services involved in the purchase; [0355] The transaction records (301) are aggregated to generate aggregated measurements (e.g., variable values (321)) that are not specific to a particular transaction, such as frequencies of purchases made with different merchants or different groups of merchants, the amounts spent with different merchants or different groups of merchants, and the number of unique purchases across different merchants or different groups of merchants; [0372] In FIG. 3, transaction data (109) are summarized (371) using the factor solutions and cluster solutions to generate the aggregated spending profile (341)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney and Chhibber, applying the category to a transaction record that lacks item-level data, as taught by Ovick, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. It further would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Rigney and Chhibber, to include the teachings of Ovick, in order to provide intelligence information about the spending patterns, preferences and/or trends of a user based on transaction patterns (Ovick, [0088][0346-0348]).
Claim 12 recites a system comprising substantially similar limitations as claim 2. The claim is rejected under substantially similar grounds as claim 2.
Regarding Claim 13, Rigney, Chhibber and Ovick teach the limitations of claim 11.
Rigney further discloses wherein the scoring is further based on one or more user behaviors ([0051] a machine learning model will consult a database of business expenses organized by occupation and vendor to determine the probability of whether a particular receipt item is a business expense; [0052] In Step 210, the system determines a probability that the line item actually falls within the predicted category… see [0019] A “line item” is a good, service, donation, or other purpose for which money is sent).
Claim 18 recites a system comprising substantially similar limitations as claim 9. The claim is rejected under substantially similar grounds as claim 9.
Claim 21 is directed to a computer-implemented method. Claim 21 recites limitations that are substantially parallel in nature to those addressed above for claim 11 which is directed towards a system. The combination of Rigney/Chhibber/Ovick teaches the limitations of claim 11 as noted above. Rigney further discloses A computer-implemented method for categorization (Rigney: [0047]). Claim 21 is therefore rejected for the reasons set forth above in claim 1 and in this paragraph.
Claims 4-7 and 14-16 are rejected under 35 U.S.C. 103 as being unpatentable over Rigney in view of Chhibber in view of Ovick, and further in view of previously cited Calinescu (US 2022/0067545 A1).
Regarding Claim 4, Rigney, Chhibber and Ovick teach the limitations of claim 1.
Rigney further discloses wherein the scoring includes generating a category for the first item ([0053] In Step 212, the system compares this probability to a threshold… if the probability is above the threshold, the system may query the user to confirm line items; [0054] In Step 214, the system requests verification from the user that the line item was categorized correctly. The verification may present the line item and the determined categorization);
But does not explicitly disclose generating a ranking of potential item categories for the first item.
Calinescu, on the other hand, discloses generating a ranking of potential item categories for the first item ([0045] the confidence score is determined based on a comparison between the label score for the highest scored label and the label score for the second highest scored label; [0054-0055] the taxonomy module 303 receives 510 information associated with the content item to assign taxonomy labels… the first-level model determines a label score for each first-level label 110 in the first level class 150A… based on the determined label scores, the label selection module 315 selects 525 a first-level label 110 to be associated with the content item by selecting the first-level label with the highest score).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, Chhibber and Ovick, generating a ranking of potential item categories for the first item, as taught by Calinescu, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. It further would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Rigney, Chhibber and Ovick, to include the teachings of Calinescu, in order to provide consistent labels across content publishers (Calinescu, [0002]).
Regarding Claim 5, Rigney, Chhibber, Ovick and Calinescu teach the limitations of claim 4.
Rigney further discloses verifying that said scoring results in at least one potential item category which satisfies a pre-defined threshold ([0053] if the probability is above the threshold, the system may determine that the line item is worthy of query to the user. In other embodiments, the system may query the user to confirm all line items; [0054] In Step 214, the system requests verification from the user that the line item was categorized correctly. The verification may present the line item and the determined categorization).
Regarding Claim 6, Rigney, Chhibber and Ovick teach the limitations of claim 1.
Rigney further discloses wherein said labeling comprises: selecting for the first item a first item category ([0051] a machine learning model will consult a database of business expenses organized by occupation and vendor to determine the probability of whether a particular receipt item is a business expense; [0052] In Step 210, the system determines a probability that the line item actually falls within the predicted category. The probability may be based upon a degree of certainty that the item description matches a corresponding item description in the database of common expenses);
But does not explicitly disclose selecting a first item category, the first item category being a highest scoring potential item category of a plurality of potential item categories.
Calinescu, on the other hand, discloses selecting a first item category, the first item category being a highest scoring potential item category of a plurality of potential item categories ([0054-0055] the taxonomy module 303 receives 510 information associated with the content item to assign taxonomy labels… the first-level model determines a label score for each first-level label 110 in the first level class 150A… based on the determined label scores, the label selection module 315 selects 525 a first-level label 110 to be associated with the content item by selecting the first-level label with the highest score).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, Chhibber and Ovick, selecting a first item category, the first item category being a highest scoring potential item category of a plurality of potential item categories, as taught by Calinescu, for the same reasons discussed above with respect to claim 4.
Regarding Claim 7, Rigney, Chhibber and Ovick teach the limitations of claim 1.
Rigney does not explicitly disclose wherein said labeling comprises: verifying that no potential item category satisfies a pre-defined threshold; and selecting for the first item a default item category, the default item category being associated with none of the potential item categories.
Calinescu, on the other hand, discloses wherein said labeling comprises: verifying that no potential item category satisfies a pre-defined threshold ([0055] If there is a sufficient confidence level (e.g., the confidence level is above a threshold value), the first-level label with the highest score is selected and assigned to the content item. However, if the confidence level is below a threshold value, the content item may be sent for manual review); and
selecting for the first item a default item category, the default item category being associated with none of the potential item categories ([0088] if the confidence score C is not greater than the threshold value C.sub.th, the content item is flagged 850 for manual review… see [0009] if the confidence score is below the threshold value, the content item is sent for manual classification) (Examiner notes that manual review and classification is not associated with any of the potential item categories generated by the content classification system of Calinescu).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, Chhibber and Ovick, wherein said labeling comprises: verifying that no potential item category satisfies a pre-defined threshold; and selecting for the first item a default item category, the default item category being associated with none of the potential item categories, as taught by Calinescu, for the same reasons discussed above with respect to claim 4.
Claim 14 recites a system comprising substantially similar limitations as claim 4. The claim is rejected under substantially similar grounds as claim 4.
Claim 15 recites a system comprising substantially similar limitations as claim 5. The claim is rejected under substantially similar grounds as claim 5.
Claim 16 recites a system comprising substantially similar limitations as claim 6. The claim is rejected under substantially similar grounds as claim 6.
Claims 10 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Rigney in view of Chhibber in view of Ovick, and further in view of previously cited O’Neil (US 2003/0105711 A1).
Regarding Claim 10, Rigney, Chhibber and Ovick teach the limitations of claim 9.
Rigney does not explicitly disclose wherein adjusting the user experience includes binding a first instrument of the user to an instrument item category such that the first instrument is unable to complete a workflow for an item outside of said instrument item category.
O’Neil, on the other hand, discloses wherein adjusting the user experience includes binding a first instrument of the user to an instrument item category such that the first instrument is unable to complete a workflow for an item outside of said instrument item category ([0013] a transaction card 110 issued to a customer 120 by an issuer 130 usable in transactions via a POS terminal or similar device (not illustrated) of a merchant 140 in a pre-qualified category; [0022] when a customer 120 swipes the customer’s card 110 at a POS terminal of the merchant 120 in connection with a transaction, the transaction data is processed, and the category of the goods/services is interrogated to determine whether or not the transaction can be posted on the pre-qualified category card account; [0023] if the transaction does not fall within the pre-qualified category of goods/services, approval of the transaction is denied).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, Chhibber and Ovick, wherein adjusting the user experience includes binding a first instrument of the user to an instrument item category such that the first instrument is unable to complete a workflow for an item outside of said instrument item category, as taught by O’Neil, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. It further would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Rigney, Chhibber and Ovick, to include the teachings of O’Neil, in order to provide pre-defined parameters in which a customer is permitted to perform a transaction for a particular card (O’Neil, [0008][0023-0024]).
Claim 19 recites a system comprising substantially similar limitations as claim 10. The claim is rejected under substantially similar grounds as claim 10.
Claims 17 is rejected under 35 U.S.C. 103 as being unpatentable over Rigney in view of Chhibber in view of Ovick, and further in view of newly cited U.S Patent No. 6,334,110 B1 to Walter et al., hereinafter Walter.
Regarding Claim 17, Rigney, Chhibber and Ovick teach the limitations of claim 11.
Rigney does not explicitly disclose further comprising: injecting code into a browser associated with the first item, wherein the code includes one or more of an additional item offering, relevant item information, or an adjustment to a layout of a webpage associated with the first item.
Walter, on the other hand, teaches further comprising: injecting code into a browser associated with the first item, wherein the code includes one or more of an additional item offering, relevant item information, or an adjustment to a layout of a webpage associated with the first item ([Col 4 Ln 44-Col 5 Ln 53] gathering a temporal browsing/buying behavior report for a customer in order to create a temporal profile; [Col 6 Ln 12-31] based on merchandise in the user’s profile, the invention performs a matching/clustering algorithm on Willard's profile to determine if there are other people similar to Willard in their mix of product browsing and buying behaviors. These sets of people with like behaviors are called temporal segments; [Col 7 Ln 10-20] when Willard or anyone else from his temporally defined community enters the web store at the right time, the system dynamically generates an advertisement about the skiers baby carrier that is showed to the customer. If the customer browses this information, or better yet, buys the product advertised, this is viewed as a successful advertising event).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include in the system, as taught by Rigney, Chhibber and Ovick, further comprising: injecting code into a browser associated with the first item, wherein the code includes one or more of an additional item offering, relevant item information, or an adjustment to a layout of a webpage associated with the first item, as taught by Walter, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable. It further would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Rigney, Chhibber and Ovick, to include the teachings of Walter, in order to better predict the effectiveness of messaging in targeting marketing campaigns based on customer profiles, and analyze the behavior of customers in the financial industry (Walter, [Col 2 Ln 6-16][Col 8 Ln 27-36]).
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 ZACHARY R DONAHUE whose telephone number is (571)272-5850. The examiner can normally be reached M-F 8a-5p.
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, Marissa Thein can be reached at (571) 272-6764. 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.
/ZACHARY RYAN DONAHUE/Examiner, Art Unit 3689
/MARISSA THEIN/Supervisory Patent Examiner, Art Unit 3689
Prosecution Insights