DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Amendment
Claims 1, 9-10, 16, 22-23, and 29 are amended. Claims 1-29 are pending.
Response to Arguments
Applicant's arguments filed 12/15/2025 have been fully considered and are partially persuasive.
Regarding the objection to claim 16, and as noted by Applicant on page 9 of the response, the amendment to claim 16 overcomes the objection, which is withdrawn.
Regarding the rejections of claims 9-17 and 22 under 112(b), and as persuasively explained by Applicant on page 10 of the response, the amendments to claims 9, 10 and 22 overcome the rejections, which are withdrawn.
Regarding the rejections of claims 1-29 under 101, Applicants arguments are persuasive.
Regarding the rejections of independent claims 23 and 29 under 102, the Examiner agrees with Applicant’s contention on page 12 of the response, that the amendments overcome the rejections of claims 23 and 29 under 102, which are withdrawn. However, in view of further search and consideration, new grounds for rejecting claims 23 and 29 under 103 are set forth herein.
Regarding the rejection of claim 1 under 102, the Examiner respectfully disagrees with Applicant contention on pages 11-12 of the response, that Denli does not disclose “the economic parameter specifies a cost for drilling a borehole proximate the location.” The Examiner had previously indicated in the Examiner Interview conducted with Applicant’s representative that, based on a relatively brief review of Denli’s disclosure that the model parameters disclosed by Denli do not appear to expressly convey a cost. However, upon further analysis of the manner in which “cost” may be interpreted in view of Applicant’s specification and in view of a closer analysis of what aspects of Denli’s disclosure may fall within such interpretation, the Examiner has determined that Denli teaches “wherein the economic model parameter specifies a cost for drilling a borehole proximate the location ([0036] implementation of the embodiment (generating graph network data) entails identifying geologic information including relationships such as stratigraphic relationships; [0045]-[0046] graph information includes connectivity between geologic and fluid objects that may be characterized in part by distance; [0079] graph network conveys lithology. Examiner notes that stratigraphic and lithographic information as well as underground distance information effectively indicates cost of drilling in terms of distances to drill and type of material to drill through).”
Therefore, the rejection of claim 1 under 102 as anticipated by Denli is maintained.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-29 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Although the specification recites determining an economic model parameter that relates to a variety of tasks, the specification as filed does not explicitly support in claim 1 “placing a drilling rig of the well system at a location”, claim 23 “a geo-steering system located downhole of the well system utilizes the economic model parameter to alter a borehole path”, or claim 29 “directing a drilling controller to alter a hydraulic fracturing location and an intensity”. Therefore claims 1-29 contain new matter.
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
Claims 23-28 are rejected under 35 U.S.C. 112(b) as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor regards as the invention.
Claim 23 is rendered indefinite due to the unclear characterization of “a geo-steering system located downhole of the well system” within the claim. In particular, it is unclear from the language and structure of claim 23, whether the “geo-steering” system forms part of the overall recited “computing system” of claim 23, or whether this component along with its described function “utilizes the economic model parameter to alter a borehole path” convey an intended use rather than being a positively recited limitation to the structure function of the recited “computing system.” In one aspect, the structure of the claim seems to indicate that the “computing system” includes two main components – “a data receiver” and “one or more processors.” The description of the “geo-steering system” is appended to the description of the “one or more processors” in a manner that renders its relation to the overall structure of the “computing system” unclear. Based on apparent intent of Applicant to recite the “geo-steering system” and its associated functionality as a positive limitation for purposes of overcoming the 101 and prior art rejections, claim 23 is interpreted to require “a geo-steering system located downhole of the well system” that “utilizes the economic model parameter to alter a borehole path” as a component of the overall recited system.
Claims 24-28 depend from claim 23 and are likewise rejected for the same reasons.
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 1-22 and 29 are rejected under 35 U.S.C. 101 because the claimed invention in each of these claims is directed to the abstract idea judicial exception without significantly more.
Claim 1, substantially representative also of independent claim 29, recites:
“[a] method of inferring subsurface knowledge, comprising:
obtaining a geoscience knowledge system;
obtaining subsurface information at a subterranean location; and
inferring subsurface knowledge of the subterranean location from the subsurface information using the geoscience knowledge system, wherein the subsurface knowledge is utilized to calculate an economic model parameter for a well system proximate the subterranean location; and
placing a drilling rig of the well system at a location determined by the economic model parameter, wherein the economic model parameter specifies a cost for drilling a borehole proximate the location.”
The claim limitations considered to fall within in the abstract idea are highlighted in bold font above and the remaining features are “additional elements.”
Step 1 of the subject matter eligibility analysis entails determining whether the claimed subject matter falls within one of the four statutory categories of patentable subject matter identified by 35 U.S.C. 101: process, machine, manufacture, or composition of matter. Claim 1 recites a method, and each of claims 23 and 29 recites an apparatus and therefore each falls within a statutory category.
Step 2A, Prong One of the analysis entails determining whether the claim recites a judicial exception such as an abstract idea. Under a broadest reasonable interpretation, the highlighted portions of claim 1 fall within the abstract idea judicial exception. Specifically, under the 2019 Revised Patent Subject matter Eligibility Guidance, the highlighted subject matter falls within the mental processes category (including an observation, evaluation, judgment, opinion) and the mathematical concepts category (mathematical relationships, mathematical formulas or equations, mathematical calculations). MPEP § 2106.04(a)(2).
The recited function “inferring subsurface knowledge of the subterranean location from the subsurface information” may be performed via mental processes (e.g., evaluation of subsurface information by a field expert to infer via judgement subsurface knowledge), as may “wherein the subsurface knowledge is utilized to calculate an economic model parameter for a well system proximate the subterranean location” “wherein the economic model parameter specifies a cost for drilling a borehole proximate the location” (e.g., evaluation of subsurface knowledge entirely mentally and/or with aid of pen-and-paper to determine/calculate an economic parameter for a well system). Furthermore, “a location determined by the economic model parameter” may also be performed via mental processes (e.g., evaluation and judgement).
The type of high-level information analysis and deduction recited in these elements has been found by the Federal Circuit to constitute patent ineligible matter (see Electric Power Group v. Alstom, S.A., 830 F.3d 1350, 1353-54, 119 USPQ2d 1739, 1741-42 (Fed. Cir. 2016), a claim to "collecting information, analyzing it, and displaying certain results of the collection and analysis," where the data analysis steps are recited at a high level of generality such that they could practically be performed in the human mind).
The recited function “inferring subsurface knowledge of the subterranean location from the subsurface information” is further determined by the Examiner as falling within the mathematical relationships sub-category of mathematical concepts (MPEP 2106.04(a)(2)) because as disclosed in Applicant’s specification (e.g., [0076]) inferring subsurface knowledge from the subsurface information may be implemented using Bayes rule, decision trees, and/or Markov's reinforcement learning each of which is fundamentally characterized by mathematical calculations/relations and therefore constitutes mathematical relationships.
Step 2A, Prong Two of the analysis entails determining whether the claim includes additional elements that integrate the recited judicial exception into a practical application. “A claim that integrates a judicial exception into a practical application will apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception, such that the claim is more than a drafting effort designed to monopolize the judicial exception” (MPEP § 2106.04(d)).
MPEP § 2106.04(d) sets forth considerations to be applied in Step 2A, Prong Two for determining whether or not a claim integrates a judicial exception into a practical application. Based on the individual and collective limitations of claim 1 and applying a broadest reasonable interpretation, the most applicable of such considerations appear to include: improvements to the functioning of a computer, or to any other technology or technical field (MPEP 2106.05(a)); applying the judicial exception with, or by use of, a particular machine (MPEP 2106.05(b)); and effecting a transformation or reduction of a particular article to a different state or thing (MPEP 2106.05(c)).
Regarding improvements to the functioning of a computer or other technology, none of the “additional elements,” individually or in combination, including “obtaining a geoscience knowledge system,” “obtaining subsurface information at a subterranean location,” and “using the geoscience knowledge system” to infer subsurface knowledge of the subterranean location from the subsurface information,” in claims 1 and 29, and “placing a drilling rig of the well system at a location” in claim 1, and “directing a drilling controller to alter a hydraulic fracturing location and an intensity” in claim 29 in any combination appear to integrate the abstract idea in a manner that technologically improves any aspect of a device or system that may be used to implement the highlighted step or a device for implementing the highlighted step such as a signal processing device or a generic computer. Obtaining a geoscience knowledge system entails routine, conventional computer system configuration (e.g., obtaining program constructs such as data modeling constructs) that constitutes insignificant extra solution activity having no particularized relation to the underlying processing functions and therefore fails to integrate the judicial exception into a practical application. Obtaining subsurface information at a subterranean location entails high-level data collection that also constitutes extra solution activity. Using the geoscience knowledge system to infer subsurface knowledge entails application of program instructions for implementing the underlying function that falls within the judicial exception and therefore also constitutes insignificant extra solution activity that fails to integrate the judicial exception into a practical application. Placing a drilling rig of the well system at a location (i.e., the placement per se) and directing a drilling controller to alter hydraulic fracturing location and intensity represents standard well operations activity having no particularized relation to the steps that fall within the judicial exception except for the determination of the location, which as explained above itself falls within the judicial exception. Directing a drilling controller to alter a hydraulic fracturing location and an intensity as set forth in claim 29 has no apparent functional relation to the steps falling within the judicial exception. Therefore, “placing a drilling rig of the well system at a location” in claim 1 and “directing a drilling controller to alter a hydraulic fracturing location and an intensity” in claim 29 constitutes insignificant extra solution activity that fails to integrate the judicial exception into a practical application.
Regarding application of the judicial exception with, or by use of, a particular machine, the additional elements are configured and implemented in a conventional rather than a particularized manner of implementing exploration and analysis of subsurface characteristics.
Regarding a transformation or reduction of a particular article to a different state or thing, claim 1 does not include any such transformation or reduction. Instead, claim 1 as a whole entails receiving input information (subsurface information obtained at a subterranean location), applying standard processing techniques (computer-implemented processing) to the information to obtain and deduce subsurface information and an economic model parameter (e.g., data such as underground structures indicating hydrocarbon distribution) with the additional elements failing to provide a meaningful integration of the abstract idea (inferring subsurface knowledge and calculating an economic model parameter) in an application that transforms an article to a different state. Instead, the additional elements represent extra-solution activity that does not integrate the judicial exception into a practical application. In view of the various considerations encompassed by the Step 2A, Prong Two analysis, claim 1 does not include additional elements that integrate the recited abstract idea into a practical application.
Therefore, claim 1 is directed to a judicial exception and requires further analysis under Step 2B.
Regarding Step 2B, and as explained in the Step 2A Prong Two analysis, the additional elements in claim 1 constitute insignificant extra solution activity and therefore do not result in the claim as a whole amounting to significantly more than the judicial exception. Furthermore, the additional elements appear to be generic and well understood as evidenced by the disclosures of Denli (US 2023/0161061 A1) and Liu (US 2020/0184374 A1), each of which teach a substantially similar method/system for collecting and interpreting subsurface information.
As explained in the grounds for rejecting claim 1 under 102, Denli teaches “obtaining a geoscience knowledge system,” “obtaining subsurface information at a subterranean location,” “using the geoscience knowledge system” for inferring subsurface knowledge, and “placing a drilling rig at a location determined by the economic model parameter” as does Liu (Abstract disclosing a method/apparatus for automated seismic interpretation including obtaining training model; FIGS. 1A and 1B; [0023] management includes identifying well locations).
Therefore, the additional elements are insufficient to amount to significantly more than the judicial exception.
Claim 1 is therefore not patent eligible.
Independent claim 29 includes substantially similar elements falling within the mental processes judicial exception as claim 1. Claim 29 further recites the additional element “computer program product having a series of operating instructions stored on a non- transitory computer-readable medium that directs a data processing apparatus when executed thereby to perform operations to infer subsurface knowledge,” which entails routine, conventional data processing means for implementing the functions falling within the judicial exception and therefore constitute insignificant extra solution activity that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception. Claim 29 further recites “directing a drilling controller to alter a hydraulic fracturing location and an intensity,” which as explained above in the Step 2A Prong 2 analysis, represents standard hydraulic fracturing activity unrelated to the steps falling within the judicial exception and therefore also constitutes extra solution activity that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception.
Independent claim 29 therefore also constitute ineligible subject matter under 101.
Claims 2-22 depending from claim 1, provide additional features/steps that are part of an expanded algorithm that includes the abstract idea of claim 1 (Step 2A, Prong One). None of dependent claims 2-22 recite additional elements that integrate the abstract idea into practical application (Step 2A, Prong Two), and all fail the “significantly more” test under the step 2B for substantially similar reasons as discussed with regards to the independent claims.
For example, claims 2-4, 6-7, and 13-14, 19, and 21 recite characterizations of the types of data used in the processing steps falling within the judicial exception and therefore fall within the same judicial exception.
Claims 17-18 recite characterization of the sources of the input data used in the processing steps falling within the judicial exception and therefore fall within the same judicial exception.
Claim 5 recites “wherein the subsurface knowledge is utilized to direct a well operation at a well site,” which constitutes post-solution type extra solution activity that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception.
Claim 8 further recites that a “machine learning” algorithm/system is used to represent subsurface knowledge, which represents routine, conventional data processing functions (machine learning) for implementing the function that falls within the judicial exception and therefore constitutes extra solution activity that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception.
Claim 9 recites “acquiring subsurface knowledge text; acquiring subsurface information, correlating the subsurface knowledge text and the subsurface information,” as part of “training the geoscience knowledge system,” which represents high-level data gathering that and conventional data processing that constitutes insignificant extra solution activity. Claim 9 further recites “training the geoscience knowledge system to identify hydrocarbon distributions utilizing the correlating,” which represents high-level program instruction preparation/configuration using the high-level data gathering and therefore itself constitutes insignificant extra solution activity that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception.
Claim 10 recites “wherein geoscience text is integrated with subsurface images in geoscience data,” which characterizes the nature of the data as including text and images and therefore does not constitute an “additional element” (i.e. the processing of the data via a judicial exception is not changed). Claim 10 further recites “the acquiring subsurface knowledge text, the acquiring subsurface information, and the correlating includes” “extract the subsurface knowledge text and the subsurface images from the geoscience data and relate respective of the extracted subsurface knowledge text and the subsurface images,” which falls within the mental processes type judicial exception because it can be performed via mental processes (e.g., evaluation of information including text and images and judgement in relating respective portions of the information). The feature “using a vision language system” to perform this function represents routine, conventional data processing means (vision language being a known processing type prior to the effective filing data) for implementing the function and therefore constitutes extra solution activity that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception.
Claim 11 recites “tokenizing the subsurface knowledge text,” and “creating training labels from the tokenized subsurface knowledge text, wherein the training includes using the tokenized subsurface knowledge text as the training labels and the extracted subsurface images as training data or using the tokenized subsurface knowledge text as the training data and the extracted subsurface images as the training labels.” Tokenizing/vectorizing input data to enable efficient machine processing such as for generating training labels represents routine, conventional data processing functions (data pre-processing to place the data in condition for a training step) and therefore constitutes extra solution activity that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception.
Claim 12 recites “wherein the subsurface images are received using a vision-language learning system,” which represents data gathering using routine, convention data processing function (receiving data) using a learning system characterized (labeled) as “vision-language” (images received by an overall system that in some respect includes a vision-language function) with no clear expression that the method including applying some form of computer vision to acquire the images. As such, this element constitutes extra solution activity (computer processing means that includes a function known in the art prior to the effective filing data that receives the data) with respect to the functions falling within the judicial exception.
Claim 15 recites using a natural language processing (NLP) learning system “to capture the subsurface knowledge text from geoscience data, convert the subsurface knowledge text to machine processable data, and vectorize the machine processable data to use as training labels for the training,” which represents routine, conventional data processing functions (learning system that in some manner integrates NLP) for preparing data inputs in a conventional manner (vectorizing and using as training labels) such that this element constitutes extra solution activity that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception.
Claim 16 recites “wherein the training includes using the machine processable data as the training labels and subsurface images as training data, or using the subsurface images as the training labels and the machine processable data as the training data,” which characterizes the type of data to be processed as either training labels or training data and therefore falls and does not significantly limit the scope of the claim with respect to the training process, which as explained in the grounds for rejecting claim 9, constitutes extra solution activity (preparing program instructions for implementing a function that falls within the judicial exception) that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception.
Claim 20 recites “wherein the geoscience knowledge system is a machine learning system using one or more of a reinforcement learning algorithm, a meta-learning algorithm, a NLP algorithm, or an active learning algorithm,” which represents routine, conventional data processing functions (machine learning using one of several well-known types) to implement functions falling within the judicial exception and therefore constitutes extra solution activity that neither integrates the judicial exception into a practical application nor results in the claim as a whole amounting to significantly more than the judicial exception.
Claim 22 characterizes the nature/source of the “synthetic data” using a function “using the geoscience knowledge system correlated with geoscience text and the subsurface information,” that falls within the mental processes judicial exception because it can be performed via mental processes (e.g., evaluation of knowledge system a portion of which constitutes data and geoscience text and subsurface information and judgement to correlate).
Claims 2-22 are therefore also patent ineligible under 101.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-6, 8, 18-25, and 27-29 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Denli (US 2023/0161061 A1).
As to claim 1, Denli teaches “[a] method of inferring subsurface knowledge (Abstract disclosing inferring information from structured information regarding subsurface geology and fluid objects; method performed by geological reasoning system 1000 in FIG. 10), comprising:
obtaining a geoscience knowledge system (graph network depicted in FIG. 7 in combination with structured image representation model 870 and structured attribute representation model 860 depicted in FIG. 8);
obtaining subsurface information (Abstract disclosing obtaining subsurface data; FIG. 5 (obtained) seismic images 501, [0049]; FIG. 8 seismic images 810, [0068]) at a subterranean location (seismic images in FIGS. 5 and 8 inherently relate to a location for which the image data corresponds, [0068] seismic image data corresponds to geophysical observations representative of a subsurface volume); and
inferring subsurface knowledge of the subterranean location from the subsurface information using the geoscience knowledge system (Abstract disclosing inferences of geological objects and fluid objects and relationships between the geological objects and fluid objects (structured representation) from the subsurface information and further disclosing inferring information from the structured representation; FIGS. 2-4 depicting knowledge graphs that relate nodes (objects) by edges (relationships); [0043] identification of objects with the image data; FIG. 5 subprocedures 500 for (per blocks 502-505) extracting information from seismic images to generate graph; FIG. 8 depicting models 870 and 860 for implementing functions in FIG. 5; FIG. 7 graph network 700 processes graph to predict hydrocarbon accumulations for reservoir nodes; [0075] geological reasoning system 1000 (depicted in FIG. 10 as including Inference With Graph Network) performs inference to predict hydrocarbon accumulation based on structured representation 1031) , wherein the subsurface knowledge is utilized to calculate an economic model parameter for a well system proximate the subterranean location (graph network 700, [0057]-[0058] graph network processes graph to predict hydrocarbon accumulations for reservoir nodes. Examiner notes that hydrocarbon accumulation/distribution prediction (e.g., V’ in FIG. 7) is inherently an economic parameter for a prospective or currently existing well (hydrocarbon extraction being the fundamental economic factor of well operation) in relation to (proximate to) the subterranean location represented by the seismic data and that may be used for modeling; [0047]-[0048] disclosing uses of graph data for hydrocarbon exploration relative to hydrocarbon extraction; FIG. 10 inference by Inference With Graph Network used to determine Hydrocarbon Accumulations); and
placing a drilling rig of the well system at a location determined by the economic model parameter ([0030] hydrocarbon management (per [0087] for example) performed based on the graph data) includes identifying well locations), wherein the economic model parameter specifies a cost for drilling a borehole proximate the location ([0036] implementation of the embodiment (generating graph network data) entails identifying geologic information including relationships such as stratigraphic relationships; [0045]-[0046] graph information includes connectivity between geologic and fluid objects that may be characterized in part by distance; [0079] graph network conveys lithology. Examiner notes that stratigraphic and lithographic information as well as underground distance information effectively indicates cost of drilling in terms of distances to drill and type of material to drill through).”
As to claim 2, Denli teaches “[t]he method as recited in Claim 1, wherein the subsurface knowledge is a hydrocarbon distribution (FIG. 7 graph network 700 receives graph inputs 710 including structure and attributes including per FIGS. 3-4 structures such as reservoirs and sources that indicate hydrocarbon distribution).”
As to claim 3, Denli teaches “[t]he method as recited in Claim 1, wherein the subsurface knowledge is geophysical data (FIG. 7 graph network 700 receives graph inputs 710 including structure and attributes including per FIGS. 3-4 and [0040] and [0048] structures such as reservoirs and sources entailing geophysical data).”
As to claim 4, Denli teaches “[t]he method as recited in Claim 3, wherein the geophysical data is one or more of a seismic data ([0040] seismic data contains the geophysical data (e.g., geological features/fluids) or a fossil information ([0040] graph data includes data that per [0048] relates to hydrocarbons that constitute fossilized material).”
As to claim 5, Denli teaches “[t]he method as recited in Claim 1, wherein the subsurface knowledge is utilized to direct a well operation at a well site ([0030] hydrocarbon management that per [0047]-[0048] may be performed base on the graph data may include hydrocarbon extraction, production, drilling, etc. all of which are operations performed at a well site; [0087]).”
As to claim 6, Denli teaches “[t]he method as recited in Claim 1, wherein the subsurface knowledge is represented by graph data (FIG. 3 depicting graph data represented with respect to subsurface features; FIG. 4 graph 400 representing subsurface features such as reservoirs, sources, traps).”
As to claim 8, Denli teaches “[t]he method as recited in Claim 1, wherein the subsurface knowledge is represented by a machine learning algorithm ([0043] graph objects identified via classification; [0050] classification of objects performed via machine learning; [0067] graph/model generated by machine learning).”
As to claim 18, Denli teaches “[t]he method as recited in Claim 1, wherein the subsurface information is received from well log data ([0048] petrophysical interpretation (subsurface data for a current or prospective well) may be log data (i.e., received from well log)).”
As to claim 19, Denli teaches “[t]he method as recited in Claim 1, wherein the subsurface knowledge represents one or more of a spatial ([0036], [0043], and [0055] structured representation may entail spatial relationships (e.g., as depicted in FIG. 4); claim 25. Examiner further notes that the further inferences of subsurface knowledge performed by the graph network (e.g., FIGS. 7 and 10) are based on the structured/graph representation and therefore represent the spatial information) and a depth information, or the spatial and a geologic time information of subterranean formations.”
As to claim 20, Denli teaches “[t]he method as recited in Claim 1, wherein the geoscience knowledge system is a machine learning system ([0043] graph objects identified via classification; [0050] classification of objects performed via machine learning; [0067] graph/model generated by machine learning) using one or more of a reinforcement learning algorithm, a meta-learning algorithm, a NLP algorithm, or an active learning algorithm (the ML classification (labeling) of objects per [0043], [0050], and [0067] and depicted in FIG. 4 (showing labeled vertices/nodes) for downstream processing by a graph network (FIG. 7 and FIG. 10 Inference With Graph Network) defines the structured representation models (models that generate the labeled graphs such as models 870 and 860 in FIG. 8) as active learning models (i.e., models that select features/points for labeling)).”
As to claim 21, Denli teaches “[t]he method as recited in Claim 1, wherein the subsurface information is at least partially synthetic data ([0031] initial data may be synthetic; [0068] geophysical input data for model may include synthetic data).”
As to claim 22, Denli teaches “[t]he method as recited in Claim 21, wherein the synthetic data is generated using the geoscience knowledge system ([0068] simulation models (inherently of the overall system) used to generate synthetic geophysical input data) correlated with geoscience text ([0072] training data used by (and therefore correlated with) the modeling system may be based on domain experts’ annotations (recorded human annotation entails text). Examiner notes that the experts’ annotations (text) relate to geoscience in the context of collecting training data for identifying geological (and subsurface in context) objects) and the subsurface information (Abstract disclosing that the system obtains (and is therefore correlated with) subsurface data; FIG. 5 (obtained) seismic images 501, [0049]; FIG. 8 seismic images 810, [0068]).”
Claim Rejections - 35 USC § 103
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.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Denli in view of Prindle (US 2021/0318465 A1).
As to claim 7, Denli teaches “[t]he method as recited in Claim 6,” but does not appear to teach “wherein the subsurface information is denoised prior to being used for the inferring.”
Prindle discloses a method for generating subsurface models representing lithographic characteristics that includes subsurface information processed by machine learning models (Abstract) and that includes denoising subsurface information prior to further modeling processing ([0048] perform noise filtering on raw data that per FIG. 1 may include seismic data prior to sending as conditioned data for processing by machine learning models 112).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Prindle’s teaching of denoising raw subsurface information prior to modeling to the method taught by Denli such that in combination the method includes denoising the subsurface information prior to being used for inferring.
The motivation would have been to condition the data for optimal modeling performance such as inference modeling as suggested by Prindle.
Claims 9 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Denli in view of Fei, N., Lu, Z., Gao, Y. et al. Towards artificial general intelligence via a multimodal foundation model. Nat Commun 13, 3094 (2022) (Published June 2, 2022), hereinafter “Fei.”
As to claim 9, Denli teaches “[t]he method as recited in Claim 1, further comprising:
training the geoscience knowledge system ([0067] and [0072] structured representation model (e.g., neural network) trained; [0058] graph network (portion of overall geoscience knowledge system) trained), wherein the training comprises:
acquiring subsurface knowledge text ([0072] training data may be based on domain experts’ annotations (recorded human annotation entails text). Examiner notes that the experts’ annotations (text) relate to subsurface knowledge in the context of collecting training data for identifying geological (and subsurface in context) objects);
acquiring subsurface information ([0072] training data may be based on geoscientific simulations in combination with expert’s annotations);”
[combining] “the subsurface knowledge text and the subsurface information ([0072] training data may be based on geoscientific simulations in combination with expert’s annotations); and
training the geoscience knowledge system to identify hydrocarbon distributions utilizing the” [combining] ([0072] training data (i.e., data used for training the model such as structured representation models 860 and 870) may be based on geoscientific simulations in combination with expert’s annotations).”
Denli suggests that experts’ annotations (text) are used as training labels such that the combined text and image data would be correlated ([0072] and claim 21 experts’ annotations used in combination with geoscience simulation data as training data) but does not expressly teach “correlating” the subsurface knowledge text and the subsurface information and training the geoscience knowledge system to identify hydrocarbon distributions utilizing the “correlating.”
Fei discloses a method for training (termed pre-training as applied to multimodal modelling) a model in which training/learning is performed by correlating image data with text data (FIG. 1a and 1b depicting image and text data received and relationally processed (semantically correlated/labeled) in a learning phase; page 3, Results, paragraphs beginning with “Pre-training data collection. We construct …” and “Neural network visualization. Humans have the ability…” describing weakly correlated image-text datasets used for learning).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Fei’s teaching of correlating image data with text data and using the correlation as a form of labeling to the method taught by Denli such that in combination the method includes correlating the subsurface knowledge text and the subsurface information and training the geoscience knowledge system to identify hydrocarbon distributions utilizing the correlating.
The motivation would have been to leverage the semantic nature of textual training data that by its nature conveys a condition to label/classify other type of information such as image data as a useful modeling construct for training and utilizing models that may receive multimodal (e.g., image and text data) as input.
As to claim 17, the combination of Denli and Fei teaches “[t]he method as recited in Claim 9, wherein the subsurface knowledge text and subsurface images are acquired from one or more of a geoscience knowledge database (Denli: claim 27 subsurface data may be seismic images and may stored in/as knowledge bases; [0048]), a geoscience document (Denli: [0048] experts’ geoscience information that per [0072] may be experts’ annotation used for training may be recorded in documents; claim 28 subsurface data may be recorded in/as text documents), or a geoscience article.”
Claims 10-13 are rejected under 35 U.S.C. 103 as being unpatentable over Denli in view of Fei as applied to claim 9 above, and further in view of Ma (US 2023/0206041 A1).
As to claim 10, the combination of Denli and Fei teaches “[t]he method as recited in Claim 9, wherein geoscience text (Denli: [0072] domain experts’ annotations for training) is integrated with subsurface images in geoscience data (Denli: [0067] and [0072]-[0073] seismic image data including via geoscience simulations and domain experts’ annotations used in combination as training data), and the acquiring subsurface knowledge text, the acquiring subsurface information, and the correlating includes using a” “system to extract the subsurface knowledge text and the subsurface images from the geoscience data (Denli: per [0067] and [0072]-[0073] seismic images and experts’ annotations used for training and therefore in some manner are extracted from the overall body of data (geoscience data)) and relate respective of the extracted subsurface knowledge text and the subsurface images (Denli: seismic image data including via geoscience simulations and domain experts’ annotations used in combination (combined for a same overall task) for model training).”
Denli further discloses that the models may be trained and used for seismic image object detection (entails recognizing objects within an image) and furthermore that a convolutional neural network (CNN) (widely known for implementing computer vision) may be used for training the structured representation model for detecting objects in the images ([0067]).
However, Denli does not appear to explicitly teach that the CNN used for training the structured representation model includes a language-vision function and therefore does not expressly teach using “a vision language” system (system that includes a computer vision and language recognition/conversion function) for implementing one or more of the extraction and relation functions used in the training.
Ma discloses a method for implementing deep learning and discloses in the background that it is known in the field of deep learning to use CNNs for learning (training) including CNNs that implement computer vision and natural language processing ([0003]).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Ma’s disclosure that CNNs may implement computer vision and natural language processing for deep learning to the method taught by Denli as modified by Fei in which a CNN may be used for training a machine learning model in a process that entails extracting and relating subsurface knowledge text and subsurface images such that in combination the method includes using a vision language system such as one or more CNNs that implement computer vision and natural language processing for a process that entails extracting and relating the subsurface knowledge text and subsurface images.
Such a combination would amount to selecting a known design option for configuring a CNN for training a machine learning model to achieve predictable results and further motivation would have been to efficiently (systematically) process various types of training data inputs (multimodal) including text and image data such as the data types disclosed by Denli.
As to claim 11, the combination of Denli, Fei, and Ma teaches “[t]he method as recited in Claim 10,” and Fei further teaches tokenizing a knowledge text and creating training labels from the tokenized text (page 5, paragraph beginning with “Text-to-image generation. Network/neuron visualizations” describing use of tokens (text in context of text/image pairs) as vectors from a pre-trained token set (per FIG. 1b image/text pairs bilaterally label; page 10, paragraph beginning with “Text encoder. Given a sentence” describing tokenizing of text sentences by a text encoder that per FIG. 1 is used in generating/training the foundation model).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Fei’s teaching of tokenizing text data that is used for creating training labels from the tokenized text to the method taught by Denli as modified by Fei and Ma, such that in combination the method includes tokenizing the subsurface knowledge text and creating training labels from the tokenized subsurface knowledge text. In such combination, the method would consequently include the training using the tokenized subsurface knowledge text as the training labels (text training data as disclosed by Denli in [0072] in which experts’ annotations are used as descriptions (labels) for learning to identify geologic object and or, as disclosed by Fei, text used for generating foundation model (FIGS. 1a and 1b) in bilateral labeling in text/image pairs, being tokenized per Fei) and the extracted subsurface images as training data (Denli [0067] and [0072]-[0073] seismic images as training data).
The motivation for such combination of Fei’s tokenizing of text data and used of the tokenized data to be used as training labels would have been to more efficiently processing text/image pairs for generating/training multimodal/foundation models as suggested by Fei.
As to claim 12, the combination of Denli, Fei, and Ma teaches “[t]he method as recited in Claim 10, wherein the subsurface images are received using a vision-language learning system (Denli: per [0067] and [0072]-[0073] seismic images used for training and therefore are received by/using a learning system. Denli as combined with Ma in the grounds for rejecting claim 10 teaches a method in which a CNN applying vision language (i.e. computer vision and natural language processing) is used for training/learning such that the learning system is a “vision-language” learning system).”
As to claim 13, the combination of Denli, Fei, and Ma teaches “[t]he method as recited in Claim 10, wherein the subsurface images are seismic images (Denli: [0067] and [0072]-[0073] model trained to detect objects in seismic images).”
Claims 14 is rejected under 35 U.S.C. 103 as being unpatentable over Denli in view of Fei as applied to claim 9 above, and further in view of Prindle (US 2021/0318465 A1).
As to claim 14, the combination of Denli and Fe teaches “[t]he method as recited in Claim 9,” but does not appear to expressly teach that the “subsurface information” that per claim 9 is correlated with the subsurface knowledge text for training the model includes subsurface raw data and subsurface processed data, and therefore does not teach “wherein the subsurface information includes subsurface raw data and subsurface processed data” in context with claim 9.
Prindle discloses a method for generating subsurface models representing lithographic characteristics that includes subsurface information processed by machine learning models (Abstract) and that includes raw and processed subsurface data (FIG. 1 system 100 receiving geoscience data 114 including subsurface data such as seismic that is stored in memory 104 as raw data 106 and conditioned data 108 for processing by machine learning models 112; [0044] raw data used for modeling development; [0048] raw data conditioned to enhance its use for training (e.g., adding labels/tags)).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Prindle’s teaching of using raw and processed subsurface data for model training to the method taught by Denli as modified by Fei such that in combination the method includes using subsurface information that entails raw and processed data (e.g., using raw measured data in addition or as an alternative to the simulation training data disclosed by Denli) that are correlated with subsurface text for model training.
The motivation would have been to leverage the information available in raw data (e.g., seismic image) and to optimize such information by pre-processing the raw data (e.g., denoising, labeling, etc.) to optimize training processing as suggested by Prindle.
Claims 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Denli in view of Fei as applied to claim 9 above, and further in view of Prindle (US 2021/0318465 A1) and Kisra (US 2022/0372866 A1).
As to claim 15, the combination of Denli and Fei teaches “[t]he method as recited in Claim 9,” and Denli further teaches “wherein the acquiring the subsurface knowledge text includes using a” “learning system to capture the subsurface knowledge text from geoscience data (Denli: per [0067] and [0072]-[0073] experts’ annotations used for training and therefore are captured by a learning system from an overall body of data (geoscience data)), convert the subsurface knowledge text to machine processable data (converting the text to use in training as disclosed in Denli [0072] inherently entails a conversion to some form of machine processable data for the machine-implemented training), and vectorize the machine processable data (Denli: [0040] and [0045] attributes (labels for the structured graphs) may be encoded as vectors).”
Denli suggests that experts’ annotations (text) are used as training labels ([0072] and claim 21 experts’ annotations used in combination (to describe geological objects) with geoscience simulation data) but does not teach wherein the acquiring the subsurface knowledge text includes using a “natural language processing (NLP)” learning system to capture the subsurface knowledge text from geoscience data and vectorize the machine processable data “to use as training labels for the training.”
Prindle discloses a method for generating subsurface models representing lithographic characteristics that includes subsurface information processed by machine learning models (Abstract) and that includes obtaining/generating conditioned data that may be labelled (i.e., used for training) in which the conditioned data may be extracted from published (text) documents ([0048]). Prindle further teaches that the labels themselves may be assigned by user manual entry of data, which effectively constitutes using text as labels ([0048] disclosing user entry of labels), which inherently entails the user entered data being converted to machine-processable inputs for the training processing such as via API ingestion ([0048]).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Prindle’s teaching of obtaining/generating text data to be used as training labels (text converted for machine processing) to the method taught by Denli as modified by Fei such that in combination the method includes wherein the acquiring the subsurface knowledge text includes using a learning system to capture the subsurface knowledge text from geoscience data, convert the subsurface knowledge text to machine processable data, and use the machine processable data as training labels for the training.
The motivation would have been to leverage textual information, such as experts’ annotations disclosed by Denli and/or label data entered by a user as disclosed by Prindle, for data labeling purposes since such information may be directly relevant and particularly useful for labeling as suggested by each of Denli and Prindle.
Prindle further teaches that data used in the training process may be extracted from published (text) documents ([0048]) and further teaches that computer vision may be used for labeling ([0048]). However, none of Denli, Fei, or Prindle expressly teaches using a “natural language processing (NLP)” learning system for the capture, convert, and vectorizing functions and in which the vectorized machine processable data is used as training labels for the training.
Natural language processing for converting text input to machine processable data was well known prior to the effective filing date. For example, Kisra discloses a method for using machine learning for extracting well-related information for use in subsequent processing (Abstract) that includes using natural language processing for extracting text information and converting the text information into machine processable data (Abstract NLP algorithm used for extracting topics data from raw data such as raw comments (text); [0025]-[0026] extracted data may be from text formatted documents such as pdf and Word; FIG. 3 blocks 306 and 314 NPL models extract topics for subsequent processing (aggregation into calibration points per comment, [0054]-[0056]) in which the machine processable data has been vectorized (encompasses being tokenized as per Applicant’s specification) for downstream processing (FIG. 3 blocks 302 and 304 depicting extraction of raw data from drilling reports and preprocessing of the raw data including tokenization; [0054]-[0055] pre-processing entails converting raw comment data into a defined format for subsequent steps such as by converting raw comments into tokens).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Kisra’s teaching of using natural language processing for extracting and converting textual information for further processing, including vectorizing/tokenizing the data with the method taught by Denli as modified by Fei and Prindle in which text information may be used for training data labelling such that in combination the learning system is a NLP learning system that implements the extraction, conversion, and vectorization of the text data.
Such a combination would amount to applying a known design option for extracting and converting text input (NLP) and a known design option for formatting the data for downstream machine processing (vectorizing/tokenizing) to achieve predictable results. A further motivation would have been to accurately recognize textual input that may be received by a training/learning system and format the otherwise unstructured data to enable efficient machine processing of the label data.
As to claim 16, the combination of Denli, Fei, Prindle, and Kisra teaches “[t]he method as recited in Claim 15, wherein the training includes using the machine processable data as the training labels (as combined in the grounds for rejecting claim 15, the machine processable data (experts’ annotations disclosed by Denli in [0072] and/or user manual input disclosed by Prindle in [0048]) is used as training labels) and subsurface images as training data (as combined in the grounds for rejecting claim 15, the subsurface images (raw data including seismic data disclosed by Prindle in FIG. 1 and [0048] and/or seismic images disclosed by Denli [0067] and [0072]-[0073]) are the training data), or using the subsurface images as the training labels and the machine processable data as the training data.”
Claims 23-25 and 27-28 are rejected under 35 U.S.C. 103 as being unpatentable over Denli (US 2023/0161061 A1) in view of Samuel (US 2018/0334897 A1).
As to claim 23, as best understood in view of the grounds for rejecting claim 23 under 112(b), Denli teaches “[a] computing system (FIG. 8 system 800 and FIG. 10 system 1000 (computer-implemented as indicated by processing of input data by models 870 and 860); FIG. 11 depicting example computer system), comprising:
a data receiver (FIG. 5 depicting input of seismic image input to subprocedures (the interface and subprocedures constituting a receiver); FIG. 7 depicting interface for receiving inputs 710 to graph network 700; FIG. 8 depicting input interface (arrow) from seismic images 810 to structured image representation model 870; FIG. 10 depicting input interfaces from seismic image 1010 to structured representation model 1070 and from the structured representation model (blocks 1070 and Construct Graph) to Inference with Graph Network (graph network model). Examiner notes that the data receiver comprising each of the nodes/interfaces receiving the data constitute computer processing nodes. Further disclosure of a “data receiver” includes computer-implemented seismic data analysis system 9900 in FIG. 11 for executing the disclosed methods, and which include an external data receiver (communications adapter 9922) as well as internal depicted data receivers (each of CPU 9902, RAM 9906, and GPU 9914 configured to receive various data via bus 9904 and internal data receivers (e.g., executable program constructs within memory (e.g., RAM 9906 and CPU 9902) such as depicted in FIGS. 5, 7-8, and 10), capable of receiving input parameters, a geoscience knowledge system, and subsurface information, where the subsurface information is at a subterranean location (the processing interfaces nodes depicted in FIGS. 5, 7-8, and 10 are computer processing nodes that are “capable of” receiving any input data including input parameters, a geoscience knowledge system (modeling data inputs that become part of the model and therefore part of the “knowledge system”), and subsurface information that is at a subterranean location); and
one or more processors to perform operations (FIG. 11 CPU 9902, [0082]), wherein the operations include communicating with the data receiver (processing implemented by processors (e.g., CPU 9902) inherently entails communications between processing nodes (e.g., seismic image inputs to structured representation models and structured graph inputs to graph network as depicted in FIGS. 8 and 10), and inferring subsurface knowledge of the subterranean location using the subsurface information processed using the geoscience knowledge system (Abstract disclosing inferences of geological objects and fluid objects and relationships between the geological objects and fluid objects (structured representation) from the subsurface information and further disclosing inferring information from the structured representation; FIGS. 2-4 depicting knowledge graphs that relate nodes (objects) by edges (relationships); [0043] identification of objects with the image data; FIG. 5 subprocedures 500 for (per blocks 502-505) extracting information from seismic images to generate graph; FIG. 8 depicting models 870 and 860 for implementing functions in FIG. 5; FIG. 7 graph network 700 processes graph to predict hydrocarbon accumulations for reservoir nodes; [0075] geological reasoning system 1000 (depicted in FIG. 10 as including Inference With Graph Network) performs inference to predict hydrocarbon accumulation based on structured representation 1031), where the subsurface knowledge is utilized to calculate an economic model parameter for a well system proximate the subterranean location (graph network 700, [0057]-[0058] graph network processes graph to predict hydrocarbon accumulations for reservoir nodes. Examiner notes that hydrocarbon accumulation/distribution prediction (e.g., V’ in FIG. 7) is inherently an economic parameter for a prospective or currently existing well (hydrocarbon extraction being the fundamental economic factor of well operation) in relation to (proximate to) the subterranean location represented by the seismic data and that may be used for modeling; [0047]-[0048] disclosing uses of graph data for hydrocarbon exploration relative to hydrocarbon extraction; FIG. 10 inference by Inference With Graph Network used to determine Hydrocarbon Accumulations),” “utilizes the economic model parameter to alter a borehole path ([0030] hydrocarbon management (per [0087] for example) performed based on the graph data) includes drilling (inherently entails altering a borehole path at least in terms of distance) and determining the “direction” of operations such as drilling).”
Denli does not appear to disclose a downhole geo-steering system or that the economic model parameter specifies a cost for altering the geo-steering system proximate the subterranean location.
Samuel discloses that it was known in the art to determine a well path followed by a drill bit based on survey measurements that indicate formation structure and properties including drilling operations that use geosteering of a BHA (steering implicitly entails changing directions and the geosteering is inherently performed in some relative proximity to a formation) ([0001]) and further teaches a method/system for controlling drilling operations based on modeled formation properties (Abstract) in which cost is associated with geo-steering in terms of the brittleness and corresponding rate of penetration ([0073] and [0076] brittleness and corresponding rate of penetration used for geosteering. Examiner notes that formation lithology (e.g., brittleness) and rate of penetration are parameters that generally indicate a cost of steering (drilling along a path having these characteristics)).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Samuel’s teaching that geosteering is a well management function that may be controlled based on formation properties that themselves convey a cost of the geosteeing operation (formation properties such as brittleness and corresponding rate of penetration for portions of the formation) to the system taught by Denli such that in combination the system is configured to include a geo-steering system located downhole of the well system that utilizes the economic model parameter to alter a borehole path and in which the economic model parameter specifies a cost for altering the geo-steering system proximate the subterranean location.
Such a combination would amount to selecting a known design option for wellbore formation (Samuel’s disclosed use of geosteering) and known design options for formation properties (lithological properties such as brittleness and corresponding rate of penetration) to achieve predictable results in terms of optimizing economic/cost factors when planning and/or implementing a drilling path.
As to claim 24, the combination of Denli and Samuel teaches “[t]he computing system as recited in Claim 23, further comprising:
a result transceiver (Denli: FIG. 11 communications adapter 9922 is widely known component of computer system that adapts communications to and from the computer (seismic data analysis system 9900)), capable of communicating the subsurface knowledge to a well planning system (Denli: [0087] output of geological graph network used for drilling a well and/or causing a well to be drilled (well planning) – effectively disclosing a well planning system using (and therefore having received) the subsurface knowledge), a reservoir planning system, or a user (Denli: communications adapter of system that generates the subsurface knowledge is “capable of” transmitting any information including the subsurface knowledge from the computer to a destination node/computer; [0087] output of geological graph network used for drilling a well and/or causing a well to be drilled (well planning)).”
As to claim 25, the combination of Denli and Samuel teaches “[t]he computing system as recited in Claim 23, wherein the one or more processors utilize a machine learning system to infer the subsurface information (Denli: [0043] graph objects identified via classification; [0050] classification of objects performed via machine learning; [0067] graph/model generated by machine learning model (structured representation model); [0057]-[0058] graph network e.g., FIGS. 7 and 10 may be a graph neural network) using the geoscience knowledge system (Denli: FIG. 8 structured representation models 870 and 860 form a part of the overall knowledge system; FIG. 10 Inference With Graph Network (graph network) form part of overall knowledge system) and the subsurface information (Denli: FIG. 5 classification/generation of the objects in blocks 502, 503, and 504 performed using seismic image input 501; FIG. 8 generation of graphs (encoded subsurface information) by model 870 based on seismic image input 810; FIG. 10 graph representation 1031 generated base on seismic input information).”
As to claim 27, the combination of Denli and Samuel teaches “[t]he computing system as recited in Claim 25, wherein the machine learning system utilizes synthetic (Denli: [0068] simulation models may be utilized to generate synthetic geophysical input data (for the structured representation models 870 and 860)) and non-synthetic subsurface information (Denli: [0068] seismic image data from surveys may be used in combination with the synthetic data as the initial input data) received from a database (Denli: [0068] input data may be obtained from a library of data from previous surveys and/or simulations. Examiner notes that a library of collected results constitutes a database in accordance with a broadest reasonable interpretation in view of Applicant’s specification), a lab, a corporate environment, or well logs.”
As to claim 28, the combination of Denli and Samuel teaches “[t]he computing system as recited in Claim 23, wherein the one or more processors are part of a reservoir controller (Denli: [0087] disclosing that output of the geological graph network may be used for “producing hydrocarbons using the well” (extraction/pumping) such that the processors used for generating and processing structured graphs at least constructively form part of a reservoir controller).”
Claim 26 is rejected under 35 U.S.C. 103 as being unpatentable over Denli in view of Samuel as applied to claim 25, and in further view of Ma (US 2023/0206041 A1).
As to claim 26, the combination of Denli and Samuel teaches “[t]he computing system as recited in Claim 25, wherein the machine learning system is trained using the subsurface information (Denli: [0072] training date for structured representation model may include experts’ annotations (in context of geological object identification would constitute subsurface information) and geoscientific simulations (Examiner notes with reference to antecedent relation to the “subsurface information” in claim 25, that the simulation data disclosed in [0072] is also disclosed in [0068] as subsurface data that may be used for post-training modeling such that the collection of subsurface data for training is entailed within the overall “subsurface data” disclosed by Denli); [0073] learning (training) utilizes segmented images as inputs; [0058] graph network trained based on geoscientific knowledge)).”
Denli further discloses that the models may be trained and used for seismic image object detection (entails recognizing objects within an image) and furthermore that a convolutional neural network (CNN) (widely known for implementing computer vision) may be used for training the structured representation model for detecting objects in the images ([0067]).
However, Denli does not appear to explicitly teach that the CNN used for training the structured representation model includes a language-vision function and therefore does not expressly teach training the machine learning system using “a vision-language” learning system (learning system that includes a computer vision and language recognition/conversion function).
Ma discloses a method for implementing deep learning and discloses in the background that it is known in the field of deep learning to use CNNs for learning (training) including CNNs that implement computer vision and natural language processing ([0003]).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Ma’s disclosure that CNNs may implement computer vision and natural language processing for deep learning to the method taught by Denli as modified by Samuel in which a CNN may be used for training a machine learning model such that in combination the method includes training the machine learning model using a vision-language learning system such as one or more CNNs that implement computer vision and natural language processing.
Such a combination would amount to selecting a known design option for configuring a CNN for training a machine learning model to achieve predictable results and further motivation would have been to efficiently (systematically) process various types of training data inputs (multimodal) including text and image data such as the data types disclosed by Denli.
Claim 29 is rejected under 35 U.S.C. 103 as being unpatentable over Denli (US 2023/0161061 A1) in view of Dyadechko (US 2023/0204816 A1).
As to claim 29, Denli teaches “[a] computer program product having a series of operating instructions stored on a non-transitory computer-readable medium that directs a data processing apparatus when executed thereby to perform operations (FIG. 8 system 800 and FIG. 10 system 1000 (computer-implemented as indicated by processing of input data by models 870 and 860. Examiner notes that computer-implemented processes inherently entail use of operating instructions stored on a non-transitory computer-readable medium (e.g., memory storing executable instructions, processor instructions)); FIG. 11 depicting example computer system including CPU 9902, GPU 9914, RAM 9906) to infer subsurface knowledge (Abstract disclosing inferring information from structured information regarding subsurface geology and fluid objects; FIG. 10 Detect Objects 1070 and Inference With Graph Network), the operations comprising:
obtaining a geoscience knowledge system (graph network depicted in FIG. 7 in combination with structured image representation model 870 and structured attribute representation model 860 depicted in FIG. 8);
obtaining subsurface information (Abstract disclosing obtaining subsurface data; FIG. 5 (obtained) seismic images 501, [0049]; FIG. 8 seismic images 810, [0068]) at a subterranean location (seismic images in FIGS. 5 and 8 inherently relate to a location for which the image data corresponds, [0068] seismic image data corresponds to geophysical observations representative of a subsurface volume);
inferring the subsurface knowledge of the subterranean location from the subsurface information using the geoscience knowledge system (Abstract disclosing inferences of geological objects and fluid objects and relationships between the geological objects and fluid objects (structured representation) from the subsurface information and further disclosing inferring information from the structured representation; FIGS. 2-4 depicting knowledge graphs that relate nodes (objects) by edges (relationships); [0043] identification of objects with the image data; FIG. 5 subprocedures 500 for (per blocks 502-505) extracting information from seismic images to generate graph; FIG. 8 depicting models 870 and 860 for implementing functions in FIG. 5; FIG. 7 graph network 700 processes graph to predict hydrocarbon accumulations for reservoir nodes; [0075] geological reasoning system 1000 (depicted in FIG. 10 as including Inference With Graph Network) performs inference to predict hydrocarbon accumulation based on structured representation 1031), wherein the subsurface knowledge is utilized to calculate an economic model parameter for a well system proximate the subterranean location (graph network 700, [0057]-[0058] graph network processes graph to predict hydrocarbon accumulations for reservoir nodes. Examiner notes that hydrocarbon accumulation/distribution prediction (e.g., V’ in FIG. 7) is inherently an economic parameter for a prospective or currently existing well (hydrocarbon extraction being the fundamental economic factor of well operation) in relation to (proximate to) the subterranean location represented by the seismic data and that may be used for modeling; [0047]-[0048] disclosing uses of graph data for hydrocarbon exploration relative to hydrocarbon extraction; FIG. 10 inference by Inference With Graph Network used to determine Hydrocarbon Accumulations).”
Denli further teaches controlling/managing drilling operations ([0030] and [0087]) which inherently includes location of whatever type of well operation that drilling is performed in furtherance but does not specifically hydraulic fracturing as a type of well for which planning is conducted and therefore does not expressly teach “directing a drilling controller to alter a hydraulic fracturing location and an intensity, wherein the economic model parameter specifies a cost for performing a hydraulic fracturing task proximate the hydraulic fracturing location.”
Dyadechko discloses a system/method for implementing geological modeling ([0029]-[0034]) for planning well operations including hydraulic fracturing operations (Abstract; [0050]) that includes optimizing (directing controlling via determining/altering) hydraulic fracturing well locations and location of perforations based on sub-surface modeling ([0049]-[0050]), directing controlling hydraulic fracturing intensity ([0070] operation plans include parameter ranges for hydraulic fracture operations including flow rates and volumes, proppant concentrations, fluid concentrations), and generating an economic model parameter that specifies a cost for performing a hydraulic fracturing task proximate the hydraulic fracturing location ([0034] modeling may include stratigraphic model providing a spatial representation of sequences of rock types in the subsurface. Examiner notes that stratigraphic and lithographic (rock type) information as well as underground distance information effectively indicates cost of drilling (hydraulic fracture operations inherently entail drilling [0036], [0048]-[0049]) in terms of distances to drill and type of material to drill through).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Dyadechko’s teaching of applying subsurface modeling including modeling that includes parameters that convey a cost of performing hydraulic fracturing operations that include controlling (directing a controller) to alter a hydraulic fracturing location and an intensity to the computer program product taught by Denli such that in combination, the program product is configured for directing a drilling controller to alter a hydraulic fracturing location and an intensity, wherein the economic model parameter specifies a cost for performing a hydraulic fracturing task proximate the hydraulic fracturing location.
Such a combination would amount to applying known modeling techniques taught by Denli to a particular category of well and well operations (hydraulic fracturing) that as disclosed by Dyadechko may be more optimally planned/implemented using subsurface properties modeling to achieve predictable results.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MATTHEW W BACA whose telephone number is (571)272-2507. The examiner can normally be reached Monday - Friday 8:00 am - 5:30 pm.
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, Catherine T Rastovski can be reached at (571) 270-0349. 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.
/MATTHEW W. BACA/Examiner, Art Unit 2863
/Catherine T. Rastovski/Supervisory Primary Examiner, Art Unit 2863