DETAILED ACTION
This office action is in response to the communication filed on July 07, 2025. Claims 1-20 are currently pending.
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 .
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.
At step 1:
Independent claims 1, 15, and 20 respectively recite a method, an electronic device, and a non-transitory computer-readable storage medium, which are directed to a statutory category such as a process, machine, or an article of manufacture.
At step 2A, prong one:
Independent claim 1 and similarly independent claims 15 and 20 recite the limitations:
“determining, by using a first…model and respectively based on a first prompt element and description information of each of the plurality of content items, a plurality of content item embedding representations respectively corresponding to the plurality of content items, the first prompt element indicating extraction of a corresponding content item embedding representation from the description information of each content item”;
A person can mentally or using a pen and paper determine, by using a first model and respectively based on a first prompt element and description information of each of a plurality of content items, a plurality of content item embedding representations respectively corresponding to the plurality of content items, the first prompt element indicating extraction of a corresponding content item embedding representation from the description information of each content item.
“determining, by using a second…model and based on at least the plurality of content item embedding representations, a recommended content item to be recommended to the target user”;
A person can mentally or using a pen and paper determine, by using a model and based on at least a plurality of content item embedding representations, a recommended content item to be recommended to a target user.
The limitations, as recited above, are processes that, under their broadest reasonable interpretation, cover steps that can be performed in the human mind or by a human using a pen and paper, but for recitation of generic computer components.
If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claims recite an abstract idea.
At step 2A, prong two:
This judicial exception is not integrated into a practical application.
Independent claim 1 and similarly independent claims 15 and 20 recite the limitations:
“obtaining a content item sequence associated with historical behavior data of a target user, the content item sequence comprising a plurality of content items for which the target user sequentially performs conversion behavior”, which is a step of obtaining data. The step is recited at a high level of generality, and amounts to mere data gathering, which is a form of insignificant extra-solution activity (MPEP 2106.05(g)).
The additional elements “by using a first machine learning model” and “by using a second machine learning model” in the steps in claim 1 are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
The additional elements “an electronic device, comprising: at least one processor; and at least one memory, wherein the at least one memory is coupled to the at least one processor and stores instructions for execution by the at least one processor, and the instructions, when executed by the at least one processor, cause the device to perform acts comprising:”, “by using a first machine learning model” and “by using a second machine learning model” in the steps in claim 15 are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
The additional elements “a non-transitory computer-readable storage medium having a computer program stored thereon, the computer program, when executed by a processor, implementing acts comprising:”, “by using a first machine learning model” and “by using a second machine learning model” in the steps in claim 20 are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
At step 2B:
Independent claims 1, 15, and 20 recite the same additional elements as identified in step 2A prong two above. These additional elements are not sufficient to amount to significantly more than the judicial exception.
Independent claim 1 and similarly independent claims 15 and 20 recite the limitations:
“obtaining a content item sequence associated with historical behavior data of a target user, the content item sequence comprising a plurality of content items for which the target user sequentially performs conversion behavior”, which is a step of obtaining data, and is recognized as a well understood, routine, and conventional activity within the field of computer functions as an element of storing and retrieving information in memory (MPEP 2106.05(d)(II)(iv)).
Accordingly, the additional limitations are not sufficient to amount to significantly more than the judicial exception. Therefore, the claims are directed to an abstract idea and are not patent eligible.
Dependent claim 2 and similarly dependent claim 16 recites additional limitations, such as:
wherein determining the plurality of content item embedding representations respectively corresponding to the plurality of content items comprises:
for each content item of the plurality of content items,
“generating, based on the first prompt element and the description information of the content item, a first input sequence for the first machine learning model”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claims 1 and 15, because a person can mentally or using a pen and paper determine a plurality of content item embedding representations respectively corresponding to a plurality of content items by mentally or using a pen and paper generating, for each content item of the plurality of content items, based on a first prompt element and a description information of the content item, a first input sequence for a first machine learning model, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“obtaining, by using the first machine learning model to process the first input sequence, a first output sequence of the first machine learning model, the first output sequence comprising a content item embedding representation”, which is a step of obtaining data.
At step 2A prong two, the step is recited at a high level of generality, and amounts to mere data gathering, which is a form of insignificant extra-solution activity.
At step 2B, the step is recognized as a well understood, routine, and conventional activity within the field of computer functions as an element of storing and retrieving information in memory (MPEP 2106.05(d)(II)(iv)).
The additional elements “for the first machine learning model”, “by using the first machine learning model”, and “of the first machine learning model” in the steps are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 3 and similarly dependent claim 17 recites additional limitations, such as:
“wherein the first prompt element is placed after the description information of the content item in the first input sequence”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claims 1 and 15, because a person can mentally or using a pen and paper place a first prompt element after a description information of a content item in a first input sequence, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 4 and similarly dependent claim 18 recites additional limitations, such as:
wherein determining, by using the second machine learning model and based on at least the plurality of content item embedding representations, the recommended content item to be recommended to the target user comprises:
“generating, based on the plurality of content item embedding representations, a second input sequence for the second machine learning model”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claims 1 and 15, because a person can mentally or using a pen and paper determine, by using a model and based on at least a plurality of content item embedding representations, recommended content item to be recommended to a target by mentally or using a pen and paper generating, based on the plurality of content item embedding representations, a second input sequence for a second machine learning model, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“obtaining, by using the second machine learning model to process the second input sequence, a second output sequence of the second machine learning model, an output unit at a given position in the second output sequence indicating a content item embedding representation predicted at the given position based on a content item embedding representation before the given position in the second input sequence”, which is a step of obtaining data.
At step 2A prong two, the step is recited at a high level of generality, and amounts to mere data gathering, which is a form of insignificant extra-solution activity.
At step 2B, the step is recognized as a well understood, routine, and conventional activity within the field of computer functions as an element of storing and retrieving information in memory (MPEP 2106.05(d)(II)(iv)).
“determining the recommended content item based on a content item embedding representation indicated by a last output unit in the second output sequence”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claims 1 and 15, because a person can mentally or using a pen and paper determine a recommended content item based on a content item embedding representation indicated by a last output unit in a second output sequence, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
The additional elements “by using the second machine learning model”, “for the second machine learning model”, and “of the second machine learning model” in the steps are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 5 and similarly dependent claim 19 recites additional limitations, such as:
wherein determining the recommended content item based on the content item embedding representation indicated by the last output unit in the second output sequence comprises:
“selecting, based on a similarity between the content item embedding representation indicated by the last output unit and content item embedding representations corresponding to the plurality of candidate content items, the recommended content item from a plurality of candidate content items”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claims 1 and 15, because a person can mentally or using a pen and paper determine recommended content item based on content item embedding representation indicated by a last output unit in a second output sequence by mentally or using a pen and paper selecting, based on a similarity between the content item embedding representation indicated by the last output unit and content item embedding representations corresponding to a plurality of candidate content items, the recommended content item from a plurality of candidate content items, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 6 recites additional limitations, such as:
wherein determining, by using the second machine learning model and based on at least the plurality of content item embedding representations, the recommended content item to be recommended to the target user comprises:
“generating, based on a second prompt element and the plurality of content item embedding representations, a third input sequence for the second machine learning model, the second prompt element indicating extraction of a user embedding representation for the target user from the plurality of content item embedding representations”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine, by using a second model and based on at least a plurality of content item embedding representations, a recommended content item to be recommended to a target user by mentally or using a pen and paper generating, based on a second prompt element and the plurality of content item embedding representations, a third input sequence for a second machine learning model, the second prompt element indicating extraction of a user embedding representation for the target user from the plurality of content item embedding representations, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“obtaining, by using the third machine learning model to process the third input sequence, a third output sequence of the third machine learning model, the third output sequence comprising the user embedding representation”, which is a step of obtaining data.
At step 2A prong two, the step is recited at a high level of generality, and amounts to mere data gathering, which is a form of insignificant extra-solution activity.
At step 2B, the step is recognized as a well understood, routine, and conventional activity within the field of computer functions as an element of storing and retrieving information in memory (MPEP 2106.05(d)(II)(iv)).
“determining, based on the user embedding representation and content item embedding representations of at least one candidate content item, a probability of each of the at least one candidate content item being recommended to the target user”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine, based on a user embedding representation and content item embedding representations of at least one candidate content item, a probability of each of the at least one candidate content item being recommended to a target user, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“determining, based on the probability, the recommended content item from the at least one candidate content item”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine, based on a probability, a recommended content item from at least one candidate content item, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
The additional elements “by using the second machine learning model”, “for the second machine learning model”, “by using the third machine learning model”, and “of the third machine learning model” in the steps are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 7 recites additional limitations, such as:
“wherein the second prompt element is placed after the plurality of content item embedding representations in the third input sequence”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper place a second prompt element after a plurality of content item embedding representations in a third input sequence, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 8 recites additional limitations, such as:
wherein determining, by using the second machine learning model and based on at least the plurality of content item embedding representations, the recommended content item to be recommended to the target user comprises:
“generating, for each candidate content item of at least one candidate content item and based on the plurality of content item embedding representations and a content item embedding representation of the candidate content item, a fourth input sequence for the second machine learning model”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine, by using a second model and based on at least a plurality of content item embedding representations, a recommended content item to be recommended to a target user by mentally or using a pen and paper generating, for each candidate content item of at least one candidate content item and based on a plurality of content item embedding representations and a content item embedding representation of a candidate content item, a fourth input sequence for a second machine learning model, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“obtaining, by using the second machine learning model to process the fourth input sequence, a fourth output sequence of the second machine learning model”, which is a step of obtaining data.
At step 2A prong two, the step is recited at a high level of generality, and amounts to mere data gathering, which is a form of insignificant extra-solution activity.
At step 2B, the step is recognized as a well understood, routine, and conventional activity within the field of computer functions as an element of storing and retrieving information in memory (MPEP 2106.05(d)(II)(iv)).
“determining, based on the fourth output sequence generated for the at least one candidate content item, the recommended content item from the at least one candidate content item”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine, based on a fourth output sequence generated for at least one candidate content item, a recommended content item from the at least one candidate content item, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
The additional elements “by using the second machine learning model”, “for the second machine learning model”, and “of the second machine learning model” in the steps are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 9 recites additional limitations, such as:
“wherein the first machine learning model and the second machine learning model are language models”, which are additional elements recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 10 recites additional limitations, such as:
wherein the first machine learning model and the second machine learning model are trained by:
“determining, by using a first machine learning model and respectively based on a first prompt element and description information of each of the plurality of content items, a plurality of content item embedding representations respectively corresponding to the plurality of content items, the first prompt element indicating extraction of a corresponding content item embedding representation from the description information of each content item”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine, by using a first model and respectively based on a first prompt element and description information of each of the plurality of content items, a plurality of content item embedding representations respectively corresponding to the plurality of content items, the first prompt element indicating extraction of a corresponding content item embedding representation from the description information of each content item, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“obtaining, by using the second machine learning model to process a first number of sample content item embedding representations, a first sample output sequence of the second machine learning model”, which is a step of obtaining data.
At step 2A prong two, the step is recited at a high level of generality, and amounts to mere data gathering, which is a form of insignificant extra-solution activity.
At step 2B, the step is recognized as a well understood, routine, and conventional activity within the field of computer functions as an element of storing and retrieving information in memory (MPEP 2106.05(d)(II)(iv)).
“determining a first loss function based on a sample output unit at a given position in the first sample output sequence and a sample content item embedding representation at a position after the given position in the first number of sample content item embedding representations”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine a first loss function based on a sample output unit at a given position in a first sample output sequence and a sample content item embedding representation at a position after the given position in the first number of sample content item embedding representations, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“…reducing or minimizing a value of the first loss function”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper reduce or minimize a value of a first loss function, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
The additional elements “wherein the first machine learning model and the second machine learning model are trained by:”, “by using a first machine learning model”, “by using the second machine learning model to process a first number of sample content item embedding representations”, “of the second machine learning model”, and “training the first machine learning model and the second machine learning model by” in the steps are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 11 recites additional limitations, such as:
wherein when the model parameters of the first machine learning model remain unchanged, the second machine learning model is further trained by:
“obtaining, by using the second machine learning model to process a second number of sample content item embedding representations, a second sample output sequence of the second machine learning model, the second number is greater than the first number”, which is a step of obtaining data.
At step 2A prong two, the step is recited at a high level of generality, and amounts to mere data gathering, which is a form of insignificant extra-solution activity.
At step 2B, the step is recognized as a well understood, routine, and conventional activity within the field of computer functions as an element of storing and retrieving information in memory (MPEP 2106.05(d)(II)(iv)).
“determining the first loss function based on the sample output unit at the given position in the second sample output sequence and the sample content item embedding representation at the position following the given position in the second number of sample content item embedding representations”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine a first loss function based on a sample output unit at a given position in a second sample output sequence and a sample content item embedding representation at a position following the given position in a second number of sample content item embedding representations, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“…reducing or minimizing a value of the first loss function”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper reduce or minimize a value of a first loss function, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
The additional elements “wherein when the model parameters of the first machine learning model remain unchanged, the second machine learning model is further trained by:”, “by using the second machine learning model”, “of the second machine learning model”, and “training the second machine learning model by” in the steps are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 12 recites additional limitations, such as:
wherein the first machine learning model and the second machine learning model are trained by:
“determining, by using the first machine learning model and based on a first sample prompt element and description information of each of the plurality of sample content items, a plurality of sample content item embedding representations respectively corresponding to a plurality of sample content items, the first sample prompt element indicating extraction of a corresponding sample content item embedding representation from the description information of each sample content item”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine, by using a first model and based on a first sample prompt element and description information of each of a plurality of sample content items, a plurality of sample content item embedding representations respectively corresponding to a plurality of sample content items, the first sample prompt element indicating extraction of a corresponding sample content item embedding representation from the description information of each sample content item, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“obtaining, by using the second machine learning model to process a first number of sample content item embedding representations, a first sample user embedding representation for a first sample user”, which is a step of obtaining data.
At step 2A prong two, the step is recited at a high level of generality, and amounts to mere data gathering, which is a form of insignificant extra-solution activity.
At step 2B, the step is recognized as a well understood, routine, and conventional activity within the field of computer functions as an element of storing and retrieving information in memory (MPEP 2106.05(d)(II)(iv)).
“determining, according to the first sample user embedding representation and a content item embedding representation of the first sample candidate content item, a first probability of a first sample candidate content item being recommended to the first sample user”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine, according to a first sample user embedding representation and a content item embedding representation of a first sample candidate content item, a first probability of a first sample candidate content item being recommended to the first sample user, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“determining a second loss function based on a difference between a label of the first sample candidate content item and the first probability”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine a second loss function based on a difference between a label of a first sample candidate content item and a first probability, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“…at least reducing or minimizing a value of the second loss function”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper reduce or minimize at least a value of a second loss function, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
The additional elements “wherein the first machine learning model and the second machine learning model are trained by:”, “by using a first machine learning model”, “by using the second machine learning model to process a first number of sample content item embedding representations”, “of the second machine learning model”, and “training the first machine learning model and the second machine learning model by” in the steps are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 13 recites additional limitations, such as:
wherein when the model parameters of the first machine learning model remain unchanged, the second machine learning model is further trained by:
“obtaining, by using the second machine learning model to process a second number of sample content item embedding representations, a second sample user embedding representation for a second sample user, wherein the second number is greater than the first number”, which is a step of obtaining data.
At step 2A prong two, the step is recited at a high level of generality, and amounts to mere data gathering, which is a form of insignificant extra-solution activity.
At step 2B, the step is recognized as a well understood, routine, and conventional activity within the field of computer functions as an element of storing and retrieving information in memory (MPEP 2106.05(d)(II)(iv)).
“determining, according to the second sample user embedding representation and a content item embedding representation of the second sample candidate content item, a probability of a second sample candidate content item being recommended to the second sample user”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine, according to the second sample user embedding representation and a content item embedding representation of a second sample candidate content item, a probability of a second sample candidate content item being recommended to the second sample user, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“determining the second loss function based on a difference between a label of the second sample candidate content item and the second probability”;
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine a second loss function based on a difference between a label of second sample candidate content item and a second probability, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
“…at least reducing or minimizing a value of the second loss function”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper reduce or minimize at least a value of a second loss function, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
The additional elements “wherein when the model parameters of the first machine learning model remain unchanged, the second machine learning model is further trained by:”, “by using the second machine learning model”, and “training the second machine learning model by” in the steps are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Dependent claim 14 recites additional limitations, such as:
“wherein an embedding representation of the first prompt element and/or an embedding representation of the second prompt element are determined during a training process of the first machine learning model and the second machine learning model”.
These limitations are directed to the same abstract idea under the mental processes grouping as independent claim 1, because a person can mentally or using a pen and paper determine an embedding representation of a first prompt element and/or an embedding representation of a second prompt element during a training process of a first machine learning model and a second machine learning model, and because the limitations do not recite any additional elements that are sufficient to amount to significantly more.
The additional elements “of the first machine learning model and the second machine learning model” in the steps are recited at a high-level of generality, such that it amounts to no more than mere instructions to apply the exception using generic computer components.
Accordingly, the additional elements, individually or in combination, do not integrate the abstract idea into a practical application, even viewing the claims a whole, because it does not impose any meaningful limits on practicing the abstract idea.
Accordingly, dependent claims 2-14 and 16-19 are also directed to abstract idea without significantly more and are not patent eligible.
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) 1-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over in Sathiamoorthy (US Pub 2025/0200440, claims priority to provisional application 63/610,866 filed on 12/15/23) in view of Muthu (US Pub 2025/0209543).
With respect to claim 1, Sathiamoorthy discloses a method for content recommendation, comprising:
obtaining a content item sequence associated with historical behavior data of a target user, the content item sequence comprising a plurality of content items for which the target user sequentially performs conversion behavior (Sathiamoorthy in [0007] and [0008] discloses a recommendation system including a sequence processing model trained on auxiliary prompts, auxiliary prompt include a prompt input and a prompt output, prompts encode recommendation-related knowledge about plurality of items, obtaining item dataset describing a plurality of items, item dataset describing for each of the plurality of items attribute values such as title, category, descriptions, specifying historical user interactions, prompts include loss reduction prompts; Sathiamoorthy in [0029] and [0030] discloses natural language prompting capture distinct historical interactions and context, loss reduction prompt generated based on information on user choices between positive and negative item);
determining, by using a first machine learning model and respectively based on a first prompt element and description information of each of the plurality of content items, a plurality of content item embedding representations respectively corresponding to the plurality of content items… (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
determining, by using a second machine learning model and based on at least the plurality of content item embedding representations, a recommended content item to be recommended to the target user (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
Sathiamoorthy discloses determining based on a prompt comprising prompt element and description information of content items, however, Sathiamoorthy does not explicitly disclose:
the first prompt element indicating extraction of a corresponding content item embedding representation from the description information of each content item;
The Muthu reference discloses a first prompt element indicating extraction of a corresponding content item embedding representation from a description information of each content item (Muthu in [0016] and [0066] discloses one or more machine learning models used to generate content based on data in documents, one or more prompts are dynamically generate and provided to one or more ML models in order to extract information from documents and to generate content based on the extracted information; Muthu in [0020] discloses a first machine learning model trained to extract information from forms and/or other documents based on embeddings, machine learning model comprising a natural language processing model such as a large language model, ML model provided with an embedding of a form and a prompt that instructs it to extract a type of information from the form, extracting fields of the form in response to the prompt; Muthu in [0026] and [0053] discloses generating prompts based on an indication of desired intent for a ML model, prompt generated based on information extracted from documents, prompt comprising instructions to extract, based on an embedding of a form, a type of information from the form; Muthu in [0042] discloses providing training inputs to a ML model, ML model generating output, comparing outputs to known labels associated with training inputs, such as historical data that is verified to determine accuracy of the ML model, adjusting parameters of the ML model, such as cost or loss function for optimizing model accuracy).
Therefore, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, having the teachings of Sathiamoorthy and Muthu, to have combined Sathiamoorthy and Muthu. The motivation to combine Sathiamoorthy and Muthu would be to improve and accelerate the process of content generation using artificial intelligence and prompt engineering (Muthu: [0001]).
With respect to claim 2, Sathiamoorthy in view of Muthu discloses the method of claim 1, wherein determining the plurality of content item embedding representations respectively corresponding to the plurality of content items comprises:
for each content item of the plurality of content items, generating, based on the first prompt element and the description information of the content item, a first input sequence for the first machine learning model (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
obtaining, by using the first machine learning model to process the first input sequence, a first output sequence of the first machine learning model, the first output sequence comprising a content item embedding representation (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 3, Sathiamoorthy in view of Muthu discloses the method of claim 2, wherein the first prompt element is placed after the description information of the content item in the first input sequence (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 4, Sathiamoorthy in view of Muthu discloses the method of claim 1, wherein determining, by using the second machine learning model and based on at least the plurality of content item embedding representations, the recommended content item to be recommended to the target user comprises:
generating, based on the plurality of content item embedding representations, a second input sequence for the second machine learning model (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
obtaining, by using the second machine learning model to process the second input sequence, a second output sequence of the second machine learning model, an output unit at a given position in the second output sequence indicating a content item embedding representation predicted at the given position based on a content item embedding representation before the given position in the second input sequence (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
determining the recommended content item based on a content item embedding representation indicated by a last output unit in the second output sequence (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 5, Sathiamoorthy in view of Muthu discloses the method of claim 4, wherein determining the recommended content item based on the content item embedding representation indicated by the last output unit in the second output sequence comprises:
selecting, based on a similarity between the content item embedding representation indicated by the last output unit and content item embedding representations corresponding to the plurality of candidate content items, the recommended content item from a plurality of candidate content items (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 6, Sathiamoorthy in view of Muthu discloses the method of claim 1, wherein determining, by using the second machine learning model and based on at least the plurality of content item embedding representations, the recommended content item to be recommended to the target user comprises:
generating, based on a second prompt element and the plurality of content item embedding representations, a third input sequence for the second machine learning model, the second prompt element indicating extraction of a user embedding representation for the target user from the plurality of content item embedding representations (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
obtaining, by using the third machine learning model to process the third input sequence, a third output sequence of the third machine learning model, the third output sequence comprising the user embedding representation (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
determining, based on the user embedding representation and content item embedding representations of at least one candidate content item, a probability of each of the at least one candidate content item being recommended to the target user (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
determining, based on the probability, the recommended content item from the at least one candidate content item (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 7, Sathiamoorthy in view of Muthu discloses the method of claim 6, wherein the second prompt element is placed after the plurality of content item embedding representations in the third input sequence (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 8, Sathiamoorthy in view of Muthu discloses the method of claim 1, wherein determining, by using the second machine learning model and based on at least the plurality of content item embedding representations, the recommended content item to be recommended to the target user comprises:
generating, for each candidate content item of at least one candidate content item and based on the plurality of content item embedding representations and a content item embedding representation of the candidate content item, a fourth input sequence for the second machine learning model (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
obtaining, by using the second machine learning model to process the fourth input sequence, a fourth output sequence of the second machine learning model (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
determining, based on the fourth output sequence generated for the at least one candidate content item, the recommended content item from the at least one candidate content item (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 9, Sathiamoorthy in view of Muthu discloses the method of claim 1, wherein the first machine learning model and the second machine learning model are language models (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 10, Sathiamoorthy in view of Muthu discloses the method of claim 1, wherein the first machine learning model and the second machine learning model are trained by:
determining, by using a first machine learning model and respectively based on a first prompt element and description information of each of the plurality of content items, a plurality of content item embedding representations respectively corresponding to the plurality of content items, the first prompt element indicating extraction of a corresponding content item embedding representation from the description information of each content item (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
obtaining, by using the second machine learning model to process a first number of sample content item embedding representations, a first sample output sequence of the second machine learning model (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
determining a first loss function based on a sample output unit at a given position in the first sample output sequence and a sample content item embedding representation at a position after the given position in the first number of sample content item embedding representations (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
training the first machine learning model and the second machine learning model by reducing or minimizing a value of the first loss function (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 11, Sathiamoorthy in view of Muthu discloses the method of claim 10, wherein when the model parameters of the first machine learning model remain unchanged, the second machine learning model is further trained by:
obtaining, by using the second machine learning model to process a second number of sample content item embedding representations, a second sample output sequence of the second machine learning model, the second number is greater than the first number (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
determining the first loss function based on the sample output unit at the given position in the second sample output sequence and the sample content item embedding representation at the position following the given position in the second number of sample content item embedding representations (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
training the second machine learning model by reducing or minimizing a value of the first loss function (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 12, Sathiamoorthy in view of Muthu discloses the method of claim 1, wherein the first machine learning model and the second machine learning model are trained by:
determining, by using the first machine learning model and based on a first sample prompt element and description information of each of the plurality of sample content items, a plurality of sample content item embedding representations respectively corresponding to a plurality of sample content items, the first sample prompt element indicating extraction of a corresponding sample content item embedding representation from the description information of each sample content item (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
obtaining, by using the second machine learning model to process a first number of sample content item embedding representations, a first sample user embedding representation for a first sample user (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
determining, according to the first sample user embedding representation and a content item embedding representation of the first sample candidate content item, a first probability of a first sample candidate content item being recommended to the first sample user (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
determining a second loss function based on a difference between a label of the first sample candidate content item and the first probability (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
training the first machine learning model and the second machine learning model by at least reducing or minimizing a value of the second loss function (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 13, Sathiamoorthy in view of Muthu discloses the method of claim 12, wherein when the model parameters of the first machine learning model remain unchanged, the second machine learning model is further trained by:
obtaining, by using the second machine learning model to process a second number of sample content item embedding representations, a second sample user embedding representation for a second sample user, wherein the second number is greater than the first number (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
determining, according to the second sample user embedding representation and a content item embedding representation of the second sample candidate content item, a probability of a second sample candidate content item being recommended to the second sample user (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
determining the second loss function based on a difference between a label of the second sample candidate content item and the second probability (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
training the second machine learning model by at least reducing or minimizing a value of the second loss function (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 14, Sathiamoorthy in view of Muthu discloses the method of claim 10, wherein an embedding representation of the first prompt element and/or an embedding representation of the second prompt element are determined during a training process of the first machine learning model and the second machine learning model (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 15, Sathiamoorthy discloses an electronic device (Sathiamoorthy in [0179] discloses computing device including one or more processors and a memory, memory including one or more non-transitory computer-readable storage media, memory storing instructions executed by processor to cause computing device to perform operations), comprising:
at least one processor (Sathiamoorthy in [0179] discloses computing device including one or more processors and a memory, memory including one or more non-transitory computer-readable storage media, memory storing instructions executed by processor to cause computing device to perform operations); and
at least one memory, wherein the at least one memory is coupled to the at least one processor and stores instructions for execution by the at least one processor, and the instructions, when executed by the at least one processor, cause the device to perform acts (Sathiamoorthy in [0179] discloses computing device including one or more processors and a memory, memory including one or more non-transitory computer-readable storage media, memory storing instructions executed by processor to cause computing device to perform operations) comprising:
obtaining a content item sequence associated with historical behavior data of a target user, the content item sequence comprising a plurality of content items for which the target user sequentially performs conversion behavior (Sathiamoorthy in [0007] and [0008] discloses a recommendation system including a sequence processing model trained on auxiliary prompts, auxiliary prompt include a prompt input and a prompt output, prompts encode recommendation-related knowledge about plurality of items, obtaining item dataset describing a plurality of items, item dataset describing for each of the plurality of items attribute values such as title, category, descriptions, specifying historical user interactions, prompts include loss reduction prompts; Sathiamoorthy in [0029] and [0030] discloses natural language prompting capture distinct historical interactions and context, loss reduction prompt generated based on information on user choices between positive and negative item);
determining, by using a first machine learning model and respectively based on a first prompt element and description information of each of the plurality of content items, a plurality of content item embedding representations respectively corresponding to the plurality of content items… (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
determining, by using a second machine learning model and based on at least the plurality of content item embedding representations, a recommended content item to be recommended to the target user (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
Sathiamoorthy discloses determining based on a prompt comprising prompt element and description information of content items, however, Sathiamoorthy does not explicitly disclose:
the first prompt element indicating extraction of a corresponding content item embedding representation from the description information of each content item;
The Muthu reference discloses a first prompt element indicating extraction of a corresponding content item embedding representation from a description information of each content item (Muthu in [0016] and [0066] discloses one or more machine learning models used to generate content based on data in documents, one or more prompts are dynamically generate and provided to one or more ML models in order to extract information from documents and to generate content based on the extracted information; Muthu in [0020] discloses a first machine learning model trained to extract information from forms and/or other documents based on embeddings, machine learning model comprising a natural language processing model such as a large language model, ML model provided with an embedding of a form and a prompt that instructs it to extract a type of information from the form, extracting fields of the form in response to the prompt; Muthu in [0026] and [0053] discloses generating prompts based on an indication of desired intent for a ML model, prompt generated based on information extracted from documents, prompt comprising instructions to extract, based on an embedding of a form, a type of information from the form; Muthu in [0042] discloses providing training inputs to a ML model, ML model generating output, comparing outputs to known labels associated with training inputs, such as historical data that is verified to determine accuracy of the ML model, adjusting parameters of the ML model, such as cost or loss function for optimizing model accuracy).
Therefore, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, having the teachings of Sathiamoorthy and Muthu, to have combined Sathiamoorthy and Muthu. The motivation to combine Sathiamoorthy and Muthu would be to improve and accelerate the process of content generation using artificial intelligence and prompt engineering (Muthu: [0001]).
With respect to claim 16, Sathiamoorthy in view of Muthu discloses the electronic device of claim 15, wherein determining the plurality of content item embedding representations respectively corresponding to the plurality of content items comprises: for each content item of the plurality of content items,
generating, based on the first prompt element and the description information of the content item, a first input sequence for the first machine learning model (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
obtaining, by using the first machine learning model to process the first input sequence, a first output sequence of the first machine learning model, the first output sequence comprising a content item embedding representation (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 17, Sathiamoorthy in view of Muthu discloses the electronic device of claim 16, wherein the first prompt element is placed after the description information of the content item in the first input sequence (Sathiamoorthy in [0007] and [0008] discloses a recommendation system including a sequence processing model trained on auxiliary prompts, auxiliary prompt include a prompt input and a prompt output, prompts encode recommendation-related knowledge about plurality of items, obtaining item dataset describing a plurality of items, item dataset describing for each of the plurality of items attribute values such as title, category, descriptions, specifying historical user interactions, prompts include loss reduction prompts; Sathiamoorthy in [0029] and [0030] discloses natural language prompting capture distinct historical interactions and context, loss reduction prompt generated based on information on user choices between positive and negative item).
With respect to claim 18, Sathiamoorthy in view of Muthu discloses the electronic device of claim 15, wherein determining, by using the second machine learning model and based on at least the plurality of content item embedding representations, the recommended content item to be recommended to the target user comprises:
generating, based on the plurality of content item embedding representations, a second input sequence for the second machine learning model (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output);
obtaining, by using the second machine learning model to process the second input sequence, a second output sequence of the second machine learning model, an output unit at a given position in the second output sequence indicating a content item embedding representation predicted at the given position based on a content item embedding representation before the given position in the second input sequence (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
determining the recommended content item based on a content item embedding representation indicated by a last output unit in the second output sequence (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 19, Sathiamoorthy in view of Muthu discloses the electronic device of claim 18, wherein determining the recommended content item based on the content item embedding representation indicated by the last output unit in the second output sequence comprises:
selecting, based on a similarity between the content item embedding representation indicated by the last output unit and content item embedding representations corresponding to the plurality of candidate content items, the recommended content item from a plurality of candidate content items (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
With respect to claim 20, Sathiamoorthy discloses a non-transitory computer-readable storage medium having a computer program stored thereon, the computer program, when executed by a processor, implementing acts (Sathiamoorthy in [0179] discloses computing device including one or more processors and a memory, memory including one or more non-transitory computer-readable storage media, memory storing instructions executed by processor to cause computing device to perform operations) comprising:
obtaining a content item sequence associated with historical behavior data of a target user, the content item sequence comprising a plurality of content items for which the target user sequentially performs conversion behavior (Sathiamoorthy in [0007] and [0008] discloses a recommendation system including a sequence processing model trained on auxiliary prompts, auxiliary prompt include a prompt input and a prompt output, prompts encode recommendation-related knowledge about plurality of items, obtaining item dataset describing a plurality of items, item dataset describing for each of the plurality of items attribute values such as title, category, descriptions, specifying historical user interactions, prompts include loss reduction prompts; Sathiamoorthy in [0029] and [0030] discloses natural language prompting capture distinct historical interactions and context, loss reduction prompt generated based on information on user choices between positive and negative item);
determining, by using a first machine learning model and respectively based on a first prompt element and description information of each of the plurality of content items, a plurality of content item embedding representations respectively corresponding to the plurality of content items… (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output); and
determining, by using a second machine learning model and based on at least the plurality of content item embedding representations, a recommended content item to be recommended to the target user (Sathiamoorthy in [0008] and [0024] discloses prompts including item embedding prompts that encode knowledge about a plurality of items, natural language prompts encode different types of recommendation-related knowledge, such as item attributes and user preferences, encode various operations and losses used to impart recommendation knowledge to a sequence processing model, including item embedding, using loss reduction; Sathiamoorthy in [0025] discloses models perform recommendation tasks such as retrieval, ranking, and prediction; Sathiamoorthy in [0028] and [0113] discloses encoding item embedding into prompt, embedded items represented in a common vector space; Sathiamoorthy in [0083] and [0085] discloses obtaining training datasets, training data can be labeled or unlabeled, evaluating using a loss function, compare model output to prompt output, evaluate using labels; Sathiamoorthy in [0098] and [0099] discloses sequence processing models can include one or multiple machine-learned model components configured to ingest, generate, or reason over sequences of information, models in text domain are large language models, project data into an input space with prediction layers via embedding; Sathiamoorthy in [0105] and [0110] discloses prediction layers evaluate associations between portions of input sequence and particular output element, assigning probability, generating output sequences, output of prediction layers passed through output layers, output sequence generated by sampling and context window; Sathiamoorthy in [0133] and [0163] discloses plurality of machine-learned models, a first model can process information about a task and output a input for a second model, ML models process natural language data to generate an output).
Sathiamoorthy discloses determining based on a prompt comprising prompt element and description information of content items, however, Sathiamoorthy does not explicitly disclose:
the first prompt element indicating extraction of a corresponding content item embedding representation from the description information of each content item;
The Muthu reference discloses a first prompt element indicating extraction of a corresponding content item embedding representation from a description information of each content item (Muthu in [0016] and [0066] discloses one or more machine learning models used to generate content based on data in documents, one or more prompts are dynamically generate and provided to one or more ML models in order to extract information from documents and to generate content based on the extracted information; Muthu in [0020] discloses a first machine learning model trained to extract information from forms and/or other documents based on embeddings, machine learning model comprising a natural language processing model such as a large language model, ML model provided with an embedding of a form and a prompt that instructs it to extract a type of information from the form, extracting fields of the form in response to the prompt; Muthu in [0026] and [0053] discloses generating prompts based on an indication of desired intent for a ML model, prompt generated based on information extracted from documents, prompt comprising instructions to extract, based on an embedding of a form, a type of information from the form; Muthu in [0042] discloses providing training inputs to a ML model, ML model generating output, comparing outputs to known labels associated with training inputs, such as historical data that is verified to determine accuracy of the ML model, adjusting parameters of the ML model, such as cost or loss function for optimizing model accuracy).
Therefore, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, having the teachings of Sathiamoorthy and Muthu, to have combined Sathiamoorthy and Muthu. The motivation to combine Sathiamoorthy and Muthu would be to improve and accelerate the process of content generation using artificial intelligence and prompt engineering (Muthu: [0001]).
Remarks
The relevant prior art of record that is not used in claim rejections but is pertinent to the claims or disclosure is:
Puget (US Pub 2025/0328945), which discloses machine learning models for generating recommendations using embeddings.
Contact Information
Any inquiry concerning this communication or earlier communications from the examiner should be directed to REZWANUL MAHMOOD whose telephone number is (571)272-5625. The examiner can normally be reached M-F 9-5:30.
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, Ann J. Lo can be reached at 571-272-9767. 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.
/R.M/Examiner, Art Unit 2159 /MARC S SOMERS/Primary Examiner, Art Unit 2159
Prosecution Insights