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 .
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 04/15/2026 has been entered.
Response to Amendment
Applicant’s “Amendment” filed on 03/10/2026 has been considered.
Claims 1, 5, 8, and 15 are amended. Claims 1-20 remain pending in this application and an action on the merits follow.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 USC 101. The claimed invention is directed to non-statutory subject matter because claims 1, 8, and 15 are directed to an abstract idea without significantly more. Claims 2-7, 9-14, and 16-20 fail to remedy these deficiencies.
101 Rejection:
Claim 1 (Reproduced and Parsed)
A system comprising:
· [1a] a processor; and
· [1b] a memory comprising computer program code, the memory and the computer program code configured to, with the processor, cause the processor to:
. [1b-i] obtain historical transaction data comprising transactions and corresponding actual transaction class codes;
. [1b-ii] provide the obtained historical transaction data as input to a classification model without the actual transaction class codes, causing the classification model to generate a suggested transaction class code for each of the transactions for which the historical transaction data was provided;
. [1b-iii] adjust weights of the classification model based on differences between the suggested transaction class codes and the actual transaction class codes to produce a trained classification model;
. [1b-iv] receive a notification, at a point-of-sale (POS) interface, that a transaction associated with an account has been initiated, the notification including transaction data associated with the transaction;
. [1b-v] provide the transaction data as input to the trained classification model, causing the trained classification model to generate a plurality of suggested transaction class codes, each suggested transaction class code having an associated probability value indicating a correctness of the suggested transaction class code;
. [1b-vi] send a prompt for collection, from a user, of a transaction class code for the transaction, the prompt including a human-readable description associated with each of the plurality of the suggested transaction class codes for selection by the user, wherein two or more of the plurality of suggested transaction class codes are dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy;
. [1b-vii] receive a transaction class code in response to the prompt from the user;
. [1b-viii] associate the received transaction class code with the transaction;
. [1b-ix] upon completion of the transaction, record a transaction record of the transaction including the received transaction class code and transaction data of the transaction; and
. [1b-x] retrain the classification model using the transaction data and the received transaction class code to produce an improved classification model.
STEP 1: STATUTORY CATEGORY DETERMINATION
Finding: Claim 1 recites a “system” with a processor and memory configured to perform computational steps. Under Beauregard, a system claim that includes a processor and memory storing computer program code falls within the statutory categories of “machine” under 35 U.S.C. § 101(a).
Conclusion on Step 1: Claim 1 is directed to a statutory category (machine/system).
STEP 2A, PRONG 1: IDENTIFICATION OF JUDICIAL EXCEPTIONS
Analysis
Claim 1 recites the following judicial exceptions:
A. Abstract Idea – Mathematical Concepts and Methods of Organizing Human Activity
Mathematical Concepts:
· [1b-ii]: “provide the obtained historical transaction data as input to a classification model without the actual transaction class codes, causing the classification model to generate a suggested transaction class code” – This recites the mathematical operation of training a machine learning classifier, which is a mathematical algorithm.
· [1b-iii]: “adjust weights of the classification model based on differences between the suggested transaction class codes and the actual transaction class codes” – This explicitly recites weight adjustment, a core mathematical operation in neural networks and machine learning (e.g., gradient descent, backpropagation). This is a fundamental mathematical concept.
· [1b-v]: “provide the transaction data as input to the trained classification model, causing the trained classification model to generate a plurality of suggested transaction class codes, each suggested transaction class code having an associated probability value” – This recites applying a mathematical model to generate probabilistic outputs, another mathematical concept.
Methods of Organizing Human Activity:
· [1b-vi]: “send a prompt for collection, from a user, of a transaction class code for the transaction” – This recites collecting user input to organize and classify transactions,
which falls within the grouping of “fundamental economic practices” and “commercial/legal interactions” (organizing financial transactions and their categorization).
· [1b-viii]: “associate the received transaction class code with the transaction” – This recites tagging/labeling, a data organization method.
· [1b-ix]: “record a transaction record of the transaction including the received transaction class code and transaction data of the transaction” – This recites recording and organizing transaction data, which is a method of organizing human activity (financial record-keeping).
Offending Clauses (Direct Quotes):
1. “adjust weights of the classification model based on differences between the suggested transaction class codes and the actual transaction class codes” – Mathematical concept (weight adjustment).
2. “provide the obtained historical transaction data as input to a classification model without the actual transaction class codes, causing the classification model to generate a suggested transaction class code” – Mathematical concept (machine learning algorithm).
3. “provide the transaction data as input to the trained classification model, causing the trained classification model to generate a plurality of suggested transaction class codes, each suggested transaction class code having an associated probability value” – Mathematical concept (applying a trained model to generate probabilistic classifications).
4. “send a prompt for collection, from a user, of a transaction class code for the transaction” – Method of organizing human activity (collecting user input for transaction classification).
5. “associate the received transaction class code with the transaction” and “record a transaction record of the transaction including the received transaction class code and transaction data” – Methods of organizing human activity (organizing and recording financial transactions).
Conclusion on Step 2A, Prong 1: Claim 1 recites judicial exceptions, specifically:
· Abstract ideas encompassing mathematical concepts (machine learning model training, weight adjustment, probabilistic classification)
· Abstract ideas encompassing methods of organizing human activity (transaction classification, record-keeping, user input collection)
Per the 2019 PEG and Alice v. CLS Bank, these are well-established judicial exceptions.
STEP 2A, PRONG 2: INTEGRATION INTO A PRACTICAL APPLICATION
Analysis
The critical question is whether the claim integrates the recited abstract ideas into a “practical application” that provides a meaningful limitation beyond merely applying the exception.
Applicant’s Position (from Specification):
The specification argues (¶[0015]) that the disclosure operates in an “unconventional manner” by:
· Using “the transaction initiation process on a POS interface to prompt a user to provide the transaction class code”
· Leveraging “the interaction between the user and the POS interface that is required to complete the transaction to collect the class code”
· Avoiding “resource usage, bandwidth usage, and other computation costs associated with categorizing or classifying the transaction through other means”
· Enhancing “the accuracy with which class codes are assigned to transactions”
The specification further argues (¶[0016]) that machine learning-generated suggested class codes:
· Streamline the process
· Improve accuracy and speed
· Reduce likelihood of inaccurate code selection
· Reduce time required for class code selection
· Reduce computation resources needed
Examiner’s Assessment:
1. Is there an improvement to computer functionality or another technology?
The claim does not recite any specific technological improvement. The specification’s assertions about “reducing resource usage,” “bandwidth usage,” and “computation costs” are conclusory and unsupported by technical detail. Specifically:
· No claim language specifies how the system reduces bandwidth or computation costs.
· No technical mechanism is claimed that would improve network latency, memory utilization, or processor efficiency.
· The claim does not recite any novel hardware architecture, data structure, or algorithmic optimization.
· The specification provides no comparative performance metrics, benchmarks, or technical evidence demonstrating that the claimed method reduces resource consumption versus conventional approaches.
Per Berkheimer v. HP, Inc., 881 F.3d 1360 (Fed. Cir. 2018), assertions of technological improvement without factual support are insufficient.
2. Is there a transformation of an article or application to a particular machine?
· The claim recites a generic “processor” and “memory” – standard components of any general-purpose computer.
· The claim does not specify a particular machine architecture, specialized hardware, or novel technical configuration.
· The claim does not recite transformation of a physical article or material.
· The POS interface is mentioned as a location where the process occurs, not as a particular machine to which the abstract idea is applied in a non-routine manner.
Per Diamond v. Diehr, 450 U.S. 175 (1981), and Bilski v. Kappos, 561 U.S. 593 (2010), mere application of an abstract idea on a generic computer does not constitute integration into a practical application.
3. Are there meaningful limitations beyond the abstract idea itself?
Analyzing each limitation:
· [1b-i] “obtain historical transaction data” – Data collection; routine.
· [1b-ii] “provide the obtained historical transaction data as input to a classification model…causing the classification model to generate a suggested transaction class code” – This is the abstract idea itself (applying a mathematical model). It is not a limitation beyond the exception; it is the exception.
· [1b-iii] “adjust weights of the classification model based on differences” – This is the abstract idea itself (weight adjustment in machine learning). Standard machine learning training.
· [1b-iv] “receive a notification…that a transaction associated with an account has been initiated” – Data input; routine.
· [1b-v] “provide the transaction data as input to the trained classification model…to generate a plurality of suggested transaction class codes” – This is the abstract idea itself (applying a trained mathematical model). Routine application of a standard machine learning technique.
· [1b-vi] “send a prompt for collection…of a transaction class code” – User interface interaction; routine.
· [1b-vii] “receive a transaction class code in response to the prompt” – Data collection; routine.
· [1b-viii] “associate the received transaction class code with the transaction” – Data tagging; routine.
· [1b-ix] “record a transaction record” – Data storage; routine.
· [1b-x] “retrain the classification model” – This is the abstract idea itself (retraining a mathematical model). Standard machine learning practice.
Finding: The claim does not recite any element that constitutes a meaningful limitation beyond the abstract ideas of:
· Training a machine learning classifier
· Adjusting weights based on error
· Applying a trained model to generate predictions
· Collecting and organizing user input
The additional elements (receiving notifications, sending prompts, recording data) are insignificant extra-solution activity – they are routine steps necessary to implement the abstract idea on a computer, not technological improvements or novel applications.
Per Mayo v. Prometheus Labs., 566 U.S. 66 (2012), and the 2019 PEG, merely adding routine data-gathering steps or generic computer implementation does not integrate an abstract idea into a practical application.
4. Field-of-Use Limitation?
The claim’s reference to “point-of-sale (POS) interface” and “transaction” is a field-of-use limitation – it narrows the context in which the abstract idea is applied but does not provide a technological improvement or novel application. Per the 2019 PEG (Step 2A, Prong 2), field-of-use limitations do not suffice to integrate an exception into a practical application.
Conclusion on Step 2A, Prong 2: Claim 1 does NOT integrate the recited judicial exceptions into a practical application. The claim:
· Does not recite a specific technological improvement with technical support
· Does not apply the abstract idea to a particular machine in a non-routine manner
· Does not recite meaningful limitations beyond the abstract ideas themselves
· Relies on field-of-use limitations (POS interface, transaction context), which are insufficient
The claim therefore fails Step 2A, Prong 2.
STEP 2B: SIGNIFICANTLY MORE THAN THE EXCEPTION (WURC ANALYSIS)
Because Claim 1 fails Step 2A, Prong 2, the analysis proceeds to Step 2B: whether the additional claim elements constitute “significantly more” than the exception and are not well-understood, routine, conventional (WURC).
Analysis
Elements Beyond the Abstract Idea:
1. Processor and memory – Generic computer components; WURC.
2. Receiving a notification at a POS interface – Standard data input operation; WURC.
3. Sending a prompt to a user – Standard user interface operation; WURC.
4. Recording a transaction record – Standard data storage operation; WURC.
5. Associating a class code with a transaction – Standard data tagging; WURC.
Evidentiary Support for WURC Finding:
Per Berkheimer, the applicant bears the burden of establishing that an element is not WURC. The specification provides:
· No technical details on the processor architecture, memory configuration, or specialized hardware.
· No novel data structures or algorithms beyond standard machine learning techniques.
· No performance metrics, benchmarks, or comparative analysis demonstrating non-routine implementation.
· No evidence that the combination of elements represents a non-conventional technological solution.
The specification’s assertions about “reducing resource usage” and “improving accuracy” are conclusory and unsupported by:
· Technical specifications
· Comparative performance data
· Implementation details that would distinguish the claimed approach from conventional practice
Machine Learning as WURC:
By 2024 (the publication date of this application), machine learning classification models, weight adjustment, and model retraining are well-understood, routine, and conventional in the art. See Alice v. CLS Bank, 717 F.3d 1269 (Fed. Cir. 2014) (generic computer implementation of abstract idea is WURC).
Conclusion on Step 2B: The additional elements recited in Claim 1 are WURC. They do not constitute “significantly more” than the abstract idea. The claim therefore fails Step 2B.
Therefore, a 101 rejection appears to be proper here. The claims 1, 8, and 15 are not patent eligible.
The claims 2, 4, 9, 11, 16, and 18 recite receiving a request for a transaction statement, generating a transaction statement, providing the generated transaction statement, generating a suggested transaction code, providing the generated transaction code, selecting the suggested transaction class code steps as drafted, are processes that under broadest reasonable interpretation, cover performance of managing personal behavior, but for the recitation of generic computer components. That is, other than reciting “a processor; and a memory comprising computer program code, the memory and the computer program code configured to, with the processor, cause the processor to”, nothing in the claim element precludes the steps from practically being performed by organizing human activity. For example, but for the “the processor and the memory” in the context of these claims encompasses a person manually requests for a transaction statement, manually generates/forms the transaction statement, manually provides/displays/presents the transaction statement, manually generates a suggested transaction code, manually provides the generated transaction code, and manually selects the suggested transaction class code. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation by managing personal behavior but for the recitation of generic computer components, then it falls within the “Certain Methods of Organizing Human Activity” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
This judicial exception is not integrated into a practical application because the claims as a whole merely describe how to generally “apply” the concept of receiving, generating, and providing in a computer environment. The claimed computer components such as the processor and the memory are recited at a high level of generality and are merely invoked as tools to perform receiving, generating, providing, and selecting steps. Simply implementing the abstract idea on a generic computer is not a practical application of the abstract idea. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claims 2, 4, 9, 11, 16, and 18 are directed to an abstract idea.
The claims 2, 4, 9, 11, 16, and 18 do not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional elements of using the processor and the memory to perform receiving, generating, providing, and selecting steps amount to no more than mere instructions to apply the exception using generic computer components. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. Therefore, the claim does not amount to significantly more than the recited abstract idea (Step 2B: NO). The claims 2, 4, 9, 11, 16, and 18 are not patent eligible.
Claims 3, 5-7, 10, 12-14, 17, and 19-20 disclose insignificant helpful content to further describe content, such as class descriptions are associated with the transaction class codes, the minimum threshold is 0.6, the suggested transaction codes initiation and displaying interface, and a hierarchical structure of class codes and related sub-class codes, which are merely descriptive content to further limit the abstract idea but not make it less abstract. Thus, the claims 3, 5-7, 10, 12-14, 17, and 19-20 are directed to an abstract idea.
This judicial exception is not integrated into a practical application because descriptive content in claims 3, 5-7, 10, 12-14, 17, and 19-20 further limit the abstract idea but not make it less abstract. Thus, the claims 3, 5-7, 10, 12-14, 17, and 19-20 are directed to an abstract idea.
There are no additional claim element limitations recited in the claims 3, 5-7, 10, 12-14, 17, and 19-20. Therefore, the claim does not amount to significantly more than the recited abstract idea (Step 2B: NO). The claims 3, 5-7, 10, 12-14, 17, and 19-20 are not patent eligible.
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 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 1-2, 4-13, 15-16, and 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent No. 8,554,647 to Grigg et al., U.S. Patent Application Publication No. 2024/0135236 to Mandala et al., in view of U.S. Patent Application Publication No. 2022/0343329 to Wang, and further in view of U.S. Patent Application Publication No. 2023/0306279 to Bar Elivahu et al.
With regard to claims 1 and 15, Grigg discloses a system comprising:
a processor (Fig. 5, a processor 510); and
a memory comprising computer program code, the memory and the computer program code configured to, with the processor, cause the processor to (Fig. 5, a memory 520):
receive a notification, at a point-of-sale (POS) interface, that a transaction associated with an account has been initiated, the notification including transaction data associated with the transaction (col. 10, lines 44-50, col. 11, lines 1-4, the system prompts the user to categorize at least one component of the transaction before authorizing the transaction. a point-of-sale device and a mobile device of the user wirelessly connects to provide real-time display of products or services that are being purchased in the transaction.);
provide the transaction data as input to the trained classification model, causing the trained classification model to generate a plurality of suggested transaction class codes (abstract, col. 14, lines 18-23, Systems, methods, and computer program products are provided for providing suggested transaction categories based on location. access a database including a plurality of locations and associated transaction categories; determine at least one category associated with the location of the user using the database; and provide the at least one category to the user so that the user may categorize a transaction. For example, the categorization database may receive input from users that are using the systems disclosed herein. As the user provides an input and category for a product or location, the system may determine the user's location and store the location and associated category in the database. In this way, the system learns which categories other users or the specific user is categorizing at a specific location and can improve recommendations for the location.);
send a prompt for collection, from a user, of a transaction class code for the transaction, the prompt including a human-readable description associated with each of the plurality of the suggested class codes for selection by the user (abstract, col. 10, lines 31-32, col. 11, lines 29-35, Systems, methods, and computer program products are provided for providing suggested transaction categories based on location. The system determines whether the transaction meets predefined criteria for when the system will prompt the user to categorize at least one component of the transaction. In some cases, a suggested category is displayed on the user's mobile device. Examiner notes that a suggested category is displayed to the user on the mobile device for the user to make selections, which is considered as “the prompt including a human-readable description associated with each of the plurality of suggested transaction class codes for selection by the user”);
receive a transaction class code in response to the prompt from the user (col. 15, lines 45-52, the system 300 receives input from the user selecting a category from the at least one category provided to the user.);
associate the received transaction class code with the transaction (col. 15, lines 62-col. 16, lines 12, the system 300 stores the selected category in association with the transaction); and
upon completion of the transaction, record a transaction record of the transaction including the received transaction class code and transaction data of the transaction (col. 15, lines 62-col. 16, lines 12, In an embodiment, the system stores the category in association with the transaction in a categorization database).
However, Grigg does not disclose obtain historical transaction data comprising transactions and corresponding actual transaction class codes; provide the obtained historical transaction data as input to a classification model without the actual transaction class codes, causing the classification model to generate a suggested transaction class code for each of the transactions for which the historical transaction data was provided; adjust weights of the classification model based on differences between the suggested transaction class codes and the actual transaction class codes to produce a trained classification model; each suggested transaction class code having an associated probability value indicating a correctness of the suggested transaction class code, wherein two or more of the plurality of suggested transaction class codes are dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy; and retrain the classification model using the transaction data and the received transaction class code to produce an improved classification model.
However, Mandala teaches obtain historical transaction data comprising transactions and corresponding actual transaction class codes (The executable instructions may be configured to receive a training set of transaction data. The training set of transaction data may include two or more training transactions. Each training transaction comprising may include: a business name, a short description, and a transaction category code. paragraphs 3 and 9); provide the obtained historical transaction data as input to a classification model without the actual transaction class codes, causing the classification model to generate a suggested transaction class code for each of the transactions for which the historical transaction data was provided (The program may divide the set into a training set of classified transactions and a test set of unclassified transactions. The program may analyze the training set to generate a document term matrix. The program may analyze the test set and compare the analyzed terms to the document term matrix to predict a classification. The test set may include two or more test transactions. Each test transaction may include: a business name, a short description, an “OTHER” transaction category code, as well as other data. The instructions may compare, through one or more comparison artificial intelligence/machine learning (“cAI/ML”) algorithms, each test unigram to the document term matrix to determine a predicted transaction category code for each test transaction. abstract, paragraphs 13 and 16); and
retrain the classification model using the transaction data and the received transaction class code to produce an improved classification model (the intelligent transaction classifying computer program product may add the classified test set of transaction data to the training set, creating a new and larger training set. In an embodiment, one of the one or more cAI/ML algorithms may be or include a recurrent neural network model. A recurrent neural network may allow an output result from one node in the model to affect an input to the same node, creating a cycle. In other words, recurrent neural networks may incorporate memories of past calculations or inputs to determine new inputs, paragraphs 18 and 68).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Grigg to include, obtain historical transaction data comprising transactions and corresponding actual transaction class codes; provide the obtained historical transaction data as input to a classification model without the actual transaction class codes, causing the classification model to generate a suggested transaction class code for each of the transactions for which the historical transaction data was provided; and retrain the classification model using the transaction data and the received transaction class code to produce an improved classification model, as taught in Mandala, in order to intelligently classify unclassified transactions (Mandala, paragraph 1).
However, Wang teaches adjust weights of the classification model based on differences between the suggested transaction class codes and the actual transaction class codes to produce a trained classification model (feedback module 410 provides adjusted weights 412 to machine learning model 470. For example, if a majority of the predictions 474 output by model 470 for transactions included in final embedded matrix 162 (e.g., embedded versions of transactions in training set 132) differ from known labels for transactions in matrix 162, then feedback module 410 adjust weights of model 470 accordingly. paragraph 49).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the machine learning transaction classification model of the combination of Grigg and Mandala to include, adjust weights of the classification model based on differences between the suggested transaction class codes and the actual transaction class codes to produce a trained classification model, as taught in Wang, in order to improves classification accuracy (Wang, paragraph 36).
However, Bar Bliyahu teaches each suggested transaction class code having an associated probability value indicating a correctness of the suggested transaction class code (In some cases, historical transaction categorization data of a plurality of users can be used to learn how certain types of transactions tend to be categorized. For example, a machine learning model may be trained, based on historical transaction categorization data of a plurality of users, to output one or more recommended accounts from a user's chart of accounts in response to one or more inputs describing a transaction. Outputs from the machine learning model may include confidence scores indicating likelihoods that a given transaction corresponds to each of a plurality of accounts, which may be used to determine whether to categorize the given transaction into an account of the plurality of accounts. Embodiments of the present disclosure additionally avoid incorrect categorizations (e.g., by selecting a category only when its confidence score exceeds a threshold). The updated confidence scores are then compared to the threshold to determine whether a category can be determined (e.g., if a confidence score for the category exceeds the threshold) or if another question from question decision tree 148 should be selected based on the updated confidence scores. In some cases, a category 156 of the transaction may be provided to client device 130 once a confidence score for the category that is output by the model exceeds the threshold. The category 156 may be provided in the form of an indication of an automated categorization, as a recommended category, and/or the like. paragraphs 18, 25, and 38), and wherein two or more of the plurality of suggested transaction class codes are dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy ( the confidence scores are compared to a threshold to determine if a confidence score for any one category exceeds the threshold. For example, if two categories have confidence scores within a threshold distance of each other and a question in the decision tree would clarify whether the transactions should be assigned to one of the two categories. Examiner notes that two categories are dynamically determined based on the confidence scores within a threshold distance of each other and are exceeded the threshold, which is considered as “two or more of the plurality of suggested transaction class codes are dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy”, Paragraph 23).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the machine learning transaction classification model of the combination of Grigg, Mandala, and Wang to include, each suggested transaction class code having an associated probability value indicating a correctness of the suggested transaction class code, the selection comprising transaction class codes with the associated probability values above a threshold for consideration during a response to the prompt, and wherein two or more of the plurality of suggested transaction class codes are dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy, as taught in Bar Bliyahu, in order to automatically categorize a transaction with improved confidence (Bar Bliyahu, paragraph 20).
With regard to claims 2, 9, and 16, Grigg discloses the memory and the computer program code are configured to, with the processor, further cause the processor to: receive a request for a transaction statement including the transaction and a plurality of other transactions for which transaction records have been recorded with associated transaction class codes; generate a transaction statement including transaction data associated with the transaction and the plurality of other transactions, wherein the transaction data is organized according to transaction class codes with which the transaction and the plurality of other transactions are associated; and provide the generated transaction statement in response to the received request (col. 16, lines 13-28, The user is provided with a summary of transaction and categorization information. the user is provided the summary upon the user's request. The review includes all transactions and associated categories, but in another embodiment the summary includes only a subset of the transactions and categorization information. For example, the subset may be based on the transaction amount, e.g., if it is about a pre-defined level, a specific category, a specific location, or a specific merchant).
With regard to claims 4, 11, and 18, Grigg discloses sending the prompt for collection of the transaction class code of the transaction includes: generating a suggested transaction class code for the transaction from a plurality of transaction class codes of the account; and providing the generated suggested transaction class code with the prompt for collection of the transaction class code of the transaction, causing the user to select, at the POS interface, the suggested transaction class code as the transaction class code for the transaction (col. 15, lines 29-52, the suggested categories are provided to the user after the transaction. In an embodiment, the user uses a mobile device to provide an input to the system regarding whether the user accepts the provided category, wants to apply the provided category to the transaction, or whether the user would like to reject the provided category and select an alternative category. ).
With regard to claims 5, 12, and 19, the combination of references discloses wherein a value of the minimum threshold of probable accuracy is 0.6 (Mandala, paragraphs 63, 83, and 84, The comparison may be to a pre-determined threshold of accuracy, such as 75% or 90%. the pre-determined threshold level of confidence may be modifiable by an administrator or the program. In an embodiment, an artificial intelligence/machine learning (“AI/ML”) algorithm within the program may modify the pre-determined threshold level. For example, if a significant number of transactions are returning still classified as “OTHER”, the AI/ML algorithm may determine that the threshold level of accuracy is too high and lower the threshold level. For example, if 50% of transactions are classified as “OTHER” at a 90% threshold level, the AI/ML (or an administrator) may lower the threshold level in 5% increments, until only 10% of transactions are classified as “OTHER”. Examiner notes that a pre-determined threshold of accuracy can be modified in 5% increments. Therefore, it would be obvious that a pre-determined threshold of accuracy can be modified to 0.6, which is considered as “a value of the minimum threshold of probable accuracy is 0.6”.).
With regard to claim 6, Grigg discloses the transaction is initiated at the POS interface and the suggested transaction class codes are displayed on the POS interface (col. 11, lines 1-4, and col 14, lines 18-23).
With regard to claim 7, the combination of references discloses the received transaction class code includes a hierarchical structure of multiple levels of transaction class code having a plurality of sub-class codes, each of the sub-class codes including a further plurality of sub-sub-class codes, wherein values of the plurality of sub-class codes lie in a range that comprises numerical values between a first value higher than a numeric part of the transaction class code and a second value higher than the numeric part, the second value being more than the first value (Mandala, paragraph 84).
With regard to claim 8, Grigg discloses a system comprising:
a processor (Fig. 5, a processor 510); and
a memory comprising computer program code, the memory and the computer program code configured to, with the processor, cause the processor to (Fig. 5, a memory 520):
receive a notification, during initiation of a transaction at a point-of-sale (POS) interface to leverage interaction between a user and the POS interface that is required to complete the transaction, that a transaction associated with an account has been initiated, the notification including transaction data associated with the transaction (col. 10, lines 44-50, col. 11, lines 1-4, the system prompts the user to categorize at least one component of the transaction before authorizing the transaction. a point-of-sale device and a mobile device of the user wirelessly connects to provide real-time display of products or services that are being purchased in the transaction.);
provide the transaction data as input to the trained classification model, causing the trained classification model to generate a suggested transaction class code (abstract, col. 14, lines 18-23, access a database including a plurality of locations and associated transaction categories; determine at least one category associated with the location of the user using the database; and provide the at least one category to the user so that the user may categorize a transaction. For example, the categorization database may receive input from users that are using the systems disclosed herein. As the user provides an input and category for a product or location, the system may determine the user's location and store the location and associated category in the database. In this way, the system learns which categories other users or the specific user is categorizing at a specific location and can improve recommendations for the location.);
send a prompt for collection, from a user, of a transaction class code for the transaction, the prompt including a human-readable description associated with each of the plurality of suggested transaction class codes for selection by the user (col. 10, lines 31-32, col. 11, lines 29-35, the system determines whether the transaction meets predefined criteria for when the system will prompt the user to categorize at least one component of the transaction. In some cases, a suggested category is displayed on the user's mobile device. Examiner notes that a suggested category is displayed to the user on the mobile device for the user to make selections, which is considered as “the prompt including a human-readable description associated with each of the plurality of suggested transaction class codes for selection by the user”);
receive a transaction class code in response to the prompt (col. 15, lines 45-52, the system 300 receives input from the user selecting a category from the at least one category provided to the user.);
associate the received transaction class code with the transaction (col. 15, lines 62-col. 16, lines 12, the system 300 stores the selected category in association with the transaction); and
upon completion of the transaction, record a transaction record of the transaction including the associated transaction class code and transaction data of the transaction (col. 15, lines 62-col. 16, lines 12, In an embodiment, the system stores the category in association with the transaction in a categorization database).
However, Grigg does not disclose obtain historical transaction data comprising transactions and corresponding actual transaction class codes; provide the obtained historical transaction data as input to a classification model without the actual transaction class codes, causing the classification model to generate a suggested transaction class code for each of the transactions for which the historical transaction data was provided; adjust weights of the classification model based on differences between the suggested transaction class codes and the actual transaction class codes to produce a trained classification model; and wherein a quantity of the plurality of suggested transaction class codes is dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy.
However, Mandala teaches obtain historical transaction data comprising transactions and corresponding actual transaction class codes (The executable instructions may be configured to receive a training set of transaction data. The training set of transaction data may include two or more training transactions. Each training transaction comprising may include: a business name, a short description, and a transaction category code. paragraphs 3 and 9); provide the obtained historical transaction data as input to a classification model without the actual transaction class codes, causing the classification model to generate a suggested transaction class code for each of the transactions for which the historical transaction data was provided (The program may divide the set into a training set of classified transactions and a test set of unclassified transactions. The program may analyze the training set to generate a document term matrix. The program may analyze the test set and compare the analyzed terms to the document term matrix to predict a classification. The test set may include two or more test transactions. Each test transaction may include: a business name, a short description, an “OTHER” transaction category code, as well as other data. The instructions may compare, through one or more comparison artificial intelligence/machine learning (“cAI/ML”) algorithms, each test unigram to the document term matrix to determine a predicted transaction category code for each test transaction. abstract, paragraphs 13 and 16).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Grigg to include, obtain historical transaction data comprising transactions and corresponding actual transaction class codes; provide the obtained historical transaction data as input to a classification model without the actual transaction class codes, causing the classification model to generate a suggested transaction class code for each of the transactions for which the historical transaction data was provided, as taught in Mandala, in order to intelligently classify unclassified transactions (Mandala, paragraph 1).
However, Wang teaches adjust weights of the classification model based on differences between the suggested transaction class codes and the actual transaction class codes to produce a trained classification model (feedback module 410 provides adjusted weights 412 to machine learning model 470. For example, if a majority of the predictions 474 output by model 470 for transactions included in final embedded matrix 162 (e.g., embedded versions of transactions in training set 132) differ from known labels for transactions in matrix 162, then feedback module 410 adjust weights of model 470 accordingly. paragraph 49).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the machine learning transaction classification model of the combination of Grigg and Mandala to include, adjust weights of the classification model based on differences between the suggested transaction class codes and the actual transaction class codes to produce a trained classification model, as taught in Wang, in order to improves classification accuracy (Wang, paragraph 36).
However, Bar Bliyahu teaches wherein a quantity of the plurality of suggested transaction class codes is dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy (confidence scores 216 may be used to determine which one or more categories are most likely, and a question from the question decision tree that relates to these one or more categories may be selected. For example, if a confidence score for the category of “long term liability” is 0.5 and a confidence score for the category of “other current liability” is 0.3, but neither of these confidence scores exceeds a threshold indicated by the confidence condition (e.g., a threshold of 0.8), then a question that distinguishes between these two categories may be selected. Examiner notes that a quantity of suggested category (i.e., none) is dynamically determined based on the confidence score threshold (i.e, 0.8), which is considered as “a quantity of the plurality of suggested transaction class codes is dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy”, Paragraph 48).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the machine learning transaction classification model of the combination of Grigg, Mandala, and Wang to include, wherein a quantity of the plurality of suggested transaction class codes is dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy, as taught in Bar Bliyahu, in order to automatically categorize a transaction with improved confidence (Bar Bliyahu, paragraph 20).
With regard to claim 10, the combination of references discloses the plurality of suggested transaction class codes is arranged based on relative probability values of the plurality of suggested transaction class codes (Mandala, paragraph 62, the cAI/ML algorithm or algorithms may determine that there is a different probability for each of these three categories, Bar Bliyahu, paragraph 48).
With regard to claim 13, the combination of references discloses the trained classification model is periodically retrained with additional transaction records/ a set of most recent transaction records causing continuously improved performance of the trained classification model in generating suggested class code (Mandala, paragraphs 18 and 68).
With regard to claim 20, the combination of references discloses the trained classification model is periodically retrained with additional transaction records/ a set of most recent transaction records causing continuously improved performance of the trained classification model in generating suggested class code (Mandala, paragraphs 18 and 68).
Claim 3 is rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent No. 8,554,647 to Grigg et al., U.S. Patent Application Publication No. 2024/0135236 to Mandala et al., U.S. Patent Application Publication No. 2022/0343329 to Wang, and U.S. Patent Application Publication No. 2023/0306279 to Bar Elivahu et al., and further in view of U.S. Patent Application Publication No. 2016/0125025 to Blanco.
With regard to claim 3, the combination of references substantially discloses the transaction class codes in the selection comprising the human-readable description (Grigg, abstract, col. 10, lines 31-32, col. 11, lines 29-35), however, the combination of references does not disclose transaction class codes in the selection comprising a numeric value, wherein the user inputs two or more digits of the numeric value and one or more complete transaction class codes are displayed to the user, allowing the user to select the transaction class code from the displayed transaction class codes.
However, Blanco teaches transaction class codes in the selection comprising a numeric value, wherein the user inputs two or more digits of the numeric value and one or more complete transaction class codes are displayed to the user, allowing the user to select the transaction class code from the displayed transaction class codes (The list 800 can include one or more entries 802, where each of the entries 802 includes a numeric classification code 804 and a corresponding description 806. Fig. 8, paragraph 68).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the combination of references to include, transaction class codes in the selection comprising a numeric value, wherein the user inputs two or more digits of the numeric value and one or more complete transaction class codes are displayed to the user, allowing the user to select the transaction class code from the displayed transaction class codes, as taught in Blanco, in order to provide a selection of a most likely classification code (Blanco, paragraph 20).
Claims 14 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent No. 8,554,647 to Grigg et al., U.S. Patent Application Publication No. 2024/0135236 to Mandala et al., U.S. Patent Application Publication No. 2022/0343329 to Wang, and U.S. Patent Application Publication No. 2023/0306279 to Bar Elivahu et al., and further in view of U.S. Patent Application Publication No. 2005/0098623 to Kim.
With regard to claim 14, the combination of references substantially discloses the claimed invention, however, the combination of references does not disclose a plurality of transaction class codes of the account includes a hierarchical structure of class codes and related sub-class codes; and wherein receiving the transaction class code in response to the prompt includes receiving a class code and a related sub-class code indicating a more specific classification than the received transaction class code alone.
However, Kim teaches the plurality of transaction class codes of the account includes a hierarchical structure of class codes and related sub-class codes; and wherein receiving the transaction class code in response to the prompt includes receiving a class code and a related sub-class code indicating a more specific classification than the received transaction class code alone (FIG. 4 shows a structure of transaction classification codes including transaction classification codes using preset binary classification criteria. First, transactions of an enterprise are classified into a transaction with an external side and that with an internal side. Second, every transaction of the enterprise can be classified into a transaction involving profit or loss and a transaction not involving profit and loss. Examiner notes that the multi-aspects of transaction classification structure can be considered as “a hierarchical structure of class codes and related sub-class codes”, Fig. 4, Fig. 7, Fig. 10, paragraphs 81).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the combination of references to include, the plurality of transaction class codes of the account includes a hierarchical structure of class codes and related sub-class codes; and wherein receiving the transaction class code in response to the prompt includes receiving a class code and a related sub-class code indicating a more specific classification than the received transaction class code alone, as taught in Kim, in order to systematically integrate multi-aspects of transaction classification including fundamental characteristics of transaction (Kim, paragraph 18).
With regard to claim 17, the combination of references substantially discloses the claimed invention, however, the combination of references does not disclose the transaction data of the generated transaction statement includes class descriptions for each of the transaction and the plurality of other transactions, wherein the class descriptions are associated with transaction class codes based on a defined class code map.
However, Kim teaches the transaction data of the generated transaction statement includes class descriptions for each of the transaction and the plurality of other transactions, wherein the class descriptions are associated with transaction class codes based on a defined class code map (in case a transaction outline, e.g., `merchandise credit purchase` matched with a transaction classification code "1121200" illustrated in the table 2, is selected by a user and transmitted from the control unit 183 to the journalizing processing unit 182, the journalizing processing unit 182 can automatically extract accounts of D/C from transaction outline information. For example, "merchandise" and "credit purchase" are respectively extracted as a debit and a credit account corresponding to the transaction classification code "1121200" from the automatic processing part in the table 2. Accordingly, accounting information is generated and stored in the accounting information DB 110. The control unit 183 can provide a result of the automatic journalizing to the user through an information displaying screen of a corresponding user interface., Fig. 4, Fig. 7, Fig. 10, paragraphs 92 and 109).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the combination of references to include, the transaction data of the generated transaction statement includes class descriptions for each of the transaction and the plurality of other transactions, wherein the class descriptions are associated with transaction class codes based on a defined class code map, as taught in Kim, in order to systematically integrate multi-aspects of transaction classification including fundamental characteristics of transaction (Kim, paragraph 18).
Response to Arguments
Applicants' arguments filed on 03/10/2026 have been fully considered but they are not fully persuasive especially in light of the previously references applied in the rejections.
Applicants remark that “the combination of references does not disclose the prompt including a human-readable description associated with each of the plurality of the suggested transaction class codes for selection by the user, wherein two or more of the plurality of suggested transaction class codes are dynamically determined based on the associated probability value of each of the plurality of suggested transaction class codes being above or equal to a minimum threshold of probable accuracy; and wherein a value of the minimum threshold of probable accuracy is 0.6”.
Examiner directs Applicants' attention to the office action above.
Conclusion
Please refer to form 892 for cited references.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ARIEL J YU whose telephone number is (571)270-3312. The examiner can normally be reached 11AM - 7PM (M-F).
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, Obeid Fahd A can be reached on 571-270-3324. 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.
/ARIEL J YU/Primary Examiner, Art Unit 3627
Prosecution Insights