Detailed Action
Status of Claims
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This Action is in reply to the Amendment filed on 6/17/2026.
Claims 1, 3-11, 13-20 are currently pending and have been examined. Claims 2 & 12 have been cancelled. Claims 1, 3, 8, 11, 13, 17, 20 have been amended. The prior art rejection has been overcome by amendment.
Priority
Applicant’s claim of priority to provisional application 63513541is acknowledged. The disclosure does not provide support for the limitations of Claims 7-9, 17-18. Therefore, claims 7-9, 17-18 are afforded an effective filing date of 7/15/2024. Claims 11-6, 10-16, and 19-20 are afforded an effective filing date of 7/13/2023.
Claim Rejection - 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, 3-11, 13-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
First, it is determined whether the claims are directed to a statutory category of invention. In the instant case, claims 1, 3-10 are directed to a process, claims 11, 13-19 are directed to an article of manufacture, and claim 20 is directed to a machine. Therefore, claims 1, 3-11, 13-20 are directed to statutory subject matter under Step 1 as described in MPEP 2106 (Step 1: YES).
The claims are then analyzed to determine whether the claims are directed to a judicial exception. In determining whether the claims are directed to a judicial exception, the claims are analyzed to evaluate whether the claims recite a judicial exception (Prong One of Step 2A), as well as analyzed to evaluate whether the claims recite additional elements that integrate the judicial exception into a practical application of the judicial exception (Prong Two of Step 2A).
Claims 1, 11, and 20 recite at least the following limitations that are believed to recite an abstract idea:
accessing user interaction data for a user describing a plurality of interactions by the user with a set of items of a system, wherein the user interaction data describes, for each of the plurality of interactions, an interaction of the user with an item of the set of items;
transmitting the user interaction data to a model;
receiving, from the model, a first set of item vectors, the first set of item vectors comprising an item vector for each item of the set of items, wherein the item vectors of the first set of item vectors are generated by transmitting a prompt to the model, wherein the prompt comprises the user interaction data;
concatenating the first set of item vectors to generate a user vector array for the user;
applying a transformer process to the user vector array to generate a user vector describing the user, wherein the user vector is in a vector space;
accessing a second set of item vectors, the second set of item vectors comprising an item vector for each candidate item in a set of candidate items of the system, wherein each item vector of the second set of item vectors is in the vector space;
generating, for each candidate item, an interaction score for the candidate item by comparing the item vector for the candidate item to the user vector;
selecting a candidate item of the set of candidate items to present to the user based on the generated interaction scores; and
transmitting information describing the selected candidate item to the user for display to the user.
The above limitations recite the concept of personalized recommendations. These limitations, under their broadest reasonable interpretation, fall within the “Certain Methods of Organizing Human Activity” grouping of abstract ideas, enumerated in MPEP 2106, in that they recite commercial interactions, e.g. sales activities/behaviors, and managing personal behavior or relationships or interactions between people, e.g., following rules or instructions. Accordingly, under Prong One of Step 2A, claims 1-20 recite an abstract idea (Step 2A, Prong One: YES).
Prong Two of Step 2A is the next step in the eligibility analysis and looks at whether the abstract idea is integrated into a practical application. This requires an additional element or combination of additional elements in the claims to apply, rely on, or user the judicial exception in a manner that imposes a meaningful limit on the judicial exception, such that the claim is more than a drafting effort designed to monopolize the exception.
In this instance, the claims recite the additional elements of:
A computer system comprising a processor and a computer-readable medium
An online system
A generative language model
Embeddings
A transformer network
Latent space
A client device
A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform steps
A system comprising: a processor; and a non-transitory computer-readable medium storing instructions that, when executed by the processor, cause the processor to perform steps
However, these elements do not amount to an improvement in the functioning of a computer or any other technology or technical field; apply the judicial exception with, or by use of, a particular machine; or apply or use the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort to monopolize the exception.
In addition, the recitations are recited at a high level of generality and also do not amount to an improvement in the functioning of a computer or any other technology or technical field; apply the judicial exception with, or by use of, a particular machine; or apply or use the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort to monopolize the exception.
The dependent claims also fail to recite elements which amount to an improvement in the functioning of a computer or any other technology or technical field; apply the judicial exception with, or by use of, a particular machine; or apply or use the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort to monopolize the exception. For example, claims 4-8, 14-17, and are directed to the abstract idea itself and do not amount to an integration according to any one of the considerations above. As for claims 3, 9-10, 13, 18-19 these claims are similar to the independent claims except that they recite the further additional elements of central processing unit (CPU) memory, graphics processing unit (GPU) memory, and training with training examples. These additional elements are recited at a high level of generality and also do not amount to an improvement in the functioning of a computer or any other technology or technical field; apply the judicial exception with, or by use of, a particular machine; or apply or use the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort to monopolize the exception. Therefore, the dependent claims do not create an integration for the same reasons.
Step 2B is the next step in the eligibility analyses and evaluates whether the claims recite additional elements that amount to an inventive concept (i.e., “significantly more”) than the recited judicial exception. According to Office procedure, revised Step 2A overlaps with Step 2B, and thus, many of the considerations need not be re-evaluated in Step 2B because the answer will be the same.
In Step 2A, several additional elements were identified as additional limitations:
A computer system comprising a processor and a computer-readable medium
An online system
A generative language model
Embeddings
A transformer network
Latent space
A client device
A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to perform steps
A system comprising: a processor; and a non-transitory computer-readable medium storing instructions that, when executed by the processor, cause the processor to perform steps
These additional limitations, including the limitations in the dependent claims, do not amount to an inventive concept because they were already analyzed under Step 2A and did not amount to a practical application of the abstract idea. Therefore, the claims lack one or more limitations which amount to an inventive concept in the claims.
For these reasons, the claims are rejected under 35 U.S.C. 101.
Allowable over Prior Art of Record
Claims 1, 3-11, 13-20 are allowable over prior art though rejected on other grounds [e.g. 35 USC §101] as discussed above. The combination of elements of the claim as a whole are not found in the prior art.
Claims 1, 3-11, 13-20 would be allowable if rewritten to overcome the rejections under 35 USC §101 as set forth in this Office Action, and to include all of the limitations of the base claim and any intervening claims.
Upon review of the evidence at hand, it is hereby concluded that the totality of the evidence, alone or in combination, neither anticipates, reasonably teaches, nor renders obvious the below noted features of the Applicant’s invention.
In the present application, claims 1, 3-11, 13-20 are allowable over prior art. The most related prior art patent of record is Yoon et al (US20190179915A1) hereinafter Yoon, and Jalan et al (WO2024201209A1) hereinafter Jalan.
Yoon teaches a system that stores usage history on items purchased or used by a user in a database [0053], such as viewing time and user-provided ratings [0061]. The user may enter the interaction data through an interface [0054]. A recommendation predicting part uses user-specific information from the stored usage history to predict a latent user vector [0049] and generates latent vectors using a machine-learning natural-language-processing algorithm, such as a neural network. [0060] This vector is weighted by the item vectors of items corresponding to the user based on the number of times each item is used by the user [0086], generating a vector that constitutes and embedding for a ML model [0080-0081]. Based on the metadata vector and the usage history, the system generates a second vector [0046-0050], the metadata vector being based on stored item data including details of the items [0061]. This includes consideration of scores of a plurality of items for the specific user, which are calculated to generate recommendations [0088]. These scores are calculated using the user and item vectors [0052], and used to calculate a score for each possible recommendable item for the specific user [0095]. A list of recommendations is generated using a ranking based on these scores [0100], and the recommendation list is displayed to the user through an interface [0054].
While Yoon teaches receiving user interaction data as an input to a ML model and specific processing steps to generate new vectors as part of the recommendation process, it does not teach receiving, from the generative language model, a first set of item embeddings, the first set of item embeddings comprising an item embedding for each item of the set of items, wherein the item embeddings of the first set of item embeddings are generated by transmitting a prompt to the generative language model, wherein the prompt comprises the user interaction data; concatenating the first set of item embeddings to generate a user embedding array for the user; applying a transformer network to the user embedding array to generate a user embedding describing the user, wherein the user embedding is in a latent space.
Jalan teaches an embeddings-based recommendation system [Abstract] that operates on online servers [0024] to generate a second embedding through the concatenation of an array, by performing a convolution summation of user history and item similarity [0060]. This embedding associates items with each item in user history based on history information [0059], and is fed into a transformer model as an input. [0060] The transformer model, such as a machine learning encoder model [0089], creating an inference corresponding to the user [0053]. The ML model may be a large language model trained on masked language modeling [0025], i.e. a generative-language model.
While Jalan teaches a generative language model that provides the array outputted by a concatenation to a transformer model of a network of transformers, it does not teach receiving, from the generative language model, a first set of item embeddings, the first set of item embeddings comprising an item embedding for each item of the set of items, wherein the item embeddings of the first set of item embeddings are generated by transmitting a prompt to the generative language model, wherein the prompt comprises the user interaction data; concatenating the first set of item embeddings to generate a user embedding array for the user; applying a transformer network to the user embedding array to generate a user embedding describing the user, wherein the user embedding is in a latent space.
Further relevant prior art includes WO2024137122A1 which teaches concatenation of data values and a generative transformer model or LLM with an encoder and decoder, as well as US 20210174023 A1 which teaches a transformer encoder that can encode input into a concatenation sequence, and which inputs user activity/conversational history into the neural network.
However, each of these references fail to disclose or render obvious at least the combination of the limitations of: receiving, from the generative language model, a first set of item embeddings, the first set of item embeddings comprising an item embedding for each item of the set of items, wherein the item embeddings of the first set of item embeddings are generated by transmitting a prompt to the generative language model, wherein the prompt comprises the user interaction data; concatenating the first set of item embeddings to generate a user embedding array for the user; applying a transformer network to the user embedding array to generate a user embedding describing the user, wherein the user embedding is in a latent space.
Ultimately, the particular combination of limitations as claimed, is not anticipated nor rendered obvious in view of the cited references, and the totality of the prior art. While certain references may disclose more general concepts and parts of the claim, the prior art available does not specifically disclose the particular combination of these limitations.
The references, however, do not teach or suggest, alone or in combination, the claimed invention. Examiner emphasizes that the prior art/additional art would only be combined and deemed obvious based on knowledge gleaned from the applicant’s disclosure. Such a reconstruction is improper (i.e. hindsight reasoning). See In re McLaughlin, 443 F.2d 1392, 170 USPQ 209 (CCPA 1971).
The Examiner further emphasizes the claims as a whole and hereby asserts that the totality of the evidence fails to set forth, either explicitly or implicitly, an appropriate rationale for further modification of the evidence at hand to arrive at the claimed invention. The combination of features as claimed would not be obvious to one of ordinary skill in the art as combining various references from the totality of evidence to reach the combination of features as claimed would be a substantial reconstruction of Applicant’s claimed invention relying on improper hindsight bias.
It is thereby asserted by Examiner that, in light of the above and further deliberation over all of the evidence at hand, that the claims are allowable over prior art (though rejected under 35 USC §101) as the evidence at hand does not anticipate the claims and does not render obvious any further modification of the references to a person of ordinary skill in the art.
Response to Arguments
Applicant’s arguments filed 6/17/2026 have been fully considered but are not persuasive.
Claim Rejection – 35 §USC 101
Applicant argues “conventional embedding models require structured data as input and must be trained by the deploying system,” and that “the amended claims address this limitation directly. …By transmitting the raw user interaction data as a prompt to the generative language model, the system generates item embeddings from free-text interaction records without a prior structured-data conversion step and without a separately trained embedding model.” Applicant points to the Specification’s recitation that “LLMs do not require structured data as input. Thus, the online system does not need to expend resources converting unstructured data to structured data,” and argues that “this is a concrete improvement to the embedding pipeline itself: the recited prompt-based invocation is the system’s mechanism for operating on interaction data that a traditional embedding model could not accept directly.” Applicant argues that “the improvement lies in how the computer system performs embedding generation, not in the quality of the downstream recommendation output.” Applicant concludes that the claim therefore “improves the technical field of machine learning embedding generation by reciting that item embeddings are generated by transmitting a prompt to the generative language model, wherein the prompt comprises the user interaction data,” which is “a specific pipeline step that addresses the structured-data preprocessing burden of conventional embedding models and constitutes a concrete improvement to how the computer system performs embedding generation.”
Examiner respectfully disagrees. Firstly, the claims do not recite an LLM specifically, nor does the disclosure make clear that the interaction data is considered “unstructured data” for a hypothetical alternative embedding model that would require additional steps to process a particular unclaimed type of data. At best, the claims are reciting a single generative model that both pre-processes the interaction data and generates embeddings, as opposed to two separate software models that each perform one of the tasks. As best understood, this is at most a business improvement stemming solely from the abstract idea’s recitation of a single system for providing interaction data and processing it into data vectors, as opposed to two systems, each handling one of the two tasks, rather than a technical improvement stemming from the particular arrangement of additional elements. These elements are recited at a high level of generality, and are invoked to mere instructions to apply the abstract idea to a technological environment [MPEP 2106.05(f)]. The claims do not provide support that the generative language model performs operations in such a way as to offer a technological improvement over the data-handling capabilities of prior-art systems as alleged.
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 THOMAS J SULLIVAN whose telephone number is (571)272-9736. The examiner can normally be reached Mon - Fri 9-5 ET.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Marissa Thein can be reached at 571-272-6764. 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.
/T.J.S./Examiner, Art Unit 3689
/MARISSA THEIN/Supervisory Patent Examiner, Art Unit 3689
Prosecution Insights