DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This Final Office Action is in response to the application 19/198,387 filed on 07/23/2026.
Status of Claims:
Claims 1-20 are pending in this Office Action.
Response to Arguments
Rejection of Claims under 35 U.S.C. §112(a) & 35 U.S.C. §112(b)
After reviewing the Applicant’s arguments filed in the remarks filed 07/23/2026 (pg. 6 and 7) regarding to rejection of claims under 35 U.S.C. §112(a) and 35 U.S.C. §112(b), the Examiner respectfully submits that the rejections made in the previous office action under 35 U.S.C. §112(a) and 35 U.S.C. §112(b) are now withdrawn.
Rejection of Claims under 35 U.S.C. §103
After reviewing the Applicant’s arguments filed in the remarks filed 07/23/2026 (pg. 7-10) regarding to claims 1, 9, and 15, the Examiner respectfully submits that the arguments are not persuasive.
The applicant argues that “neither cited prior art reference appears to be related to the process of defining/discovering (i.e., retaining) new data features in an input dataset via the use of LLM prompting”. The Examiner respectfully disagrees and submits that Lev discloses “[0090]: “The exemplary process 300 starts, as shown in 302, with receiving a dataset having a plurality of original text items”…[0092]: “The exemplary process 300 continues, as shown in 304, with receiving a plurality of prompts, each prompt comprising instructions for a language model to rephrase a text item”… [0093]: “The prompts may be designed in order to instruct one or more generative AI models, also known as large language models, or conversational language models, to rephrase one or more textual contents, optionally with adjustment of model parameters”. The system of Lev is directed to receive input data and corresponding prompts to rephrase or create variations the input data using a large language model. Under BRI, a new data feature of a dataset can correspond to any piece of data that is related to or describes the dataset. The claim does not define or limit in what “new data features” can be. The generation of rephrased input based on a prompt can correspond to defining new data features for an input data based on a prompt. In addition, the claim is silent of “via the use of LLM prompting” or language that associates with large language model. Therefore, Lev at least teaches “a language model configured to generate instructions for defining new data features for an input dataset based on a prompt”.
The applicant also argues “Lev "rephrases" existing text items, which may create new data samples (i.e., analogous to new rows in the dataset), but not new data features (i.e., analogous to new columns in the dataset). In other words, nothing in the teachings of Lev describes the defining of a new feature in a dataset.”. However, the in dependent claims do not claim that the new feature data is limited to anything such as data analogous to new columns in the dataset. Lev’s rephrasing of text items can create rephrased text items and that can correspond to new data feature because the rephrased text item is associated with the original text item and can be represented as a feature of the original text item. The examiner struggles to determine where in the claim the new data features are limited to features analogous to new columns in the dataset and where in Lev that describes the rephrased data is limited to data analogous to new rows in the dataset.
The applicant further argues that “Liu iterates a model and holds the data constant (i.e., a small language model is trained to imitate an LLM, and the training dataset is never modified). By contrast, the Present Application iterates on the dataset and holds the models constant (i.e., each round of iteration adds retained data feature columns, and no model is fine-tuned)… The alleged "evaluation" performed in Liu involves the LLM adjudging how well the small model imitates it. By contrast, the evaluation of the Present Application is empirical, i.e., the generated code is executed, and then the downstream predictor's performance change on the augmented dataset is measured. In other words, there is no LLM judgment in the validation path of the Present Application.” However, the contrasts explained by the applicant is not reflected in the claim. The examiner struggles to determine where in the claim that have associations or describes “iterates on the dataset and holds the models constant (i.e., each round of iteration adds retained data feature columns, and no model is fine-tuned)” and “empirical, i.e., the generated code is executed, and then the downstream predictor's performance change on the augmented dataset is measured”.
The Examiner respectfully submits that Liu discloses “[0060]: “The system initializes the natural language generation system with the first natural language output generated by the large language model”…[0061]: “Subsequently, at operation 510, the system generates a second natural language output imitating the first natural language output of the large language model using the natural language generation system”…[0062]: “Then, at operation 512, the system evaluates the second natural language output generated by the natural language system using the large language model to generate a feedback input and an output score”. The system of Liu is directed to evaluating data that associates with an original data such as second natural language that associates with the first natural language and subsequently, generating feedback input and score relating to the evaluation in a LLM environment. This can correspond to evaluating and retaining new data features for an input dataset as claimed by the application. Therefore, Lev in view of Liu at least teaches “a validation module configured to execute the instructions generated by the language model to evaluate and retain one or more new data features for the input dataset”.
Liu also discloses “[0032]: “The feedback input can enable the large language model to fine tune the natural language generation system through many iterations such that the second natural language output gradually aligns with the tendencies of the large language model”… [0043]: “Based on the imitation learning protocol 306, the small language model 318 generates a second natural language output 320 that mimics the first natural language output... The second natural language output 320 can then be returned to the large language model 302 for evaluation against various scoring criteria 322”. The system of Liu generates data features such as second natural language outputs and uses those outputs as input to the model for further evaluations. Thus, the system uses the revised dataset such as the second natural language outputs to use as the new input to the LLM model. Therefore, Lev in view of Liu at least teaches “iterative feedback loop configured to revise the input dataset with the retained new data features and recursively provide the revised dataset as a new input dataset to the language model”.
Lastly, the applicant argues “neither cited prior art reference executes model-generated instructions as code”. However, the examiner struggles to determine which limitation in the claims that describe or has association with the limitation as mentioned by the applicant.
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier. Such claim limitations are: “a validation module configured to execute” in claim 1 and “validation module is further configured to retain” in claim 7.
Because these claim limitations are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Lev (US PGPUB 20250021766) “Lev” in view of Liu et al. (US PGPUB 20240346254) “Liu”.
Regarding claim 1, Lev teaches a system for automated feature engineering, comprising: a language model configured to generate instructions for defining new data features for an input dataset based on a prompt ([0090]: “The exemplary process 300 starts, as shown in 302, with receiving a dataset having a plurality of original text items”…[0092]: “The exemplary process 300 continues, as shown in 304, with receiving a plurality of prompts, each prompt comprising instructions for a language model to rephrase a text item”… Examiner’s note: The language model receives dataset and prompts to rephrase/defining new data features of the dataset based on the prompts).
Lev does not explicitly teach a validation module configured to execute the instructions generated by the language model to evaluate and retain one or more new data features for the input dataset; an iterative feedback loop configured to revise the input dataset with the retained new data features and recursively provide the revised dataset as a new input dataset to the language model, wherein the new data feature definition process is repeated iteratively until a specified performance improvement threshold is no longer met.
Liu teaches a validation module configured to execute the instructions generated by the language model to evaluate and retain one or more new data features for the input dataset ([0060]” The system initializes the natural language generation system with the first natural language output generated by the large language model”…[0061]: “Subsequently, at operation 510, the system generates a second natural language output imitating the first natural language output of the large language model using the natural language generation system”…[0062]: “Then, at operation 512, the system evaluates the second natural language output generated by the natural language system using the large language model to generate a feedback input and an output score”… Examiner’s note: The system uses or retains new data features such as the natural language outputs of the system and evaluates the outputs such as second natural language output to generate feedback and score.); and an iterative feedback loop configured to revise the input dataset with the retained new data features and recursively provide the revised dataset as a new input dataset to the language model ([0032]: “The feedback input can enable the large language model to fine tune the natural language generation system through many iterations such that the second natural language output gradually aligns with the tendencies of the large language model”… [0043]: “Based on the imitation learning protocol 306, the small language model 318 generates a second natural language output 320 that mimics the first natural language output... The second natural language output 320 can then be returned to the large language model 302 for evaluation against various scoring criteria 322”... Examiner’s note: The system replaces the first natural language output with the second natural langue output and processes the second natural language output through many iterations. The second natural language output is fed back into the system for many iterations thus can correspond to recursively provide the revised dataset as a new input dataset to the language model), wherein the new data feature definition process is repeated iteratively until a specified performance improvement threshold is no longer met ([0044]: “Over many iterations of modifying the second natural language output 320 and feedback inputs 324, the small language model 318 may reach an output score 326 that satisfies a threshold output score. In response, the system 300 can determine that the small language model 318 has achieved a satisfactory level of performance and can begin receiving new inputs”). It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the Liu teachings in the Lev system. Skilled artisan would have been motivated to incorporate using the output data over many iterations to reach a satisfaction threshold for a system taught by Liu in the Lev system to create a system that would generate desired results, thus can increase the accuracy of the future results. This close relation between both of the references highly suggests an expectation of success.
Regarding claim 2, Lev in view of Liu teaches all of the limitations of claim 1. Lev further teaches wherein the prompt is a context-aware prompt ([0092]: “The exemplary process 300 continues, as shown in 304, with receiving a plurality of prompts, each prompt comprising instructions for a language model to rephrase a text item… [0093]: “The prompts may be designed in order to instruct one or more generative AI models, also known as large language models, or conversational language models, to rephrase one or more textual contents, optionally with adjustment of model parameters”… Examiner’s note: A prompt can comprise instructions for a language model to rephrase a text item, thus can correspond to a context-aware prompt.).
Regarding claim 3, Lev in view of Liu teaches all of the limitations of claim 1. Lev further teaches wherein the prompt encapsulates a natural language description comprising one or more of: input dataset characteristics, a prediction objective, or domain-specific knowledge related to the input dataset ([0092]: “The exemplary process continues, with receiving a plurality of prompts, each prompt comprising instructions for a language model to rephrase a text item”… [0093]: “The prompts may be designed in order to instruct one or more generative AI models, also known as large language models, or conversational language models, to rephrase one or more textual contents, optionally with adjustment of model parameters”… Examiner’s note: A prompt can contain instructions for a language model to rephrase one or more textual contents thus can associate with input dataset characteristics, a prediction objective, or domain-specific knowledge related to the input dataset).
Regarding claim 4, Lev in view of Liu teaches all of the limitations of claim 1. Lev further teaches wherein the language model is based on any pre-trained model capable of processing and generating natural language instructions ([0085]: “The language model may be designed to analyze a natural language text and generate responses such as those expected in human conversation”.).
Regarding claim 5, Lev in view of Liu teaches all of the limitations of claim 1. Lev does not explicitly teach wherein the evaluation of the one or more new data features is based, at least in part, on an effect of the one or more new data features on a performance of a data processing model.
Liu teaches evaluation of the one or more new data features is based, at least in part, on an effect of the one or more new data features on a performance of a data processing model ([0044]: “The reward model can configure the small language model to seek actions (e.g., the second natural language output) that maximize the output score. Over many iterations of modifying the second natural language output and feedback inputs, the small language model may reach an output score that satisfies a threshold output score. In response, the system can determine that the small language model has achieved a satisfactory level of performance and can begin receiving new inputs”... Examiner’s note: The system evaluates the new data features such as the second natural language output for score that satisfies a threshold output score, thus can correspond to effect of the data features on a performance of a data processing model). Please refer to claim 1 for the motivational statement.
Regarding claim 6, Lev in view of Liu teaches all of the limitations of claim 5. Lev further teaches wherein the data processing model comprises one or more of: a statistical model; a machine learning model; or a deep learning network ([0085]: “The conversational language model may be an artificial intelligence module designed to generate text based on a given prompt or input. The model may be an incorporated in house or third party machine learning model”).
Regarding claim 7, Lev in view of Liu teaches all of the limitations of claim 1. Lev does not explicitly teach wherein the validation module is further configured to retain new data features based, at least in part, on a performance improvement criterion being met.
Liu teaches validation module is further configured to retain new data features based, at least in part, on a performance improvement criterion being met ([0044]: “The reward model can configure the small language model to seek actions (e.g., the second natural language output) that maximize the output score . Over many iterations of modifying the second natural language output and feedback inputs, the small language model may reach an output score that satisfies a threshold output score”… Examiner’s note: The system retains a new data feature such as modified natural language output, based on a criterion being met such as an output score that satisfies a threshold output score). Please refer to claim 1 for the motivational statement.
Regarding claim 8, Lev in view of Liu teaches all of the limitations of claim 7. Lev does not explicitly teach wherein the performance improvement criterion comprises one or more of: a statistical metric; or a machine learning performance metric.
Liu teaches performance improvement criterion comprises one or more of: a statistical metric; or a machine learning performance metric ([0044]: “Over many iterations of modifying the second natural language output and feedback inputs , the small language model may reach an output score that satisfies a threshold output score... Examiner’s note: The criterion such as score that satisfies a threshold output score can correspond to a statistical metric”). Please refer to claim 1 for the motivational statement.
Regarding claim 9, Lev teaches a method for automated feature engineering, comprising: providing a language model with an input dataset and a prompt ([0090]: “The exemplary process 300 starts, as shown in 302, with receiving a dataset having a plurality of original text items”…[0092]: “The exemplary process 300 continues, as shown in 304, with receiving a plurality of prompts, each prompt comprising instructions for a language model to rephrase a text item”.);employing the language model to generate instructions for defining new data features for the input dataset and based, at least in part, on the prompt ([0090]: “The exemplary process 300 starts, as shown in 302, with receiving a dataset having a plurality of original text items…[0092] The exemplary process 300 continues, as shown in 304, with receiving a plurality of prompts, each prompt comprising instructions for a language model to rephrase a text item… Examiner’s note: The language model receives dataset and prompts to rephrase/defining new data features of the dataset based on the prompts).
Lev does not explicitly teach executing the instructions generated by the language model to evaluate and retain one or more new data features for the input dataset; revising the input dataset with the retained new data features; and
recursively providing the revised dataset as a new input dataset to the language model, wherein the new data feature definition process is repeated iteratively until a specified performance improvement threshold is no longer met.
Liu teaches executing the instructions generated by the language model to evaluate and retain one or more new data features for the input dataset ([0060]: “The system initializes the natural language generation system with the first natural language output generated by the large language model…[0061] Subsequently, at operation 510, the system generates a second natural language output imitating the first natural language output of the large language model using the natural language generation system…[0062] Then, at operation 512, the system evaluates the second natural language output generated by the natural language system using the large language model to generate a feedback input and an output score… Examiner’s note: The system retains new data features such as the natural language outputs of the system and evaluates the outputs such as second natural language output to generate feedback and score); revising the input dataset with the retained new data features; and recursively providing the revised dataset as a new input dataset to the language model ([0032]: “The feedback input can enable the large language model to fine tune the natural language generation system through many iterations such that the second natural language output gradually aligns with the tendencies of the large language model”… [0043]: “Based on the imitation learning protocol 306, the small language model 318 generates a second natural language output 320 that mimics the first natural language output... The second natural language output 320 can then be returned to the large language model 302 for evaluation against various scoring criteria 322”... Examiner’s note: The system replaces the first natural language output with the second natural langue output and processes the second natural language output through many iterations. The second natural language output is fed back into the system for many iterations thus can correspond to recursively provide the revised dataset as a new input dataset to the language model), wherein the new data feature definition process is repeated iteratively until a specified performance improvement threshold is no longer met ([0044]: Over many iterations of modifying the second natural language output 320 and feedback inputs 324, the small language model 318 may reach an output score 326 that satisfies a threshold output score. In response, the system 300 can determine that the small language model 318 has achieved a satisfactory level of performance and can begin receiving new inputs.). Please refer to claim 1 for the motivational statement.
Regarding claim 10, note the rejections of claim 3. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 11, note the rejections of claim 4. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 12, note the rejections of claim 5. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 13, note the rejections of claim 6. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 14, note the rejections of claim 7. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 15, note the rejections of claim 9. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 16, note the rejections of claim 3. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 17, note the rejections of claim 4. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 18, note the rejections of claim 5. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 19, note the rejections of claim 6. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Regarding claim 20, note the rejections of claim 7. The instant claims recite substantially same limitations as the above-rejected claims and are therefore rejected under the same prior-art teachings.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to CAO DANG VUONG whose telephone number is (571)272-1812. The examiner can normally be reached on M-F 7:30-5 EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kavita Stanley can be reached at (571) 272-8352. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see https://ppair-my.uspto.gov/pair/PrivatePair. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/C.D.V./Examiner, Art Unit 2153 09/30/2026
/KAVITA STANLEY/Supervisory Patent Examiner, Art Unit 2153
