DETAILED ACTION
Notice to Applicant
1. The following is a NON-FINAL Office action upon examination of application number 18/178,132 filed on March 03, 2023, in response to Applicant’s Request for Continued Examination (RCE) filed on October 03, 2025. Claims 1-21 are pending in this application and have been examined on the merits discussed below.
2. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Priority
3. Application 18/178,132 filed 03/03/2023 claims Priority from Provisional Application 63268829, filed 03/03/2022.
Response to Amendment
4. In the response filed October 03, 2025, Applicant amended claims 1, 9, 15, and 17, and did not cancel any claim. New claim 21 was presented for examination.
5. Applicant's amendments to the claims are hereby acknowledged. The amendments are not sufficient to overcome the previously issued claim rejection under 35 U.S.C. 101; accordingly, this rejection has been maintained.
Response to Arguments
6. Applicant's arguments filed October 03, 2025, have been fully considered.
7. Applicant submits “The Claims Do Not Recite a Judicial Exception.” [Applicant’s Remarks, 10/03/2025, page 12]
The Examiner respectfully disagrees. In response, it is noted that claim 1 recites steps and concepts falling under the “Mathematical Concepts” grouping of abstract ideas set forth in MPEP 2106. The Examiner maintains that when evaluated under Step 2A Prong One, the “receive input data from a plurality of sources; normalize the received input data; analyze the normalized input data, the analyzing comprising generating an output based on a first input including the normalized input data, a second input including calculation attributes, and a third input including one or more rules; store the output; continuously monitor the output as the output is stored, wherein continuously monitoring comprises: monitoring an output value as the output is stored; determining a threshold value based on the one or more rules; and triggering a restful endpoint when the monitored output value meets or exceeds the determined threshold value: predict, based on the triggering, two or more classifications, each of the two or more classifications comprising a category associated with a raw data of each stored output, wherein the predicting is performed using a model, the model being trained to: embed the raw data of each stored input into a vector representation; recognize patterns or features indicative of particular classes associated with the raw data; generate one or more decision boundaries separating the particular classes; and predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries” are part of the abstract idea itself, i.e., are steps within the “mathematical concepts - mathematical relationships, mathematical formulas or equations, mathematical calculations” group within the enumerated groupings of abstract ideas set forth in MPEP 2106. These steps require some sort of mathematical analysis. It is clear from Applicant’s claims that the method involves mathematical concepts such as mathematical algorithms, mathematical relationships, and calculations. Therefore, this necessarily suggests that the method includes concepts related to “mathematical concepts - mathematical relationships, mathematical formulas or equations, mathematical calculations” group within the enumerated groupings of abstract ideas set forth in MPEP 2106.
Furthermore, the Examiner maintains that the claims set forth or describe steps that can be accomplished mentally such as via human observation and perhaps with the aid of pen and paper, which fall under the “Mental Processes” abstract idea grouping set forth in MPEP 2106. The 101 rejection found the limitations in claim 1 to recite an abstract idea that falls into the “mental processes” based on the limitations receive input data from a plurality of sources; normalize the received input data; analyze the normalized input data, the analyzing comprising generating an output based on a first input including the normalized input data, a second input including calculation attributes, and a third input including one or more rules; store the output; continuously monitor the output as the output is stored, wherein continuously monitoring comprises: monitoring an output value as the output is stored; determining a threshold value based on the one or more rules; and triggering a restful endpoint when the monitored output value meets or exceeds the determined threshold value: predict, based on the triggering, two or more classifications, each of the two or more classifications comprising a category associated with a raw data of each stored output, generate one or more reports based on the raw data and the predicted two or more classes of data for each entry of the raw data.” These limitations recite an abstract idea that falls into the “Mental processes — concepts performed in the human mind (including an observation, evaluation, judgment, opinion)” group within the enumerated groupings of abstract ideas set forth in MPEP 2106. As claimed, the steps can be practically performed mentally, by a human evaluating information. The above noted limitations can be accomplished mentally such as via human observation/judgement perhaps with the aid of pen and paper. These steps describe data gathering, observation, and decision making. Thus, the steps recite the abstract concept of “mental processes.” Therefore, Applicant’s arguments under Step 2A Prong 1 are not persuasive because the claims have been shown to set forth or describe activities falling under the “Mental Processes” abstract idea grouping set forth in MPEP 2106. For the reasons above, this argument is found unpersuasive. The Examiner has addressed the claim amendments in the updated rejection below, responsive to the limitations introduced by the claim amendments.
For the reasons above, Applicant’s argument is not persuasive.
8. Applicant submits “For example, at least the step of “predicting two or more classifications associated with each stored output using a natural language processing (NLP) model, the NLP model being trained to embed the raw data of each stored input into a vector representation|,] recognize patterns or features indicative of particular classes associated with the raw data[,] generate one or more decision boundaries separating the particular classes[,] and predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries,” as recited in claim 1, does not constitute mathematical relationships, formulas, equations, or calculations (e.g., predicting...classes” is not a mathematical relationship, formula, equation, or calculation). Therefore at least this feature of the claims cannot be deemed a mathematical concept. This same feature also cannot be deemed as part of a mental process because it cannot practically be performed in the mind (e.g., “using a...NLP...model” cannot be performed by a human with the aid of pen and paper). Therefore, at least this feature of the claims also cannot be deemed a mental process.” [Applicant’s Remarks, 10/03/2025, pages 12-13]
In response to Applicant’s argument that “at least the step of “predicting two or more classifications associated with each stored output using a natural language processing (NLP) model, the NLP model being trained to embed the raw data of each stored input into a vector representation|,] recognize patterns or features indicative of particular classes associated with the raw data[,] generate one or more decision boundaries separating the particular classes[,] and predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries,” as recited in claim 1, does not constitute mathematical relationships, formulas, equations, or calculations (e.g., predicting...classes” is not a mathematical relationship, formula, equation, or calculation). Therefore at least this feature of the claims cannot be deemed a mathematical concept,” it is noted that the sue of a natural language processing model involved mathematical operations such as vector embedding, pattern recognition, and decision boundary generating. These operations are well-understood algorithmic and mathematical processes. The steps recited effectively describe abstract mathematical operations applied to data, satisfying the criteria for a mathematical concept judicial exception under 35 U.S.C. 101. Accordingly, this argument is found unpersuasive.
Second, in response to Applicant’s argument that “using a...NLP...model” cannot be performed by a human with the aid of pen and paper, it is noted that the fact that NLP operations cannot be practically performed mentally by a human, does not exclude the claim from being directed to an abstract idea. The Examiner has addressed the claim amendments in the updated rejection below, responsive to the limitations introduced by the claim amendments.
9. Applicant submits “Even further, “continuously monitor[ing] the output,” “determining a threshold value based on the one or more rules,” and “triggering a restful endpoint when the monitored output value meets or exceeds the determined threshold value,” as recited in the claims, involves complex computer processing that cannot be performed mentally and also do not set forth or describe any mathematical relationships, calculations, formulas, or equations using words or mathematical symbols.” [Applicant’s Remarks, 10/03/2025, page 14]
The Examiner respectfully disagrees. While the claim recites “continuously monitoring the output,” “determining a threshold value based on the one or more rules,” and “triggering a restful endpoint,” these steps fundamentally involve numerical values, comparing them to predetermined criteria, and initiating an action once the criteria are met. Such steps can practically be performed by a human using pen and paper or simple mental calculations, for instance tracking a value over tie, checking if it exceeds a threshold, and deciding to send a notification. Therefore, these steps fall within the mental process abstract idea grouping, as they describe concepts that humas routinely perform without the need for specialized technology, The mere use of a computer to automate this process does not confer patent eligibility. Accordingly, this argument is found unpersuasive.
10. Applicant submits “Even if the pending claims were directed toward a judicial exception (which they are not), they would still be patent-eligible under Prong Two of Step 2A because they integrate any alleged exception into a practical application. See MPEP § 2106.04(d). Specifically, the pending claims present additional elements that reflect “an improvement in the functioning of a computer,” and “appl[y] or use|] the judicial exception in [a] meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception.” MPEP § 2106.04(d)(I).” [Applicant’s Remarks, 10/03/2025, page 15]
In response to Applicant’s assertion that “the claims integrate any alleged judicial exception into a practical application,” the Examiner respectfully disagrees. Under Step 2A Prong Two of the eligibility inquiry, any additional elements are evaluated individually and in combination to determine whether they integrate the judicial exception into a practical application, with consideration of the following exemplary considerations that may be indicative of a practical application: an additional element that reflects an improvement to the functioning of a computer or to any other technology or technical field, applying the exception with a particular machine, applying the judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition, effecting a transformation of a particular article to a different state or thing, and applying or using the judicial exception some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment.
In this instance, the additional elements recited in exemplary claim 1 are: a memory, at least one data storage medium, at least one processor, logic, a natural language processing model, and a user interface. These elements have been considered individually and in combination, however these computing elements amount to using a generic computer programmed with computer-executable instructions/software to perform the abstract idea, similar to adding the words “apply it” (or an equivalent), which merely serves to link the use of the judicial exception to a particular technological environment, which is not sufficient to amount to a practical application, as noted in MPEP 2106. See also MPEP 2106.05(f) and 2106.05(h). Furthermore, these additional elements fail to provide an improvement to the functioning of a computer or to any other technology or technical field, fail to apply the exception with a particular machine, fail to apply the judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition, fail to effect a transformation of a particular article to a different state or thing, and fail to apply/use the abstract idea in a meaningful way beyond generally linking the use of the judicial exception to a particular technological environment. Instead, the memory, at least one data storage medium, at least one processor, logic, a natural language processing (NLP) model, and a user interface (claim 1); amounts to using generic computing devices as tools to implement the abstract idea, which does not amount to a technological improvement or otherwise indicate a practical application. See MPEP 2106.05(f). Notably, Applicant’s Specification acknowledges that the invention may be implement with generic computing devices. See Specification, paragraphs 0076.
Even assuming arguendo that an improvement was achieved, improving the method for providing regulatory insights analysis using only generic computing devices does not improve the computing devices or any technology, but instead any incidental improvement achieved by automating the claim steps would come from the capabilities of a general-purpose computer rather than the sequence of steps/activities recited in the method itself, which does not materially alter the patent eligibility of the claim. See Bancorp Servs., L.L.C. v. Sun Life Assurance Co. of Can. (U.S.), 687 F.3d 1266, 1278 (Fed. Cir. 2012) (“[T]he fact that the required calculations could be performed more efficiently via a computer does not materially alter the patent eligibility of the claimed subject matter.”) (cited in the Federal Circuit's FairWarning decision). Accordingly, the generic computing elements do not integrate the judicial exception into a practical application.
Moreover, allowing a user to view one or more generated reports provides a benefit to the end user, not an improvement to the technology.
Lastly, in response to Applicant’s argument that “the pending claims…appl[y] or use[] the judicial exception in [a] meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception,” it is noted that 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 that the recited abstract idea is a 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 the reasons above, this argument is found unpersuasive.
11. Applicant submits “the claims integrate any alleged abstract idea into a practical application by reciting specific improvements to computer functionality because the claims recite an automated system that uses a model trained to, e.g., “process documents or portions of documents” and “classify them to a broader taxonomy or category” (i.e., “two or more different classes of data”). Even further, and based on the “predicted two or more difference classes of data,” the system may also, e.g., “generate summaries in coherent text” and explain “key drivers or factors that underly a given output metric.” See, e.g., Applicant’s specification at Paragraph [0053].” [Applicant’s Remarks, 10/03/2025, page 18]
The Examiner respectfully disagree. Applicant’s argument that the claims improve computer functionality by automating document processing, classification, and summary generation is not persuasive. These claimed functions involve organizing information and producing outputs that are conventional steps in data handling and analysis. The claims do not specify any technical improvements to computer hardware, or software systems, nor do they describe enhancement to how the computer performed these tasks. Simply implementing these abstract ideas on a generic computer does not satisfy the requirement to integrate the abstract idea into a practical application that improves computer functionality. The claim lacks limitations that would show a technical solution that advances the operation of the computer or a particular technological field. Accordingly, this argument is found unpersuasive.
Further, in response to Applicant’s argument that “the combination of vector embedding, pattern recognition, decision boundary generation, and class predictions works together with the natural language processing (NLP) model to provide a specific technological solution that goes beyond merely linking abstract ideas to a technological environment” [Remarks at page 19], it is noted that these steps describe standard machine learning techniques applied to data processing and do not specify any technical improvement to the functioning of the computer itself. Accordingly, this argument is found unpersuasive.
12. Applicant submits “the specific arrangement…provides meaningful limits that go beyond merely linking abstract ideas to a technological environment. Thus, rather than reciting the steps at a high level of generality, amended claim 1 provides a specific, discrete, and interconnected implementation that leverages machine learning and continuous monitoring to solve the technical problem of proactively identifying regulatory compliance issues before they materialize. For these additional reasons, the claims also amount to “significantly more” than any abstract idea, and the rejection of claim 1 based on an abstract idea under 35 U.S.C. § 101 should be withdrawn additionally for this reason.” [Applicant’s Remarks, 10/03/2025, page 22]
The Examiner respectfully disagrees. Applicant’s assertion that the specific arrangement of machine learning and continuous monitoring provides meaningful limits beyond the abstract ideas is not persuasive. The claimed combination recites routine functions, machine learning for classification and threshold-based monitoring, tata are widely used in various technical and non-technical fields. The claims does not recite any specific, unconventional configuration or improvement to the underlying technology or computer architecture. Merely applying generic machine learning techniques alongside continuous monitoring does not constitute “significantly more” under 35 U.S.C. 101. Without concrete technical improvements, the claim remains directed to an abstract idea. Accordingly, this argument is found unpersuasive.
13. Applicant submits “Soleimani also fails to teach any “predict[ing of classifications], based on the triggering…” Soleimani fails to teach any causal relationship between a triggering event and the prediction of classifications. While Soleimani may generally describe classification functionality, it does not specifically teach “predicting” in response to a “triggering.” Soleimani lacks any disclosure of the temporal or causal relationship where a classification prediction is initiated as a direct result of “triggering a restful endpoint,” as recited in claim 1.” [Applicant’s Remarks, 10/03/2025, pages 24-25]
The Examiner respectfully disagrees. In response to Applicant’s argument it is noted that Soleimani discloses the use of trained machine learning models to analyze data and provide classification outputs (col. 29, lines 18-26), including classifying textual datasets (col. 31, lines 27-50). Soleimani further teaches that such models may be used in various application including natural langue processing and real-time or near-real time processing (cl.. 27, lines 58-57), which reasonably compromises the type of prediction and classification recited in claim 1.
Soleimani clearly describes applying a trained model upon receiving new input data and generating classification in response. Under the broadest reasonable interpretation, the act of receiving input ad initiating the classification process can be understood as a triggering event. Thus, given the broadest reasonable interpretation consistent with the specification in construing the claimed invention, it is Examiner’s position that the disclosure of Soleimani teaches the disputed limitation. Accordingly, this argument is found unpersuasive.
14. Applicant submits “Soleimani does not specifically teach training an NLP model to “recognize patterns or features indicative of a particular class” or to “predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries,” as recited in claim 1.” [Applicant’s Remarks, 10/03/2025, page 25]
The Examiner respectfully disagrees. In response to Applicant’s argument that “Soleimani does not specifically teach training an NLP model to recognize patterns or features indicative of a particular class,” it is noted that Soleimani expressly discusses machine learning models that “learn, categorize, and make predictions about data” and that can “identify patterns or trends in input data” (col. 27, lines 26-57). Soleimani further explains that during training, input data is iteratively supplied to the machine learning model “to enable the machine-learning model to identify patterns related to the input data or to identify relationships between the input data and output data” (col. 28, lines 15-29).
Additionally, Soleimani describes that the model may be trained using supervised or unsupervised learning technique, including transformer and other neural network architectures suitable for natural language processing (col. 31, lines 27-43). Under the broadest reasonable interpretation, identifying patterns during training for classification purposes reasonably encompasses recognizing features indicative of a particular class, as recited in claim 1. Thus, given the broadest reasonable interpretation consistent with the specification in construing the claimed invention, it is Examiner’s position that the disclosure of Soleimani teaches the disputed limitation. Accordingly, this argument is found unpersuasive.
Last, in response to the Applicant’s argument “Soleimani does not specifically teach training an NLP model to “predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries,” as recited in claim 1,” the Examiner notes the limitation being argued by Applicant as being newly amended to the claims in the response filed 10/03/2025, which has been addressed in the updated rejection below. Applicant’s argument has been considered, but it pertains to amendments to independent claim 1 that are believed to be addressed via the updated ground of rejection under §103 set forth in the instant Office action, which incorporates new citations to address the amended limitations in claim 1 and supports a conclusion of obviousness of the amended claims.
15. Applicant submits “the applied prior art fails to teach or suggest training the NLP model to “embed the raw data of each stored input into a vector representation; recognize patterns or features indicative of particular classes associated with the raw data; generate one or more decision boundaries separating the particular classes; and predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries” as also recited in amended claim 1.” [Applicant’s Remarks, 10/03/2025, pages 25-26]
In response to the Applicant’s argument “the applied prior art fails to teach or suggest training the NLP model to “embed the raw data of each stored input into a vector representation; recognize patterns or features indicative of particular classes associated with the raw data; generate one or more decision boundaries separating the particular classes; and predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries” as also recited in amended claim 1,” it is noted that this argument is a mere allegation of patentability by the Applicant with no supporting rationale or explanation. Merely stating that the claims do not teach a feature does not offer any insight as to why the specific sections of the prior art relied upon by the Examiner fail to disclose the claimed features. Applicant's arguments amount to a general allegation that the claims define a patentable invention without specifically pointing out how the language of the claims patentably distinguishes them from the references. Moreover, the Examiner notes the limitations being argued by Applicant as being newly amended to the claims in the response filed 10/03/2025, which have been addressed in the updated rejection below. Applicant’s argument has been considered, but it pertains to amendments to independent claim 1 that are believed to be addressed via the updated ground of rejection under §103 set forth in the instant Office action, which incorporates new citations to address the amended limitations in claim 1 and supports a conclusion of obviousness of the amended claims.
16. Applicant’s remaining arguments either logically depend from the above-rejected arguments, in which case they too are unpersuasive for the reasons set forth above, or they are directed to features which have been newly added via amendment. Therefore, this is now the Examiner's first opportunity to consider these limitations and as such any arguments regarding these limitations would be inappropriate since they have not yet been examined. A full rejection of these limitations will be presented later in this Office Action.
Examiner’s Note
17. The amendment document filed on October 03, 2025 is considered non-compliant because it has failed to meet the requirements of 37 CFR 1.121 or 1.4. Specifically, the claim amendments are not properly indicated. The status indicator of claim 15 should indicate “Currently Amended” instead of “Previously Presented” since claim 15 was amended. Examiner points out that all claims being currently amended in an amendment paper shall be presented in the claim listing, indicate a status of “currently amended,” and be submitted with markings to indicate the changes that have been made relative to the immediate prior version of the claims. The text of any added subject matter must be shown by underlining the added text (i.e. the text of the original subject matter should not be underlined). The text of any deleted matter must be shown by strike-through except that double brackets placed before and after the deleted characters may be used to show deletion of five or fewer consecutive characters. The text of any deleted subject matter must be shown by being placed within double brackets if strike-through cannot be easily perceived. Only claims having the status of “currently amended,” or “withdrawn” if also being amended, shall include markings. For further amendments and for further explanation of the amendment format required by 37 CFR 1.121, see MPEP § 714.
Claim Rejections - 35 USC § 101
18. 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.
19. Claims 1-21 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
20. Claims 1-21 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The eligibility analysis in support of these findings is provided below, in accordance with MPEP 2106.
With respect to Step 1 of the eligibility inquiry (as explained in MPEP 2106), it is first noted that the system (claims 1-8, 21), method (claims 9-16), and non-transitory computer readable medium (claims 17-20), and is directed to at least one potentially eligible category of subject matter (i.e., machine, process, and article of manufacture, respectively). Thus, Step 1 of the Subject Matter Eligibility test for claims 1-21 is satisfied.
With respect to Step 2A Prong One, it is next noted that the claims recite abstract ideas that fall into the “Mathematical Concepts” such as mathematical relationships, formulas and calculations and “Mental Processes” or concepts performed in the human mind such as via observation, evaluation, and judgment, as set forth in the enumerated groupings of abstract ideas set forth in MPEP 2106 the limitations for managing personal behavior or interactions or following rules or instructions for assigning human agents to tasks. With respect to independent claim 1, the limitations reciting the abstract idea are indicated in bold below: a memory, at least one data storage medium, and at least one processor configured to: receive input data from a plurality of sources; normalize the received input data; analyze the normalized input data, the analyzing comprising using logic for generating an output based on a first input including the normalized input data, a second input including calculation attributes, and a third input including one or more rules; store the output; continuously monitor the output as the output is stored, wherein continuously monitoring comprises: monitoring an output value as the output is stored; determining a threshold value based on the one or more rules; and triggering a restful endpoint when the monitored output value meets or exceeds the determined threshold value: predict, based on the triggering, two or more classifications, each of the two or more classifications comprising a category associated with a raw data of each stored output, wherein the predicting is performed using a natural language processing (NLP) model, the NLP model being trained to: embed the raw data of each stored input into a vector representation; recognize patterns or features indicative of particular classes associated with the raw data; generate one or more decision boundaries separating the particular classes; and predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries; generate one or more reports based on the raw data and the predicted two or more classes of data for each entry of the raw data; and receive, from a user and via a user interface: additional input data; a request to view the one or more generated reports; or a request for an additional output. These steps can be performed by a human with the aid of pen and paper, and are therefore a mental step, and also are performable as mathematical relationships, formulas, equations, and/or calculations. Considered together, these steps set forth an abstract idea falling under the “Mathematical Concepts” and “Mental Processes” abstract idea groupings set forth in MPEP 2106.
Because the above-noted limitations recite steps falling within the “Mathematical Concepts” abstract idea grouping and the “Mental Processes” abstract idea grouping, they have been determined to recite at least one abstract idea when evaluated under Step 2A Prong One of the eligibility inquiry. Independent claims 9 and 17 recite similar limitations as those recited in claim 1 and therefore are found to recite the same abstract idea(s) as claim 1.
With respect to Step 2A Prong Two, the judicial exception is not integrated into a practical application. With respect to independent claims 1, 9, and 17, the additional elements are: a memory, at least one data storage medium, at least one processor, logic, a natural language processing (NLP) model, the NLP model being trained, and a user interface (claim 1); logic, a natural language processing (NLP) model, the NLP model being trained, and a user interface (claim 9), instructions, at least one processor, logic, a natural language processing (NLP) model, the NLP model being trained, and a user interface (claim 17). These additional elements have been evaluated, but fail to integrate the abstract idea into a practical application because they amount to using generic computing elements or computer-executable instructions (software) to perform the abstract idea, similar to adding the words “apply it” (or an equivalent), and merely serve to link the use of the judicial exception to a particular technological environment. See MPEP 2106.05(f) and 2106.05(h). Even if the “receiving” steps are evaluated as additional elements, these steps amount at most to insignificant extra-solution activity, which is not indicative of a practical application, as noted in MPEP 2106.05(g). In addition, these limitations fail to provide an improvement to the functioning of a computer or to any other technology or technical field, fail to apply the exception with a particular machine, fail to apply the judicial exception to effect a particular treatment or prophylaxis for a disease or medical condition, fail to effect a transformation of a particular article to a different state or thing, and fail to apply/use the abstract idea in a meaningful way beyond generally linking the use of the judicial exception to a particular technological environment.
Accordingly, because the Step 2A Prong One and Prong Two analysis resulted in the conclusion that the claims are directed to an abstract idea, additional analysis under Step 2B of the eligibility inquiry must be conducted in order to determine whether any claim element or combination of elements amount to significantly more than the judicial exception.
With respect to Step 2B of the eligibility inquiry, it has been determined that the claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. With respect to independent claims 1, 9, and 17, the additional elements are: a memory, at least one data storage medium, at least one processor, logic, a natural language processing (NLP) model, the NLP model being trained, and a user interface (claim 1); logic, a natural language processing (NLP) model, the NLP model being trained, and a user interface (claim 9), instructions, at least one processor, logic, a natural language processing (NLP) model, the NLP model being trained, and a user interface (claim 17). These elements have been considered individually and in combination, but fail to add significantly more to the claims because they amount to using generic computing elements or instructions (software) to perform the abstract idea, similar to adding the words “apply it” (or an equivalent), and merely serve to link the use of the judicial exception to a particular technological environment and does not amount to significantly more than the abstract idea itself. Notably, Applicant’s Specification describes that generic computer devices that may be used to implement the invention, which cover virtually any computing device under the sun (Specification at paragraph [0076]). Accordingly, the generic computer involvement in performing the claim steps merely serves to generally link the use of the judicial exception to a particular technological environment, which does not add significantly more to the claim. See, e.g., Alice Corp., 134 S. Ct. 2347, 110 USPQ2d 1976.).
With respect to the “receiving” steps, even if considered as additional elements, these steps at most amounts to receiving/transmitting data, which has been recognized as well-understood, routine, and conventional, and thus insufficient to add significantly more to the abstract idea. See MPEP 2106.05(d) - Receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016) (using a telephone for image transmission); OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015) (sending messages over a network); buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014).
With respect to reliance on “natural language processing (NLP),” this activity is recognized as well-understood, routine, and conventional in the art, which does not amount to significantly more than the abstract idea itself. See, e.g., Morsa, Pub. No.: US 2006/0085408 (paragraph 0144: well -known-to-the-arts natural language processing (NLP) (computational linguistics) or some other method as is well known to the arts may be used). See also, Szabo, Patent No. 5,966,126 (col. 6, lines 57-62 and col. 28, lines 16-19: e.g., definitions may be produced in known manner, such as by explicit definition, or through use of assistive technologies, such as natural language translators; user defines a search using prior known techniques, such as natural language searching). See also, Schutt et al., Pub. No.: US 2023/0208880 (paragraph 0025: a conventional or other natural language processing model may be applied to the unstructured text to extract the identities of the components and/or relationships between components.).
In addition, when taken as an ordered combination, the ordered combination adds nothing that is not already present as when the elements are taken individually. There is no indication that the combination of elements integrates the abstract idea into a practical application. Their collective functions merely provide generic computer implementation. Therefore, when viewed as a whole, these additional claim elements do not provide meaningful limitations to transform the abstract idea into a practical application of the abstract idea or that, as an ordered combination, amount to significantly more than the abstract idea itself.
Dependent claims 2-8, 10-16, and 18-21 recite the same abstract idea as recited in the independent claims, and when evaluated under Step 2A Prong One are found to merely recite details that serve to narrow the same abstract idea recited in the independent claims accompanied by the same generic computing elements or software as those addressed above in the discussion of the independent claims, which is not sufficient to amount to a practical application or add significantly more, or other additional elements that fail to amount to a practical application or add significantly more, as noted above. In particular, dependent claims 2-8 and 21 recite “store the normalized input data prior to analyzing the normalized input data,” “generate the calculation attributes based on the normalized input data from at least one of the plurality of sources; and store the calculation attributes,” “receive the one or more rules as configured by the user; and store the one or more rules,” “using the stored output or the predicted two or more classifications,” “wherein triggering the restful endpoint provides at least one additional function based on the monitored output value,” “distribute the stored output or the predicted two or more classifications,” “provide insights based on the stored output or the predicted two or more classifications using at least one of a model or a pipeline generated,” “wherein the one or more reports include an interactive visualization based on the continuously monitored output and the predicted two or more classes of data for each entry of the raw data,” however these limitations are part of the same abstract idea as addressed in the independent claims that falls within the “Mathematical Concepts” and “Mental Processes” abstract idea groupings. The other dependent claims have been evaluated as well, but similar to dependent claims 2-8 and 21, recite details/steps that merely refine the same abstract idea recite in the independent claims. Accordingly, these steps are part of the same abstract idea(s) set forth in the independent claims. Dependent claim 2, 5, 7-8, 10, 13, 15-16, 18-19, and 21 recite additional elements of a transient data storage, build at least one of a data mart or a data lake, multiple client-side devices, a machine learning platform, and the trained NLP model. However, when evaluated under Step 2A Prong Two and Step 2B, the additional elements do not amount to a practical application or significantly more since they merely require generic computing devices (or computer-implemented instructions/code) which as noted in the discussion of the independent claims above is not enough to render the claims as eligible.
The ordered combination of elements in the dependent claims (including the limitations inherited from the parent claim(s)) add nothing that is not already present as when the elements are taken individually. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide generic computer implementation. Accordingly, the subject matter encompassed by the dependent claims fails to amount to a practical application or significantly more than the abstract idea itself.
For more information, see MPEP 2106.
Claim Rejections - 35 USC § 103
21. 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.
22. 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.
23. The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
24. This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
25. Claims 1, 3, 5-6, 8-9, 11, 13-14, 16-18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Hanson et al., Pub. No.: US 2023/0267105 A1, [hereinafter Hanson], in view of Cai et al., Pub. No.: US 2019/0347282 A1, [hereinafter Cai], in further view of Soleimani et al., Patent No.: US 11,501,084 B1, [hereinafter Soleimani].
As per claim 1, Hanson teaches a system for providing regulatory insight analysis (paragraphs 0002, 0033), comprising:
a memory (paragraph 0071: “The techniques described herein may be implemented in one or more computer programs executing on (or executable by) a programmable computer or electronic device having any combination of any number of the following: a processor, a storage medium readable and/or writable by the processor (including, for example, volatile and non-volatile memory and/or storage elements), an input device, an output device, and a display.”; paragraph 0072),
at least one data storage medium (paragraph 0071: “The techniques described herein may be implemented in one or more computer programs executing on (or executable by) a programmable computer or electronic device having any combination of any number of the following: a processor, a storage medium readable and/or writable by the processor (including, for example, volatile and non-volatile memory and/or storage elements), an input device, an output device, and a display; paragraph 0078), and
at least one processor (paragraph 0071: “The techniques described herein may be implemented in one or more computer programs executing on (or executable by) a programmable computer or electronic device having any combination of any number of the following: a processor, a storage medium readable and/or writable by the processor…”; paragraphs 0072, 0074) configured to:
receive input data from a plurality of sources (paragraph 0007: “The present invention is directed to a data aggregation and normalization system for aggregating data from disparate data sources…”; paragraph 0040, discussing a data aggregation and normalization system for collecting, collating or aggregating data, such as for example financial and non-financial data, from a variety of different data sources…As shown, the data aggregation and normalization system can include a plurality of data sources…The data acquired by the data sources can be conveyed through any suitable data connection);
normalize the received input data (paragraph 0034: “a data aggregation and normalization system for aggregating data from disparate data sources, processing the data to clean the data and to normalize or standardize the data using one or more data models”; paragraph 0042, discussing that the common data model can serve to conform, organize, and normalize elements of data…; paragraph 0054, discussing that the data aggregation and normalization system employs the illustrated units and modules to form a complete, efficient, and robust data normalization unit for automatically extracting, cleaning, and normalizing data for subsequent use by a reporting unit);
analyze the normalized input data (paragraph 0034, discussing a data aggregation and normalization system for aggregating data from disparate data sources, processing the data to clean the data and to normalize or standardize the data using one or more data models, and then applying one or more discrete machine learning techniques to the normalized data to provide meaningful data insights and predictions…; paragraph 0054, discussing that the data aggregation and normalization system employs the illustrated units and modules to form a complete, efficient, and robust data normalization unit for automatically extracting, cleaning, and normalizing data for subsequent use by a reporting unit),
the analyzing comprising using logic for generating an output based on a first input including the normalized input data, a second input including calculation attributes, and a third input including one or more rules (paragraph 0034, discussing a data aggregation and normalization system for aggregating data from disparate data sources, processing the data to clean the data and to normalize or standardize the data using one or more data models, and then applying one or more discrete machine learning techniques to the normalized data to provide meaningful data insights and predictions. The normalized data can also be processed by one or more reporting modules to provide one or more customized reports to an end user; paragraph 0038, discussing that the term “financial unit, “financial subsystem,” “financial system” or “financial infrastructure” is intended to include any unit implemented in hardware, software or a combination thereof that applies financial rules and models to data of any type, including financial data and environmental data, so as generate one or more financial reports; paragraph 0043, discussing that the data preprocessing and enrichment unit can also employ an assessment unit for assessing the data quality of the cleaned data in the common data model by determining or identifying the data that is anomalous. This can be performed by analyzing historical data and then detecting discrepancies, or can employ if desired data from third party data sources that can be employed to detect anomalies in the cleaned data…; paragraph 0044, discussing that the data lineage unit can also be configured to employ one or more business rules for applying selected different business rules to the cleaned data…Business rules modify the raw data to prepare the data for later applications by both updating exiting data and calculating new data based on the predefined business rules. The cleaned and enriched data can then be stored in the data storage unit; paragraph 0063, discussing that the entropy of the data can be measured and quantified according to known techniques, and can be calculated into an entropy value that typically ranges between 0 and 1 or higher, where the higher the number corresponds to a higher amount of disorder in the data or system. The entropy values can be collated to form a distribution of entropy values. For example, each of the data segments or tiers of data segments can be represented as a node in the decision tree, and an entropy value can be calculated for the subpopulation distribution; paragraphs 0046, 0059, 0064);
store the output (paragraph 0040, discussing that the data storage unit can also be configured to store processed data in addition to the raw data. The data storage unit can be constructed as a single data store for storing raw and processed data that can be subsequently used for tasks such as reporting, visualization, advanced analytics, machine learning, and the like. The data storage unit can employ, according to one practice, multiple different data buckets that provides a place to store extracted data (e.g., raw data), a place to store cleaned data, provides a workspace for AI/ML modeling processing and a storage area for machine language models, prediction units, and data associated therewith or generated thereby);
generate one or more reports (paragraph 0015, discussing applying one or more selected artificial intelligence and machine learning (AI/ML) techniques to selected portions of the cleaned data to form machine language data, wherein the one or more selected artificial intelligence and machine learning (AI/ML) techniques is stored in a machine language module having a plurality of predefined machine learning units, transforming the machine language data into a selected reporting format, and generating with a reporting unit one or more reports from the data in the reporting format; paragraph 0053, discussing that the transformed data can then be conveyed to one or more reporting or visual representation software applications stored in or which form part of the reporting unit via the API layer. The API layer allows the transformed data and other system software applications to communicate with the applications of the reporting unit, as well as with external third party applications. The reporting unit can employ one or more reporting applications that can be configured for generating one or more reports, including financial reports, based on the transformed data; paragraph 0034); and
receive, from a user and via a user interface: additional input data; a request to view the one or more generated reports; or a request for an additional output (paragraph 0034, discussing that the normalized data can also be processed by one or more reporting modules to provide one or more customized reports to an end user; paragraph 0048, discussing that the reporting unit can include an application programming interface for enabling selected reporting software applications to interface with the transformed data. The reporting software applications can include any selected commercially available or custom reporting applications that generate selected user interfaces for reporting and displaying selected information; paragraph 0053, discussing that the transformed data can then be conveyed to one or more reporting or visual representation software applications stored in or which form part of the reporting unit via the API layer. The API layer allows the transformed data and other system software applications to communicate with the applications of the reporting unit, as well as with external third party applications. The reporting unit can employ one or more reporting applications that can be configured for generating one or more reports, including financial reports, based on the transformed data. Further, a system user can interface with the reporting unit so as to construct a selected report; paragraphs 0049, 0068).
While Hanson describes follow-up analysis (paragraph 0068), Hanson does not explicitly teach continuously monitor the output as the output is stored, wherein continuously monitoring comprises: monitoring an output value as the output is stored; determining a threshold value based on the one or more rules; and triggering a restful endpoint when the monitored output value meets or exceeds the determined threshold value; predict, based on the triggering, two or more classifications, each of the two or more classifications comprising a category associated with a raw data of each stored output, wherein the predicting is performed using a natural language processing (NLP) model, the NLP model being trained to: embed the raw data of each stored input into a vector representation; recognize patterns or features indicative of particular classes associated with the raw data; generate one or more decision boundaries separating the particular classes; and predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries; generate one or more reports based on the raw data and the predicted two or more classes of data for each entry of the raw data. Cai in the analogous at of data analytics systems teaches:
continuously monitor the output as the output is stored (paragraph 0140, discussing that the platform can have comprehensive documentation support. The platform can facilitate ease of use and model customization. The platform can provide in-database analytics and comprehensive user/data/project access/storage management at the enterprise level. The platform can provide model live scoring as a service. The platform can provide model output monitoring and alerting; paragraph 0150, discussing that the visual elements can be displayed interface application. The user can interact with different components of the interface application drill down on different kinds of information or events. This can allow for faster monitoring and can also include visual elements representing different predictive metrics),
wherein continuously monitoring comprises: monitoring an output value as the output is stored (paragraph 0140, discussing that the platform can have comprehensive documentation support. The platform can facilitate ease of use and model customization. The platform can provide in-database analytics and comprehensive user/data/project access/storage management at the enterprise level. The platform can provide model live scoring as a service. The platform can provide model output monitoring and alerting; paragraph 0150, discussing that the visual elements can be displayed interface application. The user can interact with different components of the interface application drill down on different kinds of information or events. This can allow for faster monitoring and can also include visual elements representing different predictive metrics);
determining a threshold value based on the one or more rules (paragraph 0073, discussing that incident management platform can use machine learning processes to identify hidden relationships or patterns connecting different data points and trigger execution on future similar scenarios. Incident management platform can enable both business and IT users to augment human capabilities. Incident management platform uses operational risk models to predict operational risk events that could cause impact from a financial, reputational, operational or regulatory perspective. Incident management platform models predictive models and event detection to detect user access anomalies or intrusion detection for infrastructure resources; paragraph 0096, discussing that the processor is configured to output the prescriptive solution to a virtual agent. In this example, the virtual agent may then provide a graphical user interface or other output to indicate to the user what a potential solution is, and the data structure of potential solutions can be traversed to provide potential alternate solutions, provided such solutions have confidence scores that are above a pre-defined threshold. For example, the virtual agent may receive an incident string, and upon traversing historical incident tickets, respond with “There have been three successful resolutions of similar problems. The most likely solution is to allocate additional memory to the page file”); and
triggering a restful endpoint when the monitored output value meets or exceeds the determined threshold value (paragraph 0096, discussing that the processor is configured to output the prescriptive solution to a virtual agent. In this example, the virtual agent may then provide a graphical user interface or other output to indicate to the user what a potential solution is, and the data structure of potential solutions can be traversed to provide potential alternate solutions, provided such solutions have confidence scores that are above a pre-defined threshold. For example, the virtual agent may receive an incident string, and upon traversing historical incident tickets, respond with “There have been three successful resolutions of similar problems. The most likely solution is to allocate additional memory to the page file”; paragraph 0154, discussing that the interface includes an overall security compliance indicia to trigger the display of visual elements for security compliance reporting for an application server or database. The interface includes graph data export indicia to export data for the visual elements of the network topology diagram in different formats; paragraph 0082, discussing that the platform can implement an Incident Ticket Volume Prediction process using predictive models. The Prediction Process can include instructions to implement the following operations: receive data from service management repository for all incident tickets from the past week or other time period at a folder with transfer trigger by schedule).
Hanson is directed towards systems and methods for aggregating, enriching and normalizing data. Cai is directed towards a data processing system. Therefore they are deemed to be analogous as they both are directed towards data processing systems. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hanson with Cai because the references are analogous art because they are both directed to solutions for data processing, which falls within applicant’s field of endeavor (system for providing regulatory insight analysis), and because modifying Hanson to include Cai’s features for including continuously monitor the output as the output is stored, wherein continuously monitoring comprises: monitoring an output value as the output is stored; determining a threshold value based on the one or more rules; and triggering a restful endpoint when the monitored output value meets or exceeds the determined threshold value, in the manner claimed, would serve the motivation of facilitating a culture of data driven decision making (Cai at paragraph 0075); or in the pursuit of properly and efficiently capturing, aggregating and organizing data so that the data can be later used in meaningful ways (Hanson at paragraph 0003); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
The Hanson-Cai combination does not explicitly teach predict, based on the triggering, two or more classifications, each of the two or more classifications comprising a category associated with a raw data of each stored output, wherein the predicting is performed using a natural language processing (NLP) model, the NLP model being trained to: embed the raw data of each stored input into a vector representation; recognize patterns or features indicative of particular classes associated with the raw data; generate one or more decision boundaries separating the particular classes; and predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries; and generate one or more reports based on the raw data and the predicted two or more classes of data for each entry of the raw data. However, Soleimani in the analogous at of data analytics systems teaches:
predict, based on the triggering, two or more classifications, each of the two or more classifications comprising a category associated with a raw data of each stored output, wherein the predicting is performed using a natural language processing (NLP) model (col. 27, lines 58-67 & col. 28, lines 1-5, discussing that different machine-learning models may be used interchangeably to perform a task. Examples of tasks that can be performed at least partially using machine-learning models include various types of scoring; bioinformatics; cheminformatics; software engineering; fraud detection; customer segmentation; generating online recommendations; adaptive websites; determining customer lifetime value; search engines; placing advertisements in real time or near real time; classifying DNA sequences; affective computing; performing natural language processing and understanding; object recognition and computer vision…; col. 29, lines 18-26, discussing that the trained machine-learning model is used to analyze the new data and provide a result. For example, the new data can be provided as input to the trained machine-learning model. The trained machine-learning model can analyze the new data and provide a result that includes a classification of the new data into a particular class, a clustering of the new data into a particular group, a prediction based on the new data, or any combination of these; col. 31, lines 27-43, discussing that a processor provides a textual dataset as input to a machine-learning model. In some examples, the textual dataset can be a long document. A document can be any file that includes textual data...The machine-learning model can be of any suitable type and may have been previously trained using any suitable training technique, such as supervised or unsupervised learning. For example, the machine-learning model can be a transformer model, such as a Bidirectional Encoder Representations from Transformers (BERT) model designed for natural language processing. Other examples of the machine-learning model can include attention models, recurrent neural networks, convolutional neural networks, linear classifiers, or any combination of these; col. 31, lines 44-50, discussing that the processor receives an output from the machine-learning model indicating an overall classification for the textual dataset),
the NLP model being trained to: embed the raw data of each stored input into a vector representation (col. 27, lines 26-57, discussing a process for generating and using a machine-learning model according to some aspects. Machine learning is a branch of artificial intelligence that relates to mathematical models that can learn from, categorize, and make predictions about data. Such mathematical models, which can be referred to as machine-learning models, can classify input data among two or more classes; cluster input data among two or more groups; predict a result based on input data; identify patterns or trends in input data; identify a distribution of input data in a space; or any combination of these. Examples of machine-learning models can include (i) neural networks; (ii) decision trees, such as classification trees and regression trees; (iii) classifiers, such as naïve bias classifiers, logistic regression classifiers, ridge regression classifiers, random forest classifiers, least absolute shrinkage and selector (LASSO) classifiers, and support vector machines; (iv) clusterers, such as k-means clusterers, mean-shift clusterers, and spectral clusterers; (v) factorizers, such as factorization machines, principal component analyzers and kernel principal component analyzers; and (vi) ensembles or other combinations of machine-learning models…; col. 28, lines 41-53, discussing that a machine-learning model is trained using the training data. The machine-learning model can be trained in a supervised, unsupervised, or semi-supervised manner. In supervised training, each input in the training data is correlated to a desired output. This desired output may be a scalar, a vector, or a different type of data structure.. This may enable the machine-learning model to learn a mapping between the inputs and desired outputs. In unsupervised training, the training data includes inputs, but not desired outputs, so that the machine-learning model has to find structure in the inputs on its own. In semi-supervised training, only some of the inputs in the training data are correlated to desired outputs; col. 30, lines 15-44, discussing that in some examples, the neural network operates by receiving a vector of numbers from one layer; transforming the vector of numbers into a new vector of numbers using a matrix of numeric weights, a nonlinearity, or both; and providing the new vector of numbers to a subsequent layer of the neural network. Each subsequent layer of the neural network can repeat this process until the neural network outputs a final result at the output layer. For example, the neural network can receive a vector of numbers as an input at the input layer. The neural network can multiply the vector of numbers by a matrix of numeric weights to determine a weighted vector. The matrix of numeric weights can be tuned during the training of the neural network. The neural network can transform the weighted vector using a nonlinearity, such as a sigmoid tangent or the hyperbolic tangent);
recognize patterns or features indicative of particular classes associated with the raw data (col. 6, lines 46-60, discussing that the system can apply a machine-learning model to an input textual dataset to determine a particular category corresponding to the textual dataset. The particular category can be selected by the machine-learning model from among a set of N predefined categories; col. 27, lines 26-57, discussing a process for generating and using a machine-learning model according to some aspects. Machine learning is a branch of artificial intelligence that relates to mathematical models that can learn from, categorize, and make predictions about data. Such mathematical models, which can be referred to as machine-learning models, can classify input data among two or more classes; cluster input data among two or more groups; predict a result based on input data; identify patterns or trends in input data; identify a distribution of input data in a space; col. 28, lines 15-29, discussing that machine-learning models can be constructed through an at least partially automated process called training. During training, input data can be iteratively supplied to a machine-learning model to enable the machine-learning model to identify patterns related to the input data or to identify relationships between the input data and output data; col. 31, lines 27-43, discussing that a processor provides a textual dataset as input to a machine-learning model. In some examples, the textual dataset can be a long document. A document can be any file that includes textual data…The machine-learning model can be of any suitable type and may have been previously trained using any suitable training technique, such as supervised or unsupervised learning. For example, the machine-learning model can be a transformer model, such as a Bidirectional Encoder Representations from Transformers (BERT) model designed for natural language processing. Other examples of the machine-learning model can include attention models, recurrent neural networks, convolutional neural networks, linear classifiers, or any combination of these; col. 36, lines 13-35, discussing that he machine-learning model can output a group of probabilities for each token in the textual dataset, where the group of probabilities indicates the likelihood of the token corresponding to each category option. For example, the output from the machine-learning model can include four probability values for each token. The four probability values can correspond to the topics “sports,” “finance,” “technology,” and “movies.” Each probability value can indicate the likelihood that the token corresponds to one of those four categories);
generate one or more decision boundaries separating the particular classes (col. 27, lines 26-57, discussing a process for generating and using a machine-learning model according to some aspects. Machine learning is a branch of artificial intelligence that relates to mathematical models that can learn from, categorize, and make predictions about data. Such mathematical models, which can be referred to as machine-learning models, can classify input data among two or more classes; cluster input data among two or more groups; predict a result based on input data; identify patterns or trends in input data; identify a distribution of input data in a space; or any combination of these. Examples of machine-learning models can include (i) neural networks; (ii) decision trees, such as classification trees and regression trees; (iii) classifiers, such as naïve bias classifiers, logistic regression classifiers, ridge regression classifiers, random forest classifiers, least absolute shrinkage and selector (LASSO) classifiers, and support vector machines; (iv) clusterers, such as k-means clusterers, mean-shift clusterers, and spectral clusterers; (v) factorizers, such as factorization machines, principal component analyzers and kernel principal component analyzers; and (vi) ensembles or other combinations of machine-learning models…; col. 31, lines 27-43, discussing that a processor provides a textual dataset as input to a machine-learning model. In some examples, the textual dataset can be a long document. A document can be any file that includes textual data...The machine-learning model can be of any suitable type and may have been previously trained using any suitable training technique, such as supervised or unsupervised learning. For example, the machine-learning model can be a transformer model, such as a Bidirectional Encoder Representations from Transformers (BERT) model designed for natural language processing. Other examples of the machine-learning model can include attention models, recurrent neural networks, convolutional neural networks, linear classifiers, or any combination of these; col. 31, lines 44-50, discussing that the processor receives an output from the machine-learning model indicating an overall classification for the textual dataset. For example, the machine-learning model can be a sentiment classifier. So, the machine-learning model can generate an output indicating that the text document has an overall positive sentiment or an overall negative sentiment; col. 33, lines 32-44, discussing that the processor selects a subset of the tokens (of the textual dataset) that have classification scores exceeding a predefined threshold. To do so, the processor can compare each token's classification score to the predefined threshold to identify which tokens have classification scores that exceed the predefined threshold. In some examples, the predefined threshold may be selected by a user. This can allow the user to customize the level of granularity and visual complexity of the resulting graphical visualization. In other examples, the predefined threshold may be selected in another way. For example, the predefined threshold may be set as the mean value of the classification scores for some or all of the tokens; col. 33, lines 45-55, discussing one example of an algorithm for performing the thresholding process...The algorithm 1700 accepts 4 inputs: P which corresponds to the hierarchical information, T which corresponds to the tokens, L which corresponds to the overall classification determined by the machine-learning model, and λ which corresponds to the predefined threshold. P is a n-by-m matrix, where n refers to number of hierarchical levels and m refers to the number of tokens. The algorithm can output the classification scores for the most important features and their corresponding tokens); and
predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries (col. 29, lines 18-26, discussing that the trained machine-learning model is used to analyze the new data and provide a result. For example, the new data can be provided as input to the trained machine-learning model. The trained machine-learning model can analyze the new data and provide a result that includes a classification of the new data into a particular class, a clustering of the new data into a particular group, a prediction based on the new data, or any combination of these; col. 31, lines 51-62, discussing that the machine-learning model can provide the output in any suitable format to convey the determined classification. For example, the output may be a numerical value corresponding to the overall classification determined for the textual document—e.g., a 0 for positive sentiment or a 1 for negative sentiment. Alternatively, the output may include one or more probabilities indicating the likelihood that the textual document falls into each of multiple candidate classifications. For instance, the output can be {0.2, 0.8} to indicate that the textual document has a 20% probability of having a negative sentiment and an 80% probability of having a positive sentiment; col. 33, lines 32-44, discussing that the processor selects a subset of the tokens (of the textual dataset) that have classification scores exceeding a predefined threshold. To do so, the processor can compare each token's classification score to the predefined threshold to identify which tokens have classification scores that exceed the predefined threshold. In some examples, the predefined threshold may be selected by a user. This can allow the user to customize the level of granularity and visual complexity of the resulting graphical visualization. In other examples, the predefined threshold may be selected in another way. For example, the predefined threshold may be set as the mean value of the classification scores for some or all of the tokens); and
generate one or more reports based on the raw data and the predicted two or more classes of data for each entry of the raw data (col. 6, lines 16-45, discussing that one example of a graphical visualization described can include a hierarchically arranged list of tokens extracted from an input textual dataset, where the tokens are color coded to indicate how much they contributed to an overall classification of the textual dataset determined by a machine-learning model. A token can include a word, a portion of a word, or a punctuation element in the input textual dataset. More specifically, a user can supply a textual dataset of to the system. The textual dataset can include a relatively large set of tokens, such as thousands of tokens. The system can apply a machine-learning model to the textual dataset to determine an overall classification for the textual dataset. The machine-learning model may have been previously trained for this classification purpose. The overall classification may be a sentiment classification (e.g., positive, negative, or neutral) or another type of classification…The system can then determine a subset of the tokens from the input textual dataset that contributed most heavily to the overall classification determined by machine-learning model. Using this information, the system can generate a graphical visualization that organizes that subset of tokens in a hierarchical format and color codes them to indicate their relative weights in the machine-learning model's decision-making process. This may allow a user or developer to understand why the machine-learning model selected that particular overall classification. This type of visualization may be particularly useful for situations involving a relatively small number of possible classifications, such as three or fewer classifications; col. 27, lines 26-57, discussing that FIG. 11 is a flow chart of an example of a process for generating and using a machine-learning model according to some aspects. Machine learning is a branch of artificial intelligence that relates to mathematical models that can learn from, categorize, and make predictions about data. Such mathematical models, which can be referred to as machine-learning models, can classify input data among two or more classes; cluster input data among two or more groups; predict a result based on input data; identify patterns or trends in input data; identify a distribution of input data in a space; or any combination of these; col. 29, lines 18-26, discussing that the trained machine-learning model is used to analyze the new data and provide a result. For example, the new data can be provided as input to the trained machine-learning model. The trained machine-learning model can analyze the new data and provide a result that includes a classification of the new data into a particular class, a clustering of the new data into a particular group, a prediction based on the new data, or any combination of these; col. 31, lines 51-62, discussing that the machine-learning model can provide the output in any suitable format to convey the determined classification. For example, the output may be a numerical value corresponding to the overall classification determined for the textual document—e.g., a 0 for positive sentiment or a 1 for negative sentiment. Alternatively, the output may include one or more probabilities indicating the likelihood that the textual document falls into each of multiple candidate classifications. For instance, the output can be {0.2, 0.8} to indicate that the textual document has a 20% probability of having a negative sentiment and an 80% probability of having a positive sentiment; col. 33, lines 32-44, discussing that the processor selects a subset of the tokens (of the textual dataset) that have classification scores exceeding a predefined threshold. To do so, the processor can compare each token's classification score to the predefined threshold to identify which tokens have classification scores that exceed the predefined threshold. In some examples, the predefined threshold may be selected by a user. This can allow the user to customize the level of granularity and visual complexity of the resulting graphical visualization. In other examples, the predefined threshold may be selected in another way. For example, the predefined threshold may be set as the mean value of the classification scores for some or all of the token).
The Hanson-Cai combination describes features related to data processing and analysis. Soleimani is directed towards data analysis systems. Therefore they are deemed to be analogous as they both are directed towards data processing systems. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Hanson-Cai combination with Hunt because the references are analogous art because they are both directed to solutions for data processing, which falls within applicant’s field of endeavor (system for providing regulatory insight analysis), and because modifying the Hanson-Cai combination to include Soleimani’s features for including predicting, based on the triggering, two or more classifications, each of the two or more classifications comprising a category associated with a raw data of each stored output, wherein the predicting is performed using a natural language processing (NLP) model, the NLP model being trained to: embed the raw data of each stored input into a vector representation; recognize patterns or features indicative of particular classes associated with the raw data; generate one or more decision boundaries separating the particular classes; and predict two or more different classes of data for each entry of the raw data based on the vector representation and the one or more decision boundaries; and generate one or more reports based on the raw data and the predicted two or more classes of data for each entry of the raw data, in the manner claimed, would serve the motivation of improving the reliability of a system that relies on the live or real-time processing of the data streams (Soleimani at col. 26, lines 52-54); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 3, the Hanson-Cai-Soleimani combination teaches the system of claim 1. Hanson further teaches wherein the at least one processor is further configured to: generate the calculation attributes based on the normalized input data from at least one of the plurality of sources (paragraph 0044, discussing that while business rules are generally conceptual in nature, in a software application they are usually implemented by some fragments or snippets of source code, which enforce the validations or execute the associated calculations. According to one practice, the data can be employed by any selected combination of business rules so as to process the data in a predefined manner and according to a predefined technique. Business rules modify the raw data to prepare the data for later applications by both updating exiting data and calculating new data based on the predefined business rules. The cleaned and enriched data can then be stored in the data storage unit; paragraph 0059, discussing that the prediction unit can further include a user feature extraction unit for processing user data, determining and identifying selected relevant or important user features or elements, and then generating a plurality of user feature scores or values that can be weighted relative to each other…the term “user feature” is intended to include specific relevant traits or attributes of a user or a set of users that can function as variables when employed in a machine learning technique...The user feature extraction unit can identify the important or primary relevant features in the user data by applying a selection reduction technique, such as for example a principal component analysis technique, to reduce the dimensionality of the user data by identifying the primary features or principal components in a dataset defined by the user data…The user feature extraction unit can then determine and identify the important or primary user features, and can then apply a weighting technique to the user features so as to weight the user features relative to each other…; paragraph 0060, discussing that prediction unit can also include a product feature extraction unit for processing the product data, determining and identifying selected important or primary product features or elements, and then generating a plurality of product feature scores or values that can be weighted relative to each other; paragraphs 0042, 0063); and
store the calculation attributes (paragraph 0042, discussing that data models can include a set of standardized, extensible data schemas that employ a defined set of data entities, data attributes, relationships, and semantic metadata; paragraph 0061, discussing that the scoring unit can employ a neural network technique for processing and managing the input score values, and preferably can employ a feed forward neural network. The final product interest score can have a value associated therewith between the range of 0 and 1. The final product interest score can be conveyed to a ranking unit for selecting and ranking the final product interest scores associated with a number of different products. The ranking unit can rank the scores in any selected manner or fashion, and preferably ranks the scores from highest to lowest scores. The ranking unit can then generate rank data 126 indicative of the product rankings. The ranking unit also allows for additional business consideration to be incorporated with rankings output from the machine learning unit to prepare a final set of recommendations. For example, the scoring unit can output the top best recommendations for a customer as a series of products. If the scoring unit is preset to provide the top five recommendations…The ranking unit, however, may incorporate a business preference to sell more of a selected product, and change the product ranking to reflect this business preference…Business considerations may also prioritize related products in a group in the ranking. For example, the scoring unit may output product 1a, product 2a, product 1b, product 3, product 2b. But the ranking unit prioritizes related products and changes the ranking to product 1a, product 1b, product 2a, product 2b, and product 3. The ranking data thus serves as predictions regarding the products and product features that the user have interest. The rank data forms part of the machine language data that can be stored in the data storage unit. This collection of units allows for a flexible approach which can begin to work when a client has a minimal amount of data, and becomes more sophisticated as more data becomes available; paragraph 0062).
As per claim 5, the Hanson-Cai-Soleimani combination teaches the system of claim 1. Hanson further teaches wherein the at least one processor is further configured to build at least one of a data mart or a data lake using the stored output or the predicted two or more classifications (paragraph 0040, discussing a data aggregation and normalization system for collecting, collating or aggregating data, such as for example financial and non-financial data, from a variety of different data sources, and then cleaning and enriching the data for subsequent use in a variety of different ways. The data aggregation and normalization system can include a plurality of data sources…The data acquired by the data sources can be conveyed through any suitable data connection to a data extraction unit. The illustrated data extraction unit can extract, transform and load (ETL) the extracted data into a data storage unit. Specifically, the data extraction unit is configured to copy the data from the data sources, transform the data by converting the file or format structure of the source data into another usable form or suitable format, and then load the data in the data storage unit. The data extraction unit thus serves as one or more extract, transform and load (ETL) data pipelines between the data sources and the data storage unit...The data storage unit can be configured to store the extracted data in any suitable form or format. The data storage unit can be in essence a data lake or a data warehouse; paragraph 0051, discussing that the data storage unit can function as a data lake with multiple different data buckets providing a place to land or store the extracted data, the cleaned data, the intermediate results, and the trusted or machine language data).
As per claim 6, the Hanson-Cai-Soleimani combination teaches the system of claim 1. Although not explicitly taught by Hanson, Cai in the analogous at of data analytics systems teaches wherein triggering the restful endpoint provides at least one additional function based on the monitored output value (paragraph 0096, discussing that the processor is configured to output the prescriptive solution to a virtual agent. In this example, the virtual agent may then provide a graphical user interface or other output to indicate to the user what a potential solution is, and the data structure of potential solutions can be traversed to provide potential alternate solutions, provided such solutions have confidence scores that are above a pre-defined threshold. For example, the virtual agent may receive an incident string, and upon traversing historical incident tickets, respond with “There have been three successful resolutions of similar problems. The most likely solution is to allocate additional memory to the page file”; paragraph 0154, discussing that the interface includes an overall security compliance indicia to trigger the display of visual elements for security compliance reporting for an application server or database. The interface includes graph data export indicia to export data for the visual elements of the network topology diagram in different formats; paragraph 0082, discussing that the platform can implement an Incident Ticket Volume Prediction process using predictive models. The Prediction Process can include instructions to implement the following operations: receive data from service management repository for all incident tickets from the past week or other time period at a folder with transfer trigger by schedule).
Hanson is directed towards systems and methods for aggregating, enriching and normalizing data. Cai is directed towards a data processing system. Therefore they are deemed to be analogous as they both are directed towards data processing systems. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Hanson with Cai because the references are analogous art because they are both directed to solutions for data processing, which falls within applicant’s field of endeavor (system for providing regulatory insight analysis), and because modifying Hanson to include Cai’s feature for including wherein triggering the restful endpoint provides at least one additional function based on the monitored output value, in the manner claimed, would serve the motivation of facilitating a culture of data driven decision making (Cai at paragraph 0075); or in the pursuit of properly and efficiently capturing, aggregating and organizing data so that the data can be later used in meaningful ways (Hanson at paragraph 0003); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 8, the Hanson-Cai-Soleimani combination teaches the system of claim 1. Hanson further teaches wherein the at least one processor is further configured to provide, via the user interface, insights based on the stored output or the predicted two or more classifications using at least one of a model or a pipeline generated via a machine learning platform (paragraph 0007, discussing a data aggregation and normalization system for aggregating data from disparate data sources, processing the data to clean the data and to normalize or standardize the data using one or more common data models, and then applying one or more discrete machine learning techniques or prediction units to the normalized data to provide data insights and predictions; paragraph 0015, discussing that the method of the present invention also includes applying one or more selected artificial intelligence and machine learning (AI/ML) techniques to selected portions of the cleaned data to form machine language data, wherein the one or more selected artificial intelligence and machine learning (AI/ML) techniques is stored in a machine language module having a plurality of predefined machine learning units, transforming the machine language data into a selected reporting format, and generating with a reporting unit one or more reports from the data in the reporting format; paragraph 0045, discussing that the data aggregation and normalization system can also employ a machine language module that employs a set of predefined machine learning units for applying one or more selected artificial intelligence and machine learning (AI/ML) models or techniques to selected portions of the cleaned data. The machine language module can also employ one or more separate prediction units for generating predictions and/or insights from the cleaned and enriched data. The machine learning techniques can be custom or commonly available artificial intelligence and machine learning methodologies that have been proven to work with large volumes of data and are able to capture and identify intricate or detailed patterns in the data…; paragraph 0061).
Claims 9 and 17 recite substantially similar limitations that stand rejected via the art citations and rationale applied to claim 1, as discussed above. Further, as per claim 9 the Hanson-Cai-Soleimani combination teaches a method for providing regulatory insight analysis (Hanson, paragraph 0002). As per claim 17, the Hanson-Cai combination teaches a non-transitory computer readable medium containing instructions that when executed by at least one processor, cause the at least one processor to perform operations for providing regulatory insight analysis (paragraph 0071).
Claim 11 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 3, as discussed above.
Claims 13 and recite substantially similar limitations that stand rejected via the art citations and rationale applied to claim 5, as discussed above.
Claims 14 and recite substantially similar limitations that stand rejected via the art citations and rationale applied to claim 6, as discussed above.
Claims 16 and 18 recite substantially similar limitations that stand rejected via the art citations and rationale applied to claim 8, as discussed above.
Claim 20 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 14, as discussed above.
26. Claims 2, 10, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Hanson in view of Cai, in view of Soleimani, in further view of Ducato, Pub. No.: US 2004/0036902 A1, [hereinafter Ducato].
As per claim 2, the Hanson-Cai-Soleimani combination teaches the system of claim 1. Hanson further teaches wherein the at least one processor is further configured to store the normalized input data in a data storage prior to analyzing the normalized input data (paragraph 0008, discussing a data storage unit for storing the extracted data, and a data preprocessing and enrichment unit for processing and enriching the extracted data to form cleaned data that is stored in the data storage unit; paragraph 0050, discussing that the data aggregation and normalization system can employ a data preprocessing and enrichment unit or preprocessing the extracted raw data to form cleaned data. The cleaned data can also be stored in the data storage unit; paragraphs 0015, 0044).
The Hanson-Cai combination does not explicitly teach store the normalized input data in a transient data storage. However, Ducato in the analogous art of data processing systems teaches this concept. Ducato teaches:
store the normalized input data in a transient data storage. (paragraph 0059, discussing that the primary data normalized are intermediately stored in a temporary data memory…The allocated information about the position of the intermediately stored data relative to other data is deposited in the temporary data memory and is additionally deposited in the sorting table as corresponding entry. The page manager thereby assumes a central control and coordination task for the allocation, intermediate storage and sorting of the primary, variable data).
The Hanson-Cai-Soleimani combination describes features related to data processing and analysis. Ducato is directed towards a method and system for data processing. Therefore they are deemed to be analogous as they both are directed towards data processing systems. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Hanson-Cai-Soleimani combination with Ducato because the references are analogous art because they are both directed to solutions for data processing, which falls within applicant’s field of endeavor (system for providing regulatory insight analysis), and because modifying the Hanson-Cai-Soleimani combination to include Ducato’s feature for including storing the normalized input data in a transient data storage, in the manner claimed, would serve the motivation of improving the data flow (Ducato at paragraph 0071); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Claims 10 and 19 recite substantially similar limitations that stand rejected via the art citations and rationale applied to claim 2, as discussed above.
27. Claims 4, 7, and 12 are rejected under 35 U.S.C. 103 as being unpatentable over Hanson in view of Cai, in view of Soleimani, in further view of Hunt et al., Pub. No.: US 2009/0018996 A1, [hereinafter Hunt].
As per claim 4, the Hanson-Cai-Soleimani combination teaches the system of claim 1. Hanson further teaches wherein the at least one processor is further configured to: receive the one or more rules as configured by the user via the user interface (paragraph 0044, discussing that a business rule is intended to mean a particular predefined manner or way in which a software application performs, processes or treats data, and which has a business connotation….According to one practice, the data can be employed by any selected combination of business rules so as to process the data in a predefined manner and according to a predefined technique. Business rules modify the raw data to prepare the data for later applications by both updating exiting data and calculating new data based on the predefined business rules. The cleaned and enriched data can then be stored in the data storage unit; paragraph 0050, discussing that the data preprocessing and enrichment unit can also employ a common data model unit for mapping or placing the data in a common data model. The common data model can have a set of defined attributes and entities for organizing the data in a standardized data format. The data in the common data model can then be processed by an assessment unit for assessing the quality of the data…The data from the assessment unit can then be processed by a data lineage unit for determining and then displaying a data lineage map or graph of selected data. The data lineage unit can also apply or overlay one or more business rules to the data; paragraph 0045).
The Hanson-Cai combination does not explicitly teach store the one or more rules. However, Hunt in the analogous art of data management systems teaches this concept. Hunt teaches:
store the one or more rules (paragraph 0439, discussing that the Master Data Management Hub (MDMH) may receive data, cleanse the data, standardize attribute values of the data, and so on. The data may comprise facts, which the MDMH may be associated with dimensional information. The MDMH may receive, generate, store, or otherwise access hierarchies of information and may process the data so as to produce an output that comprises the data in association with hierarchy. The MDMH may provide syntactic and/or semantic integration, may synchronize definitions, may store domain rules, and so on. In embodiments, the MDMH may utilize a federated data warehouse or any and all other kinds of data warehouse in which there persists a common definition of a record and, perhaps or perhaps not, the record itself; paragraph 0453, discussing that the data manipulation and structuring facility may perform operations, procedures, methods and systems including data cleansing, data standardization, keying, scrubbing data, validating data, transforming data, storing data values in a standardized format, mapping and/or keying standardized data to a canonical view, or some other data manipulation or structuring procedure, method or system).
The Hanson-Cai-Soleimani combination describes features related to data processing and analysis. Hunt is directed towards an analytic platform. Therefore they are deemed to be analogous as they both are directed towards data processing systems. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Hanson-Cai-Soleimani combination with Hunt because the references are analogous art because they are both directed to solutions for data processing, which falls within applicant’s field of endeavor (system for providing regulatory insight analysis), and because modifying the Hanson-Cai-Soleimani combination to include Hunt’s feature for including storing the one or more rules, in the manner claimed, would serve the motivation of providing better business intelligence (Hunt at paragraph 0009); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
As per claim 7, the Hanson-Cai-Soleimani combination teaches the system of claim 1. Hanson further teaches wherein the at least one processor is further configured to distribute the stored output or the predicted two or more classifications to a client-side device (paragraph 0034, discussing a data aggregation and normalization system for aggregating data from disparate data sources, processing the data to clean the data and to normalize or standardize the data using one or more data models, and then applying one or more discrete machine learning techniques to the normalized data to provide meaningful data insights and predictions. The normalized data can also be processed by one or more reporting modules to provide one or more customized reports to an end user; paragraph 0040, discussing that the data storage unit can be implemented in hardware and software on premises (i.e., within the data centers of an enterprise), distributed between multiple different locations or premises, or can be hosted in the cloud using known cloud hosting services from vendors such as Amazon, Microsoft, Amazon, Google, and the like; paragraph 0049, discussing that the data storage unit can include multiple storage units that can be located in a single location or can be dispersed throughout a network. Further, the data storage unit can also include remote storage resources that are cloud hosted by one or more cloud storage providers; paragraph 0046).
The Hanson-Cai combination does not explicitly teach to multiple client-side devices. However, Hunt in the analogous art of data management systems teaches this concept. Hunt teaches:
distribute the stored output or the predicted two or more classifications to multiple client-side devices (paragraph 0504, discussing that the sales plan performance facility may be designed to provide users with critical information and insights to facilitate efficient and effective sales execution; paragraph 0517, discussing that the analytic platform may provide for a sales performance analyzer, an on-demand software application for consumer packaged goods (CPG) manufacturing sales. The analytic platform may help maximize sales performance and improve attainment of revenue growth goals by giving sales management the ability to see the marketplace and their customers through hierarchies that represent their organization and that of their customers. It may provide sales executives within the CPG industry the ability to perform detailed analysis of revenue and sales team performance in a manner that is directly aligned with sales organization structure and user-defined territories. The sales performance analyzer may include workflows for benchmarking and trend analysis that may provide faster and more accurate response to sales activity; paragraph 1645, discussing that the analytic platform may enable multi-user collaboration, report sharing, dynamic filtering, attribute filtering, sorting, ranking, and the like; paragraphs 0510, 0513, 1513).
The Hanson-Cai-Soleimani combination describes features related to data processing and analysis. Hunt is directed towards an analytic platform. Therefore they are deemed to be analogous as they both are directed towards data processing systems. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Hanson-Cai-Soleimani combination with Hunt because the references are analogous art because they are both directed to solutions for data processing, which falls within applicant’s field of endeavor (system for providing regulatory insight analysis), and because modifying the Hanson-Cai-Soleimani combination to include Hunt’s feature for including distributing the stored output or the predicted two or more classifications to multiple client-side devices, in the manner claimed, would serve the motivation of providing better business intelligence (Hunt at paragraph 0009); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Claim 12 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 4, as discussed above.
Claim 15 recites substantially similar limitations that stand rejected via the art citations and rationale applied to claim 7, as discussed above.
28. Claim 21 is rejected under 35 U.S.C. 103 as being unpatentable over Hanson in view of Cai, in view of Soleimani, in further view of Xia et al., Pub. No.: US 2021/0034980 A1, [hereinafter Xia].
As per claim 21, the Hanson-Cai-Soleimani combination teaches the system of claim 1. Although not explicitly taught by the Hanson-Cai combination, Soleimani in the analogous art of data analytics systems teaches wherein the one or more reports include an interactive visualization generated by the trained NLP model based on the predicted two or more classes of data for each entry of the raw data (col. 5, lines 43-58, discussing providing a graphical user interface that includes graphical visualizations designed to indicate contributing factors that influenced a particular output from a machine-learning model. For example, the graphical user interface can include a graphical visualization indicating which words in a textual dataset were most heavily relied upon by a machine-learning model in producing a particular sentiment classification. By identifying which factors contributed most heavily to a particular output from a machine-learning model, it may help the users better understand why the model produced a certain result. This may help such users ensure that they are complying with applicable laws and regulations, as well as help them understand how certain inputs influence the model's outputs; col. 6, lines 16-45, discussing that one example of a graphical visualization can include a hierarchically arranged list of tokens extracted from an input textual dataset, where the tokens are color coded to indicate how much they contributed to an overall classification of the textual dataset determined by a machine-learning model. A token can include a word, a portion of a word, or a punctuation element in the input textual dataset. More specifically, a user can supply a textual dataset of to the system. The textual dataset can include a relatively large set of tokens, such as thousands of tokens. The system can apply a machine-learning model to the textual dataset to determine an overall classification for the textual dataset. The machine-learning model may have been previously trained for this classification purpose. The overall classification may be a sentiment classification (e.g., positive, negative, or neutral) or another type of classification…The system can then determine a subset of the tokens from the input textual dataset that contributed most heavily to the overall classification determined by machine-learning model. Using this information, the system can generate a graphical visualization that organizes that subset of tokens in a hierarchical format and color codes them to indicate their relative weights in the machine-learning model's decision-making process. This may allow a user or developer to understand why the machine-learning model selected that particular overall classification. This type of visualization may be particularly useful for situations involving a relatively small number of possible classifications, such as three or fewer classifications; col. 33, discussing that a predefined threshold may be selected by a user. This can allow the user to customize the level of granularity and visual complexity of the resulting graphical visualization; col. 36, lines 56-67, discussing that the processor selects a subset of the tokens (of the textual dataset) that have category scores exceeding a predefined threshold. To do so, the processor can compare each token's category score to the predefined threshold to identify which tokens have category scores that exceed the predefined threshold. In some examples, the predefined threshold may be selected by a user, which can allow the user to customize the level of granularity of the resulting graphical visualization; col. 34, lines 41-61).
The Hanson-Cai combination describes features related to data processing and analysis. Soleimani is directed towards data analysis systems. Therefore they are deemed to be analogous as they both are directed towards data processing systems. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Hanson-Cai combination with Hunt because the references are analogous art because they are both directed to solutions for data processing, which falls within applicant’s field of endeavor (system for providing regulatory insight analysis), and because modifying the Hanson-Cai combination to include Soleimani’s feature for including wherein the one or more reports include an interactive visualization generated by the trained NLP model based on the predicted two or more classes of data for each entry of the raw data, in the manner claimed, would serve the motivation of improving the reliability of a system that relies on the live or real-time processing of the data streams (Soleimani at col. 26, lines 52-54); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
The Hanson-Cai-Soleimani combination does not explicitly teach wherein the one or more reports include an interactive visualization based on the continuously monitored output. However, Xia in the analogous art of machine learning models teaches this concept. Xia teaches:
wherein the one or more reports include an interactive visualization based on the continuously monitored output (paragraph 0025, discussing an example system environment in which real time visualizations of various characteristics of complex machine learning models may be provided to clients…The machine learning service comprises a model testing and training node pool with numerous execution platforms. At a given point in time, a subset or all of the execution platforms may be allocated to train respective variants or instances of a machine learning model using a training data set stored at one or more data sources. Several different types of data sets may be available from the data sources, including continuously collected data streams and/or static data collections…In some embodiments a given data set used to train a set of model variants may be quite large…In at least one embodiment the raw data records may be stored at a storage service of a provider network, and the data sources may comprise pointers to or identifiers of the objects which contain the raw data at the storage service. For some machine learning algorithms, raw data may be pre-processed before it is used to train models; paragraph 0030, discussing that a respective local log may be maintained to track the training and/or testing operations being performed; paragraph 0031, discussing that clients may issue programmatic requests to the visualization manager 134, indicating the particular model variant or variants for which visualizations are to be provided. In some implementations clients may use interactive control elements of the interface to indicate the particular layer or feature they wish to inspect visually, to zoom in on a particular iteration's details, and so on. In at least some embodiments, the visualizations may be provided in real time or near real time—for example, within a few seconds of the completion of a particular training iteration, the value of the loss function value corresponding to that iteration may be displayed. Each model variant may have an associated identifier in the depicted embodiment, and clients may use such identifiers to indicate the particular subset of model variants for which data is to be displayed. In at least one embodiment, a client may be able to view the rate of change of a particular parameter or attribute of a model; paragraph 0040, discussing that the visualization manager may be configured to collect log entries from the different training nodes, process the metadata indicated in the entries, and provide easy-to-understand visualizations of the data in the depicted embodiment. The visualization manager may comprise a number of subcomponents, such as an iteration correlator,…, a real-time display updater and a recommendations generator; paragraph 0042, discussing that the real-time display updater may be responsible for efficiently generating the visual layout of the information to be provided to the client regarding the various model variants as new information becomes available, responding to input received from the clients via various types of controls to zoom in and out of various subsets of the data, and so on. In various embodiments the visualization manager or tool used to provide insights into the complex machine learning models being trained/tested may include subcomponents other than those shown in FIG. 3, or may not include some of the depicted subcomponents; paragraph 0056).
The Hanson-Cai-Soleimani combination describes features related to data processing and analysis. Xia is directed towards real-time visualization of machine learning models. Therefore they are deemed to be analogous as they both are directed towards data processing systems. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the Hanson-Cai-Soleimani combination with Xia because the references are analogous art because they are both directed to solutions for data processing, which falls within applicant’s field of endeavor (system for providing regulatory insight analysis), and because modifying the Hanson-Cai-Soleimani combination to include Xia’s feature for including wherein the one or more reports include an interactive visualization based on the continuously monitored output, in the manner claimed, would serve the motivation of efficiently generating the visual layout of the information to be provide as new information becomes available (Xia at paragraph 0042); and further obvious because the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Roberts et al., Pub. No.: US 2022/0357821 A1 – describes an interactive graphical user interface for monitoring computer models.
Reynolds et al., Pub. No.: US 2021/0109629 A1 – describes interactive interfaces as computerized tools to present summarization data of dataset attributes for collaborative datasets.
Akhigbe, Okhaide, et al. "GoRIM: a model-driven method for enhancing regulatory intelligence." Software and Systems Modeling 21.4 (2022): 1613-1641 – introduces the goal-oriented regulatory intelligence method, which enables effective management of regulations through modelling and data analytics.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DARLENE GARCIA-GUERRA whose telephone number is (571) 270-3339. The examiner can normally be reached M-F 7:30a.m.-5:00p.m. 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, Brian M. Epstein can be reached on (571) 270-5389. 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.
/Darlene Garcia-Guerra/
Primary Examiner, Art Unit 3625
Prosecution Insights