Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 12/29/2025 has been entered.
Status of Claims
This communication is a Non-Final Office Action in response to Applicant’s RCE for application number 18/711,386 received on 03/11/2026.
In accordance with Applicant’s amendment, claims 1/19/20 are amended. Claims 24-25 are new. Claims 1-25 are currently pending and have been examined.
Priority
Applicant’s claim for the benefit of a prior-filed application under 35 U.S.C. 119 and/or 35 U.S.C. 120 is acknowledged.
Response to Amendment
Applicant’s amendment necessitated the new ground(s) of rejection set forth in this Office Action.
Upon review of Applicant’s amendment, the previously applied §103 rejections are withdrawn.
Response to Arguments
Response to §101 arguments – Applicant’s arguments with respect to the §101 rejections previously applied to the claims have been considered and are not persuasive.
Applicant argues (Remarks at pg. 11): “The Office Action asserted that the claims recite "mental processes" because certain high-level steps (e.g., "receiving remarks," "processing the remarks") are activities a human could theoretically perform. However, as amended, the claims now expressly require a specific hybrid computational architecture comprising a rule-based dependency-matching model, a weakly-supervised text classification model (WESTCLASS), a "pseudo-document generator," and a "self-training component," all implemented by one or more processors of a computing system. These features are not mental operations. A human cannot implement a "pseudo-document generator that leverages seed information to generate pseudo-labeled documents," nor can a human perform WESTCLASS "self-training ... that bootstraps on unlabeled data," nor can a human execute the fallback control flow in which the system automatically routes inputs from the rule-based detector to a weakly-supervised classifier only upon failure of dependency-matching ("subsequently and responsive to the dependency matching of the rule-based model failing to detect the event" ... "an output of the rule-based model being an input to the machine learning model, the machine learning model comprising a weakly-supervised text classification (WESTCLASS) model" and "first using the rule-based model to predict the event and, in response to no event being detected by the rule-based model, directing the remarks to the WESTCLASS model to generate a prediction"). These operations require specialized machine learning structures and computing resources, and thus cannot reasonably be performed mentally.”. In response, Examiner respectfully disagrees and notes that the mere recitation of additional elements does not constitute an integration into a practical application, or significantly more. That is determined in Step 2A, Prong 2 and Step 2B. Applicant’s claims fall under the “Mental Processes” abstract idea grouping because the claims recite limitations that can be performed in the human mind via observation, evaluation, judgement, opinion, or with the help of pen and paper, as documented in the §101 analysis of the instant office action, as well as the office action dated 02/09/2026. Examiner notes that the abstract limitations recited in Applicant’s claims are not limitations that “merely involve an exception”. For example, the limitation for “processing the remarks to detect an event” is a critical step in Applicant’s method because it determines what happens next.
Applicant argues (Remarks at pg. 14): “The amendments therefore integrate the alleged exception into a technological process for NLP-based event detection, consistent with the reasoning in McRO and Enfish.”. In response, Examiner respectfully disagrees and notes that the present claims do not provide an analogous improvement to the computer to that of Enfish, specifically because the present claims do not improve the computer itself. The improvements of a self-referential table provide a specific benefit to the functioning of the computer, which is not the case in the claims of the instant application. The present claims are directed to processing remarks, which is not an improvement to the computer itself. Rather, this is an improvement to the abstract idea associated with “Mental Processes. In addition, Examiner notes that the present claims do not provide an analogous improvement to the computer to that of McRO. The present claims are not directed to the automation of specific animation tasks that previously could only be performed subjectively by humans. Therefore, Examiner respectfully disagrees with Applicant's assertion that the present claims are directed to statutory subject matter in view of McRO.
Applicant argues (Remarks at pgs. 15-16): “Accordingly, the amended claims are subject-matter eligible under the legal rule set forth in Finjan, Inc. v. Blue Coat Sys., Inc., 879 F.3d 1299 (Fed. Cir. 2018) and McRO, Inc. v. Bandai Namco Games America Inc., 837 F.3d 1299 (Fed. Cir. 2016) (claims that recite specifically limited steps or elements that effect a technological improvement or useful result are not abstract), under the legal rule set forth in Visual Memory LLC v. NVIDIA Corp., 867 F.3d 1253 (Fed. Cir. 2017) (claims directed towards a technological improvement are not abstract), the rule set forth in Data Engine Techs. LLC v. Google LLC, 906 F.3d 999 (Fed. Cir. 2018) and Enfish LLC v. Microsoft Corp., 822 F.3d 1327 (Fed. Cir. 2016) (claims directed towards an improvement in the functioning or operation of a computer or computer network are not abstract), and the legal rule set forth in Weisner v. Google LLC, 51 F.4th 1073 (Fed. Cir. 2022) and DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d 1245 (Fed. Cir. 2014) (claims directed towards a technical solution to a technical problem necessarily recite more than an abstract idea). As the foregoing makes clear, the claims recite patent eligible subject matter under the 2019 Guidance and comply with the two-step Alice test under controlling Federal Circuit case law. Therefore, Applicants respectfully requests withdrawal of the foregoing rejections of claims 1-20 under 35 U.S.C. § 101.”. In response, Examiner respectfully disagrees and notes that Applicant’s claims are not analogous to any of the cited cases because the additional elements recited amount to using generic computing elements or instructions (software) to perform the abstract idea, similar to adding the words “apply it” (or equivalent), which merely serves to link the use of the judicial exception to a particular technological environment and does not integrate the abstract idea into a practical application, add significantly more, or otherwise represent an improvement to technology.
Response to §103 arguments – But for the following, Applicant’s arguments with respect to the §103 rejections previously applied to the original claims are primarily raised in support of the amendments, which are believed to be fully addressed in the updated 103 rejections below.
Applicant argues (Remarks at pg. 20): “Verma’s first model is not rule-based; it is an ML model.”. In response, Examiner respectfully disagrees and notes that in at least par. [0028], Verma discloses “the model can determine that a lost circulation event is occurring after identifying indicators of a lost circulation event in a predetermined number of consecutive data rows.”. See also [0037, 0039].
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-25 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-patentable subject matter. The claims are directed to an abstract idea without significantly more. The judicial exception is not integrated into a practical application. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception as further set forth in MPEP 2106.
Step 1: The claimed invention is analyzed to determine if it falls outside one of the four statutory categories of invention. See MPEP 2106.03
Claim(s) 1-18, and 21-23 is/are directed to a method (i.e., Process), and claim(s) 19 is/are directed to a system (i.e., Machine), and claim(s) 20 is/are directed to a non-transitory computer-readable storage media (i.e., Manufacture). Therefore, claims 1-23 are directed to patent eligible categories of invention. Accordingly, the claims satisfy Step 1 of the eligibility inquiry.
Step 2A, Prong 1: In prong one of step 2A, the claim(s) is/are analyzed to evaluate whether they recite a judicial exception. See MPEP 2106.04
Independent claims 1, 19, and 20 recite a method, a system and one or more non-transitory computer-readable storage media for processing field operations data. As drafted, the limitations recited by the independent claims fall under the “Mental Processes” abstract idea group by setting forth activities that could be performed mentally by a human (including an observation, evaluation, judgment, opinion) (see MPEP § 2106.04(a)(2), subsection III).
Independent claim 1 recites a method for processing field operations data with the following abstract limitations: “receiving remarks associated with one or more field operations; processing the remarks for event detection using dependency matching to detect an event among a plurality of predefined events, the event being detected by the dependency matching using one or more patterns to match the remarks to the event; detecting that the dependency matching of the rule-based model failed to detect the event; subsequently and responsive to the dependency matching failing to detect the event, further processing the remarks for event detection to detect the event; outputting at least the detected event; generate pseudo-labeled documents for model pre-training, to predict the event and, to generate a prediction.”. But for the additional elements recited in the claim, these limitations could be accomplished mentally, such as by human observation, evaluation, judgement, opinion, or with the help of pen and paper. Independent claims 19 and 20 recite a system and one or more non-transitory computer-readable storage media with limitations that are substantially similar to the limitations of independent claim 1. Therefore, the same analysis applies.
Claims 9, 13, 14, and 25 further narrow the abstract idea and introduce further additional elements for consideration. The rest of the dependent claims further narrow the abstract idea and do not introduce further additional elements for consideration.
Step 2A, Prong 2: An evaluation is made whether a claim recites any additional element, or combination of additional elements, that integrate the judicial exception into a practical application of the exception. See MPEP 2106.04(d).
Regarding the computing additional elements, namely one or more processors of a computing system from claim 1, one or more processors and memory from claim 19, and one or more non-transitory computer-readable storage media comprising processor-executable instructions from claim 20, 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 instructions (software) to perform the abstract idea, similar to adding the words “apply it” (or equivalent), which merely serves to link the use of the judicial exception to a particular technological environment (generic computing environment). See MPEP 2106.05(f) and 2106.05(h).
With respect to the limitations for using a rule-based model, using a machine learning model, an output of the rule-based model being an input to the machine learning model, the machine learning model comprising a weakly-supervised text classification (WESTCLASS) model, the WESTCLASS model being trained on labeled training data comprising labels generated for events that had previously been detected by the rule-based model; incorporating any event detected by the rule-based model into the labeled training data for the WESTCLASS model; and incorporating the any event detected by the rule-based model into the labeled training data for the WESTCLASS model, wherein using the rule-based model for event detection comprises using the rule-based model to enlarge a labeled training data set, followed by training of the WESTCLASS model based on the enlarged training set, wherein the WESTCLASS model comprises: a pseudo-document generator that leverages seed information for model pre-training, and a self-training component that bootstraps on unlabeled data for model refinement, wherein the WESTCLASS model is trained with a set of labeled training data that is ten percent or less in size of a total amount of data used for a natural language processing technique, and wherein, during an inference phase subsequent to training the WESTCLASS model, the method further comprises first using the rule-based model, in response to no event being detected by the rule-based model, directing the remarks to the WESTCLASS model from claims 1/19/20, they provide nothing more than mere instructions to implement an abstract idea on a generic computer, which does not integrate the judicial exception into a practical application. See MPEP 2106.05(f). MPEP 2106.05(f) provides the following considerations for determining whether a claim simply recites a judicial exception with the words “apply it” (or an equivalent), such as mere instructions to implement an abstract idea on a computer: (1) whether the claim recites only the idea of a solution or outcome i.e., the claim fails to recite details of how a solution to a problem is accomplished; (2) whether the claim invokes computers or other machinery merely as a tool to perform an existing process; and (3) the particularity or generality of the application of the judicial exception.
Therefore, the additional elements of the independent claims, when considered both individually and in combination, are not sufficient to prove integration into a practical application.
With respect to the limitations for the deep neural network model comprises one or more of a convolution neural network model, or recurrent neural network model from claim 9, rendering a graphical user interface to a display from claim 13, training the machine learning model using the feedback from claim 14, and generating a graphical user interface from claim 25, they provide nothing more than mere instructions to implement an abstract idea on a generic computer, which does not integrate the judicial exception into a practical application. See MPEP 2106.05(f). MPEP 2106.05(f) provides the following considerations for determining whether a claim simply recites a judicial exception with the words “apply it” (or an equivalent), such as mere instructions to implement an abstract idea on a computer: (1) whether the claim recites only the idea of a solution or outcome i.e., the claim fails to recite details of how a solution to a problem is accomplished; (2) whether the claim invokes computers or other machinery merely as a tool to perform an existing process; and (3) the particularity or generality of the application of the judicial exception.
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.
Step 2B: The claims are analyzed to determine whether any additional element, or combination of additional elements, is/are sufficient to ensure that the claims amount to significantly more than the judicial exception. This analysis is also termed as search for "inventive concept." See MPEP 2106.05.
Regarding the computing additional elements, namely one or more processors of a computing system from claim 1, one or more processors and memory from claim 19, and one or more non-transitory computer-readable storage media comprising processor-executable instructions from claim 20, these additional element(s) has/have been evaluated but fail(s) to add significantly more because it/they amount to using generic computing elements or instructions (software) to perform the abstract idea, similar to adding the words “apply it” (or equivalent), which merely serves to link the use of the judicial exception to a particular technological environment (generic computing environment). See MPEP 2106.05(f) and 2106.05(h). Therefore, the computing additional elements merely describe generic computing elements or computer-executable instructions (software) merely serve to tie the abstract idea to a particular operating environment, which does not add significantly more to the abstract idea. See, e.g., Alice Corp., 134 S. Ct. 2347, 110 USPQ2d 1976; Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015).
With respect to the limitations for using a rule-based model, using a machine learning model, an output of the rule-based model being an input to the machine learning model, the machine learning model comprising a weakly-supervised text classification (WESTCLASS) model, the WESTCLASS model being trained on labeled training data comprising labels generated for events that had previously been detected by the rule-based model; incorporating any event detected by the rule-based model into the labeled training data for the WESTCLASS model; and incorporating the any event detected by the rule-based model into the labeled training data for the WESTCLASS model, wherein using the rule-based model for event detection comprises using the rule-based model to enlarge a labeled training data set, followed by training of the WESTCLASS model based on the enlarged training set, wherein the WESTCLASS model comprises: a pseudo-document generator that leverages seed information for model pre-training, and a self-training component that bootstraps on unlabeled data for model refinement, wherein the WESTCLASS model is trained with a set of labeled training data that is ten percent or less in size of a total amount of data used for a natural language processing technique, and wherein, during an inference phase subsequent to training the WESTCLASS model, the method further comprises first using the rule-based model, in response to no event being detected by the rule-based model, directing the remarks to the WESTCLASS model from claims 1/19/20, they provide nothing more than mere instructions to implement an abstract idea on a generic computer, which does not add significantly more to the abstract idea. See MPEP 2106.05(f). MPEP 2106.05(f) provides the following considerations for determining whether a claim simply recites a judicial exception with the words “apply it” (or an equivalent), such as mere instructions to implement an abstract idea on a computer: (1) whether the claim recites only the idea of a solution or outcome i.e., the claim fails to recite details of how a solution to a problem is accomplished; (2) whether the claim invokes computers or other machinery merely as a tool to perform an existing process; and (3) the particularity or generality of the application of the judicial exception.
Therefore, the additional elements of the independent claims, when considered both individually and in combination, are not sufficient to add significantly more to the claims.
With respect to the limitations for the deep neural network model comprises one or more of a convolution neural network model, or recurrent neural network model from claim 9, rendering a graphical user interface to a display from claim 13, training the machine learning model using the feedback from claim 14, and generating a graphical user interface from claim 25, they provide nothing more than mere instructions to implement an abstract idea on a generic computer, which does not add significantly more to the abstract idea. See MPEP 2106.05(f). MPEP 2106.05(f) provides the following considerations for determining whether a claim simply recites a judicial exception with the words “apply it” (or an equivalent), such as mere instructions to implement an abstract idea on a computer: (1) whether the claim recites only the idea of a solution or outcome i.e., the claim fails to recite details of how a solution to a problem is accomplished; (2) whether the claim invokes computers or other machinery merely as a tool to perform an existing process; and (3) the particularity or generality of the application of the judicial exception.
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. Their collective functions merely provide generic computer implementation. Therefore, when viewed as a whole, these additional claim elements do not provide meaningful limitations to amount to significantly more than the abstract idea itself. The ordered combination of elements in the 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 significantly more than the abstract idea itself.
Accordingly, claims 1-25 are rejected under 35 USC 101.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 2, 10-12, and 22-23 are rejected under 35 U.S.C. 103 as being unpatentable over Verma et al. (US 20210285321 A1, hereinafter “Verma”), in view of Kryscinski et al. (US 20210124876 A1, hereinafter “Kryscinski”), in further view of Narlikar et al. (US 20210406601 A1, hereinafter “Narlikar”), in further view of Tang et al., "Leveraging Large-Scale Weakly Labeled Data for Semi-Supervised Mass Detection in Mammograms," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 3854-3863 (hereinafter “Tang”).
Regarding claims 1/19/20: Verma teaches a hybrid computer-implemented method, a hybrid computer-implemented system comprising one or more processors, and a memory with processor-executable instructions, and one or more non-transitory computer-readable storage media ([0040] In some aspects, a system, a method, or a non-transitory computer-readable medium for detecting and warning for a lost circulation event are provided; [0020] The processing device 202 can execute instructions stored in the memory device 208 to perform the operations.) with limitations for:
receiving remarks associated with one or more field operations; ([0036] Real-time drilling data 602 can be received and undergo a data ingestion process 604. The data ingestion process 604 can pre-process the real-time drilling data 602 using operations such as temporary caching 610, resampling and cleaning 612, and feature selection and scaling 614 based on training parameters. The ingested data can also undergo a real-time outlier check process 606.; [0028] the data can be received in data rows with each row corresponding to a time value);
processing the remarks, using a rule-based model, for event detection ([0028] the data can be received in data rows with each row corresponding to a time value and the model can determine that a lost circulation event is occurring after identifying indicators of a lost circulation event in a predetermined number of consecutive data rows.; [0037] The pre-processed data can be used in an automated mud loss detection process 608. The automated mud loss detection process 608 may include applying a first model 622 to make a judgment if the data suggests that a lost circulation event has begun. If the detection is made continuously for a set amount of consecutive data rows, an alarm can be raised to alert the drilling operator 624.);
using dependency matching to detect an event among a plurality of predefined events, the event being detected by the dependency matching using one or more patterns to match the remarks to the event; ([0037] If the detection is made continuously for a set amount of consecutive data rows, an alarm can be raised to alert the drilling operator 624.);
detecting that the dependency matching of the rule-based model failed to detect the event; ([0037] If the analysis on the data suggests a normal operation, the data can be passed to a second model 626 to identify early warning precursors in the data that may indicate that a lost circulation event is about to occur.);
subsequently and responsive to the dependency matching of the rule-based model failing to detect the event, further processing the remarks for event detection using a machine learning model to detect the event, an output of the rule-based model being an input to the machine learning model, ([0037] If the analysis on the data suggests a normal operation, the data can be passed to a second model 626 to identify early warning precursors in the data that may indicate that a lost circulation event is about to occur. The second model 626 may be more probabilistic than the first model 622 as it is predicting whether a lost circulation event is about to occur.);
comprising labels generated for events that had previously been detected by the rule-based model; ([0035] The outlier removed data can be used in a model training process 508. The model training process 508 may be a semi-supervised learning process that uses unsupervised spectral clustering based automated mud loss zone warning 526 or unsupervised spectral clustering based precursor learning 528, which can then be used to train classifiers. For example, unlabeled drilling data can be segmented into different zones that can correspond to normal zones and lost circulation zones. The identification can occur in two steps: (1) segmenting between normal zone data and event data and (2) apply the clustering to the normal zone data and identify precursors across the aggregated dataset. The resulting dataset can be used to train supervised classifiers that can detect mud loss and the associated precursors.);
outputting at least the detected event; ([Abstract] An alarm can then be outputted to indicate a lost circulation event is occurring for the drilling operation);
incorporating any event detected by the rule-based model into the labeled training data for the WESTCLASS model; ([0035] The identification can occur in two steps: (1) segmenting between normal zone data and event data and (2) apply the clustering to the normal zone data and identify precursors across the aggregated dataset. The resulting dataset can be used to train supervised classifiers that can detect mud loss and the associated precursors. After a system testing process 510, the models can be deployed 512 for raising alarms and predicting lost circulation events based on incoming real-time drilling data.).
and wherein, during an inference phase subsequent to training the WESTCLASS model, the method further comprises first using the rule-based model to predict the event and, in response to no event being detected by the rule-based model, directing the remarks to the WESTCLASS model to generate a prediction. ([0037] automated mud loss detection process 608 may include applying a first model 622 to make a judgment if the data suggests that a lost circulation event has begun. If the detection is made continuously for a set amount of consecutive data rows, an alarm can be raised to alert the drilling operator 624. If the analysis on the data suggests a normal operation, the data can be passed to a second model 626 to identify early warning precursors in the data that may indicate that a lost circulation event is about to occur.).
Verma doesn’t explicitly teach:
the machine learning model comprising a weakly-supervised text classification (WESTCLASS) model, the WESTCLASS model being trained on labeled training data
and incorporating the any event detected by the rule-based model into the labeled training data for the WESTCLASS model,
wherein using the rule-based model for event detection comprises using the rule-based model to enlarge a labeled training data set, followed by training of the WESTCLASS model based on the enlarged training set,
wherein the WESTCLASS model comprises: a pseudo-document generator that leverages seed information to generate pseudo-labeled documents for model pre-training, and a self-training component that bootstraps on unlabeled data for model refinement,
wherein the WESTCLASS model is trained with a set of labeled training data that is ten percent or less in size of a total amount of data used for a natural language processing technique,
and wherein, during an inference phase subsequent to training the WESTCLASS model, the method further comprises first using the rule-based model to predict the event and, in response to no event being detected by the rule-based model, directing the remarks to the WESTCLASS model to generate a prediction.
Kryscinski teaches:
the machine learning model comprising a weakly-supervised text classification (WESTCLASS) model, the WESTCLASS model being trained on labeled training data ([Abstract] A weakly-supervised, model-based approach is provided for verifying or checking factual consistency and identifying conflicts between source documents and a generated summary. In some embodiments, an artificially generated training dataset is created by applying rule-based transformations to sentences sampled from one or more unannotated source documents of a dataset.);
and incorporating the any event detected by the rule-based model into the labeled training data for the WESTCLASS model, ([Abstract] A weakly-supervised, model-based approach is provided for verifying or checking factual consistency and identifying conflicts between source documents and a generated summary. In some embodiments, an artificially generated training dataset is created by applying rule-based transformations to sentences sampled from one or more unannotated source documents of a dataset.);
wherein using the rule-based model for event detection comprises using the rule-based model to enlarge a labeled training data set, followed by training of the WESTCLASS model based on the enlarged training set, ([0022] an artificially generated training dataset is created by applying rule-based transformations to sentences sampled from one or more unannotated source documents of a dataset.; [0024] The unannotated source documents and the labeled, transformed sentences can be provided to a neural network language model for training on checking or verifying factual consistency. It is demonstrated that training with this weak supervision substantially improves over using the strong supervision provided by previously developed datasets for NLI and fact-checking.);
wherein the WESTCLASS model comprises: a pseudo-document generator that leverages seed information to generate pseudo-labeled documents for model pre-training, ([Abstract] an artificially generated training dataset is created by applying rule-based transformations to sentences sampled from one or more unannotated source documents of a dataset. Each of the resulting transformed sentences can be either semantically variant or invariant from the respective original sampled sentence, and labeled accordingly. In some embodiments, the generated training dataset is used to train a factual consistency checking model.);
and wherein, during an inference phase subsequent to training the WESTCLASS model, the method further comprises first using the rule-based model to predict the event and, in response to no event being detected by the rule-based model, directing the remarks to the WESTCLASS model to generate a prediction. ([0022] According to some embodiments, the present disclosure provides a weakly-supervised, model-based approach for verifying or checking factual consistency and identifying conflicts between source documents and a generated summary.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine Verma with Kryscinski’s feature(s) listed above. One would’ve been motivated to do so in order to evaluate the factual consistency of abstractive text summarization (Kryscinski; [0003]). By incorporating the teachings of Kryscinski, one would’ve been able to use a westclass model for detection.
Kryscinski doesn’t explicitly teach:
and a self-training component that bootstraps on unlabeled data for model refinement,
wherein the WESTCLASS model is trained with a set of labeled training data that is ten percent or less in size of a total amount of data used for a natural language processing technique,
Narlikar teaches:
and a self-training component that bootstraps on unlabeled data for model refinement, ([0073] Model Training. An end model can be trained using both the weakly supervised data in the new modality and fully supervised data of existing modalities.; [0133] In the feature generation steps, data of both new (e.g., unlabeled) and old (e.g., labelled) modalities can be first transformed into a common feature space by the data processing system 105.; [0101] An unlabeled data point that shares edges with labeled data points can be assigned a weighted combination of its neighbor's labels. The algorithm can iteratively update this assignment until convergence; [0049] However, as described herein, it is possible to bootstrap data-limited cross-modal training pipelines by augmenting the split architecture.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Narlikar’s feature(s) listed above. One would’ve been motivated to do so, so that live traffic can be sampled to generate unlabeled data independent of previously labeled image data (Narlikar; [0147]). By incorporating the teachings of Narlikar, one would’ve been able to conduct self-training by performing bootstrapping on unlabeled data.
Narlikar doesn’t explicitly teach:
wherein the WESTCLASS model is trained with a set of labeled training data that is ten percent or less in size of a total amount of data used for a natural language processing technique,
Tang teaches:
wherein the WESTCLASS model is trained with a set of labeled training data that is ten percent or less in size of a total amount of data used for a natural language processing technique, ([Section 2. Related Work] Weakly/Semi-supervised learning for object detection or segmentation.; [Fig. 2] The framework consists of a supervised learning model (Stage 1) with fully labeled images and an alternate and iterative training procedure (Stage 2.1 and 2.2) using a mixture of fully labeled images and weakly labeled images by the NLP technique.; [4.4 Evaluation] Even though the baseline model trained with a small number (e.g., 10%) of labeled images is less accurate, adding large-scale weakly labeled data immensely increases the performance.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Tang’s feature(s) listed above. One would’ve been motivated to do so in order to effectively reduce the effort of manual annotations from medical experts (Tang; [4.4 Evaluation]). By incorporating the teachings of Tang, one would’ve been able to train the model with 10% or less of the data used for an NLP technique.
Regarding claim 2: Verma teaches:
wherein the dependency matching comprises discriminating, the discriminating comprising excluding one or more of negations or forecasts in the remarks such that, responsive to the dependency matching finding a match for the remarks that is negated or predicted as corresponding to a possible future event, the event is not detected. ([0044] resampling and cleaning the real-time data to remove data that is unrelated to detecting the lost circulation event. Examiner notes that one of ordinary skill in the art would reasonably consider excluding one or more negations or forecasts in the remarks as equivalent to ignoring data that is not indicative of an event or potential event, equivalent to removing data that is unrelated to detecting the lost circulation event, as disclosed by Verma.).
Regarding claim 10: Verma teaches
wherein the one or more field operations comprise at least one oil and gas field operation. ([0016] FIG. 1 is a schematic diagram of a drilling rig 100 for drilling a wellbore 102 into a subterranean formation 101 with fractures according to one example of the present disclosure. In this example, drilling rig 100 is depicted for a well, such as an oil or gas well, for extracting fluids from a subterranean formation 101.).
Regarding claim 11: Verma teaches
wherein the remarks comprise drilling remarks. ([0014] the trained system can be used in connection with a drilling operation by receiving real-time drilling data and other parameters about the drilling operation.).
Regarding claim 12: Verma teaches
wherein the remarks comprise drilling fluid remarks. ([0012] Certain aspects and examples of the present disclosure relate to identifying a lost circulation event occurring and identifying the likelihood of a lost circulation event occurring in connection with a drilling operation by using a machine-learning process and monitoring real-time data in connection with the drilling operation. Lost circulation includes the loss to the formation being drilled of drilling fluids that are used in connection with the drilling operation.).
Regarding claim 22: Verma teaches:
wherein the finding a match for the remarks that is negated comprises distinguishing between the event not occurring and the failing to detect the event. ([0037] The pre-processed data can be used in an automated mud loss detection process 608. The automated mud loss detection process 608 may include applying a first model 622 to make a judgment if the data suggests that a lost circulation event has begun. If the detection is made continuously for a set amount of consecutive data rows, an alarm can be raised to alert the drilling operator 624. If the analysis on the data suggests a normal operation, the data can be passed to a second model 626 to identify early warning precursors in the data that may indicate that a lost circulation event is about to occur. The second model 626 may be more probabilistic than the first model 622 as it is predicting whether a lost circulation event is about to occur. If the signature is found to be above a pre-set threshold that may be based on the system training phase, an early warning indicator can be outputted 628 so that the drilling operator can determine the problem and take preventive issues. The system may provide some indications, such as by comparing the signature to historic precursors, as to why the alarm is raised. The system may use the incoming data to update its models so that the system can continuously improve the models. If no precursors are detected, the system moves to the next row of data 630.).
Regarding claim 23: Verma teaches:
wherein the finding a match for the remarks that is negated comprises determining that a matched event did not occur. ([Fig. 6] Step 630: If no precursors detected, move to next data row.).
Claims 3, 8, 9, 13, 15, and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Verma et al. (US 20210285321 A1, hereinafter “Verma”), in view of Kryscinski et al. (US 20210124876 A1, hereinafter “Kryscinski”), in further view of Narlikar et al. (US 20210406601 A1, hereinafter “Narlikar”), in further view of Tang et al., "Leveraging Large-Scale Weakly Labeled Data for Semi-Supervised Mass Detection in Mammograms," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 3854-3863 (hereinafter “Tang”), as applied to claims 1 and 2 above, in further view of Wheatley et al. (WO 2020072720 A1, hereinafter “Wheatley”).
Regarding claim 3: Verma teaches
wherein the forecasts pertain to possible occurrence of a future event recorded in the remarks ([0037] The second model 626 may be more probabilistic than the first model 622 as it is predicting whether a lost circulation event is about to occur.).
Verma doesn’t explicitly teach:
using temporal language.
Wheatley teaches:
using temporal language. ([00204] a method can include utilizing a combination of time- series and model data with one or more of equipment, asset, and reservoir specific details. For example, consider an analysis that assesses pressures in view of what data are available about a reservoir (e.g., as to possible damage to a reservoir, as to pump performance, expected duration of artificial lift employment, etc.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Wheatley’s feature(s) listed above. One would’ve been motivated to do so in order to detect an event and/or associate an action with an event (Wheatley; [00204]). By incorporating the teachings of Wheatley, one would’ve been able to detect events by analyzing temporal remarks (e.g., time series).
Regarding claim 8: Verma doesn’t explicitly teach:
wherein the machine learning model comprises a deep neural network model.
Wheatley teaches:
wherein the machine learning model comprises a deep neural network model. ([00185] As an example, a convolution neural network system (CNNS) can be implemented using one or more platforms. Examiner notes that CNNs are known examples of deep neural networks as they include multiple hidden layers in the neural network.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Wheatley’s feature(s) listed above. One would’ve been motivated to do so in order to include an analysis engine that can include one or more features of the TENSORFLOW platform (Wheatley; [00185]). By incorporating the teachings of Wheatley, one would’ve been able to use deep neural networks to detect events.
Regarding claim 9: Verma doesn’t explicitly teach:
wherein the deep neural network model comprises one or more of a convolution neural network model or a recurrent neural network model.
Wheatley teaches:
wherein the deep neural network model comprises one or more of a convolution neural network model or a recurrent neural network model. ([00185] As an example, a convolution neural network system (CNNS) can be implemented using one or more platforms. Examiner notes that CNNs are known examples of deep neural networks as they include multiple hidden layers in the neural network.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Wheatley’s feature(s) listed above. One would’ve been motivated to do so in order to include an analysis engine that can include one or more features of the TENSORFLOW platform (Wheatley; [00185]). By incorporating the teachings of Wheatley, one would’ve been able to use deep neural networks to detect events.
Regarding claim 13: Verma doesn’t explicitly teach:
rendering a graphical user interface to a display, wherein the graphical user interface comprises a remark among the remarks and the detected event.
Wheatley teaches:
rendering a graphical user interface to a display, wherein the graphical user interface comprises a remark among the remarks and the detected event. ([Fig. 25] GUI 2500 – Comment Log, Events).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Wheatley’s feature(s) listed above. One would’ve been motivated to do so in order to commence a workflow, issue one or more control actions, etc. (Wheatley; [00197]). By incorporating the teachings of Wheatley, one would’ve been able to use a GUI to display information.
Regarding claim 15: Verma doesn’t explicitly teach:
further comprising adjusting at least one ongoing field operation based at least in part on the detected event.
Wheatley teaches:
further comprising adjusting at least one ongoing field operation based at least in part on the detected event. ([00173] one or more displays can provide content (e.g., adjust ESP speed set point or choke position, etc.) that can depend on particulars of a workflow. In such an example, the content may be associated with a selected event such as one or more actions to mitigate the event.; [00204] As an example, a method can include utilizing a combination of time- series and model data with one or more of equipment, asset, and reservoir specific details. For example, consider an analysis that assesses pressures in view of what data are available about a reservoir (e.g., as to possible damage to a reservoir, as to pump performance, expected duration of artificial lift employment, etc.); or, for example, an analysis that assesses motor temperature in view of the maximum operating temperature of the motor (e.g., where a higher operating temperature may shorten operational life, shorten time to maintenance, etc.); or, for example, an analysis that assesses pump operating point in view of a manufacturer’s operating curve (e.g., deviation from manufacturer’s operating curve, etc.). Such approaches can, for example, help to detect an event and/or associate an action with an event.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Wheatley’s feature(s) listed above. One would’ve been motivated to do so, so that a sandface pressure related event that can be mitigated (Wheatley; [00204]). By incorporating the teachings of Wheatley, one would’ve been able to adjust operations based on a detected event.
Regarding claim 16: Verma doesn’t explicitly teach:
wherein the detected event comprises a stuck pipe event.
Wheatley teaches:
wherein the detected event comprises a stuck pipe event. (([0062] As an example, one or more portions of a drillstring may become stuck. One of ordinary skill in the art would recognize a pipe as part of a drillstring.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Wheatley’s feature(s) listed above. One would’ve been motivated to do so in order to rotate pipe or lower it back into a bore (Wheatley; [0062]). By incorporating the teachings of Wheatley, one would’ve been able to detect a stuck pipe event.
Regarding claim 18: Verma doesn’t explicitly teach:
rendering a graphical user interface to a display, wherein the graphical user interface comprises at least one graphic that associates the detected event with one or more of a location or a rig.
Wheatley teaches:
rendering a graphical user interface to a display, wherein the graphical user interface comprises at least one graphic that associates the detected event with one or more of a location or a rig. ([Fig. 17] AED Type (e.g., pump off, insufficient lift), Client, Field, Well).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Wheatley’s feature(s) listed above. One would’ve been motivated to do so in order to allow a user to discern differences in wells given triage scores where a user may understand components that make up a triage score (Wheatley; [00201]). By incorporating the teachings of Wheatley, one would’ve been able to detect a stuck pipe event.
Claims 4 and 5 are rejected under 35 U.S.C. 103 as being unpatentable over Verma et al. (US 20210285321 A1, hereinafter “Verma”), in view of Kryscinski et al. (US 20210124876 A1, hereinafter “Kryscinski”), in further view of Narlikar et al. (US 20210406601 A1, hereinafter “Narlikar”), in further view of Tang et al., "Leveraging Large-Scale Weakly Labeled Data for Semi-Supervised Mass Detection in Mammograms," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 3854-3863 (hereinafter “Tang”), as applied to claim 1 above, in further view of New (US 20180322111 A1, hereinafter “New”).
Regarding claim 4: Verma doesn’t teach:
wherein the dependency matching comprises parsing phrases in the remarks to generate a pattern,
the parsing comprising examining dependency structures by identifying a head word, a dependent word, and a dependency link in the remarks.
Narlikar teaches:
wherein the dependency matching comprises parsing phrases in the remarks to generate a pattern, ([0095] text (e.g., string or pattern matchers, etc.)).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Narlikar’s additional feature(s) listed above. One would’ve been motivated to do so in order to define predicates that operate over certain data points (Narlikar; [0095]). By incorporating the teachings of Narlikar, one would’ve been able to use a parser.
Narlikar doesn’t explicitly teach:
the parsing comprising examining dependency structures by identifying a head word, a dependent word, and a dependency link in the remarks.
New teaches:
the parsing comprising examining dependency structures by identifying a head word, a dependent word, and a dependency link in the remarks. ([0009] All words of an utterance are contained in one or more of these dependent relationships. In each of these dependency links, one word is dependent on the other. The dependent word can come before or after the word it depends on. When an utterance is parsed with this type of parse, a single word is said to be the head of the utterance. All other words in the utterance directly depend on the head word or indirectly depend on the head word by depending on one or more intermediary words that depend on the head word. The term Dependency Grammar has been used to describe such a system for parsing utterances.; [0006] In one type of parser, hand written rules are created that describe how words can be joined together to form phrases and sub-phrases. Additional rules describe how phrases can be joined into larger phrases and sentences. An utterance is parsed by finding one or more words that match one of the rules and linking them together. This process is repeated until all words have been matched to rules. The words are thus grouped together into sub-phrases. The sub-phrases are then grouped together into phrases. Each time a rule is applied to group words and/or phrases, a new node is created that contains the elements that are grouped. Ultimately all words and phrases are grouped into a single node that represents the entire utterance. The term Context Free Grammar (CFG) has been used to describe such a system of rules for parsing.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with New’s feature(s) listed above. One would’ve been motivated to do so, so that dependent links can be found between all words of an utterance (New; [0009]). By incorporating the teachings of New, one would’ve been able to use patterns with a head word, a dependent word, and a dependency link.
Regarding claim 5: Verma doesn’t explicitly teach:
wherein the pattern comprises a head word, a dependent wordy and a dependency link.
New teaches:
wherein the pattern comprises a head word, a dependent wordy and a dependency link. ([0009] All words of an utterance are contained in one or more of these dependent relationships. In each of these dependency links, one word is dependent on the other. The dependent word can come before or after the word it depends on. When an utterance is parsed with this type of parse, a single word is said to be the head of the utterance. All other words in the utterance directly depend on the head word or indirectly depend on the head word by depending on one or more intermediary words that depend on the head word. The term Dependency Grammar has been used to describe such a system for parsing utterances.
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with New’s feature(s) listed above. One would’ve been motivated to do so, so that dependent links can be found between all words of an utterance (New; [0009]). By incorporating the teachings of New, one would’ve been able to use patterns with a head word, a dependent word, and a dependency link.
Claims 6 and 7 are rejected under 35 U.S.C. 103 as being unpatentable over Verma et al. (US 20210285321 A1, hereinafter “Verma”), in view of Kryscinski et al. (US 20210124876 A1, hereinafter “Kryscinski”), in further view of Narlikar et al. (US 20210406601 A1, hereinafter “Narlikar”), in further view of Tang et al., "Leveraging Large-Scale Weakly Labeled Data for Semi-Supervised Mass Detection in Mammograms," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 3854-3863 (hereinafter “Tang”), as applied to claim 1 above, in further view of Katz et al. (US 20210117456 A1, hereinafter “Katz”).
Regarding claim 6: Verma doesn’t explicitly teach:
wherein the dependency matching comprises stemming and lemmatization.
Katz teaches:
wherein the dependency matching comprises stemming and lemmatization. (([0035] Lemmatization may also incorporate stemming or other types of analysis that reduce inflectional forms and derivatively related forms to a common base form, which may vary based on the language to which the analysis applies. For example, lemma replacement controller 250 may apply a Porter Stemmer lemmatization analysis for identifying a lemma of a head word in an answer in English.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Katz’s feature(s) listed above. One would’ve been motivated to do so in order to identify the lemma of the head word in the answer (Katz; [0035]). By incorporating the teachings of Katz, one would’ve been able to analyze text data using stemming and lemmatization.
Regarding claim 7: Verma doesn’t explicitly teach:
wherein the dependency matching uses the stemming and lemmatization to replace a word with its root form.
Katz teaches:
wherein the dependency matching uses the stemming and lemmatization to replace a word with its root form. (([0035] lemma replacement controller 250 may apply a more complex lemmatization analysis for more morphologically complex languages, such as a lookup table from surface forms to root forms. [0037] lemma replacement controller 250 replaces the non-matching answer with the identified surface form in replaced answers 254).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Katz’s additional feature(s) listed above. One would’ve been motivated to do so in order to modify answers to match the grammatical properties of the question (Katz; [0037]). By incorporating the teachings of Katz, one would’ve been able to replace words with its root form.
Claim 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over Verma et al. (US 20210285321 A1, hereinafter “Verma”), in view of Kryscinski et al. (US 20210124876 A1, hereinafter “Kryscinski”), in further view of Narlikar et al. (US 20210406601 A1, hereinafter “Narlikar”), in further view of Tang et al., "Leveraging Large-Scale Weakly Labeled Data for Semi-Supervised Mass Detection in Mammograms," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 3854-3863 (hereinafter “Tang”), in further view of Wheatley et al. (WO 2020072720 A1, hereinafter “Wheatley”), as applied to claim 13 above, in further view of Basu et al. (US 10364662 B1, hereinafter “Basu”),
Regarding claim 14: Verma doesn’t explicitly teach:
comprising receiving input via the graphical user interface as feedback
and comprising training the machine learning model using the feedback.
Wheatley teaches:
comprising receiving input via the graphical user interface as feedback ([Fig. 25] GUI 2500 – Comment Log; [00260] The GUI 2500 also shows various comments in a comment log, which can include various concerns, learnings, etc.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Wheatley’s additional feature(s) listed above. One would’ve been motivated to do so in order to guide a user in deciding whether or not to cause the system to issue the control instruction(s) (Wheatley; [00260]). By incorporating the teachings of Wheatley, one would’ve been able to receiving input via the GUI as feedback.
Wheatley doesn’t explicitly teach:
and comprising training the machine learning model using the feedback.
Basu teaches:
and comprising training the machine learning model using the feedback. ([Column 28, Lines 54-62] the system can include a report analysis engine 5710 trained to analyze drill reports, such as historical drill reports or current drill reports. Such a report analysis engine 5710 can provide feedback to further enhance the field specific model 5708 or to provide information useful by other engines in determining parameters and recipes to be used on a given well or throughout a given field. The report analysis engine 5710 can be trained as described above and can incorporate structures, such as a neural network or classification.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Basu’s additional feature(s) listed above. One would’ve been motivated to do so in order to include a field specific model and engines derived from such a field specific model (Basu; [Column 28, Lines 32-33]). By incorporating the teachings of Basu, one would’ve been able to fine tune the model with feedback.
Claim 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Verma et al. (US 20210285321 A1, hereinafter “Verma”), in view of Kryscinski et al. (US 20210124876 A1, hereinafter “Kryscinski”), in further view of Narlikar et al. (US 20210406601 A1, hereinafter “Narlikar”), in further view of Tang et al., "Leveraging Large-Scale Weakly Labeled Data for Semi-Supervised Mass Detection in Mammograms," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 3854-3863 (hereinafter “Tang”), as applied to claim 1 above, in further view of Basu et al. (US 10364662 B1, hereinafter “Basu”),
Regarding claim 17: Verma doesn’t explicitly teach:
further comprising planning at least one field operation based at least in part on the detected event.
Basu teaches:
further comprising planning at least one field operation based at least in part on the detected event. ([Column 20, Lines 29-33] The proposed system incorporates data, regardless of source, structure, size, or format, to prescribe actionable recipes for drilling, completing, and producing wells that maximize their economic value at every point over the course of their serviceable lifetimes.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Basu’s additional feature(s) listed above. One would’ve been motivated to do so in order to help operators anticipate and improve wells (Basu; [Column 20, Lines 28-29]). By incorporating the teachings of Basu, one would’ve been able to plan operations based on detected events.
Claim 21 is/are rejected under 35 U.S.C. 103 as being unpatentable over Verma et al. (US 20210285321 A1, hereinafter “Verma”), in view of Kryscinski et al. (US 20210124876 A1, hereinafter “Kryscinski”), in further view of Narlikar et al. (US 20210406601 A1, hereinafter “Narlikar”), in further view of Tang et al., "Leveraging Large-Scale Weakly Labeled Data for Semi-Supervised Mass Detection in Mammograms," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 3854-3863 (hereinafter “Tang”), as applied to claim 1 above, in further view of Wolverton et al. (US 20140136187 A1, hereinafter “Wolverton”).
Regarding claim 21: Verma doesn’t explicitly teach:
wherein the rule-based model and the machine learning model each operate on a phrase-by-phrase basis or a sentence-by- sentence basis to process the remarks.
Wolverton teaches:
wherein the rule-based model and the machine learning model each operate on a phrase-by-phrase basis or a sentence-by- sentence basis to process the remarks. (Par. [0046] teaches an input recognizer/interpreter 130 may include a standard (now existing or later-developed) natural language processor. The natural language processor may apply syntactic, grammar, and/or semantic rules; [0052] The input classifier 134 analyzes the computer-readable representations of the inputs 102 as prepared by the input recognizer/interpreter 130, and classifies the inputs 102 according to rules or templates that may be stored in the vehicle-specific conversation model 132 or the vehicle context model 116; [0053] Such classifications are, in general, based on the degree of specificity of the inputs 102 as well as the type of inquiry associated with the inputs 102. For example, direct, fact-based questions for which an answer can be provided fairly quickly (such as questions that have a yes or no answer, or "what is?" questions) may be classified differently than questions that require a more complicated explanation (such as "how-to" questions) or questions that are possibly ambiguous unless further information is obtained. The manner in which inputs 102 are classified may be used to determine whether a suitable response can be quickly found in the vehicle-specific user's guide knowledge base 140, or whether sources in the vehicle-related search realm 142 may need to be searched, alternatively or in addition to the knowledge base 140.; [0054] If the classifier 134 determines that an input 102 is "situation-aware;" that is, the input 102 may have a different meaning depending on the current context, the reasoner 136 considers the current context as determined by the context analyzer 126, described below, and incorporates the most likely relevant aspects of the current context into the interpretation of the input 102 to infer the user's most probable intended meaning of the input 102. To do this, the reasoner 136 applies a probabilistic or statistical model (using, e.g., Bayesian modeling). The probabilistic or statistical model includes data relating to the likelihood, probability, or degree of confidence or certainty with which particular inputs 102 are associated with particular meanings based on the context.).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Wolverton’s feature(s) listed above. One would’ve been motivated to do so in order to process conversational and other inputs generated by a person (Wolverton; [0031]). By incorporating the teachings of Wolverton, one would’ve been able to process the natural language remarks by 2 separate models.
Claim 24 is/are rejected under 35 U.S.C. 103 as being unpatentable over Verma et al. (US 20210285321 A1, hereinafter “Verma”), in view of Kryscinski et al. (US 20210124876 A1, hereinafter “Kryscinski”), in further view of Narlikar et al. (US 20210406601 A1, hereinafter “Narlikar”), in further view of Tang et al., "Leveraging Large-Scale Weakly Labeled Data for Semi-Supervised Mass Detection in Mammograms," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 3854-3863 (hereinafter “Tang”), as applied to claim 1 above, in further view of Inagaki et al. (US 20200342354 A1, hereinafter “Inagaki”).
Regarding claim 24: Verma doesn’t teach
wherein the processing the remarks using the rule-based model further comprises, before event detection using dependency matching: defining a dictionary comprising one or more of: terminologies, acronyms, or common misspellings;
and replacing matching words in the remarks with predefined standard spellings using the dictionary.
Inagaki teaches:
wherein the processing the remarks using the rule-based model further comprises, before event detection using dependency matching: defining a dictionary comprising one or more of: terminologies, acronyms, or common misspellings; ([0078] This method is also called natural language processing by rules.; [0040] the dictionary may include a plurality of generic terms, where each generic term is correlated to a plurality of specific terms (e.g., product names, specific examples, etc.).);
and replacing matching words in the remarks with predefined standard spellings using the dictionary. ([0040] term replacement may be performed utilizing a dictionary. For example, the dictionary may include a plurality of generic terms, where each generic term is correlated to a plurality of specific terms (e.g., product names, specific examples, etc.).).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Inagaki’s feature(s) listed above. One would’ve been motivated to do so in order to replace one or more terms within the single instance of training data with a genericized term (Inagaki; [0040]). By incorporating the teachings of Inagaki, one would’ve been able to use a dictionary to process remarks and replace words in the remarks using the dictionary.
Claim 25 is/are rejected under 35 U.S.C. 103 as being unpatentable over Verma et al. (US 20210285321 A1, hereinafter “Verma”), in view of Kryscinski et al. (US 20210124876 A1, hereinafter “Kryscinski”), in further view of Narlikar et al. (US 20210406601 A1, hereinafter “Narlikar”), in further view of Tang et al., "Leveraging Large-Scale Weakly Labeled Data for Semi-Supervised Mass Detection in Mammograms," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA, 2021, pp. 3854-3863 (hereinafter “Tang”), as applied to claim 1 above, in further view of Saha et al. Integration of 1D Geomechanics Modeling and High-Performance Water-Based Mud HPWBM System Design, Improving Cost-Effective Drilling of High-Angle Wells Through Cauvery Shale Sequence: A Case Study from Cauvery Basin, Offshore India. (2019) (hereinafter “Saha”), in further view of Kashi “Maps And Graphs Explain Why The US Is World’s Largest Producer Of Hydrocarbons”, International Business Times (2013) (hereinafter “Kashi”).
Regarding claim 25: Verma doesn’t teach:
computing, from the detected events, an event count distribution by rig;
and generating a graphical user interface in which, for each rig: a stacked bar represents counts of different types of events; and a line indicates a number of wells drilled by the rig.
Saha teaches:
computing, from the detected events, an event count distribution by rig; ([Fig. 3] Drilling Days);
and generating a graphical user interface in which, for each rig: a stacked bar represents counts of different types of events; ([Fig. 3] Drilling Days, Non-Productive Time (NPT));
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Saha’s feature(s) listed above. One would’ve been motivated to do so in order to display total drilling days required in shale section for each offset well (Saha; [Page 5]). By incorporating the teachings of Saha, one would’ve been able to count events and display event data in a graph.
Saha doesn’t teach:
and a line indicates a number of wells drilled by the rig.
Kashi teaches:
and a line indicates a number of wells drilled by the rig. ([Page 3] New-well oil production per rig, New-well gas production per rig).
It would have been obvious to one of ordinary skill in the art, at the time of applicant’s invention, to combine modified Verma with Kashi’s feature(s) listed above. One would’ve been motivated to do so in order to take into account both the time it takes to drill a new well and the productivity of that well (Kashi; [Page 2]). By incorporating the teachings of Kashi, one would’ve been able to display the wells per rig in a line graph.
Accordingly, claims 1-25 are rejected under 35 USC 103.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Selman et al. (US 8838390 B1), which discloses a system for creating a log during gas detection and monitoring and is relevant because it discloses a series of dashboards used by operators and executives to display drilling operation information relevant to their respective roles, including events counts, alarms, and progress.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GABRIEL J TORRES CHANZA whose telephone number is (571)272-3701. The examiner can normally be reached Monday thru Friday 8am - 5pm ET.
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 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.
/G.J.T./Examiner, Art Unit 3625
/BRIAN M EPSTEIN/Supervisory Patent Examiner, Art Unit 3625
Prosecution Insights