DETAILED ACTION
This action is in response to the filing on 12/15/2025. Claims 1, 3-4, 6, 8, 11, and 19-33, are pending and have been considered below.
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 .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 11/20/2025 has been entered.
Claim Objections
Claims 1, 23, 25, and 29 objected to because of the following informalities:
Claims 1, 25, and 29 recite “inserting the training test data an iterative training and testing loop to predict a target variable;”, should recite -- inserting the training test data into an iterative training and testing loop to predict a target variable; -- (emphasis added).
Claim 23, line 1, recites “wherein the at least one processor processor:”, should recite -- wherein the at least one processor: --.
Appropriate correction is required.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 3-4, 6, 8, 11, 19-20, and 23-32 are rejected under 35 U.S.C. 103 as being unpatentable over Zhao (US 11,436,119 B1, first cited in the IDS filed 02/01/2024), hereinafter Zhao, in view of Porter et al. (US 11,611,653 B1), hereinafter Porter, and further in view of Joa et al. (US 8,290,951 B1), hereinafter Joa, and further in view of Simpson et al. (US 2008/0021813 A1), hereinafter Simpson, and further in view of Yeri et al. (US 2013/0325713 A1), hereinafter Yeri, and further in view of Rozin et al. (US 10,943,217 B1), hereinafter Rozin, and further in view of Guminy et al. (US 2014/0019225 A1), hereinafter Guminy.
Regarding claim 1, Zhao teaches A computing system, comprising (A data management system predicts whether users will continue using the data management system. [see Zhao, Abstract]):
an interaction database including data for interactions between users of an entity and nodes over multiple interaction entity channels, wherein the data in the interaction database includes (Zhao discloses a user data management database containing data related to services provided to users [see Zhao Col. 14, lines 15-25]) information indicating types of actions that the users perform (Each of the data fields represents a particular type of activity or action that can occur in relation to an account of the user with the data management system during a given period of time. [see Zhao, Col. 11, lines 5-8]), and action history of the users (There are N user activity vectors 211 each representing the activity of the user with the data management system during a respective period of time T.sub.1-N. Thus, there are N periods of time in the example FIG. 3. [see Zhao, Col. 10, line 67–Col. 11, line 4]);
a master database including identification information of the users (The user data management database 608 includes the user data management data 622. The user data management data 622 can include user data management data for all of the users of the data management system 602. Thus, the user data management database 608 can include a vast amount of data related to the data management services provided to users. In one embodiment, when the user utilizes the user interface module 606 to view interface content data 620, the interface content data 620 includes user data management data 622 related to the user as retrieved from the user data management database 608. [see Zhao, Col. 14, lines 15-25]);
a back-end server operatively coupled with the interaction database and the master database, said back-end server including: at least one processor; a communication interface communicatively coupled to the at least one processor; and a memory device storing executable code that, when executed, causes the processor to: (The data management system 602 includes computing resources 614. The computing resources 614 include processing resources 630 and memory resources 632. The processing resources 630 include one or more processors. The memory resources 632 include one or more memories configured as computer readable media capable of storing software instructions and other data. The processing resources 630 are capable of executing software instructions stored on the computer readable media. In one embodiment, the various components, modules, databases, and engines of the data management system utilize the computing resources 614 to assist in performing their various functions. Alternatively, or additionally, the various components, modules, databases, and engines can utilize other computing resources. [see Zhao, Col. 16, lines 4-18]);
collect interaction data from the multiple interaction entity channels between the users and the nodes; store the collected interaction data in the interaction database; collect user data corresponding to the users; store the collected user data in the master database (Zhao discloses a user data management database containing data related to services provided to users and user management data related to the user [see Zhao Col. 14, lines 15-25]);
access both the interaction database and the master database to process the stored interaction data and the stored user data, the stored interaction data and the stored user data being ingested data, (Zhao discloses a user data management database containing data related to services provided to users and user management data related to the user, wherein the user utilizes a user interface to view the data [see Zhao Col. 14, lines 15-25]);
train, using training test data of the preprocessed ingested data the machine learning model to predict a likelihood of user attrition and trigger a retention action, the training including: (The machine learning process trains the recurrent neural network and the deep learning cross network to predict attrition and to predict what steps should be taken to retain a user. The data management system then takes the recommended actions to retain the user. [see Zhao, Col. 2, lines 63-67]);
inserting the training test data an iterative training and testing loop to predict a target variable; and repeatedly predicting a value for the target variable during a plurality of iterations of the training and testing loop, each iteration of the plurality of iterations having differing weights applied to one or more nodes of the machine learning model, each of the differing weights being updated with each respective iteration of the training and testing loop to reduce an error in predicting the value for the target variable for respective iteration relative to the error in predicting the value for the target variable for remaining iterations (Zhao discloses an iterative training and testing loop for training the model to predict user retention, comprising calculating an error between prediction and labels then adjusting nodes and relationships between data values based on the error until the model can consistently generate predictions [see Zhao, Col. 15, lines 18-37]);
predict, via the machine learning model, a likelihood of user attrition for a given user, the likelihood being based on a probability of the user attrition; and trigger, based on the likelihood of the user attrition exceeding a predetermined threshold, the retention action (The machine learning process trains the recurrent neural network and the deep learning cross network to predict attrition and to predict what steps should be taken to retain a user. The data management system then takes the recommended actions to retain the user. [see Zhao, Col. 2, lines 63-67]).
It would have been obvious to one of ordinary skill, in the art at the time before the effective filing date of the invention to incorporate deploy the trained machine learning model and predict, via the deployed machine learning model because although Zhao does not explicitly disclose deploying the model to the user retention system, they disclose the system using the trained model [see Zhao, Col. 6, lines 58-67]. Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date to deploy the model and to use the deployed model.
However, Zhao fails to teach wherein the data includes on-line presence data of the users, places that the users have visited inside the entity, types of accounts that the users have with the entity, balance levels of user accounts, whether deposits have been made into accounts, whether the users have made withdrawals from their accounts, how long the users have been with the entity, how often the users call a user help center, and whether the users have filed any complaints; and perform preprocessing on the data, the preprocessing comprising (i) performing data cleaning on the ingested data, (ii) transforming the ingested data from a plurality of data formats into a standardized training format for training a machine learning model.
In the same field of endeavor, Porter teaches:
perform preprocessing on the data, the preprocessing comprising (i) performing data cleaning on the ingested data, (ii) transforming the ingested data from a plurality of data formats into a standardized training format for training a machine learning model (Porter discloses cleaning and formatting the ingested data records [see Porter, Col. 7, lines 59-61] by performing data preprocessing on the ingested data and transforming the data from a plurality of original formats into a common standard format that can be used for training the machine learning model [see Porter, Col. 6, lines 28-36 and Col. 17, lines 19-29 and FIG. 3]).
It would have been obvious to one of ordinary skill, in the art at the time before the effective filing date of the invention to incorporate perform preprocessing on the data, the preprocessing comprising (i) performing data cleaning on the ingested data, (ii) transforming the ingested data from a plurality of data formats into a standardized training format for training a machine learning model as suggested in Porter into Zhao because both methods are directed to machine learning in user retention environments (see Zhao, Abstract; see Porter, Col. 4, lines 36-38). Incorporating the teachings of Porter into Zhao would format data according to a number of data formats into a standardized data format that the system can use (see Porter, Col. 6, lines 28-36 and Col. 17, lines 19-29, and FIG. 3).
However, the combination of Zhao and Porter fails to teach wherein the data includes on-line presence data of the users, places that the users have visited inside the entity, types of accounts that the users have with the entity, balance levels of user accounts, whether deposits have been made into accounts, whether the users have made withdrawals from their accounts, how long the users have been with the entity, how often the users call a user help center, and whether the users have filed any complaints.
In the same field of endeavor, Joa teaches:
wherein the data and information includes whether the users have filed any complaints (In accordance with various aspects of the invention, methods, computer-readable media, and apparatuses are disclosed in which data is bridged from business support units, e.g., call centers and marketing operations, with a data warehouse to augment and enrich pre-existing customer information. Bridged data may include audio data, image data, category of call or customer interaction (e.g., account servicing, account closure, and complaints), and contextual analysis keywords. [see Joa, Col. 3, lines 7-14]).
It would have been obvious to one of ordinary skill, in the art at the time before the effective filing date of the invention to incorporate wherein the data and information includes whether the users have filed any complaints as suggested in Joa into the combination of Zhao, Simpson, Yeri, and Rozin because both systems disclose databases for data management (see Zhao, Col. 13, lines 43-57; see Joa, Col. 8, lines 19-24). Incorporating the teaching of Joa into the combination of Zhao and Porter would improve pre-existing data integration methods and may create a secure, accurate, and replicable data bridge to join data from a data warehouse to data contained in servers located in business support units (see Joa, Col. 11, lines 5-8).
However, the combination of Zhao, Porter, and Joa fails to teach wherein the data includes on-line presence data of the users, places that the users have visited inside the entity, types of accounts that the users have with the entity, balance levels of user accounts, whether deposits have been made into accounts, whether the users have made withdrawals from their accounts, how long the users have been with the entity, how often the users call a user help center.
In the same field of endeavor, Simpson teaches:
wherein the data includes on-line presence data of the users, types of accounts that the users have with the entity, and how long the users have been with the entity (In various embodiments, the components used in the retention score may include one or more of the following components (and/or other desired components): whether or not the account has automated clearing house (ACH) credits; whether or not an account has ACH debits; number of ACH debits/credits per month; whether or not Internet banking is used; number of Internet banking logons per month; whether or not electronic bill payment is used; number of bills paid electronically per month; whether or not the customer uses a voice response unit (VRU) to check the account; number of VRU accesses per month; whether or not the customer has multiple accounts (loans, checking, savings, certificates of deposit (CDs)), types of accounts; numbers of accounts; age of the account; number of transactions per month; ties to a pay-day lender; whether or not the account has debit card transactions; number of debit card transactions per month; and type of debit card transactions (signature or personal identification number (PIN) transactions). [see Simpson, para. 33]).
It would have been obvious to one of ordinary skill, in the art at the time before the effective filing date of the invention to incorporate wherein the data includes on-line presence data of the users, types of accounts that the users have with the entity, and how long the users have been with the entity as suggested in Simpson into the combination of Zhao, Porter, and Joa because both methods use a plurality of data and information from a database to predict the likelihood of retention (see Zhao, Abstract; see Simpson, para. 27). Thus it would have been obvious to one of ordinary skill in the art at the time of invention, to incorporate the data used to predict user retention as suggested in Simpson, to achieve the predictable result of wherein the data includes types of actions that users perform, action history of the users, on-line presence data of the users, types of accounts that the users have with the entity, and how long the users have been with the entity.
However, the combination of Zhao, Porter, Joa, and Simpson fails to teach wherein the data includes places that the users have visited inside the entity, balance levels of user accounts, whether regular deposits have been made into the users accounts, whether the users have made large withdrawals from their accounts, how often the users call a user help center.
In the same field of endeavor, Yeri teaches:
wherein the data includes whether regular deposits have been made into the users accounts, (FIG. 7 is a flowchart providing an overview of a system and method 700 for retaining users in their existing relationship with a financial institution are provided. Based on past and current account activity, a decrease in account activity such as a slowdown or stoppage of direct deposits, reductions in deposit patterns, or slowdown in payments using specific accounts is identified. [see Yeri, para. 78; FIG. 14A-14B]; A "deposit reduction 1" trigger (DR1) runs on the 8.sup.th calendar day of every month and identifies a pattern for the beginning of the current month and the end of the previous month incoming deposits while a "deposit reduction 2" trigger (DR2) runs on the 22.sup.nd calendar day of every month and identifies end of the current month incoming deposits. The difference of the total value or total number of deposits for DR1 and the total value or total number of the deposits for DR2 can be calculated to determine that the total value or total number of the DR1 deposits is greater than the total value or total number of the DR2 deposits. Moreover, the pattern identified in DR1 may identify a pattern of decreased activity where the value or number of deposits that occur at the beginning of the current month are lower than the value or number of deposits that occur at the end of the previous month. [see Yeri, para. 81; FIG. 14A-14B]) and whether the users have made large withdrawals from their accounts (FIG. 9 is a flowchart providing an overview of a system and method 900 for reviewing accounts to enhance user relationships and prevent account loss. Account activity is reviewed in depth to timely identify account activity that signifies "off-us" transactions. For example, off-us transactions include a large withdrawal, opening a new account with a competitor or other third party entity, making a third party credit card payment, or any other account activity that signifies third party transactions. Timely identification of off-us activity can be used to avoid losing the user to a competitor and enhance the relationship with the user. [see Yeri, para. 96]; A "Large withdrawal" trigger (LWD) of Trigger Table 3 (FIG. 14E) includes outbound transactions each having an amount greater than $2,500, where the transaction amount is the product of the average total amount of all transactions of the previous six months and a predetermined constant, e.g., 2.5, and where the tenure of the one or more accounts is greater than 90 days. [see Yeri, para. 99; FIG. 14E]).
It would have been obvious to one of ordinary skill, in the art at the time before the effective filing date of the invention to incorporate wherein the data includes whether regular deposits have been made into the users accounts, and whether the users have made large withdrawals from their accounts as suggested in Yeri into the combination of Zhao and Simpson because both methods are directed to retaining users (see Zhao, Abstract; see Yeri, Abstract). Incorporating the teaching of Yeri into the combination of Zhao, Porter, Joa, and Simpson would be a cost effective and profitable strategy because it costs must less to retain customers than to acquire new ones (see Yeri, para. 78) and can be used to avoid losing the user to a competitor and enhance the relationship with the user (see Yeri, para. 96).
However, the combination of Zhao, Porter, Joa, Simpson, and Yeri fails to teach wherein the data includes places that the users have visited inside the entity, balance levels of user accounts, how often the users call a user help center.
In the same field of endeavor, Rozin teaches:
wherein the data includes balance levels of user accounts (Users 415 may have various numbers and types of accounts 432 with FIs 435 including a checking account, savings account, money market account, credit card account, an account for a loan such as a mortgage, car loan or education loan. Such accounts 432 may have different types of account data 433 depending on the type of account 432. For example, account data 433 may include balances, transactions, payments, pending payments or payments in process and deposits and account summaries. [see Rozin, Col. 9, lines 18-26]).
It would have been obvious to one of ordinary skill, in the art at the time before the effective filing date of the invention to incorporate wherein the data includes balance levels of user accounts as suggested in Rozin into the combination of Zhao, Simpson, and Yeri because both methods are directed to financial data management systems (see Zhao, Col. 13, lines 43-57; see Rozin, Abstract). Incorporating the teaching of Rozin into the combination of Zhao, Simpson, and Yeri would identify recurring or repeating online banking actions or patterns (see Rozin, Col. 1, lines 18-22).
However, the combination of Zhao, Porter, Joa, Simpson, Yeri, and Rozin fails to teach wherein the data includes places that the users have visited inside the entity, how often the users call a user help center.
In the same field of endeavor, Guminy teaches:
wherein the data includes places that the users have visited inside the entity, and how often the users call a user help center (The behavior collector module 218 also records the behavior of the customer/user himself or herself in various touch points. This information may include how many times the customer/user calls a call center, what physical store the customer/user visits most often, information regarding whether the customer/user ever shops through mobile devices, etc. [see Guminy, para. 48]).
Joa discloses bridging data from business support units, e.g., call centers and marketing operations, with a data warehouse to augment and enrich pre-existing customer information (see Joa, Abstract). While Guminy discloses integrating multiple "entity access channels" also known as "touch points" to correlate user influence of support by others within social networks across different business interaction venues (see Guminy, para. 17) and a plurality of touch points such as a call center (see Guminy, para. 31; FIG. 1). Thus, it would have been obvious to incorporate the collected data from the multiple touch points as disclosed by Guminy with the data bridging techniques of Joa because incorporating the data from touch points of Guminy would allow businesses to learn more about their customers and the effectiveness of their marketing initiatives, and allows businesses to learn more about the effectiveness of social networks (see Guminy, para. 14) and would extend the utility of Joa for improving pre-existing data integration methods and may create a secure, accurate, and replicable data bridge to join data from a data warehouse to data contained in servers located in business support units (see Joa, Col. 11, lines 5-8).
Regarding claim 4, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 1 and further teaches:
wherein the probability is a percentage estimate of user attrition (In one example, if the analysis model 102 predicts an 80% probability of retention, then the analysis model 102 predicts a 20% probability of attrition. [see Zhao, Col. 16, lines 30-33]).
Regarding claim 6, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 1 and further teaches:
further comprising a former user information source that stores data and information about former users that have left the entity, wherein the machine learning model uses the data and information from the former user information source (The machine learning processes utilize historical user data that includes time dependent user data broken up into periods of time and static user data. Machine learning processes also include labels indicating whether the historical users remained paying customers or became converted at selected dates subsequent to the historical time dependent user data time periods. [see Zhao, Col. 6, lines 12-18]).
Regarding claim 11, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 1 and further teaches:
wherein the entity is a bank and the users are clients of the bank (The customer account database 10 may be maintained by the financial institution or a financial institution service provider, and may be updated as new accounts are opened and/or customer transactions are processed. For example, the customer account database 10 may include data identifying each account, as well as account activity data such as deposits, withdrawals, checks cleared, interest earned or charged, fees charged, etc. The account data may also include other information, such as the overdraft score, retention score, and/or profit score for each account. For brevity, the financial institution will be referred to in this description as a "bank", but any financial institution may implement the system described herein in various embodiments. [see Simpson, para. 25]).
Regarding claim 3, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 11 and further teaches:
wherein the entity channels include a website, mobile applications (Possible recommended actions can include sending an email to the user, prompting the user to try a particular feature of the bookkeeping system, offering the user a discounted or promotional rate, recommending the user add a mobile bookkeeping application associated with the bookkeeping system, or recommending the user visit a website of the bookkeeping system. [see Zhao, Col. 10, lines 39-45]), bank branch (In some embodiments, statistical analysis may be performed periodically on the account database 10 to correlate components to the retention experience at a particular bank or bank branch, and the analysis may be performed by the retention scorer 14. [see Simpson, para. 28]), online activities (Possible recommended actions can include sending an email to the user, prompting the user to try a particular feature of the bookkeeping system, offering the user a discounted or promotional rate, recommending the user add a mobile bookkeeping application associated with the bookkeeping system, or recommending the user visit a website of the bookkeeping system. [see Zhao, Col. 10, lines 39-45]) and service center calls (whether the user has called customer support to receive assistance [see Zhao, Col. 11, lines 17-18]).
Regarding claim 19, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 1 and further teaches:
wherein the machine learning model employs at least one neural network (During the machine learning process, the behavior of the user for each previous time period is represented by a respective vector and fed into a recurrent neural network. The recurrent neural network processes the vectors and generates a series of processed vectors. The processed vectors, as well as static vectors that represent the static characteristics of the user, are passed to a deep learning cross network that concatenates the processed vectors with the static vectors and then further processes the concatenated vectors. The deep learning cross network then outputs a prediction regarding whether the user is likely to abandon or continue with the data management system during a selected time period. The machine learning process trains the recurrent neural network and the deep learning cross network to predict attrition and to predict what steps should be taken to retain a user. [see Zhao, Col. 2, lines 51-67]).
Regarding claim 8, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 19 and further teaches:
wherein the at least one neural network is a convolutional neural network (CNN) or a recurrent neural network (RNN) (During the machine learning process, the behavior of the user for each previous time period is represented by a respective vector and fed into a recurrent neural network. The recurrent neural network processes the vectors and generates a series of processed vectors. The processed vectors, as well as static vectors that represent the static characteristics of the user, are passed to a deep learning cross network that concatenates the processed vectors with the static vectors and then further processes the concatenated vectors. The deep learning cross network then outputs a prediction regarding whether the user is likely to abandon or continue with the data management system during a selected time period. The machine learning process trains the recurrent neural network and the deep learning cross network to predict attrition and to predict what steps should be taken to retain a user. [see Zhao, Col. 2, lines 51-67]).
Regarding claim 20, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 1 and further teaches:
wherein the at least one processor collects former client data and information about former clients that have left the entity (The data management system 602 includes computing resources 614. The computing resources 614 include processing resources 630 and memory resources 632. The processing resources 630 include one or more processors. The memory resources 632 include one or more memories configured as computer readable media capable of storing software instructions and other data. The processing resources 630 are capable of executing software instructions stored on the computer readable media. In one embodiment, the various components, modules, databases, and engines of the data management system utilize the computing resources 614 to assist in performing their various functions. Alternatively, or additionally, the various components, modules, databases, and engines can utilize other computing resources. [see Zhao, Col. 16, lines 4-18]).
Regarding claim 23, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 1 and further teaches:
wherein the at least one processor processor: collects (a) additional interaction data from multiple interaction entity channels between the users and the nodes, and (b) additional user data (Porter discloses continuously collecting new data and retraining the machine learning model [see Porter, Col. 3, lines 36-41]. Thus, it would have been obvious to collect interaction data from the channels between users and the nodes and user data as disclosed by Zhao [see Zhao, Col. 14, lines 15-25] continuously);
based on the collecting, retrains the machine learning model (Porter discloses continuously collecting new data and retraining the machine learning model [see Porter, Col. 3, lines 36-41]).
Regarding claim 24, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 1 and further teaches:
wherein the predicting of the likelihood of user attrition reduces an amount of trash data stored to system databases (A data management system predicts whether users will continue using the data management system. The data management system includes an analysis model that generates user retention prediction data based on time dependent user data and static user data. The analysis model also generates recommended actions to be taken by the data management system to increase the probability of retaining the user. [see Zhao, Abstract]).
Regarding claim 25, claim 25 contains substantially similar limitations to those found in claim 1 above. Consequently, claim 25 is rejected for the same reasons.
Regarding claim 29, claim 29 contains substantially similar limitations to those found in claim 1. Therefore it is rejected for the same reason as claim 1 above. Additionally, the combination of Zhao, Porter, Joe, Simpson, Yeri, Rozin, and Guminy further teaches:
A non-transitory computer-readable storage medium, the computer-readable storage medium including instructions that when executed by a processor, cause the processor to: (The data management system 602 includes computing resources 614. The computing resources 614 include processing resources 630 and memory resources 632. The processing resources 630 include one or more processors. The memory resources 632 include one or more memories configured as computer readable media capable of storing software instructions and other data. The processing resources 630 are capable of executing software instructions stored on the computer readable media. [see Zhao, Col. 16, lines 4-12]).
Regarding claims 26 and 30, claims 26 and 30 contains substantially similar limitations to those found in claim 11 above. Consequently, claims 26 and 30 are rejected for the same reasons.
Regarding claims 27 and 31, claims 27 and 31 contains substantially similar limitations to those found in claim 23 above. Consequently, claims 27 and 31 are rejected for the same reasons.
Regarding claims 28 and 32, claims 28 and 32 contains substantially similar limitations to those found in claim 24 above. Consequently, claims 28 and 32 are rejected for the same reasons.
Claims 21 and 33 are rejected under 35 U.S.C. 103 as being unpatentable over Zhao (US 11,436,119 B1, first cited in the IDS filed 02/01/2024), hereinafter Zhao, in view of Porter et al. (US 11,611,653 B1), hereinafter Porter, and further in view of Joa et al. (US 8,290,951 B1), hereinafter Joa, and further in view of Simpson et al. (US 2008/0021813 A1), hereinafter Simpson, and further in view of Yeri et al. (US 2013/0325713 A1), hereinafter Yeri, and further in view of Rozin et al. (US 10,943,217 B1), hereinafter Rozin, and further in view of Guminy et al. (US 2014/0019225 A1), hereinafter Guminy, as applied in claim 1 above, and further in view of Hart et al. (US 10,672,005 B1), hereinafter Hart, and further in view of Ally Bank, Press Release "Ally Bank eliminates all overdraft fees, ending centuries-old industry practice and lifting consumer burden", hereinafter Ally Bank.
Regarding claim 21, the combination of Zhao, Porter, Joa, Simpson, Yeri, Rozin, and Guminy as applied in claim 1 above teaches all the limitations of claim 21 and further teaches:
wherein the retention action is associated with retaining the user as a client of the entity by offering education about entity features and programs that the user may be interested in or benefit from (Zhao discloses the system includes actions that are selected to improve the probability of retaining a user [see Zhao, Col. 5, lines 44-56] such as prompting the user to try a particular feature [see Zhao, Col. 5, lines 44-56] or any other possible action such as offering incentives [see Zhao, Col. 12, lines 43-53].).
However, the combination of Zhao, Simpson, Yeri, Rozin, Joa, and Guminy fails to teach wherein the retention action includes cash back on a credit card, reduction in fees, avoidance of overdraft fees, or addressing false fraud occurrences on a credit card.
In the same field of endeavor, Hart teaches:
wherein the action includes cash back on a credit card, education about entity features and programs that the user may be interested in or benefit from, and addressing false fraud occurrences on a credit card (Hart discloses performing one or more actions in response to a false fraud occurrence on a credit card to minimize damaged customer relations and prevent losing the user due to the false fraud transaction [see Hart, Col. 8, lines 47-53; Col. 9, lines 16-21]. The one or more actions include providing communication to the user about the false fraud [see Hart, Col. 8, lines 47-53] and increasing a cash back offer [see Hart, Col. 9, lines 16-21]).
It would have been obvious to one of ordinary skill, in the art at the time before the effective filing date of the invention to incorporate wherein the retention action consists of better cash back on a credit card, education about entity features and programs that the user may be interested in or benefit from, and addressing false fraud occurrences on a credit card as suggested in Hart into the combination of Zhao, Simpson, Yeri, Rozin, Joa, and Guminy because both methods are directed to financial data management systems (see Zhao, Col. 13, lines 43-57; see Hart, Col. 3, lines 14-17). Incorporating the teaching of Hart into the combination of Zhao, Simpson, Yeri, Rozin, Joa, and Guminy would minimizes damaged customer relations with the user (see Hart, Col. 8, lines 47-53) and prevents losing the user due to the denied transaction and improves customer service (see Hart, Col. 9, lines 16-21).
However, the combination of Zhao, Simpson, Yeri, Rozin, Joa, Guminy, and Hart fails to teach wherein the retention action includes reduction in fees, and avoiding overdraft fees.
In the same field of endeavor, Ally Bank teaches:
wherein the retention action includes reduction in fees, avoiding overdraft fees (Ally Bank discloses eliminating overdraft fees for all accounts and that all customers are eligible [see Ally Bank, pg. 1, para. 1], and that for over a decade they have not charged maintenance fees and ACH transfer fees, and have a no-fee ATM network [see Ally Bank, pg. 1, para. 6]).
It would have been obvious to one of ordinary skill, in the art at the time before the effective filing date of the invention to incorporate wherein the retention action consists of reduction in fees, and avoiding overdraft fees as suggested in Ally Bank into the combination of Zhao, Simpson, Yeri, Rozin, Joa, and Guminy because both systems are directed to financial management systems (see Zhao, Col. 13, lines 43-57; see Ally Bank, pg. 1, para. 1). It would have been obvious to one of ordinary skill in the art to eliminate overdraft fees as a retention action because "overdraft fees are a pain point for many consumers" (see Ally Bank, pg. 1, para. 2) and "Eliminating these fees helps keep people from falling further behind and feeling penalized as they catch up" (see Ally Bank, pg. 1, para. 3).
Regarding claim 33, claim 33 contains substantially similar limitations to those found in claim 21 above. Consequently, claim 33 is rejected for the same reasons.
Response to Amendment
Applicant’s amendment cancelling claim 22 has overcome the rejection claim 22 under 35 U.S.C. 112(a), and the rejection is respectfully withdrawn.
Applicant’s amendments to the claims have overcome the rejection of claims 1, 3-4, 6, 8, 11, and 19-21 under 35 U.S.C. 112(b), and the rejections are respectfully withdrawn.
Response to Arguments
Applicant’s arguments traversing the rejections of claims 1, 3-4, 6, 8, 11, and 19-22 under 35 U.S.C. 101 have been fully considered and are persuasive, the rejections are respectfully withdrawn.
Applicant’s arguments traversing the rejections of claims 1, 3-4, 6, 8, 11, and 19-22 under 35 U.S.C. 103 have been fully considered and are not persuasive. Applicant argues the prior art of record (Zhao, Simpson, Yeri, Rozin, Joa, and Guminy) fails to suggest the newly amended limitation of claim 1 directed to training, that the rejection relies upon impermissible hindsight which requires knowledge of Applicant’s disclosure to arrive at the claimed subject matter, and that the asserted benefits that would result from combing the cited references lack a sufficient rationale. Examiner respectfully disagrees.
With respect to Applicant’s argument that the prior art of record fails to suggest “train, using training test data of the preprocessed ingested data the machine learning model to predict a likelihood of user attrition and trigger a retention action, the training including: inserting the training test data an iterative training and testing loop to predict a target variable; and repeatedly predicting a value for the target variable during a plurality of iterations of the training and testing loop, each iteration of the plurality of iterations having differing weights applied to one or more nodes of the machine learning model, each of the differing weights being updated with each respective iteration of the training and testing loop to reduce an error in predicting the value for the target variable for a respective iteration relative to the error in predicting the value for the target variable for remaining iterations.”, Zhao as mapped in the above 35 U.S.C. 103 section was found to render obvious the claim limitation in view of at least Col. 15, lines 18-37.
In response to applicant's argument that the examiner's conclusion of obviousness is based upon improper hindsight reasoning, it must be recognized that any judgment on obviousness is in a sense necessarily a reconstruction based upon hindsight reasoning. But so long as it takes into account only knowledge which was within the level of ordinary skill at the time the claimed invention was made, and does not include knowledge gleaned only from the applicant's disclosure, such a reconstruction is proper. See In re McLaughlin, 443 F.2d 1392, 170 USPQ 209 (CCPA 1971).
In response to applicant's argument that the examiner has combined an excessive number of references, reliance on a large number of references in a rejection does not, without more, weigh against the obviousness of the claimed invention. See In re Gorman, 933 F.2d 982, 18 USPQ2d 1885 (Fed. Cir. 1991).
With respect to Applicant’s argument that the asserted benefits from combining the references lack a sufficient rationale underpinning as required by KSR v. Teleflex and MPEP § 2143, as the proposed benefits appear generic and not specifically tied to the claimed combination, and that each reference addresses different problems in different contexts; MPEP § 2143(I) gives examples of rationales that Examiners may rely upon to support a conclusion of obviousness, the rationales used herein rely upon these examples. However, MPEP § 2143(I) does not suggest that the rationale for all art must be the same, such that each reference cannot address different problems. With respect to the Graham factual inquiries, each combination of prior art herein outlines the scope and content of the prior art; identifies the difference between the prior art by stating what each prior art teaches alone, what the prior art is missing prior to combination, and how incorporating art in combination would arrive at the claimed invention; and resolves the ordinary level of skill in the art by incorporating art directed to the same environment of user retention, thus, one of ordinary skill in the art would know types of problems in the art and prior art solutions to those problems. The motivation to combine for each prior art then relies upon one of the examples outlined in MPEP § 2143(I), specifically, (A) and (G), such that the rationale to combine is supported by prior art elements that reflect the claim language. The rationale to combine with Porter falls under (G), such that their method of preprocessing data for use by a machine learning system would format data in a common standardized format that the system can use. The rationale to combine with Joa falls under (G), such that their method of collecting data and incorporating the collected data would improve pre-existing data integration methods and join data contained in servers located in business support units, the data relied upon specifically is user complaints from business support units. The rationale to combine with Simpson falls under (A), specifically, combining the data collected by financial institutions as suggested in Simpson according to the known method of Zhao to yield the predictable result of collecting and incorporating data for use by a machine learning system for user retention. The rationale to combine with Yeri falls under (G), such that collecting and incorporating the data suggested in Yeri can be used to avoid losing the user to a competitor and enhance the relationship with the user, which is desirable because it costs must less to retain customers than to acquire new ones. The rationale to combine with Rozin falls under (G), such that that data collected and incorporated as suggested in Rozin would identify recurring or repeating online banking actions or patterns, which would be desirable in the context of a user retention system such as Zhao, including in a financial institution’s user retention system. The rationale to combine with The rationale to combine with Guminy falls under (A), specifically, combining the collected data of Guminy according to the known method of Joa to yield the predictable result of incorporating data from business support units for use by a machine learning system for user retention. Thus, the rationales do not propose generic benefits, and instead are tied to prior art elements which are part of the prior art disclosure and that in combination reflect the cited benefits and yielded predictable results, such that the combination reflects the claim language.
For at least the aforementioned reasons, claim 1 is rendered obvious in view of the prior art of record and the rejection under 35 U.S.C. 103 is respectfully maintained. Claims 25 and 29 are also rejected under 35 U.S.C. 103 in view of the prior art of record as they recite substantially similar limitations and are rendered obvious for similar reasons. Claims 3-4, 6, 8, 11, 19-21, 23-24, 26-28, and 30-33 are also rendered obvious in view of the prior art of record and rejected under 35 U.S.C. 103.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
White et al. (US 2021/0271934 A1) discloses a method of preprocessing data such that ingested data is brought to a common, standardized format that can be easily ingested and acted upon by a further computational workflow downstream.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JAKE BREEN whose telephone number is (571)272-0456. The examiner can normally be reached Monday - Friday, 7:00 AM - 3:00 PM EST.
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, Jennifer Welch can be reached at (571) 272-7212. 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.
/J.T.B./Examiner, Art Unit 2143
/JENNIFER N WELCH/Supervisory Patent Examiner, Art Unit 2143