DETAILED ACTION
This Action is in response to Claims filed 08/29/2024.
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 .
Drawings
Receipt of Drawings filed 08/29/2024 is acknowledged. These Drawings are acceptable.
Status of the Claims
Claims 1-20 are currently pending.
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 rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more; and because the claims as a whole, considering all claim elements both individually and in combination, do not amount to significantly more than the abstract idea, see Alice Corporation Pty. Ltd. v. CLS Bank International, et al, 573 U.S. (2014). In determining whether the claims are subject matter eligible, the Examiner applies the 2019 USPTO Patent Eligibility Guidelines. (2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50, Jan. 7, 2019.)
Step 1: All Claims
Claims 1-9 recite a method, which falls under the statutory category of a process. Claims 10-18 recite a non-transitory computer readable storage medium storing instructions, which falls under the statutory category of a manufacture. Claims 19-20 recite a computer system, which falls under the statutory category of a machine.
Step 2A – Prong 1:
Claim 1 recites an abstract idea, law of nature, or natural phenomenon. The limitations “evaluating, by the computer processor, each of the set of text artifacts to identify performance of each of the set of text artifacts;”, “iteratively improving the set of text artifacts…”, “updating, by the computer processor, the prompt based on the performance of each of the set of text artifacts to obtain a new prompt;”, and “evaluating, by the computer processor, the new set of text artifacts to identify performance of each of the new set of text artifacts;” under the broadest reasonable interpretation, cover a mental process including an observation, evaluation, judgment or opinion that could be performed in the human mind or with the aid of pencil and paper. These limitations therefore fall within the mental process group.
The high level of generality recited in the above limitations do not preclude a human mind with or without the aid of pen and paper from evaluating text artifacts, improve the artifacts, or updating a text prompt.
Step 2A – Prong 2:
The additional elements of claim 1 do not integrate the abstract idea into a judicial exception. The claim recites the additional elements “A method”, “a computer processor”, and “text artifacts” which are recognized as a generic computer components recited at a high level of generality. Although they have and execute instructions to perform the abstract idea itself, this also does not serve to integrate the abstract idea into a practical application as it merely amounts to instructions to "apply it." (See MPEP 2106.04(d)(2) indicating mere instructions to apply an abstract idea does not amount to integrating the abstract idea into a practical application).
The additional elements “a machine learned language model” and “a search engine” are recognized as non-generic computing components, but are recited at a level of generality so as to merely link the claims to a particular technological area or field of use (See MPEP 2106.05(h)).
The additional elements “receiving, by a computer processor, a prompt for a machine learned language model, the prompt configured to request the machine learned language model to generate text artifacts;” and “and transmitting text artifacts from the set of text artifacts to a search engine for storing, wherein the search engine provides the text artifacts to users along with search results returned in response to search queries processed by the search engine.” are found to be mere data gathering or data transmittal extra-solution activity steps (See MPEP 2106.05(g)).
The additional element “executing, by the computer processor, the machine learned language model using the prompt to generate a set of text artifacts;” is found to be mere data gathering or data transmittal extra-solution activity steps (generic model output is generically generated) (See MPEP 2106.05(g)).
The additional element “executing, using the new prompt, the machine learned language model to generate a new set of text artifacts;” amount to instructions to apply the abstract idea mental process steps of iteratively improving and updating the prompts (See MPEP 2106.05(f)).
Step 2B:
The only limitation on the performance of the described method is limitations reciting “A method”, “a computer processor”, and “text artifacts”. These elements are insufficient to transform a judicial exception to a patentable invention because the recited element is considered insignificant extra-solution activity (generic computer system, processing resources, links the judicial exception to a particular, respective, technological environment). The claim thus recites computing components only at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components; mere instructions to apply an exception using a generic computer component cannot provide an inventive concept (see MPEP 2106.05(f)).
The additional elements “a machine learned language model” and “a search engine” are recognized as non-generic computing components, but are recited at a level of generality so as to merely link the claims to a particular technological area or field of use (See MPEP 2106.05(h)).
The additional elements “receiving, by a computer processor, a prompt for a machine learned language model, the prompt configured to request the machine learned language model to generate text artifacts;” and “and transmitting text artifacts from the set of text artifacts to a search engine for storing, wherein the search engine provides the text artifacts to users along with search results returned in response to search queries processed by the search engine.” are found to be generally data transmittal and are acknowledged to be well-understood, routine, conventional activity (see, e.g., court recognized WURC examples in MPEP 2106.05(d)(I)(ii)).
The additional element “executing, by the computer processor, the machine learned language model using the prompt to generate a set of text artifacts;” is found to be generally data transmittal and are acknowledged to be well-understood, routine, conventional activity (see, e.g., court recognized WURC examples in MPEP 2106.05(d)(I)(ii)).
The additional element “executing, using the new prompt, the machine learned language model to generate a new set of text artifacts;” amount to instructions to apply the abstract idea mental process steps of iteratively improving and updating the prompts (See MPEP 2106.05(f)).
Taken alone or in ordered combination, these additional elements do not amount to significantly more than the above-identified abstract idea. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide conventional computer implementation.
For the reasons above, claim 1 is rejected as being directed to non-patentable subject matter under §101. This rejection applies equally to independent Claims 10 and 19.
Claim 10 recites similar limitations to Claim 1, save for “A non-transitory computer readable storage medium storing instructions that when executed by one or more computer processors cause the one or more computer processors to perform steps comprising…” (generic computer components); therefore, both claims are similarly rejected.
Claim 19 recites similar limitations to Claim 1, save for “A computer system comprising: one or more computer processors; and a non-transitory computer readable storage medium storing instructions that when executed by one or more computer processors cause the one or more computer processors to perform steps comprising…” (generic computer components); therefore, both claims are similarly rejected.
Dependent Claims:
Claim 2 (claims 11 and 20) recites abstract idea mental process step “generating a second prompt requesting the machine learned language model to identify performance of each of the new set of text artifacts;” and instructions to apply it “executing the machine learned language model using the second prompt to identify performance of each of the new set of text artifacts.” (See MPEP 2106.05(f)).
Claim 3 (claim 12) recites abstract idea mental process step “determining a performance metric by executing a machine learning based language model trained to receive a text artifact as input and output a score indicating a performance of the text artifact.”
Claim 4 (claim 13) recites abstract idea mental process step “comparing a performance metric of the text artifact with a threshold value.”
Claim 5 (claim 14) recites abstract idea mental process step “removing one or more text artifacts from the new set of text artifacts responsive to identifying that the performance of each of the one or more text artifacts is below a threshold value.”
Claim 6 (claim 15) recites abstract idea mental process step “adding one or more text artifacts to the new set of text artifacts responsive to identifying that the performance of each of the one or more text artifacts is at least a threshold value.”
Claim 7 (claim 16) recites abstract idea mental process step “replacing a subset of text artifacts from the new set of text artifacts with a set of new text artifacts generated by the machine learned language model, wherein the subset of text artifacts replaced are lowest performing artifacts of the new set of text artifacts and the set of new text artifacts are best performing text artifacts generated by the machine learned language model.”
Claim 8 (claim 17) recites refinements to the data operated on.
Claim 9 (claim 18) recites mere data gathering or data transmittal “providing text artifacts from the new set of text artifacts to a search engine for providing to users along with search results returned in response to search queries processed by the search engine.” (See MPEP 2106.05(g)).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claim(s) 1-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Duke et al. (US 2019/0392487 A1), hereinafter Duke and Zhou et al. (LARGE LANGUAGE MODELS ARE HUMAN-LEVEL PROMPT ENGINEERS, 2022), hereinafter Zhou.
In regards to Claim 1: The present invention claims: “A method comprising…a machine learned language model… the machine learned language model to generate text artifacts;” Duke teaches “For example, a Search Module operates to search for previous ads of that client; or in some embodiments, the client may provide an already-prepared historical set of their advertising work along with any related performance data or performance metrics; or such data may be fetched or downloaded or obtained or read from a repository or data silo or a cloud-based silo of the client itself and/or of an advertising network and/or of a third-party provider of online advertisement platform.” ([0016]) and “previous ads…may be obtained or collected automatically from search engine results (e.g., the first 75 results if an image search via an Internet search engine with a search query of "Coca Cola banner ads 2018"), or may be otherwise obtained.” ([0018]).
“executing, by the computer processor, the machine learned language model…to generate a set of text artifacts;” Duke teaches “An Ad Generator Module generates one or more proposed ads for that client ( or, for that product or service), by selecting one or more ad components from the database, and applying the rules, and optionally by incorporating digital content that injects a fresh proposal or promotion into the digital ad ( e.g., a new particular offering);” and “as well as using machine learning (ML) algorithms, Natural Language Generation (NLG), or deep learning algorithms or Artificial Intelligence (AI) algorithms to generate any needed textual element.” ([0016]).
“evaluating, by the computer processor, each of the set of text artifacts to identify performance of each of the set of text artifacts;” Duke teaches “Any performance data or performance metrics related to that ad is also stored, and/or is tagged or associated with the relevant Ad Components or Ad Elements of such ad; for example, indicating and reflecting in the database, not only that Previous Ad Number 435 had a click-through rate of 13 percent for end-users that are geo-located in the State of Florida; but also, that the Ad Components of that particular previous ad, which were Headline Number 4 7 and Product Image number 38 and Call-To-Action number 92, were each associated with that same metric when they appeared together in that same previous advertisement.” ([0023]).
“iteratively improving the set of text artifacts by performing a set of steps comprising…” Duke teaches “In accordance with the present invention, profiles are not merely templates. For example, a template sets out very specific (normally down to the pixel) variable and non-variable areas of an advertising creative execution. In contrast, with a profile, the elements are assigned general areas and will dynamically adapt to positions or locations or regions within the ad format, based on factors (e.g., dynamically changing factors) that may include image size; headline size; copy length or text length; ad rules or client rules or brand rules ( e.g., from brand guidelines), or the like. The system may dynamically bring the elements together and may dynamically determine the location or position of each component, rather than assigning each element to a predefined, pixel locked position, based on performance data that is manually and/or automatically fed back into the system, and which may be used by the system, in a continuous manner or in an iterative process, to further generate new combinations or new elements and new permutations of ad components and ad elements, thereby generating new creatives (new advertisements) and again feeding into the system their performance metrics data in order to be utilized iteratively in a next batch or next round or next iteration of ad generation, for the same product ( or service) or even for a different product ( or service) of the client.” ([0049]).
“executing…the machine learned language model to generate a new set of text artifacts;” Duke teaches “In the ad generation phase, a creative brief for generating a new advertisement is obtained (block 11 ); e.g., is provided by the user; or is generated based on one or more criteria or rules. The creative brief of the desired ad is transformed or converted to feature vectors or to feature list(s) or to feature data-set(s) that are suitable for utilization by machine learning algorithms.” ([0027]) and “The machine learning models that were created or generated in the training phase, are used (block 12) to generate the textual elements of the desired new advertisement (e.g. headline, sub-header, call to action, or the like), based on (or, taking into account at least) the feature vectors that were provided in block 8.” ([0028]).
“and evaluating, by the computer processor, the new set of text artifacts to identify performance of each of the new set of text artifacts;” See above where Duke performs its operations on previously generated advertisements and text elements of the advertisements. Duke also teaches “Other suitable insights may be deduced by the system and method of the present invention, based on ML analysis of the performance data of previous ads vis-a-vis the combinations of components of such previous ads.” ([0026]).
“and transmitting text artifacts from the set of text artifacts to a search engine for storing, wherein the search engine provides the text artifacts to users along with search results returned in response to search queries processed by the search engine.” Duke teaches “Duke teaches “An Ad Generator Module generates one or more proposed ads for that client ( or, for that product or service), by selecting one or more ad components from the database, and applying the rules, and optionally by incorporating digital content that injects a fresh proposal or promotion into the digital ad ( e.g., a new particular offering);” and “as well as using machine learning (ML) algorithms, Natural Language Generation (NLG), or deep learning algorithms or Artificial Intelligence (AI) algorithms to generate any needed textual element.” ([0016]) and “Any performance data or performance metrics related to that ad is also stored, and/or is tagged or associated with the relevant Ad Components or Ad Elements of such ad;” ([0023]).
Duke fails to explicitly teach the receipt, use, and updating of a prompt to the machine learning model as recited in:
“…receiving, by a computer processor, a prompt for [a machine learning model]…the prompt configured to request [the machine learning model to generate…]”
“…using the prompt…”
“…updating, by the computer processor, the prompt based on the performance of each of the set of text artifacts to obtain a new prompt;” and
“…using the new prompt…”
However, Zhou, in a similar field of endeavor, teaches “We consider a task specified by a dataset Dtrain = f{Q;A} of input/output demonstrations sampled from population X, and a prompted model M. The goal of natural language program synthesis is to find a single instruction _ such that, when M is prompted with the concatenation [p;Q] of instruction and a given input, M produces the corresponding output A. More formally, we frame this as an optimization problem, where we seek instruction _ that maximizes the expectation of some per-sample score f(p, Q, A) over possible (Q, A)… Note that in general, Q may be the empty string, such that we are optimizing _ as a prompt that directly produces outputs {A}. While this task has been widely attempted by humans, we have little knowledge of how compatible any particular instruction is with model M. Thus, we propose to treat this human-intractable question as a black-box optimization process guided by LLMs. Our algorithm, APE, uses LLMs in each of two key components, proposal and scoring. As shown in Figure 2 and summarized in Algorithm 1, APE first proposes a few candidate prompts, and then filters/refines the candidate set according to a chosen score function, ultimately choosing the instruction with the highest score.” (Page 3, Section 3, See also Algorithm 1). Section 3.1 details how the prompt to the machine learning model is formed, Section 3.2 details how scoring the success of the model is determined based on the prompt, and Section 3.3 details the process being an iterative one where low-scoring results are filtered while high-scoring ones are kept or replicated in subsequent iterations.
Zhou teaches “By conditioning on natural language instructions, large language models (LLMs) have displayed impressive capabilities as general-purpose computers. However, task performance depends significantly on the quality of the prompt used to steer the model, and most effective prompts have been handcrafted by humans. Inspired by classical program synthesis and the human approach to prompt engineering, we propose Automatic Prompt Engineer (APE) for automatic instruction generation and selection.” (Abstract). It would have been obvious to one of ordinary skill in the art at the time of the Applicant’s filing to the Automatic Prompt Engineering of Zhou in an automatic advertisement/text artifact system of Duke’s in order to efficiently prompt the system to generate the advertisement with natural language processing.
In regards to Claim 2: The present invention claims: “generating a second prompt requesting the machine learned language model to identify performance of each of the new set of text artifacts; and executing the machine learned language model using the second prompt to identify performance of each of the new set of text artifacts.” See the Rejection of Claim 1 where the combination of Duke performs its operations on previously generated advertisements and text elements of the advertisements and Zhou iteratively improves a prompt to a system to instruct a machine learning model. Duke also teaches “Other suitable insights may be deduced by the system and method of the present invention, based on ML analysis of the performance data of previous ads vis-a-vis the combinations of components of such previous ads.” ([0026]).
In regards to Claim 3: The present invention claims: “for each of the set of text artifacts, determining a performance metric by executing a machine learning based language model trained to receive a text artifact as input and output a score indicating a performance of the text artifact.” See above where Duke tracks the performance of the text/advertisement elements and where Zhou Section 3 and Algorithm 1 show a performance candidate results of a prompt are used.
In regards to Claim 4: The present invention claims: “for each of the set of text artifacts, comparing a performance metric of the text artifact with a threshold value.” Duke teaches “The AI unit generates a set of advertisement elements, such as logo, headline, a sub-headline, call-to-action, legal content, and an image; based on analysis of the input and detection that these particular advertisement elements correspond to previous performance results that are beyond a pre-defined threshold;” ([0005]).
In regards to Claim 5: The present invention claims: “removing one or more text artifacts from the new set of text artifacts responsive to identifying that the performance of each of the one or more text artifacts is below a threshold value.” Duke teaches “wherein said set of advertisement elements is generated based on analysis of said input and detection that said advertisement elements correspond to previous performance results that are beyond a pre-defined threshold;” ([0105]) and “In accordance with the present invention, profiles are not merely templates. For example, a template sets out very specific (normally down to the pixel) variable and non-variable areas of an advertising creative execution. In contrast, with a profile, the elements are assigned general areas and will dynamically adapt to positions or locations or regions within the ad format, based on factors (e.g., dynamically changing factors) that may include image size; headline size; copy length or text length; ad rules or client rules or brand rules ( e.g., from brand guidelines), or the like. The system may dynamically bring the elements together and may dynamically determine the location or position of each component, rather than assigning each element to a predefined, pixel locked position, based on performance data that is manually and/or automatically fed back into the system, and which may be used by the system, in a continuous manner or in an iterative process, to further generate new combinations or new elements and new permutations of ad components and ad elements, thereby generating new creatives (new advertisements) and again feeding into the system their performance metrics data in order to be utilized iteratively in a next batch or next round or next iteration of ad generation, for the same product ( or service) or even for a different product ( or service) of the client.” ([0049]).
In regards to Claim 6: The present invention claims: “adding one or more text artifacts to the new set of text artifacts responsive to identifying that the performance of each of the one or more text artifacts is at least a threshold value.” Duke teaches “The AI unit generates a set of advertisement elements, such as logo, headline, a subheadline, call-to-action, legal content, and an image; based on analysis of the input and detection that these particular advertisement elements correspond to previous performance results that are beyond a pre-defined threshold.” (Abstract).
In regards to Claim 7: The present invention claims: “replacing a subset of text artifacts from the new set of text artifacts with a set of new text artifacts generated by the machine learned language model, wherein the subset of text artifacts replaced are lowest performing artifacts of the new set of text artifacts and the set of new text artifacts are best performing text artifacts generated by the machine learned language model.” Duke teaches “In accordance with the present invention, profiles are not merely templates. For example, a template sets out very specific (normally down to the pixel) variable and non-variable areas of an advertising creative execution. In contrast, with a profile, the elements are assigned general areas and will dynamically adapt to positions or locations or regions within the ad format, based on factors (e.g., dynamically changing factors) that may include image size; headline size; copy length or text length; ad rules or client rules or brand rules ( e.g., from brand guidelines), or the like. The system may dynamically bring the elements together and may dynamically determine the location or position of each component, rather than assigning each element to a predefined, pixel locked position, based on performance data that is manually and/or automatically fed back into the system, and which may be used by the system, in a continuous manner or in an iterative process, to further generate new combinations or new elements and new permutations of ad components and ad elements, thereby generating new creatives (new advertisements) and again feeding into the system their performance metrics data in order to be utilized iteratively in a next batch or next round or next iteration of ad generation, for the same product ( or service) or even for a different product ( or service) of the client.” ([0049]).
In regards to Claim 8: The present invention claims: “wherein a size of the subset of text artifacts replaced is identified based on an aggregate measure of performance of the text artifacts generated by the machine learned language model.” See above where Duke teaches the measurement of the performance of text elements. Duke also teaches “Component Placement/Arrangement Unit 243 places or arranges the selected ad components within or on a predefined advertisement canvas or space; for example, by taking into account historic performance data of each ad component, and of particular combinations of two-or-more ad components.” ([0066]).
In regards to Claim 9: The present invention claims: “providing text artifacts from the new set of text artifacts to a search engine for providing to users along with search results returned in response to search queries processed by the search engine.” See above how Duke replaces/stores new advertisement elements and arrangements. Duke also teaches “may be obtained from Internet searches, or from the website(s) of the clients; may be obtained or collected automatically from search engine results (e.g., the first 75 results if an image search via an Internet search engine with a search query of "Coca Cola banner ads 2018"), or may be otherwise obtained.” ([0018]).
In regards to Claims 10-18: Claims 10- 18 recite similar limitations to Claim 1-9, save for “A non-transitory computer readable storage medium storing instructions that when executed by one or more computer processors cause the one or more computer processors to perform steps comprising…” of Claim 10; therefore, both claims are similarly rejected.
In regards to Claim 19-20: Claims 19-20 recites similar limitations to Claims 1-2, save for “A computer system comprising: one or more computer processors; and a non-transitory computer readable storage medium storing instructions that when executed by one or more computer processors cause the one or more computer processors to perform steps comprising…” of Claim 19; therefore, both claims are similarly rejected.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Hughes, J. Weston, Keng-hao Chang, and Ruofei Zhang. "Generating better search engine text advertisements with deep reinforcement learning." Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining. 2019.
Hamedi et al. US 2019/0034976 A1, Filed 2017
Aviyam et al. US 2021/0200943 A1, Filed 2020
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GRIFFIN T BEAN whose telephone number is (703)756-1473. The examiner can normally be reached M - F 7:30 - 4:30.
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, Li Zhen can be reached at (571) 272-3768. 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.
/GRIFFIN TANNER BEAN/ Examiner, Art Unit 2121
/Li B. Zhen/ Supervisory Patent Examiner, Art Unit 2121
Prosecution Insights