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 .
The present application claims foreign priority of INDIA 202141020933 filed on 05/08/2021. 35 USC 119 (a-d) conditions are met.
Status of Claims
Claims 1, 8, 9, 16, 21, 24, 26, 30 and 35 are currently pending and rejected.
Claims 2-7, 10-15, 17-20, 22, 23, 25, 27-29, 31-34, and 36 are cancelled.
Claim Rejection – 35 U.S.C. 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, 8, 9, 16, 21, 24, 26, 30 and 35 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter. The rationale for this finding is explained below. In the instant case, the claims are directed towards predicting chargeback behavioral of an account holder. The concept is clearly related to managing personal behavior, thus the present claims fall within the Certain Method of Organizing Human Activity grouping. Moreover, the present claims are directed to a process of accessing data, analyzing data, and providing result of the analysis. These processes can be performed in the human mind. As such, the present claims also fall within the Mental Processes grouping. The claims do not include limitations that are “significantly more” than the abstract idea because the claims do not include an improvement to another technology or technical field, an improvement to the functioning of the computer itself, or meaningful limitations beyond generally linking the use of an abstract idea to a particular technological environment. Note that the limitations, in the instant claims, are done by the generically recited computer device. The limitations are merely instructions to implement the abstract idea on a computer and require no more than a generic computer to perform generic computer functions that are well-understood, routine and conventional activities previously known to the industry. Therefore, claims 1, 8, 9, 16, 21, 24, 26, 30 and 35 are rejected under 35 U.S.C. 101 as being directed to non-statutory subject matter.
Step 1: The claims 1, 8, 9, 16, 21, 24, 26, 30 and 35 are directed to a process, machine, manufacture, or composition matter.
In Alice Corp. Pty. Ltd. v. CLS Bank Intern., 134 S. Ct. 2347 (2014), the Supreme Court applied a two-step test for determining whether a claim recites patentable subject matter. First, we determine whether the claims at issue are directed to one or more patent-ineligible concepts, i.e., laws of nature, natural phenomenon, and abstract ideas. Id. at 2355 (citing Mayo Collaborative Servs. v. Prometheus Labs., Inc., 132 S. Ct. 1289, 1296–96 (2012)). If so, we then consider whether the elements of each claim, both individually and as an ordered combination, transform the nature of the claim into a patent-eligible application to ensure that the patent in practice amounts to significantly more than a patent upon the ineligible concept itself.
Claims 1, 8, 21, and 24 are directed to a process (method). Claims 9, 26, and 30 are directed to a machine (system). Claim 16 and 35 are directed to a manufacture (non-transitory computer-readable storage).
Step 2A: The claims are directed to an abstract idea.
Prong One
The present claims are directed towards predicting chargeback behavioral of an account holder. The concept comprises using a GBDT model to compute first and second chargeback risk probability scores for distinct future time intervals, classifying the account holder accordingly, determining chargeback risk levels based on amount bands, and in real-time, transmitting a notification that modifies the issuer server’s authorization logic to approve or decline future transactions. The concept is clearly related to managing personal behavior, thus the present claims fall within the Certain Method of Organizing Human Activity grouping. Moreover, the present claims are directed to a process of accessing data, analyzing data, and providing result of the analysis. These processes can be performed in the human mind. The present claims closely follow the fact pattern of the ineligible claims in Electric Power Group v. Alstom. As such, the present claims also fall within the Mental Processes grouping. The performance of the claim limitations using generic computer components (i.e. a server comprising a processor, a memory, and a communication interface) does not preclude the claim limitation from being in the certain methods of organizing human activity grouping and the mental processes grouping. The claimed invention does not use blockchain technology any differently. Accordingly, this claim recites an abstract idea.
Prong Two
Independent claim 1 recites a server system as additional element. Independent claim 9 recites a server system comprising a communication interface, a memory, and a processor as additional elements. Independent claim 16 recites a server system comprising a non-transitory computer readable storage medium and a processor as additional elements. The additional elements are claimed to perform basic computer functions, such as using a GBDT model to compute first and second chargeback risk probability scores for distinct future time intervals, classifying the account holder accordingly, determining chargeback risk levels based on amount bands, and in real-time, transmitting a notification that modifies the issuer server’s authorization logic to approve or decline future transactions. Dependent claim 8 does not recite any additional element. Claims 21-36 do not recite any additional element that are indicative of integration into practical application. For example, claims 21 and 22 recites generating graph, which can be done by hand. Claims 23-25 and 28-36 merely define the elements in the graphs. Claim 26-27 recite a generic graph creation engine, which is well-understood, routine, and conventional in computer art. The recitation of the computer elements amounts to mere instruction to implement an abstract concept on computers. The present claims do not solve a problem specifically arising in the realm of computer networks. Rather, the present claims implement an abstract concept using existing computer technology in a networked computer environment. The present claims do not recite limitation that improve the functioning of computer, effect a physical transformation, or apply the abstract concept in some other meaningful way beyond generally linking the use of the abstract concept to a particular technological environment. As such, the present claims fail to integrate into a practical application.
With regards to the features, “aggregating the set of transaction features; converting the aggregating set of transaction features into a vector format” and “computing, by the server system via a chargeback risk prediction model, a set of chargeback risk probability scores corresponding to one or more time intervals based, at least in part, on the converted aggregated set of transaction features in the vector format”, Examiner points out that they can be performed without computer. Examiner points out that aggregating features is nothing more than recognizing and combining certain data from dataset. This step can be entirely performed in the human mind. Moreover, Examiner ran a search on converting transaction features into a vector format, and found that the feature appears to be a standard feature in machine learning. For example, Melul et al. (Pub. No.: US 2022/0198470) teaches “The transaction review software processes the transaction through the artificial neural network model, converts the transaction into a feature vector, encodes the feature vector into a compressed vector, decodes the compressed vector to a reconstructed vector, subtracts the reconstructed vector from the feature vectors, and determines a fraud indication based on a difference from the reconstructed and feature vectors” (see abstract); Assefa et al. (Pub. No.: US 2022/0036219) teaches “Each transaction record 210 in the transaction history 122 may be converted into a feature vector containing data indicative of various field-values in the transaction record 210” (see paragraph 0053). Ashiya et al. (Pub. No.: US 2019/0295085) teaches “this transaction information may be stored in a historical database of transaction information and converted into feature vectors representing each transaction as training data” (see paragraph 0036). Evidences show transaction data is routinely converted into feature vector to be processed by artificial intelligence. Claim 2 of Example 47 recites steps of discretizing continuous training data, training artificial intelligence using selected training algorithm including a backpropagation algorithm and a gradient descent algorithm. These steps are considered mere data gathering and data processing at a high level of generality, and thus are insignificant extra-solution activity. The court did no find any improvement in the functioning of processor in these steps. Similarly, aggregating transaction features and converting them into vector format, and computing a set of chargeback risk probability score based on the converted aggregated set of data do not integrate the abstract idea into a practical application.
Examiner also points to the new July 2024 Subject Matter Eligibility Examples. Claim 2 of Example 47 of that new guideline recites an artificial intelligence for processing transaction data and outputting anomaly data. Claim 2 of Example 47 is ineligible for patent, because the claim falls within the grouping of “mental processes” and there is no improvement in computer function or machine learning. Similarly, the present claims recite using machine learning technique (GBDT) to process transaction data in order to classify account holders and determine chargeback risk.
Examiner further points out that a network graph, also known as node-link diagrams, is “a visual representation of objects (nodes) and the relationships (edges) connecting them, ideal for visualizing complex data and relationships in areas like social networks”. Network graph was old and well-known in the area of fraud detection. Yao et al. (Pub. No.: US 2018/0130071) teaches generating network graph from transaction events (see paragraph 0004, claim 2 and 14, and FIG. 8 and 9). Network graph is entirely an abstract concept that can be drawn on paper with a pen. Network graph is not related to computer technology. The act of drawing a network graph on computer also does not improve computer function. Computer is merely used as an extra-solution to implement an abstract concept that can be done without computer. Moreover, entering the vector as input into a chargeback risk prediction model is merely applying existing machine learning model to a new data environment. According to the Recentive v. Fox decision, “patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under 101”. Therefore, the recitation of the network graph and the chargeback risk prediction model (machine learning model) is not sufficient to indicate improvement of computer functionality or integrate the abstract concept into a practical application.
Step 2B: The claims do not recite additional elements that amount to significantly more than the abstract idea.
As discussed earlier, the present claims only recite a processor coupled with a memory as additional elements. The additional elements are claimed to perform basic computer functions, such as receiving notification, accessing data record, updating data record, and transmitting data record. According to MPEP 2106.05(d), “performing repetitive calculations”, “receiving, processing, and storing data”, “electronically scanning or extracting data from a physical document”, “electronic recordkeeping”, “storing and retrieving information in memory”, and “receiving or transmitting data over a network, e.g., using the Internet to gather data” are considered well-understood, routine, and conventional functions of computer. The present claims do not improve the functioning of computer technology.
Applicant argued that the claimed invention, similar to the invention in DDR Holdings, does not merely recite the performance of some business practice known from the pre-Internet world. Applicant argued the present claims provide a non-conventional and non-generic mechanism that provides preemptive remedial action for payment accounts that are likely to experience chargebacks so that processing requirements for handling the chargebacks at issuers/acquirers that can be reduced or optimized. Examiner disagrees and points out that at a high level, the present claims merely utilize existing machine learning technique to gather data, process data, analyze processed data, and transmit result to a server system for payment authorization. Moreover, as discussed earlier, the amended features appear to be well-understood, routine, and conventional in the art of artificial intelligence.
Simply implementing the abstract idea on a generic computer or using a computer as a tool to perform an abstract idea cannot integrate a judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B. Therefore, the present claims are ineligible for patent.
Response to Remarks
Rejection under 35 U.S.C. 101
In the response filed on 02/09/2026, Applicant amended independent claim 1 by removing the following limitations:
“accessing merchant specific risk features over a period of time, the merchant specific risk features comprising: chargeback rates, and average chargeback amount, a total number of chargebacks, a number of cross border chargebacks, a percentage of chargebacks for card-not-present transactions, and fraud-related chargeback amounts at the merchant;
generating, for each merchant, a merchant specific risk vector based on the merchant specific risk features for the merchant and historical chargeback for the merchant; and
generating the set of transaction features for the account holder by aggregating the merchant specific risk vector according to the weighted edges of the network graph, the set of transaction features, wherein the set of transaction features are generated separately for different time intervals within the same period to reflect temporally sensitive risk profiles, and wherein the set of transaction features further includes a ratio of payment transactions that exceeds a fraud risk threshold for each of the plurality of merchants;”
“upon classifying the account holder as the risky account holder for the second time interval, determining, by the server system a probable chargeback amount band for potential future chargeback requests raised by the account holder within each of the time interval and the second time interval, based at least in part, on the first chargeback risk probability score, and second chargeback probability score and a chargeback amount prediction model, the chargeback amount prediction model being a GBDT based ranking model”; and
“transmitting, by the server system a second notification to provide an incentive to the account holder to reduce a likelihood to file chargeback”.
Applicant also added two limitations – “the issuer server declining the online payment transaction in real-time based on the first notification” and “modifying an authorization transaction based on the first notification”. Independent claims 9 and 16 are amended by removing and adding similar limitations. Examiner points out that the added feature - “the issuer server declining the online payment transaction in real-time based on the first notification” – is a standard procedure in electronic transactions, where risk scoring is used to determine whether to approve or decline a payment transaction in real-time. The other added feature - “modifying an authorization transaction based on the first notification” – appears to describe typical learning algorithm in payment authorization, where a machine learning model learns from past and current transactions and makes decision adjustments for future payment requests.
Applicant's arguments filed on 02/09/2026 have been fully considered but they are not persuasive.
Step 2A Prong One
Applicant argued Examiner’s characterization of the present claims – “predicting chargeback behavioral of an account holder” – overlooks the technical specificity of the claims, “which integrate machine learning and graph-based analysis into a predictive system for real-time payment transaction invention”. Examiner disagrees and points out that the amended claim 1 only briefly mentions about a machine learning model in one line – “the chargeback risk prediction model being a gradient boosting decision tree (GBDT) based classification model”. GBDT model was developed in the early 2000s. It was already two decades old by the time Applicant filed the present application. GBDT is a widely recognized machine learning algorithm used for both classification and regression tasks. In claim 1, GBDT is used to perform its basic function – “classifying…the account holder as a non-risky account holder for the first time interval and as a risky account holder for the second time interval based, at least in part, on the first chargeback risk probability score and the second chargeback risk probability score, respectively”. As such, claim 1 is merely applying a generic machine learning model to a new data environment without actually improving machine learning technology itself. According to the Recentive Analytics v. Fox Corp decision, “patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under 101”.
Examiner has also explained in the previous Office Actin that graph-based analysis was well-known and conventional in the field of finance and fraud detection. A network graph, also known as node-link diagrams, is “a visual representation of objects (nodes) and the relationships (edges) connecting them, ideal for visualizing complex data and relationships in areas like social networks”. Network graph was old and well-known in the area of fraud detection. Yao et al. (Pub. No.: US 2018/0130071) teaches generating network graph from transaction events (see paragraph 0004, claim 2 and 14, and FIG. 8 and 9). Network graph is entirely an abstract concept that can be drawn on paper with a pen. Network graph is not related to computer technology. The act of drawing a network graph on computer also does not improve computer function. Computer is merely used as an extra-solution to implement an abstract concept that can be done without computer.
Applicant also emphasized the “real-time” nature of the claimed invention. Examiner points to the decision of Electric Power Group v. Alstom. The court was not impressed by the “real time” feature – “The claims in this case specify what information in the power-grid field it is desirable to gather, analyze, and display, including in “real time”; but they do not include any requirement for performing the claimed functions of gathering, analyzing, and displaying in real time by use of anything but entirely conventional, generic technology. The claims therefore do not state an arguably inventive concept in the realm of application of the information-based abstract ideas”. Similarly, the “real time” feature of the present claims does not require anything other than off-the-shelf computers, thus this feature does not improve computer function or render the claims less abstract.
Step 2A Prong 2
Applicant argued that “the claims recite specific improvements to the functioning of payment processing systems by using a GBDT model to compute first and second chargeback risk probability scores for distinct future time intervals, classifying the account holder accordingly, determining chargeback risk levels based on amount bands, and in real-time, transmitting a notification that modifies the issuer server’s authorization logic to approve or decline future transactions”. Applicant’s summary of the claims is similar to the fact pattern of Electric Power Group v. Alstom – gathering data, analyzing data, and providing result of analysis. As discussed earlier, GBDT is a widely recognized machine learning algorithm used for both classification and regression tasks. Using a GBDT model to compute and classify transactions is merely applying a generic machine learning data to a new data environment. According to the Recentive Analytics v. Fox Corp decision, “patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under 101”.
Applicant argued that “unlike that example’s generic anomaly detection via neural network without specificity, the present claims specify a GBDT classification model tailored to chargeback prediction, incorporating graphical representations of risk and spend profile across merchants”. Examiner disagrees and points out that the present claims merely apply the generic GBDT model to a chargeback prediction data environment. The GBDT model itself is not being improved. The present claim 1, for example, is also different from MrRO v. Bandai. The USPTO’s November 2016 Memorandum discusses the McRO case. In McRO, the Federal Circuit held the claimed methods of automatic lip synchronization and facial expression animation using particular computer-implemented rules patent eligible under 35 U.S.C. 101, because the claims were directed to an improvement in computer-related technology (i.e. allowing computer to produce “accurate and realistic lip synchronization and facial expressions in animated characters” that previously could only be produced by human animators). As part of its analysis, the McRO court examined the specification, which described the claimed invention as improving computer animation through the use of specific rules, rather than human artists, to set morph weights and transition parameters between phonemes. As explained in the specification, human artists did not use the claimed rules, and instead relied on subjective determinations to set morph weights and manipulate the animated face to match pronounced phonemes. As such, McRO's claims are not focused on a mere automation of known manual process, but on an improvement in computer technology. The November 2016 Memorandum further discloses two indications that a claim may be directed to an improvement in computer-related technology:
1) a teaching in the specification about how the claimed invention improves a computer or other technology (e.g. McRO court relied on the specification's explanation of how the claimed rules enabled the automation of specific animation tasks that previously could not be automated when determining that the claims were directed to improvements in computer animation instead of an abstract idea). In contrast, the court in Affinity Labs of TX v. DirecTV relied on the specification’s failure to provide details regarding the manner in which the invention accomplished the alleged improvement when holding the claimed methods of delivering broadcast content to cellphones directed to an abstract idea.
2) a particular solution to a problem or a particular way to achieve a desired outcome defined by the claimed invention, as opposed to merely claiming the idea of a solution or outcome (e.g., McRO’s claims defined a specific way, namely use of particular rules to set morph weights and transitions through phonemes, to solve the problem of producing accurate and realistic lip synchronization and facial expressions in animated characters, and thus were not directed to an abstract idea). In contrast, Electric Power Group’s claimed method was directed to an abstract idea because it merely presented the results of collecting and analyzing information, without even identifying a particular tool for the presentation.
In this case, the rules or determining steps recited in claim 1 are not specific to machine. The process of analyzing transaction information to score chargeback risk level, and determining whether to approve or decline a transaction based on risk scores is the same as human thinking. Therefore, the present claims are not analogous to the claims in McRO v. Bandai.
Applicant further argued Ex Parte Desjardins “emphasizes that claims directed to advancement in artificial intelligence (AI) and machine learning (ML) should be evaluated for technological improvements under Enfish, LLC v. Microsoft Corp. Examiner points out that the present claims are unlike the claims in Enfish. The Enfish court relied on the distinction made in Alice between, on one hand, computer-functionality improvements and, on the other, uses of existing computers as tools in aid of processes focuses on "abstract ideas". In Enfish, the claims at issue focused not on asserted advances in uses to which existing computer capabilities could be put, but on a specific improvement - a particular database technique (i.e., self-referential table datastructure) - in how computers could carry out one of their basic functions of storage and retrieval of data. In contrast, the present claims do not recite any particular novel datastructure that improves the efficiency of computer storage or any computer function. Rather, the present claims recite applying a generic GBDT model to a chargeback prediction data environment without actually improving machine learning technology itself.
Finally, Applicant argued the present claims parallel Example 39 from the 2019 examples, where a neural network for facial detection was eligible due to its training method improving digital image processing without human intervention. Examiner points out that Applicant misunderstood the rationale of Example 39. In Example 39, the neural network applies several transformations to each digital facial image including mirroring, rotating, smoothing, or contrast reduction to create a modified set of digital facial images to improve training. These transformations cannot be performed by human, and are different from mental processing. The rationale is similar to McRO v. Bandai. In contrast, the present claims only require standard mathematical processes to score the transaction data, and the GBDT model only performs its basic functions – classify transactions based on input data. There is no improvement in machine learning technology. For these reasons, the amended claims do not recite features sufficient to indicate improvement in computer function. Therefore, the present claims are not integrated into practical application.
Step 2B
Applicant argued the GBDT model is not claimed in isolation, but is “used to produce multi-interval future risk scores that directly control issuer authorization decisions”. Examiner points out that all the risk scoring are described in high level of generality. There is no indication that the risk scoring steps cannot be performed mentally. Applicant then attempted to make analogy to Trading Techs v. CQG, but that decision is non-precedential. Moreover, claims in Trading Techs v. CQG recite “a specific, structured graphical user interface that improves the accuracy of trader transaction by displaying bid and asked prices in a particular manner that prevents order entry at a changed price”. The present claims are not related to graphical user interface at all.
Applicant also argued that the present claims “do not pre-empt the abstract idea but claim a specific, discrete implementation”. Preemption is not a standalone test for patent eligibility. Preemption concerns have been addressed by the examiner through the application of the two-step framework. Applicant’s attempt to show alternative uses of the abstract idea outside the scope of the claims does not change the conclusion that the claims are directed to patent ineligible subject matter. Similarly, applicant’s attempt to show that the recited abstract idea is a very narrow and specific one is not persuasive. A specific abstract idea is still an abstract idea and is not eligible for patent protection without significantly more recited in the claim.
See the July 2015 Update: Subject Matter Eligibility that explains that questions of preemption are inherent in the two-part framework from Alice Corp and Mayo and are resolved by using this framework to distinguish between preemptive claims, and ‘those that integrate the building blocks into something more…the latter pose no comparable risk of preemption, and therefore remain eligible.” The absence of complete preemption does not guarantee the claim is eligible. Therefore, “[w]here a patent’s claims are deemed only to disclose patent ineligible subject matter under the Mayo framework, as they are in this case, preemption concerns are fully addressed and made moot.” Ariosa Diagnostics, Inc. v. Sequenom, Inc., 788 F.3d 1371, 1379 (Fed. Cir. 2015). See also OIP Tech., Inc. v. Amazon.com, Inc., 788 F.3d 1359, 1362-63 (Fed Cir. 2015).
For these reasons, Examiner maintains the ground of rejection under 35 U.S.C. 101. Updated rejection is provided in this Office Action.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HAO FU whose telephone number is (571)270-3441. The examiner can normally be reached 9:00 AM - 6:00 PM PST.
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, Christine Behncke can be reached on (571) 272-8103. 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.
/HAO FU/Primary Examiner, Art Unit 3697
FEB-2026