DETAILED ACTION
This correspondence is responsive to the application filed on January 31, 2024. Claims 1-20 are pending in the case with claims 1, 15 and 18 in independent form.
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 .
Summary of Detailed Action
I. Claims 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite.
II. Claim 2 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite.
III. Claim 3 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite.
IV. Claim 4 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite.
V. Claim 6 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite.
VI. Claim 9 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claims 1-6, 15 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. and LIN, Wei et al.
Claims 7 and 9-11 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu and Lin, and further in view of Mehrotra et al.
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu, Lin and Mehrotra as applied to claim 7 above, and further in view of Scheepens et al.
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu and of Lin, and further in view of Lyle.
Claims 13 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Lin and Mehrotra, and further in view of Ross et al.
Claim Rejections - 35 USC § 112
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.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
I. Claims 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Independent claims 1, 15 and 18 recite “commissioning an artificially intelligent assistant” for the one or more large language models. It is not clear what commission an artificially intelligent assistant means or what commissioning an artificially intelligent assistant includes or does not include. For example, does commissioning an artificially intelligent assistant include designing or requesting or acquiring an artificially intelligent assistant? Or does commissioning an artificially intelligent assistant mean creating or generating an artificially intelligent assistant ? Or does commissioning an artificially intelligent assistant mean using an artificially intelligent assistant? Or does commissioning an artificially intelligent assistant mean testing or validating an artificially intelligent assistant? Or does commissioning an artificially intelligent assistant mean something else entirely? Thus, the boundaries of the independent claims 1, 15 and 18 are not clear and the independent claims are indefinite. Applicant may cancel independent claims 1, 15 and 18 or amend claims 1, 15 and 18 to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Claims 2-14, 16-17 and 19-20 respectively depend from claims 1, 15 and 18 and are rejected based on the same reasons discussed above with respect to claims 1, 15 and 18.
II. Claim 2 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Claim 2 depends from claim 1 and recites “wherein the technical documents are stored for a plurality of programs on a per-program basis, and wherein the one or more large language models are trained on a per-program basis.” It is not clear what the plurality of programs are or what the programs are for, much less how to store the technical documents on a per-program basis. For example, does “a plurality of programs” mean any program of any type, such as computer programs, technical programs, or artificially intelligent assistant programs or large language model programs, or work programs or are the programs for something else entirely? It is further unclear how to store the technical documents on the “per-program basis” because it is not clear what the programs are or what the programs are for. It is further unclear how to train the large language models on a per-program basis because it is not clear what the programs are or what the programs are for. Thus, the boundaries of the claim are not clear and the claim is indefinite. Applicant may cancel claim 2 or amend claim 2 to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
III. Claim 3 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Claim 2 depends from claim 1 and recites “wherein the one or more large language models are trained to crosslink extracted text and metadata for each work protocol.” It is not clear what crosslink extracted text and metadata for each work protocol means or includes or does not include. For example, does crosslink extracted text and metadata for each work protocol mean that extracted text and metadata are associated with each work protocol? Or does crosslink extracted text and metadata for each work protocol mean that extracted text and metadata reference each work protocol. Or does crosslink extracted text and metadata for each work protocol mean that extracted text and metadata are cross-indexed across each work protocol? Or does crosslink extracted text and metadata for each work protocol mean that extracted text and metadata are hyperlinked to each work protocol? Thus, the boundaries of the claim are not clear and the claim is indefinite. Applicant may cancel claim 3 or amend claim 3 to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
IV. Claim 4 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Claim 4 depends from claim 1 and recites the limitation wherein images are extracted from the technical documents, and wherein the extracted images are used to train “the large language models.” Claim 1 recites “the one or more large language models” and it is not clear if the large language models are the same or different as the “one or more large language models. There is insufficient antecedent basis for this limitation in the claim.
V. Claim 6 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Claim 6 depends from claim 1 and recites wherein the prompt comprises in-context progress through the first work protocol entered via the digital work environment. It is not clear what “in-context progress through the first work protocol” means or what the phrase is referring to or what the phrase includes or does not include. For example, is in-context progress through the first work protocol referring to a certain step in a process, or an application, or a project of a work protocol. Or does in-context progress through the first work protocol mean a part in an engineering or manufacturing or industrial or software work protocol? Or does the phrase in-context progress through the first work protocol mean something else entirely. Thus, the boundaries of the claim are not clear and the claim is indefinite. Applicant may cancel claim 6 or amend claim 6 to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
VI. Claim 9 is rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Claim 9 depends from claim 7 and recites wherein the artificially intelligent assistant is configured to dynamically reallocate resources based on contextual prompts from the artificially intelligent assistant. It is not clear what resources are, much less what the resources include or do not include. It is further unclear how resources are reallocated, or to where the resources are allocated, or to what the resources are allocated. Thus, the boundaries of the claim are not clear and the claim is indefinite. Applicant may cancel claim 9 or amend claim 9 to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim(s) recite(s) subject matter at a general, high-level of a method for extracting text and metadata from the technical documents; retrieve text and metadata in response to receiving a prompt; receiving a prompt related to a first work protocol; retrieving text and metadata related to the first work protocol based on the received prompt; and providing a contextual response based on the retrieved text and metadata, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III). This judicial exception is not integrated into a practical application and the claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Claims 1-20 recite one of the four statutory categories of patentable subject matter and belong to the statutory class(es) of a process (method claims 1-14, 15-17, 18-20), a machine (system/apparatus claims), and an article of manufacture (non-transitory computer readable media claims).
Claim 1 recites a method, thus a process and one of the four statutory categories of patentable subject matter. However, claim 1 further recites for extracting text and metadata from the technical documents; retrieve text and metadata in response to receiving a prompt; receiving a prompt related to a first work protocol; retrieving text and metadata related to the first work protocol based on the received prompt; and providing a contextual response based on the retrieved text and metadata, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
computer-implemented (an additional element merely recites the words “apply it” (or an equivalent) with the judicial exception, or merely includes instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. See also, MPEP 2106.05(f), MPEP 2106.04(d), 2019 Guidance, 84 FR 50 at 55, footnote 30.).
storing technical documents in a storage device, the technical documents including at least work protocols and technical data files tagged with metadata (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
training one or more large language models on at least the work protocols, extracted text, and metadata; (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
commissioning an artificially intelligent assistant for the one or more large language models, the artificially intelligent assistant configured to at least (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
providing a digital work environment including an interface for the artificially intelligent assistant; (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
at the artificially intelligent assistant, (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
via the large language model (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
via the digital work environment (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Thus, the claim is directed to the abstract idea.
Further, the additional elements, alone or in combination, do not provide significantly more than the abstract idea itself, because implementation on a computer (MPEP 2106.05(f)) cannot provide significantly more, and transmitting data over a network is well-understood, routine and conventional (MPEP 2106.05(d), and generally linking the use of the judicial exception to a particular technological field of use does not meaningfully limit the claims (MPEP 2106.04(d)) and the combination of additional elements does not provide an inventive concept. Thus, the claim is ineligible.
Claim 2, dependent on claim 1, does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
wherein the technical documents are stored for a plurality of programs on a per-program basis, (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
wherein the one or more large language models are trained on a per-program basis. (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 3, dependent on claim 1, recites additional abstract ideas to crosslink extracted text and metadata for each work protocol, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
wherein the one or more large language models are trained to (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 4, dependent on claim 1, recites additional abstract ideas for wherein images are extracted from the technical documents, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
wherein the extracted images are used to train the large language models. (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 5, dependent on claim 1, does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
wherein the prompt comprises text input by a user. (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 6, dependent on claim 1, recites additional abstract ideas for wherein the prompt comprises in-context progress through the first work protocol, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
entered via the digital work environment (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 7, dependent on claim 1, recites additional abstract ideas for wherein the prompt comprises a contextual prompt, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
from the artificially intelligent assistant (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 8, dependent on claim 7, recites additional abstract ideas for wherein the contextual prompt is related to one or more non-conformance reports, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
from the artificially intelligent assistant (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 9, dependent on claim 7, does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
wherein the artificially intelligent assistant is configured to dynamically reallocate resources based on contextual prompts from the artificially intelligent assistant (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 10, dependent on claim 9, recites additional abstract ideas for dynamically resequence the first work protocol based on the retrieved text and metadata, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
artificially intelligent assistant is configured to (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 11, dependent on claim 10, recites additional abstract ideas for wherein the retrieved text and metadata comprise inventory data related to the first work protocol, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
Claim 12, dependent on claim 1, does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
wherein the stored technical documents include computer aided engineering and/or drafting files and associated metadata (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
Claim 13, dependent on claim 9, does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
wherein the digital work environment is an augmented reality environment, and wherein the prompt is a visual prompt of a physical work environment received from one or more cameras (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 14, dependent on claim 10, does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
wherein the contextual response is a visual response presented via the augmented reality environment (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 15 recites a method, thus a process and one of the four statutory categories of patentable subject matter. However, claim 15 further recites for extracting text and metadata from the technical documents; retrieve extracted text and metadata in response to receiving a prompt; receiving a prompt to build a new work protocol; retrieving text and metadata contextually related to the new work protocol; and generating a low-granularity work protocol for the new work protocol based on the received prompt and the retrieved text and metadata, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
computer-implemented (an additional element merely recites the words “apply it” (or an equivalent) with the judicial exception, or merely includes instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. See also, MPEP 2106.05(f), MPEP 2106.04(d), 2019 Guidance, 84 FR 50 at 55, footnote 30.).
storing technical documents in a storage device, the technical documents including at least work protocols and technical data files tagged with metadata; (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
training one or more large language models on at least the work protocols, extracted text, and metadata; (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
commissioning an artificially intelligent assistant for the one or more large language models, the artificially intelligent assistant configured to at least (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
at the artificially intelligent assistant (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Thus, the claim is directed to the abstract idea.
Further, the additional elements, alone or in combination, do not provide significantly more than the abstract idea itself, because implementation on a computer (MPEP 2106.05(f)) cannot provide significantly more, and transmitting data over a network is well-understood, routine and conventional (MPEP 2106.05(d), and generally linking the use of the judicial exception to a particular technological field of use does not meaningfully limit the claims (MPEP 2106.04(d)) and the combination of additional elements does not provide an inventive concept. Thus, the claim is ineligible.
Claim 16, dependent on claim 15, recites additional abstract ideas for editing the low-granularity work protocol; retrieving contextually relevant content and presenting at least some of the retrieved contextually relevant content to the user, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
receiving input from a user (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
based on the received input (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
Claim 17, dependent on claim 15, recites additional abstract ideas for editing the low-granularity work protocol; retrieving contextually relevant content and indicating to the user potential non-conformance issues based on the retrieved contextually relevant content, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
receiving input from a user (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
based on the received input (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
with the received input (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
Claim 18 recites a method, thus a process and one of the four statutory categories of patentable subject matter. However, claim 18 further recites for extracting text and metadata from the technical documents; retrieve text and metadata in response to receiving a prompt; generating a contextual response to a prompt related to a first work protocol; responsive to a change in content of one or more of the technical documents, and generating an updated contextual response to the prompt related to the first work protocol, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
computer-implemented (an additional element merely recites the words “apply it” (or an equivalent) with the judicial exception, or merely includes instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. See also, MPEP 2106.05(f), MPEP 2106.04(d), 2019 Guidance, 84 FR 50 at 55, footnote 30.).
storing technical documents in a storage device, the technical documents including at least work protocols and technical data files tagged with metadata; (An additional element of extra-solution activity that courts have identified is well understood, routine and conventional activity for receiving or transmitting data over a network, e.g., using the internet to gather data, store data. See also, MPEP 2106.05(d)(II), MPEP 2106.05(g), 2019 Guidance, 84 FR 50 at 55, 2019 Guidance, 84 FR 50, footnote 31.).
training one or more large language models on at least the work protocols, extracted text, and metadata; (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
commissioning an artificially intelligent assistant for the one or more large language models, (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
the artificially intelligent assistant configured to at least (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
at the artificially intelligent assistant, (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
re-training the one or more large language models; (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
at the artificially intelligent assistant, (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Thus, the claim is directed to the abstract idea.
Further, the additional elements, alone or in combination, do not provide significantly more than the abstract idea itself, because implementation on a computer (MPEP 2106.05(f)) cannot provide significantly more, and transmitting data over a network is well-understood, routine and conventional (MPEP 2106.05(d), and generally linking the use of the judicial exception to a particular technological field of use does not meaningfully limit the claims (MPEP 2106.04(d)) and the combination of additional elements does not provide an inventive concept. Thus, the claim is ineligible.
Claim 19, dependent on claim 18, does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
responsive to user progress through the first work protocol, retraining the one or more large language models. (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
Claim 20, dependent on claim 19, recites additional abstract ideas for generating an updated contextual response to a prompt related to a second work protocol, which are mental processes or concepts that can be performed in the human mind, including observation, evaluation, judgment or opinion, or by a human using pen and paper. MPEP 210604(a)(2)(III).
The claim does not include any additional elements which integrate the abstract idea into a practical application since the additional elements consist of:
at the artificially intelligent assistant, (This additional element amounts to merely the words to “apply it” (or an equivalent) or are mere instructions to implement an abstract idea or other exception on a computer. MPEP 2106.05(f).) Also, this additional element amounts to no more than generally linking the use of the judicial exception to a particular technologic environment or field of use - The application or use of the judicial exception in this manner does not meaningfully limit the claim by going beyond generally linking the use of the judicial exception to a particular technological environment. MPEP 2106.05(h)).
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.
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(s) 1-6, 15 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. (Pub. No. US 2019/0236132 A1, published August 1, 2019) hereinafter Zhu and LIN, Wei et al. (WO 2025/090062 A1, Generative AI Appliance, Applicant Hitachi Vantara LLC, English Language, filed 25 October 2023) hereinafter Lin. The examiner notes that Zhu is cited on applicant’s information disclosure statement filed June 18, 2025.
Regarding claim 1, Zhu teaches:
A computer-implemented method (i.e., methods for generating natural language recommendations based on an industrial language model are provided. … methods can be implemented by one or more data processors either within a single computing system or distributed among two or more computing systems. Zhu, Fig 1, 3, para 5-6.), comprising:
storing technical documents in a storage device, the technical documents including at least work protocols and technical data files
Zhu teaches that, [0014] Providing recommendations of domain-specific natural language can be an important objective for organizations seeking to ensure written and verbal communications include natural language units, such as words, that may be contextually accurate with respect to the nature of the communication being performed as well as the particular industry or domain in which the organization operates. Use of domain-specific words or phrases in product documentation, training, and marketing materials, as well as technical data, application outputs, and search results, can greatly enhance the quality and interpretability of the materials or application outputs (storing technical documents (product documentation, training, marketing materials, technical documents) in a storage device (storing in a database storage device, see also para 50, Figs 1, 3), the technical documents including at least work protocols and technical data files, ). Zhu, para 14, 50, 19.
[0050] Additionally, the client 105 and the server 115 may share data stored in database 110 that can be used in the training system 300 in order to train a prediction model capable of generating domain-specific natural language outputs based on natural language inputs. The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can used to form an industrial language model (storing technical documents in a storage device (database 110), the technical documents including at least work protocols and technical data files In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry. Zhu, Fig 3. para 50, 14, 19. Zhu does not specifically disclose data tagged with metadata.
However, Lin teaches in the field related to a digital advisor associated with an industrial application. Lin, Abstract, para 1. Lin, which is analogous to the claimed invention because Lin is directed to generating output associated with an industrial application, extracting content and context information associated with the industrial application, and context-specific large language models teaches that, [0076] The system, in some aspects, may perform content tagging 721 (e.g., identifying the entities and relationships in a document and then assigning tags (data tagged with metadata) to the entities based on their meaning and context) and content grouping 722 (e.g., clustering documents together based on the entities and relationships that they share). The additional information related to the incontext tree 731 and knowledge graph 732 provided by the content tagging 721 and the content grouping 722, in some aspects, may be used to support other tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation. Lin, para 76.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on an large language model of Zhu using the data tagged with metadata of Lin, with a reasonable expectation of success, in order to provide to address context-specific queries (e.g. to provide answers or recommendations that are useful in a specific context and to provide support for tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation. Lin, para 2, 76. This would have provided the advantages of providing information associated with data used for natural language generation.
extracting text and metadata from the technical documents;
(i.e., [0050] … Additionally, the client 105 and the server 115 may share data stored in database 110 that can be used in the training system 300 in order to train a prediction model capable of generating domain-specific natural language outputs based on natural language inputs. The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can used to form an industrial language model (extracting text and metadata from the technical documents (using, extracting database 110 training data technical documents, files, words text and metadata to form an industrial language model (LLM), see also feature selector extracting, processing training data, Figs 1, 3, para 26-27, 52-53)). In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry. Zhu, Fig 3. para 50, 14, 19, 26-27, 52-53.
[0053] The model training system 255 includes a feature selector 260, a model training 265, and a plurality of training models 270. … The feature selector 260 receives the training data 120 from the client 105 and/or the database 110 via processor 245. The feature selector 260 can process the training data 120 into subsets of data so that downstream data processing and model training can optimize the resulting training models 270 as desired by the design of the machine learning process (extracting text and metadata from the technical documents (feature selector processing, using, extracting database 110 training data technical documents, files, words text and metadata to form an industrial language model LLM, see also Figs 1, 3, para 14, 50, 26-27, 52)). For example, a model training system 255 can be configured to include a feature selector 260 that can process the training input data 120 into categories of documents, images, domain-specific text, non-domain-specific text, numerical data, or the like. For each selected subset of data, also known as features, which can be present in the training data 120, the selected machine learning algorithm can be trained to predict natural language outputs that may be associated with the subset of features for which the selected machine learning algorithm was trained. Zhu, Figs 1, 3. para 53, 52, 50, 26-27,14, 19.)
training one or more large language models on at least the work protocols, extracted text, and metadata;
(i.e., The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can used to form an industrial language model. In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry (training one or more large language models on at least the work protocols, extracted text, and metadata (training industrial language model (large language model) using extracted processed database 110 training data technical documents, files, work protocols, words text and metadata to form an industrial language model LLM), see also Figs 1, 3, para 14, 26-27, 52-53)). Zhu, Fig 3, para 50, 14, 19, 26-27, 52-53.
[0053] The model training system 255 includes a feature selector 260, a model training 265, and a plurality of training models 270. The model training system 255 outputs trained prediction models 275. The feature selector 260 receives the training data 120 from the client 105 and/or the database 110 via processor 245. The feature selector 260 can process the training data 120 into subsets of data so that downstream data processing and model training can optimize the resulting training models 270 (training one or more large language models on at least the work protocols, extracted text, and metadata (training one or more large language model (industrial language model (LLM), para 50) on at least used, extracted, processed training data technical documents, files, work protocols, words text and metadata), see also Figs 1, 3, para 14, 50, 26-27, 52)). as desired by the design of the machine learning process. Zhu, Figs 1, 3, para 53, 52, 50, 14, 19, 26-27.)
commissioning an artificially intelligent assistant for the one or more large language models, the artificially intelligent assistant configured to at least retrieve text and metadata in response to receiving a prompt;
(i.e., As discussed above, Zhu teaches the text and metadata used to train the industrial language model (LLM) and generate natural language recommendations. Zhu, Figs 1, 3, para 50, 14, 19, 52-53, 26-27. Zhu teaches that [0021] Users may receive the domain-specific, natural language outputs in relation to providing natural language inputs to applications that can include auto-suggestion, auto-correction, translation, and interactive agents, such as chatbots (commissioning an artificially intelligent assistant (chatbot, see also Fig 4 #405) for the one or more large language models (for the domain specific trained industrial language model, see also Figs 1, 3-4, para 50, 52-53, 26-27,14, 19), the artificially intelligent assistant (chatbot) configured to at least retrieve text and metadata in response to receiving a prompt (retrieve domain specific text, metadata natural language outputs in relation and response to natural language inputs, see also Figs 1,3, Fig 4 #415 - #420, para 50, 52-53, 26-27,14, 19). The domain-specific natural language system described herein may also be used to provide images as the inputs, in addition to or separately from the natural language inputs. … The natural language outputs can enhance communications between users and ensure that specific formats of domain-specific information can be disseminated within an organization. In addition, the natural language outputs can reduce errors or inaccuracies by providing the outputs as results that include domain-specific language which can be used to aid document and/or text generation and automated report generation. By automating tedious and sometimes subjective text generation tasks, the improved recommendation system can reduce human biases in decision making processes and provide domain-specific data in a more timely manner so that production and operation tasks can be completed more efficiently in an oil and gas production or exploration environment. Zhu, Figs 1, 3-5, para 21, 62-66, 14, 19, 26-27, 50, 52-53.
[0062] FIG. 4 is a flow diagram illustrating an exemplary embodiment of a method 400 for generating natural language outputs including domain-specific recommendations based on natural language inputs using the client/server of FIGS. 1, 2A, 2B and the trained prediction model 275 generated in a machine learning process using the training system 300, as shown and described in relation to FIG. 3 (Figure 4 illustrates commissioning an artificially intelligent assistant (natural language inputs #405, chatbot, para 21) for the one or more large language models (transmitting to language model, language model determining, #410-415), the artificially intelligent assistant configured to at least retrieve text and metadata in response to receiving a prompt (natural language outputs #415-420, chatbot, para 21)). In certain aspects, embodiments of the method 400 can include greater or fewer operations than illustrated in FIG. 4 and the operations can be performed in a different order than illustrated in FIG. 4. Zhu, para 62, 63-66, 50, 52-53, 26-27, 14, 19.)
providing a digital work environment including an interface for the artificially intelligent assistant; at the artificially intelligent assistant, receiving a prompt related to a first work protocol;
(i.e., [0023] FIG. 1 is a block diagram illustrating an example architecture 100 (providing a digital work environment (digital work environment architecture), including an interface for the artificially intelligent assistant (chatbot, para 21)) for generating natural language recommendations based on an industrial language model. The architecture 100 includes clients 105, database 110, and server 115, which can be communicatively coupled over a network. Zhu, Fig 1, 3-5, para 23, 24, 21, 19, 62. [0021] Users may receive the domain-specific, natural language outputs in relation to providing natural language inputs to applications that can include auto-suggestion, auto-correction, translation, and interactive agents, such as chatbots (at the artificially intelligent assistant, receiving a prompt related to a first work protocol l, see also Figs 1, 3, 4, para 19, 26-27, 50, 52-53, 62). … In these ways, a large variety of inputs can be received and used to determine domain-specific, natural language outputs. The natural language outputs can enhance communications between users and ensure that specific formats of domain-specific information can be disseminated within an organization. In addition, the natural language outputs can reduce errors or inaccuracies by providing the outputs as results that include domain-specific language which can be used to aid document and/or text generation and automated report generation. By automating tedious and sometimes subjective text generation tasks, the improved recommendation system can reduce human biases in decision making processes and provide domain-specific data in a more timely manner so that production and operation tasks can be completed more efficiently in an oil and gas production or exploration environment. Zhu, Figs 1, 3-4, para 21, 23, 19, 26-27, 50, 52-53, 62)
retrieving text and metadata related to the first work protocol via the large language model based on the received prompt; and providing a contextual response via the digital work environment based on the retrieved text and metadata. (i.e., [ 0026] As further shown in FIG. 1, natural language inputs can be transmitted from the clients 105 and/or from the database 110 to the prediction server 115. In some embodiments, the natural language inputs includes training data 120 that is transmitted to the prediction server 115 for use in a machine learning process. The training data 120 is used to train a machine learning algorithm in a machine learning process in order to generate a training model capable of predicting recommendations that include domain-specific recommendations based on a wide variety of received natural language inputs (retrieving text and metadata related to the first work protocol via the large language model based on the received prompt (natural language inputs prompt); and providing a contextual response via the digital work environment based on the retrieved text and metadata (providing contextual domain-specific recommendations (text, metadata, para 50, 52-53,14, 19)) response via the digital work environment, see also Fig. 1, para 23, digital work environment architecture)). In some embodiments, the natural language inputs include prediction data 125 that is transmitted to a prediction server 115 as inputs to the generated model that was trained in the machine learning process using the training data 120. The natural language inputs can include inputs that may be provided to a domain-specific application, such as a report generator or search interface. For example, the natural language inputs can include textual inputs to a variety of fields displayed in an interface of the domain-specific application. A user may enter words, numbers, sentences, or even whole documents as the natural language inputs. The natural language inputs provided as prediction data 125 may include language that has a weak or poor contextual relevance to the domain or industry in which the application and the improved recommendation system may be associated. A user may enter “gammar” to a field of an application. The improved recommendation system would be trained to identify this input as related in a contextual and domain-specific manner to “gamma” which may be a data processing component called a “gamma board” that is used within the particular industry or domain for which the predictive model has been trained. As a result, the improved recommendation system may perform a spelling correction to “gamma” and/or provide the user with a suggestion of “gamma board”. Without the improved recommendation system, the input of “gammar” may be corrected to “grammar”, which provides no contextual or domain-specific relevance to a user of the application. Zhu, Fig 1, para 26, 27.)
Regarding claim 2, which depends from claim 1 and recites:
wherein the technical documents are stored for a plurality of programs on a per-program basis, and wherein the one or more large language models are trained on a per-program basis.
Zhu in view of Lin teaches the method of claim 1 from which claim 2 depends, including the technical documents are stored and wherein the one or more large language models are trained. Zhu teaches that, [0050] As shown in FIG. 3, the client 105, the database 110, and the server 115 are connected over the network 235. The client 105 and the server 115 can be configured to exchange data that can be used to determine natural language outputs and domain-specific recommendations. Such data may include text, events, actions, images, event data, event message, input pattern data, requests, responses, and commands to other devices transmitted over the network. Additionally, the client 105 and the server 115 may share data stored in database 110 that can be used in the training system 300 in order to train a prediction model capable of generating domain-specific natural language outputs based on natural language inputs. The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data (technical documents (training input technical documents) are stored for a plurality of programs (domains) on a per-program basis (domain-specific libraries), and wherein the one or more large language models are trained on a per-program basis (industrial language model trained on a specific domain basis)). … In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry. Zhu, Fig 1, 3-4, para 50, 14, 19, 52-53.
Regarding claim 3, which depends from claim 1 and recites:
wherein the one or more large language models are trained to crosslink extracted text and metadata for each work protocol.
Zhu in view of Lin teaches the method of claim 1 from which claim 3 depends,, including the one or more large language models are trained and the extracted text and metadata and each work protocol. Zhu teaches that, [0026] As further shown in FIG. 1, natural language inputs can be transmitted from the clients 105 and/or from the database 110 to the prediction server 115. In some embodiments, the natural language inputs includes training data 120 that is transmitted to the prediction server 115 for use in a machine learning process. The training data 120 is used to train a machine learning algorithm in a machine learning process in order to generate a training model capable of predicting recommendations that include domain-specific recommendations based on a wide variety of received natural language inputs. In some embodiments, the natural language inputs include prediction data 125 that is transmitted to a prediction server 115 as inputs to the generated model that was trained in the machine learning process using the training data 120. The natural language inputs can include inputs that may be provided to a domain-specific application, such as a report generator or search interface. For example, the natural language inputs can include textual inputs to a variety of fields displayed in an interface of the domain-specific application. A user may enter words, numbers, sentences, or even whole documents as the natural language inputs. The natural language inputs provided as prediction data 125 may include language that has a weak or poor contextual relevance to the domain or industry in which the application and the improved recommendation system may be associated. A user may enter “gammar” to a field of an application. The improved recommendation system would be trained to identify this input as related in a contextual and domain-specific manner to “gamma” which may be a data processing component called a “gamma board” that is used within the particular industry or domain for which the predictive model has been trained (one or more large language models are trained to crosslink extracted text and metadata for each work protocol (large language model trained to crosslink and identify contextually related extracted text and metadata for each work protocol domain for which the model has been trained, extracted text and metadata, see also para 50-53, Fig 3)). As a result, the improved recommendation system may perform a spelling correction to “gamma” and/or provide the user with a suggestion of “gamma board”. Without the improved recommendation system, the input of “gammar” may be corrected to “grammar”, which provides no contextual or domain-specific relevance to a user of the application. Zhu, Fig 1, 3-4, para 26, 50-53.
Regarding claim 4, which depends from claim 1 and recites:
wherein images are extracted from the technical documents, and wherein the extracted images are used to train the large language models.
Zhu in view of Lin teaches the method of claim 1 from which claim 4 depends,, including the technical documents used to train one or more large language models. Zhu teaches that, [0025] The architecture 100 also includes a database 110 that can store domain-specific data sets, documentation, textual data sources, and non-textual data sources, such as image libraries containing one or more images (images are extracted from the technical documents used to train the large language models (technical documents images extracted and used to train large language model (industrial language model)). In addition, user data associated with natural language inputs that users have provided to one or more applications that are configured on any of the clients 105 can be stored in the database 110. The database 110 can further store the natural language outputs that can be generated by the improved recommendation system described herein. Zhu, Figs 1, 3-4, para 25, 50, 26, 19, 14, 52-53. [0050] Additionally, the client 105 and the server 115 may share data stored in database 110 that can be used in the training system 300 in order to train a prediction model capable of generating domain-specific natural language outputs based on natural language inputs. The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data (images are extracted from the technical documents used to train the large language models (technical documents images extracted and used to train large language model (industrial language model)). The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can be used to form an industrial language model. In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry. Zhu, Figs 1, 3-4, para 50, 25, 26, 19, 14, 50-53.
Regarding claim 5, which depends from claim 1 and recites:
wherein the prompt comprises text input by a user.
Zhu in view of Lin teaches the method of claim 1 from which claim 5 depends,, including the prompt. Zhu teaches that, [0026] As further shown in FIG. 1, natural language inputs can be transmitted from the clients 105 and/or from the database 110 to the prediction server 115. … In some embodiments, the natural language inputs include prediction data 125 that is transmitted to a prediction server 115 as inputs to the generated model that was trained in the machine learning process using the training data 120. The natural language inputs can include inputs that may be provided to a domain-specific application, such as a report generator or search interface. For example, the natural language inputs can include textual inputs (prompt comprises text input by a user) to a variety of fields displayed in an interface of the domain-specific application. A user may enter words, numbers, sentences, or even whole documents as the natural language inputs. The natural language inputs provided as prediction data 125 may include language that has a weak or poor contextual relevance to the domain or industry in which the application and the improved recommendation system may be associated. A user may enter “gammar” to a field of an application. The improved recommendation system would be trained to identify this input as related in a contextual and domain-specific manner to “gamma” which may be a data processing component called a “gamma board” that is used within the particular industry or domain for which the predictive model has been trained. As a result, the improved recommendation system may perform a spelling correction to “gamma” and/or provide the user with a suggestion of “gamma board”. Without the improved recommendation system, the input of “gammar” may be corrected to “grammar”, which provides no contextual or domain-specific relevance to a user of the application. Zhu, Fig 1, 3-4, para 26.
Regarding claim 6, which depends from claim 1 and recites:
wherein the prompt comprises in-context progress through the first work protocol entered via the digital work environment.
Zhu in view of Lin teaches the method of claim 1 from which claim 6 depends,, including the prompt entered via the digital work environment. Zhu teaches that, [0050] As shown in FIG. 3, the client 105, the database 110, and the server 115 are connected over the network 235. The client 105 and the server 115 can be configured to exchange data that can be used to determine natural language outputs and domain-specific recommendations. Such data may include text, events, actions, images, event data, event message, input pattern data, requests, responses, and commands to other devices transmitted over the network (prompt (prompt see also inputs data, para 26, Fig 1) comprises in-context (in-context events, actions, event data, event message, input pattern data, requests, responses, commands) progress through the first work protocol entered via the digital work environment (digital work environment architecture, see also Figs 1, 3, para 23-27). Additionally, the client 105 and the server 115 may share data stored in database 110 that can be used in the training system 300 in order to train a prediction model capable of generating domain-specific natural language outputs based on natural language inputs. The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can be used to form an industrial language model. In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry. Zhu, Figs 1, 3-4, para 50, 14, 19.
Regarding claim 15, Zhu teaches:
A computer-implemented method (i.e., methods for generating natural language recommendations based on an industrial language model are provided. … methods can be implemented by one or more data processors either within a single computing system or distributed among two or more computing systems. Zhu, Fig 1, 3, para 5-6.), comprising:
storing technical documents in a storage device, the technical documents including at least work protocols and technical data files
Zhu teaches that, [0014] Providing recommendations of domain-specific natural language can be an important objective for organizations seeking to ensure written and verbal communications include natural language units, such as words, that may be contextually accurate with respect to the nature of the communication being performed as well as the particular industry or domain in which the organization operates. Use of domain-specific words or phrases in product documentation, training, and marketing materials, as well as technical data, application outputs, and search results, can greatly enhance the quality and interpretability of the materials or application outputs (storing technical documents (product documentation, training, marketing materials, technical documents) in a storage device (storing in a database storage device, see also para 50, Figs 1, 3), the technical documents including at least work protocols and technical data files, ). Zhu, para 14, 50, 19.
[0050] Additionally, the client 105 and the server 115 may share data stored in database 110 that can be used in the training system 300 in order to train a prediction model capable of generating domain-specific natural language outputs based on natural language inputs. The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can used to form an industrial language model (storing technical documents in a storage device (database 110), the technical documents including at least work protocols and technical data files product specification data including work protocols and metadata)) In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry. Zhu, Fig 3. para 50, 14, 19. Zhu does not specifically disclose data tagged with metadata.
However, Lin teaches in the field related to a digital advisor associated with an industrial application. Lin, Abstract, para 1. Lin, which is analogous to the claimed invention because Lin is directed to generating output associated with an industrial application, extracting content and context information associated with the industrial application, and context-specific large language models teaches that, [0076] The system, in some aspects, may perform content tagging 721 (e.g., identifying the entities and relationships in a document and then assigning tags (data tagged with metadata) to the entities based on their meaning and context) and content grouping 722 (e.g., clustering documents together based on the entities and relationships that they share). The additional information related to the incontext tree 731 and knowledge graph 732 provided by the content tagging 721 and the content grouping 722, in some aspects, may be used to support other tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation. Lin, para 76.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on a large language model of Zhu using the data tagged with metadata of Lin, with a reasonable expectation of success, in order to provide support other tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation. Lin, para 76. This would have provided the advantages of providing information associated with data used for natural language generation.
extracting text and metadata from the technical documents;
(i.e., [0050] … Additionally, the client 105 and the server 115 may share data stored in database 110 that can be used in the training system 300 in order to train a prediction model capable of generating domain-specific natural language outputs based on natural language inputs. The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can used to form an industrial language model (extracting text and metadata from the technical documents (using, extracting database 110 training data technical documents, files, words text and metadata to form an industrial language model (LLM), see also feature selector extracting, processing training data, Figs 1, 3, para 26-27, 52-53)). In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry. Zhu, Fig 3. para 50, 14, 19, 26-27, 52-53.
[0053] The model training system 255 includes a feature selector 260, a model training 265, and a plurality of training models 270. … The feature selector 260 receives the training data 120 from the client 105 and/or the database 110 via processor 245. The feature selector 260 can process the training data 120 into subsets of data so that downstream data processing and model training can optimize the resulting training models 270 as desired by the design of the machine learning process (extracting text and metadata from the technical documents (feature selector processing, using, extracting database 110 training data technical documents, files, words text and metadata to form an industrial language model LLM, see also Figs 1, 3, para 14, 50, 26-27, 52)). For example, a model training system 255 can be configured to include a feature selector 260 that can process the training input data 120 into categories of documents, images, domain-specific text, non-domain-specific text, numerical data, or the like. For each selected subset of data, also known as features, which can be present in the training data 120, the selected machine learning algorithm can be trained to predict natural language outputs that may be associated with the subset of features for which the selected machine learning algorithm was trained. Zhu, Figs 1, 3. para 53, 52, 50, 26-27,14, 19.)
training one or more large language models on at least the work protocols, extracted text, and metadata;
(i.e., The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can used to form an industrial language model. In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry (training one or more large language models on at least the work protocols, extracted text, and metadata (training industrial language model (large language model) using extracted processed database 110 training data technical documents, files, work protocols, words text and metadata to form an industrial language model LLM), see also Figs 1, 3, para 14, 26-27, 52-53)). Zhu, Fig 3, para 50, 14, 19, 26-27, 52-53.
[0053] The model training system 255 includes a feature selector 260, a model training 265, and a plurality of training models 270. The model training system 255 outputs trained prediction models 275. The feature selector 260 receives the training data 120 from the client 105 and/or the database 110 via processor 245. The feature selector 260 can process the training data 120 into subsets of data so that downstream data processing and model training can optimize the resulting training models 270 (training one or more large language models on at least the work protocols, extracted text, and metadata (training one or more large language model (industrial language model (LLM), para 50) on at least used, extracted, processed training data technical documents, files, work protocols, words text and metadata), see also Figs 1, 3, para 14, 50, 26-27, 52)). as desired by the design of the machine learning process. Zhu, Figs 1, 3, para 53, 52, 50, 14, 19, 26-27.)
commissioning an artificially intelligent assistant for the one or more large language models, the artificially intelligent assistant configured to at least retrieve extracted text and metadata in response to receiving a prompt;
(i.e., As discussed above, Zhu teaches the text and metadata used to train the industrial language model (LLM) and generate natural language recommendations. Zhu, Figs 1, 3, para 50, 14, 19, 52-53, 26-27. Zhu teaches that [0021] Users may receive the domain-specific, natural language outputs in relation to providing natural language inputs to applications that can include auto-suggestion, auto-correction, translation, and interactive agents, such as chatbots (commissioning an artificially intelligent assistant (chatbot, see also Fig 4 #405) for the one or more large language models (for the domain specific trained industrial language model, see also Figs 1, 3-4, para 50, 52-53, 26-27,14, 19), the artificially intelligent assistant (chatbot) configured to at least retrieve text and metadata in response to receiving a prompt (retrieve domain specific text, metadata natural language outputs in relation and response to natural language inputs, see also Figs 1,3, Fig 4 #415 - #420, para 50, 52-53, 26-27,14, 19). The domain-specific natural language system described herein may also be used to provide images as the inputs, in addition to or separately from the natural language inputs. … The natural language outputs can enhance communications between users and ensure that specific formats of domain-specific information can be disseminated within an organization. In addition, the natural language outputs can reduce errors or inaccuracies by providing the outputs as results that include domain-specific language which can be used to aid document and/or text generation and automated report generation. By automating tedious and sometimes subjective text generation tasks, the improved recommendation system can reduce human biases in decision making processes and provide domain-specific data in a more timely manner so that production and operation tasks can be completed more efficiently in an oil and gas production or exploration environment. Zhu, Figs 1, 3-5, para 21, 62-66, 14, 19, 26-27, 50, 52-53.
[0062] FIG. 4 is a flow diagram illustrating an exemplary embodiment of a method 400 for generating natural language outputs including domain-specific recommendations based on natural language inputs using the client/server of FIGS. 1, 2A, 2B and the trained prediction model 275 generated in a machine learning process using the training system 300, as shown and described in relation to FIG. 3 (Figure 4 illustrates commissioning an artificially intelligent assistant (natural language inputs #405, chatbot, para 21) for the one or more large language models (transmitting to language model, language model determining, #410-415), the artificially intelligent assistant configured to at least retrieve text and metadata in response to receiving a prompt (natural language outputs #415-420, chatbot, para 21)). In certain aspects, embodiments of the method 400 can include greater or fewer operations than illustrated in FIG. 4 and the operations can be performed in a different order than illustrated in FIG. 4. Zhu, para 62, 63-66, 50, 52-53, 26-27, 14, 19.) and
at the artificially intelligent assistant: receiving a prompt to
Zhu teaches that [0023] FIG. 1 is a block diagram illustrating an example architecture 100 for generating natural language recommendations based on an industrial language model (at the artificially intelligent assistant (chatbot, see para 21): receiving a prompt (natural language inputs prompt, see Fig 1) to a work protocol (see Fig. 1 natural language inputs prompt to a digital work protocol environment architecture) and generating a low-granularity work protocol for the work protocol based on the received prompt (generating a low-granularity natural language outputs recommendation work protocol for the work protocol based on the received natural language inputs prompt, see Fig 1)). The architecture 100 includes clients 105, database 110, and server 115, which can be communicatively coupled over a network. Zhu, Fig 1, 3-5, para 23, 24, 21, 19, 62. [0021] Users may receive the domain-specific, natural language outputs in relation to providing natural language inputs to applications that can include auto-suggestion, auto-correction, translation, and interactive agents, such as chatbots (at the artificially intelligent assistant, receiving a prompt related to a work protocol, see also Figs 1, 3, 4, para 19, 26-27, 50, 52-53, 62). … In these ways, a large variety of inputs can be received and used to determine domain-specific, natural language outputs. The natural language outputs can enhance communications between users and ensure that specific formats of domain-specific information can be disseminated within an organization. In addition, the natural language outputs can reduce errors or inaccuracies by providing the outputs as results that include domain-specific language which can be used to aid document and/or text generation and automated report generation. By automating tedious and sometimes subjective text generation tasks, the improved recommendation system can reduce human biases in decision making processes and provide domain-specific data in a more timely manner so that production and operation tasks can be completed more efficiently in an oil and gas production or exploration environment. Zhu, Figs 1, 3-4, para 21, 23, 19, 26-27, 50, 52-53, 62. [0026] As further shown in FIG. 1, natural language inputs can be transmitted from the clients 105 and/or from the database 110 to the prediction server 115. In some embodiments, the natural language inputs includes training data 120 that is transmitted to the prediction server 115 for use in a machine learning process. The training data 120 is used to train a machine learning algorithm in a machine learning process in order to generate a training model capable of predicting recommendations that include domain-specific recommendations based on a wide variety of received natural language inputs (retrieving text and metadata contextually related to the work protocol (retrieving and providing contextual domain-specific recommendations (text, metadata, para 50, 52-53,14, 19) contextually related to the natural language inputs prompt workflow, see also Fig. 1, para 23 digital work environment architecture)). In some embodiments, the natural language inputs include prediction data 125 that is transmitted to a prediction server 115 as inputs to the generated model that was trained in the machine learning process using the training data 120. The natural language inputs can include inputs that may be provided to a domain-specific application, such as a report generator or search interface. For example, the natural language inputs can include textual inputs to a variety of fields displayed in an interface of the domain-specific application. A user may enter words, numbers, sentences, or even whole documents as the natural language inputs. The natural language inputs provided as prediction data 125 may include language that has a weak or poor contextual relevance to the domain or industry in which the application and the improved recommendation system may be associated. A user may enter “gammar” to a field of an application. The improved recommendation system would be trained to identify this input as related in a contextual and domain-specific manner to “gamma” which may be a data processing component called a “gamma board” that is used within the particular industry or domain for which the predictive model has been trained. As a result, the improved recommendation system may perform a spelling correction to “gamma” and/or provide the user with a suggestion of “gamma board”. Without the improved recommendation system, the input of “gammar” may be corrected to “grammar”, which provides no contextual or domain-specific relevance to a user of the application. Zhu, Fig 1, para 26, 27.
Thus, Zhu teaches at the artificially intelligent assistant: receiving a prompt to a work protocol; retrieving text and metadata contextually related to the work protocol; and generating a low-granularity work protocol for the work protocol based on the received prompt and the retrieved text and metadata. Zhu does not specifically disclose prompt to build a new work protocol, and retrieve new work protocol and generate new work protocol.
However, Lin teaches that, [0082] In some aspects, there may be two core types of content that can be used to train an LLM (one or more of the technical documents work protocols content): documents and know-how. Documents, in some aspects, may be anything from product specifications to customer support. The documents may provide the LLM with a wealth of information about the domain in which it will be used. Know-how, in some aspects, may be the tacit knowledge that experts in a particular domain possess. [0102] Once a pretrained LLM has been fine-tuned and corrected (e.g., to produce a private LLM question generator 1421 and/or a private LLM answer generator 1432), in some aspects, it may be deployed in a model serving environment (e.g., a model server 1440). This allows the model to be accessed by users through a variety of channels, such as a web application, a mobile app, or a chatbot (at artificially intelligent assistant (access LLM at artificially intelligent assistant chatbot)). [0134] In some aspects, the method (or apparatus) may receive at least one of feedback regarding the response or additional content and context information associated with the particular context (receive prompt to build a new work protocol (feedback prompt to build new (technical document work protocol content, para 82) content), and retrieve new work protocol (retrieve, receive new additional (technical document work protocol content, para 82) content)). In some aspects, based on receiving one or more of the feedback and/or additional context and/or context, the method may return to 2020 where it may update the context-specific LLM corresponding to the particular context based on at least one of the feedback or the additional content and context information, identifying an additional triggering event associated with the particular context at 2030, and generating, after updating the context specific LLM corresponding to the particular context, an additional response to the additional triggering event using the context-specific LLM corresponding to the particular context at 2040 (receive prompt to build a new work protocol (feedback prompt to build new content, technical document work protocol content, see also para 82), and retrieve new work protocol (receive, retrieve new addition work protocol content, see also para 82) and generate new work protocol (generate, update, build new (technical document work protocol content, para 82) content)). For example, referring to FIGs. 1, 8, 9, 14, 18, and 19, the apparatus and/or method may identify and/or generate an additional prompt such as agent prompt 121/921, one or more user prompts 122/922, question prompt 1820 that may be associated with a particular context or, referring to FIGs. 1, 8, 9, and 19, the apparatus and/or method may receive feedback associated with the reinforcement learner 180/880/980/1980. Lin, para 134, 102, 82, 83, 85, 131.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on an large language model of Zhu using the data tagged with metadata and receiving prompt to build a new work protocol, and retrieve new work protocol and generate new work protocol of Lin, with a reasonable expectation of success, in order to provide to address context-specific queries (e.g. to provide answers or recommendations that are useful in a specific context and to provide support for tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation. Lin, para 2, 76, 134. This would have provided the advantages of providing updated information associated with data used for natural language generation.
Regarding claim 18, Zhu teaches:
A computer-implemented method (i.e., methods for generating natural language recommendations based on an industrial language model are provided. … methods can be implemented by one or more data processors either within a single computing system or distributed among two or more computing systems. Zhu, Fig 1, 3, para 5-6.), comprising:
storing technical documents in a storage device, the technical documents including at least work protocols and technical data files
Zhu teaches that, [0014] Providing recommendations of domain-specific natural language can be an important objective for organizations seeking to ensure written and verbal communications include natural language units, such as words, that may be contextually accurate with respect to the nature of the communication being performed as well as the particular industry or domain in which the organization operates. Use of domain-specific words or phrases in product documentation, training, and marketing materials, as well as technical data, application outputs, and search results, can greatly enhance the quality and interpretability of the materials or application outputs (storing technical documents (product documentation, training, marketing materials, technical documents) in a storage device (storing in a database storage device, see also para 50, Figs 1, 3), the technical documents including at least work protocols and technical data files, ). Zhu, para 14, 50, 19.
[0050] Additionally, the client 105 and the server 115 may share data stored in database 110 that can be used in the training system 300 in order to train a prediction model capable of generating domain-specific natural language outputs based on natural language inputs. The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can used to form an industrial language model (storing technical documents in a storage device (database 110), the technical documents including at least work protocols and technical data files In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry. Zhu, Fig 3. para 50, 14, 19. Zhu does not specifically disclose data tagged with metadata.
However, Lin teaches in the field related to a digital advisor associated with an industrial application. Lin, Abstract, para 1. Lin, which is analogous to the claimed invention because Lin is directed to generating output associated with an industrial application, extracting content and context information associated with the industrial application, and context-specific large language models teaches that, [0076] The system, in some aspects, may perform content tagging 721 (e.g., identifying the entities and relationships in a document and then assigning tags (data tagged with metadata) to the entities based on their meaning and context) and content grouping 722 (e.g., clustering documents together based on the entities and relationships that they share). The additional information related to the incontext tree 731 and knowledge graph 732 provided by the content tagging 721 and the content grouping 722, in some aspects, may be used to support other tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation. Lin, para 76.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on a large language model of Zhu using the data tagged with metadata of Lin, with a reasonable expectation of success, in order to provide support other tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation. Lin, para 76. This would have provided the advantages of providing information associated with data used for natural language generation.
extracting text and metadata from the technical documents;
(i.e., [0050] … Additionally, the client 105 and the server 115 may share data stored in database 110 that can be used in the training system 300 in order to train a prediction model capable of generating domain-specific natural language outputs based on natural language inputs. The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can used to form an industrial language model (extracting text and metadata from the technical documents (using, extracting database 110 training data technical documents, files, words text and metadata to form an industrial language model (LLM), see also feature selector extracting, processing training data, Figs 1, 3, para 26-27, 52-53)). In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry. Zhu, Fig 3. para 50, 14, 19, 26-27, 52-53.
[0053] The model training system 255 includes a feature selector 260, a model training 265, and a plurality of training models 270. … The feature selector 260 receives the training data 120 from the client 105 and/or the database 110 via processor 245. The feature selector 260 can process the training data 120 into subsets of data so that downstream data processing and model training can optimize the resulting training models 270 as desired by the design of the machine learning process (extracting text and metadata from the technical documents (feature selector processing, using, extracting database 110 training data technical documents, files, words text and metadata to form an industrial language model LLM, see also Figs 1, 3, para 14, 50, 26-27, 52)). For example, a model training system 255 can be configured to include a feature selector 260 that can process the training input data 120 into categories of documents, images, domain-specific text, non-domain-specific text, numerical data, or the like. For each selected subset of data, also known as features, which can be present in the training data 120, the selected machine learning algorithm can be trained to predict natural language outputs that may be associated with the subset of features for which the selected machine learning algorithm was trained. Zhu, Figs 1, 3. para 53, 52, 50, 26-27,14, 19.)
training one or more large language models on at least the work protocols, extracted text, and metadata;
(i.e., The training input, stored in the database 110, can include domain-specific libraries containing documents, images, presentations, and product specification data. The database 110 can contain files whose contents can contain words and images that correspond to a particular industry or domain, such as the oil and gas domain. The database 110 can include documents, metadata, files, or the like that form an energy exploration lexicon that can used to form an industrial language model. In some embodiments, the database 110 can contain data associated with internal business communications conducted between members of an organization. In this way, the industrial language model can be trained to generate domain-specific language or “lingo” that may be commonly used by experts in the domain or industry (training one or more large language models on at least the work protocols, extracted text, and metadata (training industrial language model (large language model) using extracted processed database 110 training data technical documents, files, work protocols, words text and metadata to form an industrial language model LLM), see also Figs 1, 3, para 14, 26-27, 52-53)). Zhu, Fig 3, para 50, 14, 19, 26-27, 52-53.
[0053] The model training system 255 includes a feature selector 260, a model training 265, and a plurality of training models 270. The model training system 255 outputs trained prediction models 275. The feature selector 260 receives the training data 120 from the client 105 and/or the database 110 via processor 245. The feature selector 260 can process the training data 120 into subsets of data so that downstream data processing and model training can optimize the resulting training models 270 (training one or more large language models on at least the work protocols, extracted text, and metadata (training one or more large language model (industrial language model (LLM), para 50) on at least used, extracted, processed training data technical documents, files, work protocols, words text and metadata), see also Figs 1, 3, para 14, 50, 26-27, 52)). as desired by the design of the machine learning process. Zhu, Figs 1, 3, para 53, 52, 50, 14, 19, 26-27.)
commissioning an artificially intelligent assistant for the one or more large language models, the artificially intelligent assistant configured to at least retrieve text and metadata in response to receiving a prompt;
(i.e., As discussed above, Zhu teaches the text and metadata used to train the industrial language model (LLM) and generate natural language recommendations. Zhu, Figs 1, 3, para 50, 14, 19, 52-53, 26-27. Zhu teaches that [0021] Users may receive the domain-specific, natural language outputs in relation to providing natural language inputs to applications that can include auto-suggestion, auto-correction, translation, and interactive agents, such as chatbots (commissioning an artificially intelligent assistant (chatbot, see also Fig 4 #405) for the one or more large language models (for the domain specific trained industrial language model, see also Figs 1, 3-4, para 50, 52-53, 26-27,14, 19), the artificially intelligent assistant (chatbot) configured to at least retrieve text and metadata in response to receiving a prompt (retrieve domain specific text, metadata natural language outputs in relation and response to natural language inputs, see also Figs 1,3, Fig 4 #415 - #420, para 50, 52-53, 26-27,14, 19). The domain-specific natural language system described herein may also be used to provide images as the inputs, in addition to or separately from the natural language inputs. … The natural language outputs can enhance communications between users and ensure that specific formats of domain-specific information can be disseminated within an organization. In addition, the natural language outputs can reduce errors or inaccuracies by providing the outputs as results that include domain-specific language which can be used to aid document and/or text generation and automated report generation. By automating tedious and sometimes subjective text generation tasks, the improved recommendation system can reduce human biases in decision making processes and provide domain-specific data in a more timely manner so that production and operation tasks can be completed more efficiently in an oil and gas production or exploration environment. Zhu, Figs 1, 3-5, para 21, 62-66, 14, 19, 26-27, 50, 52-53.
[0062] FIG. 4 is a flow diagram illustrating an exemplary embodiment of a method 400 for generating natural language outputs including domain-specific recommendations based on natural language inputs using the client/server of FIGS. 1, 2A, 2B and the trained prediction model 275 generated in a machine learning process using the training system 300, as shown and described in relation to FIG. 3 (Figure 4 illustrates commissioning an artificially intelligent assistant (natural language inputs #405, chatbot, para 21) for the one or more large language models (transmitting to language model, language model determining, #410-415), the artificially intelligent assistant configured to at least retrieve text and metadata in response to receiving a prompt (natural language outputs #415-420, chatbot, para 21)). In certain aspects, embodiments of the method 400 can include greater or fewer operations than illustrated in FIG. 4 and the operations can be performed in a different order than illustrated in FIG. 4. Zhu, para 62, 63-66, 50, 52-53, 26-27, 14, 19.)
at the artificially intelligent assistant, generating a contextual response to a prompt related to a first work protocol;
(i.e., [0023] FIG. 1 is a block diagram illustrating an example architecture 100 for generating natural language recommendations based on an industrial language model (at the artificially intelligent assistant (chatbot, para 21), generating a contextual response (natural language recommendations based on industrial language model LLM) to a prompt related to a first work protocol,(to a first work protocol related prompt see also Figs 3-4 and para 19, 26-27, 50, 52,-53, 62.)). The architecture 100 includes clients 105, database 110, and server 115, which can be communicatively coupled over a network. Zhu, Fig 1, 3-5, para 23, 24, 21, 19, 62. [0021] Users may receive the domain-specific, natural language outputs in relation to providing natural language inputs to applications that can include auto-suggestion, auto-correction, translation, and interactive agents, such as chatbots (at the artificially intelligent assistant (chatbot), generating a contextual response to a prompt related to a first work protocol, see also Figs 1, 3, 4, para 19, 23, 26-27, 50, 52-53, 62). … In these ways, a large variety of inputs can be received and used to determine domain-specific, natural language outputs. The natural language outputs can enhance communications between users and ensure that specific formats of domain-specific information can be disseminated within an organization. In addition, the natural language outputs can reduce errors or inaccuracies by providing the outputs as results that include domain-specific language which can be used to aid document and/or text generation and automated report generation. By automating tedious and sometimes subjective text generation tasks, the improved recommendation system can reduce human biases in decision making processes and provide domain-specific data in a more timely manner so that production and operation tasks can be completed more efficiently in an oil and gas production or exploration environment. Zhu, Figs 1, 3-4, para 21, 23, 19, 26-27, 50, 52-53, 62.
[0026] As further shown in FIG. 1, natural language inputs can be transmitted from the clients 105 and/or from the database 110 to the prediction server 115. In some embodiments, the natural language inputs includes training data 120 that is transmitted to the prediction server 115 for use in a machine learning process. The training data 120 is used to train a machine learning algorithm in a machine learning process in order to generate a training model capable of predicting recommendations that include domain-specific recommendations based on a wide variety of received natural language inputs (at the artificially intelligent assistant (chatbot, para 21), generating a contextual response (natural language recommendations based on industrial language model LLM, see also para 23, providing contextual domain-specific recommendations 50, 52-53,14, 19) to a prompt related to a first work protocol (natural language inputs prompt related to first work protocol, see also Fig. 1, para 23, 21, digital work environment architecture)). In some embodiments, the natural language inputs include prediction data 125 that is transmitted to a prediction server 115 as inputs to the generated model that was trained in the machine learning process using the training data 120. The natural language inputs can include inputs that may be provided to a domain-specific application, such as a report generator or search interface. For example, the natural language inputs can include textual inputs to a variety of fields displayed in an interface of the domain-specific application. A user may enter words, numbers, sentences, or even whole documents as the natural language inputs. The natural language inputs provided as prediction data 125 may include language that has a weak or poor contextual relevance to the domain or industry in which the application and the improved recommendation system may be associated. A user may enter “gammar” to a field of an application. The improved recommendation system would be trained to identify this input as related in a contextual and domain-specific manner to “gamma” which may be a data processing component called a “gamma board” that is used within the particular industry or domain for which the predictive model has been trained. As a result, the improved recommendation system may perform a spelling correction to “gamma” and/or provide the user with a suggestion of “gamma board”. Without the improved recommendation system, the input of “gammar” may be corrected to “grammar”, which provides no contextual or domain-specific relevance to a user of the application. Zhu, Fig 1, para 26, 27.)
responsive to a change in content of one or more of the technical documents, re-training the one or more large language models; and at the artificially intelligent assistant, generating an updated contextual response to the prompt related to the first work protocol.
As discussed above with respect, Zhu in view of Lin teaches technical documents content, training large language model, the artificial intelligent assistant, and generating contextual response to prompt related to the first protocol. Zhu does not specifically disclose responsive to a change in content of one or more of the technical documents, re-training the one or more large language models; and at the artificially intelligent assistant, generating an updated contextual response to the prompt.
However, Lin teaches that, [0082] In some aspects, there may be two core types of content that can be used to train an LLM (one or more of the technical documents): documents and know-how. Documents, in some aspects, may be anything from product specifications to customer support. The documents may provide the LLM with a wealth of information about the domain in which it will be used. Know-how, in some aspects, may be the tacit knowledge that experts in a particular domain possess. [0102] Once a pretrained LLM has been fine-tuned and corrected (e.g., to produce a private LLM question generator 1421 and/or a private LLM answer generator 1432), in some aspects, it may be deployed in a model serving environment (e.g., a model server 1440). This allows the model to be accessed by users through a variety of channels, such as a web application, a mobile app, or a chatbot (at artificially intelligent assistant (access LLM at artificially intelligent assistant chatbot)). [0134] In some aspects, the method (or apparatus) may receive at least one of feedback regarding the response or additional content and context information associated with the particular context (responsive to a change in content of one or more of the technical documents (of one or more technical documents content, see also para 82)). In some aspects, based on receiving one or more of the feedback and/or additional context and/or context, the method may return to 2020 where it may update the context-specific LLM corresponding to the particular context based on at least one of the feedback or the additional content and context information (re-training (update) the one or more large language models), identifying an additional triggering event associated with the particular context at 2030, and generating, after updating the context specific LLM corresponding to the particular context, an additional response to the additional triggering event using the context-specific LLM corresponding to the particular context at 2040 (at artificially intelligent assistant (at artificially intelligent assistant (at chatbot access LLM, see para 102), generating an updated contextual response to the prompt)). For example, referring to FIGs. 1, 8, 9, 14, 18, and 19, the apparatus and/or method may identify and/or generate an additional prompt such as agent prompt 121/921, one or more user prompts 122/922, question prompt 1820 that may be associated with a particular context or, referring to FIGs. 1, 8, 9, and 19, the apparatus and/or method may receive feedback associated with the reinforcement learner 180/880/980/1980. Lin, para 134, 102, 82, 83, 85, 131.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on an large language model of Zhu using the data tagged with metadata and responsive to a change in content of one or more of the technical documents, re-training the one or more large language models and at the artificially intelligent assistant, generating an updated contextual response to the prompt of Lin, with a reasonable expectation of success, in order to provide to address context-specific queries (e.g. to provide answers or recommendations that are useful in a specific context and to provide support for tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation. Lin, para 2, 76, 134. This would have provided the advantages of providing updated information associated with data used for natural language generation.
Claims 7 and 9-11 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Lin as applied to claim 1 above, and further in view of Mehrotra et al. (Pub. No. US 2025/0148477 A1, filed December 6, 2023) hereinafter Mehrotra.
Regarding Claim 7, which depends from claim 1 and recites:
wherein the prompt comprises a contextual prompt from the artificially intelligent assistant.
Zhu in view of Lin teaches the method of claim 1 from which claim 7 depends, including the prompt entered via the digital work environment. Zhu in view Lin does not specifically disclose prompt comprises a contextual prompt from the artificially intelligent assistant.
However, Mehrotra teaches in the field related to industrial automation systems, and, for example, to resolution of industrial alarm conditions or performance issues. Mehrotra, para 2. Mehrotra, which is analogous to the claimed invention because Mehrotra is directed to industrial technical support, interactive assistant and generative artificial intelligence, and a language-based generative model such as a large language model (LLM), teaches that, [0054] Once the context retrieval component 206 has enhanced the prompt 306 with relevant contextual data 316 (prompt comprises a contextual prompt from the artificially intelligent assistant) and chat history data 308, the generative AI component 208 analyzes the combined prompt 306, contextual data 316, and chat history data 308, and generates a response 302 to the prompt based on a result of this analysis. Mehrotra, Figs 2-5, para 54, 44, 47, 51, 53-55.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on an large language model of Zhu using the data tagged with metadata of Lin and the prompt comprises a contextual prompt from the artificially intelligent assistant of Mehrotra, with a reasonable expectation of success, in order to provide to address context-specific queries (e.g. to provide answers or recommendations that are useful in a specific context and to provide support for tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation and in order to save personnel time and effort finding resolutions to the unfamiliar conditions or problems. Lin, para 2, 76. Mehrotra, para 3. This would have provided the advantages of providing information associated with data used for natural language generation and the advantages of providing users with contextually relevant support information.
Regarding Claim 9, which depends from claim 7 and recites:
wherein the artificially intelligent assistant is configured to dynamically reallocate resources based on contextual prompts from the artificially intelligent assistant.
Zhu in view of Lin and Mehrotra teaches the method of claim 7 from which claim 9 depends, including the artificially intelligent assistant and contextual prompts from the artificially intelligent assistant. Zhu teaches that, [0016] In the oil and gas or energy exploration domain, a lexicon of domain-specific language may be used for communications between users or for communications with a computing device. … A user may provide words or phrases intended to relate to the domain-specific operating equipment but which may lack the necessary domain-specific language for the search engine to process the inputs (artificially intelligent assistant is configured to dynamically reallocate resources (assistant is configured to reallocate correct resources for domain specific industry, see also para 17)). … In other situations, incorrect or less-specific industry information can produce catastrophic results, for example if an incorrect part number or specification detail is communicated or otherwise implemented in a mission critical workflow. [0017] An industrial language model can be configured to interoperate with domain-specific software applications such that upon receiving natural language inputs to the applications, the inputs can be transmitted to the industrial language model and processed to determine natural language outputs that include words, sentences, and phrases found in a domain-specific lexicon, such as a lexicon that is associated with the oil and gas or energy exploration industry (artificially intelligent assistant is configured to dynamically reallocate resources (assistant is configured to reallocate correct resources for domain specific industry). The industrial language model may be further configured to, in response to receiving natural language inputs, embed new or missing words, classify categories or topics within the received inputs generate text to be included in the outputs, and rank the natural language outputs based on user preferences prior to providing the outputs. Zhu, para 16-17.
Regarding Claim 10, which depends from claim 9 and recites:
wherein the artificially intelligent assistant is configured to dynamically resequence the first work protocol based on the retrieved text and metadata.
Zhu in view of Lin and Mehrotra teaches the method of claim 9 from which claim 10 depends, including the artificially intelligent assistant. Zhu teaches that, [0016] In the oil and gas or energy exploration domain, a lexicon of domain-specific language may be used for communications between users or for communications with a computing device. … A user may provide words or phrases intended to relate to the domain-specific operating equipment but which may lack the necessary domain-specific language for the search engine to process the inputs (artificially intelligent assistant is configured to dynamically resequence the first work protocol based on the retrieved text and metadata (assistant is configured to dynamically resequence work protocol based on retrieved domain specific industry retrieved text and metadata language, see also para 17)). … In other situations, incorrect or less-specific industry information can produce catastrophic results, for example if an incorrect part number or specification detail is communicated or otherwise implemented in a mission critical workflow. [0017] An industrial language model can be configured to interoperate with domain-specific software applications such that upon receiving natural language inputs to the applications, the inputs can be transmitted to the industrial language model and processed to determine natural language outputs that include words, sentences, and phrases found in a domain-specific lexicon, such as a lexicon that is associated with the oil and gas or energy exploration industry (artificially intelligent assistant is configured to dynamically resequence the first work protocol based on the retrieved text and metadata (assistant is configured to dynamically resequence work protocol based on retrieved domain specific industry retrieved text and metadata language). The industrial language model may be further configured to, in response to receiving natural language inputs, embed new or missing words, classify categories or topics within the received inputs generate text to be included in the outputs, and rank the natural language outputs based on user preferences prior to providing the outputs. Zhu, para 16-17.
Regarding Claim 11, which depends from claim 10 and recites:
wherein the retrieved text and metadata comprise inventory data related to the first work protocol.
Zhu in view of Lin and Mehrotra teaches the method of claim 9 from which claim 11 depends, including the artificially intelligent assistant and retrieved text and metadata and first work protocol. Zhu teaches that, [0034] The system 200a also includes … The client 105 can be configured with one or more domain-specific software applications. The domain-specific applications can include web-based applications as well as applications that can be directly hosted or configured on the client 105. For example, the domain-specific software applications can include technical computing applications, modeling and simulation applications, search applications, asset management applications, as well as marketing, human resources, and inventory management applications (retrieved text and metadata comprise inventory data related to the first work protocol). In some embodiments, the domain-specific applications can include applications configured to tag or annotate images using domain-specific natural language. Zhu, para 34.
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Lin and Mehrotra as applied to claim 7 above, and further in view of Scheepens et al. (Pub. No. US 2025/0111199 A1, filed September 28, 2023) hereinafter Scheepens.
Regarding Claim 8, which depends from claim 7 and recites:
wherein the contextual prompt from the artificially intelligent assistant is related to one or more non-conformance reports.
Zhu in view of Lin and Mehrotra teaches the method of claim 7 from which claim 8 depends, including the contextual prompt from the artificially intelligent assistant. Zhu in view of Lin and Mehrotra does not specifically disclose related to one or more non-conformance reports.
However, Scheepens teaches in the field related to process mining, and more specifically, to a conformance assistant for process mining using LLMs (large language models). Scheepens, para 1. Scheepens, which is analogous to the claimed invention because Scheepens is directed to a conformance assistant and LLMs teaches that a conformance assistant is provided using a large language model for generating descriptions explaining the non-conformance (related to one or more conformance reports). Scheepens, Abstract, para 69, 29.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on an large language model of Zhu using the data tagged with metadata of Lin and the prompt comprises a contextual prompt from the artificially intelligent assistant of Mehrotra, with a reasonable expectation of success, in order to provide to address context-specific queries (e.g. to provide answers or recommendations that are useful in a specific context and to provide support for tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation and in order to save personnel time and effort finding resolutions to the unfamiliar conditions or problems and to provide an improved assistance related to determining reasons for non-conformance. Lin, para 2, 76. Mehrotra, para 3. Scheepens, para 2, 69, 29. This would have provided the advantages of providing information associated with data used for natural language generation and the advantages of providing users with improved and contextually relevant support information.
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Lin as applied to claim 1 above, and further in view of Lyle (Patent No. US 12,530,480 B2, filed October 30, 2023).
Regarding claim 12, which depends from claim 1 and recites:
wherein the stored technical documents include computer aided engineering and/or drafting files and associated metadata.
Zhu in view of Lin teaches the method of claim 1 from which claim 12 depends, including the stored technical documents and associated metadata. Zhu in view Lin does not specifically disclose computer aided engineering and/or drafting files.
However, Lyle teaches in the field related to large language models. Lyle, col 1:12-14. Lyle, which is analogous to the claimed invention because Lyle is directed to large language models, teaches that, In at least one embodiment, training pipelines 1504 may include AI-assisted annotation, as described in more detail herein with respect to at least FIG. 16B. In at least one embodiment, labeled data 1412 (e.g., traditional annotation) may be generated by any number of techniques. In at least one embodiment, labels or other annotations may be generated within a drawing program (e.g., an annotation program), a computer aided design (CAD) program ((training pipelines technical documents include computer aided engineering and/or drafting files and associated metadata), a labeling program, another type of program suitable for generating annotations or labels for ground truth, and/or may be hand drawn, in some examples. In at least one embodiment, ground truth data may be synthetically produced (e.g., generated from computer models or renderings), real produced (e.g., designed and produced from real-world data), machine-automated (e.g., using feature analysis and learning to extract features from data and then generate labels), human annotated (e.g., labeler, or annotation expert, defines location of labels), and/or a combination thereof. In at least one embodiment, for each instance of imaging data 1408 (or other data type used by machine learning models), there may be corresponding ground truth data generated by training system 1404. In at least one embodiment, AI-assisted annotation may be performed as part of deployment pipelines 1510; either in addition to, or in lieu of AI-assisted annotation included in training pipelines 1504. Lyle, col 40:60 -col 41:17.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on an large language model of Zhu using the data tagged with metadata of Lin and technical documents include computer aided engineering and/or drafting files and associated metadata of Lyle, with a reasonable expectation of success, in order to provide to address context-specific queries (e.g. to provide answers or recommendations that are useful in a specific context and to provide support for tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation and to improve an accuracy of future responses of the LLMs. Lin, para 2, 76. Lyle, col 1:38-41, col 1:24-46. This would have provided the advantages of providing information associated with data used for natural language generation and providing accurate LLM responses.
Claims 13 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Lin and Mehrotra as applied to claims 9 and 10 above, and further in view of Ross et al. (US Pub. No. 2018/0324229, published November 8, 2018) hereinafter Ross.
Regarding claim 13, which depends from claim 9 and recites:
wherein the digital work environment is an augmented reality environment, and wherein the prompt is a visual prompt of a physical work environment received from one or more cameras.
Zhu in view of Lin and Mehrotra teaches the method of claim 9 from which claim 13 depends, including the digital work environment and the prompt. Zhu in view Lin does not specifically disclose an augmented reality environment and a visual prompt of a physical work environment received from one or more cameras
However Ross teaches in the field related to relates to virtual training, collaboration or other virtual technologies. Ross, para 2. Ross, which is analogous to the claimed invention because Ross is directed to providing expert assistance to a user, teaches that, [0010] FIG. 1A and FIG. 1B depict aspects of a system on which different embodiments are implemented for providing expert assistance from a remote expert to a user operating an augmented reality device (an augmented reality environment). The system includes a virtual, augmented, and/or mixed reality platform 110 (e.g., including one or more servers) that is communicatively coupled to any number of virtual, augmented, and/or mixed reality user devices 120 such that data can be transferred between the platform 110 and each of the user devices 120 as required for implementing the functionality described in this disclosure (an augmented reality environment). Ross, Figs 1A-B, 2, para 10, 16, 23. [0016] FIG. 2 depicts a method for providing expert assistance from a remote expert to a user operating an augmented reality device. The method comprises: receiving, at a server, a remote assistance request from a first user device operated by a first user located at a first location (step 201); after receiving the remote assistance request, establishing a network connection between the first user device and a second user device operated by a second user located at a second location (step 203); receiving visual information captured by a camera of the first user device operated by the first user, wherein the visual information includes an image of a physical object in view of the first user (step 205) (a visual prompt of a physical work environment received from one or more cameras); transmitting the visual information to the second user device operated by the second user (step 207); receiving, from the second user device operated by the second user, assistance content generated by the second user using the second user device (step 209); and transmitting the assistance content to the first user device for presentation of the assistance content to the first user (step 211). Ross, Figs 1A-B, 2, para 16, 23, 10.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on an large language model of Zhu using the data tagged with metadata of Lin and the prompt comprises a contextual prompt from the artificially intelligent assistant of Mehrotra and an augmented reality environment and a visual prompt of a physical work environment received from one or more cameras of Ross, with a reasonable expectation of success, in order to provide to address context-specific queries (e.g. to provide answers or recommendations that are useful in a specific context and to provide support for tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation and in order to save personnel time and effort finding resolutions to the unfamiliar conditions or problems and in order to provide expert assistance from a remote expert to a user operating an augmented reality device. Lin, para 2, 76. Mehrotra, para 3. Ross, para 9. This would have provided the advantages of providing information associated with data used for natural language generation and the advantages of providing users with contextually relevant support information and using visual information for assistance context.
Regarding claim 14, which depends from claim 10 and recites:
wherein the contextual response is a visual response presented via the augmented reality environment.
Zhu in view of Lin and Mehrotra teaches the method of claim 10 from which claim 134 depends, including the contextual response. Zhu in view Lin does not specifically disclose a visual response presented via the augmented reality environment.
However Ross teaches in the field related to relates to virtual training, collaboration or other virtual technologies. Ross, para 2. Ross, which is analogous to the claimed invention because Ross is directed to providing expert assistance to a user, teaches that, [0010] FIG. 1A and FIG. 1B depict aspects of a system on which different embodiments are implemented for providing expert assistance from a remote expert to a user operating an augmented reality device (via the augmented reality environment). The system includes a virtual, augmented, and/or mixed reality platform 110 (e.g., including one or more servers) that is communicatively coupled to any number of virtual, augmented, and/or mixed reality user devices 120 such that data can be transferred between the platform 110 and each of the user devices 120 as required for implementing the functionality described in this disclosure (via the augmented reality environment). Ross, Figs 1A-B, 2, para 10, 16, 23. [0016] FIG. 2 depicts a method for providing expert assistance from a remote expert to a user operating an augmented reality device. The method comprises: receiving, at a server, a remote assistance request from a first user device operated by a first user located at a first location (step 201); after receiving the remote assistance request, establishing a network connection between the first user device and a second user device operated by a second user located at a second location (step 203); receiving visual information captured by a camera of the first user device operated by the first user, wherein the visual information includes an image of a physical object in view of the first user (step 205) (a visual prompt of a physical work environment received from one or more cameras); transmitting the visual information to the second user device operated by the second user (step 207); receiving, from the second user device operated by the second user, assistance content generated by the second user using the second user device (step 209); and transmitting the assistance content to the first user device for presentation of the assistance content to the first user (step 211) (a visual response presented via the augmented reality environment (via augmented reality environment, para 10, Figures 1A-B). Ross, Figs 1A-B, 2, para 16, 23, 10. [0023] In one embodiment of the method depicted in FIG. 2, the presented visual information includes the image of the physical object that is in view of the first user, the assistance content is generated for display at one or more positions relative to particular parts of the physical object, and the method further comprises: presenting the assistance content on a display of the first user device to appear at the one or more positions relative to the particular parts of the physical object (a visual response presented via the augmented reality environment (via augmented reality environment, para 10, Figures 1A-B). Ross, Figs 1A-B, 2, para 23, 16, 10.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to implement the generating natural language based on an large language model of Zhu using the data tagged with metadata of Lin and the prompt comprises a contextual prompt from the artificially intelligent assistant of Mehrotra and an augmented reality environment and a visual response presented via the augmented reality environment of Ross, with a reasonable expectation of success, in order to provide to address context-specific queries (e.g. to provide answers or recommendations that are useful in a specific context and to provide support for tasks including: questions and answers generation, entity extraction, text summarization, machine translation, and/or natural language generation and in order to save personnel time and effort finding resolutions to the unfamiliar conditions or problems and in order to provide expert assistance from a remote expert to a user operating an augmented reality device. Lin, para 2, 76. Mehrotra, para 3. Ross, para 9. This would have provided the advantages of providing information associated with data used for natural language generation and the advantages of providing users with contextually relevant support information and using visual information for assistance context.
Allowable Subject Matter
Claims 16-17 and 19-20 would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims and if the rejections as being indefinite are overcome and if the rejections as being directed to an abstract idea without significantly more are overcome.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US-20240330589-A1, US-20240370476-A1, US-20240406081-A1, US-20250150321-A1, US-20250173330-A1, US-12651128-B1, US-20250013633-A1, US-20210133234-A1, US-20250005058-A1, US-20250156644-A1
Thais Medeiroos et al., Vehicles, Analysis of Language-Model-Powered Chatbots for Query Resolution in PDF-Based Automotive Manuals, published 16 October 2023.
JIN HO KIM et al. KR 20240006248 A, Interactive Engineering Knowledge Search System And Method Using Chatbot Technology, filed 2022-07-06, English translation.
S, Kernan Freire, Harnessing Large Language Models for Factories,CUI '23: Proceedings of the 5th International Conference on Conversational User Interfaces Article No.: 44, Pages 1 - 6 https://doi.org/10.1145/3571884.3604313
Akan, Pachikura Durga Murali. "Interactive Design Automation Framework Using Large Language Model and Past Data."
Fangkai Yang, Empower Large Language Model to Perform Better on Industrial Domain-Specific Question Answering, December 2023.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BARBARA LEVEL whose telephone number is (303)297-4748. The examiner can normally be reached Monday through Friday 8:00 AM - 5:00 PM MT.
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, Mariela Reyes can be reached at (571) 270-1006. 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.
/BARBARA M LEVEL/ Examiner, Art Unit 2142
Prosecution Insights