DETAILED ACTION
Notice of 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 .
Information Disclosure Statement
The information disclosure statement submitted on 1/16/2024 has been considered.
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 1-25 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea significantly more.
Regarding Step 1 of the Alice/Mayo framework, Claims 1-10 and 21-25 are directed to a method (a process), Claims 11-15 are directed to a system (a machine), which each fall within one of the four statutory categories of inventions.
Claims 16-20 are directed to a “computer program product” having at least “one computer-readable storage medium”. Para. 0055 of the instant specification explains that “A computer readable storage medium, as that term is used in the present disclosure, is not to be construed as storage in the form of transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide, light pulses passing through a fiber optic cable, electrical signals communicated through a wire, and/or other transmission media. As will be understood by those of skill in the art, data is typically moved at some occasional points in time during normal operations of a storage device, such as during access, de-fragmentation or garbage collection, but this does not render the storage device as transitory because the data is not transitory while it is stored.” In view of this teaching in the specification, the recited “computer program product” having at least one “computer readable storage medium” is not being interpreted as covering transitory signals, and is therefore directed to a statutory category (article of manufacture).
Regarding Claim 1
Step 2A, prong 1 (Is the claim directed to a law of nature, a natural phenomenon or an abstract idea).
Claim 1 recites the following mental processes, that in each case under the broadest reasonable interpretation, covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components (e.g., “computer-implemented”).
inputting into the one or more templates information relating to the requester to provide one or more populated templates (under the broadest reasonable interpretation, a human such as a customer service agent, can take a template written on paper and populate the blanks when reading a script to a customer (a requestor), such as populating the blanks for the customer’s name, a product/service identifier, etc.)
Claim 1 further recites the following limitations that fall under the “managing personal behavior or relationships or interactions between people” judicial exception category. See MPEP 2106.04(a)(2) II.C.
generating the customized reply to the request based on the one or more populated templates (under the broadest reasonable interpretation, a human such as a customer service agent, can take a template written on paper and populate the blanks when reading a script to a customer (a requestor), and then based on such populated template, the human can read the populated template (which is a customized reply for the customer), where such conversation is merely providing information to another human being)
providing the customized reply to the requester. (under the broadest reasonable interpretation, a human such as a customer service agent, can take a template written on paper and populate the blanks when reading a script to a customer (a requestor), and then based on such populated template, the human can read the populated template to the customer (which is a customized reply), where such conversation is merely providing information to another human being)
Step 2A, prong 2 (Does the claim recite additional elements that integrate the judicial exception into a practical application?).
The judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements (e.g., “computer-implemented”) which are recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f)).
Regarding the “A computer-implemented method of facilitating processing within a computing environment, the computer-implemented method comprising” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of a generic computing environment. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (a generic computing environment). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Regarding the “receiving from a requester a request for a customized reply, the receiving the request using one or more networks of the computing environment” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)).
Regarding the “retrieving one or more templates stored in a selected location based, at least in part, on at least one ontology constructed for one or more domains associated with the request for the customized reply” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)).
Accordingly, at Step 2A, prong two, after considering all claim elements individually and as an ordered combination, it is determined that the claims do not integrate the judicial exception into a practical application.
Step 2B (Does the claim recite additional elements that amount to significantly more than the judicial exception?)
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception. As discussed above, the additional elements (e.g., “computer-implemented”) are recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f)).
Regarding the “A computer-implemented method of facilitating processing within a computing environment, the computer-implemented method comprising” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Regarding the “receiving from a requester a request for a customized reply, the receiving the request using one or more networks of the computing environment” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Regarding the “retrieving one or more templates stored in a selected location based, at least in part, on at least one ontology constructed for one or more domains associated with the request for the customized reply” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Accordingly, at Step 2B after considering all claim elements individually and as an ordered combination, it is determined that the claims do not integrate the judicial exception into a practical application.
Regarding Claim 2
Step 2A, Prong 2
Regarding the “wherein the one or more templates are one or more artificial intelligence templates” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of generic artificial intelligence. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (generic artificial intelligence). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Step 2B
Regarding the “wherein the one or more templates are one or more artificial intelligence templates” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Regarding Claim 3
Step 2A, Prong 2
Regarding the “wherein the one or more templates are created using a large language model that is tuned to isolate one or more chosen parts of the large language model and to prune one or more other parts from the large language model that are not chosen” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of a pruned large language model. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (a pruned large language model). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Step 2B
Regarding the “wherein the one or more templates are created using a large language model that is tuned to isolate one or more chosen parts of the large language model and to prune one or more other parts from the large language model that are not chosen” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Regarding Claim 4
Step 2A, Prong 2
Regarding the “wherein the retrieving the one or more templates comprises: using a key generated based on a prompt of the request to select for retrieval at least one template of the one or more templates stored in the selected location” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)). Moreover, such limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (data gathering using a look-up table with a key or index). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not integrate a judicial exception into a practical application.
Step 2B
Regarding the “wherein the retrieving the one or more templates comprises: using a key generated based on a prompt of the request to select for retrieval at least one template of the one or more templates stored in the selected location” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory"). Moreover, such limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use as explained above, which does not amount to significantly more than the judicial exception. MPEP 2106.05(h).
Regarding Claim 5
Step 2A, Prong 2
Regarding the “receiving requester feedback regarding the customized reply provided to the requester” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)).
Regarding the “modifying one or more weights of the one or more templates based on the requester feedback, wherein the one or more weights are modified based on performing reinforcement learning context generation” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation attempts to cover a solution to an identified problem with no restriction on how the result is accomplished, or provides no description of the mechanism for accomplishing the result (e.g., any weight modifications utilizing reinforcement learning are covered). Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)). Moreover, such limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (reinforcement learning). As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not integrate a judicial exception into a practical application.
Step 2B
Regarding the “receiving requester feedback regarding the customized reply provided to the requester” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Regarding the “modifying one or more weights of the one or more templates based on the requester feedback, wherein the one or more weights are modified based on performing reinforcement learning context generation” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation attempts to cover a solution to an identified problem with no restriction on how the result is accomplished, or provides no description of the mechanism for accomplishing the result. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)). Moreover, such limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use as explained above, which does not amount to significantly more than the judicial exception. MPEP 2106.05(h).
Regarding Claim 6
Step 2A, Prong 1
performing retrieval augmented generation to establish one or more context vectors that provide supporting information for generation of the customized reply to the request (under the broadest reasonable interpretation, a human can mentally retrieve information (such as by reading) and then establish context vectors on paper, such as by writing down on paper a series of information related to the customized reply, such as if a customer service agent writes down [customer name, time of call, associated product/service name, tone of caller])
establishing the one or more context vectors, the establishing comprising: (under the broadest reasonable interpretation, a human such as a customer service agent can establish context vectors on paper, such as by writing down on paper a series of information related to the customized reply, such as if a customer service agent writes down [customer name, time of call, associated product/service name, tone of caller])
combining the selected data that is retrieved with parametric data encoded by the large language model to establish the one or more context vectors. (under the broadest reasonable interpretation, a human such as a customer service agent can combine data to establish the context vectors on paper, such as by writing down on paper a series of information related to the customized reply, such as if a customer service agent writes down [customer name, time of call, associated product/service name, tone of caller], and then appending the fields output from a large language model that is listening to the call and providing information to the customer service agent)
Step 2A, Prong 2
Regarding the “retrieving selected data from one or more external data sources, the selected data being knowledge data missing from a large language model used to create the one or more templates, the knowledge data to be used to generate the customized reply” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)).
Step 2B
Regarding the “retrieving selected data from one or more external data sources, the selected data being knowledge data missing from a large language model used to create the one or more templates, the knowledge data to be used to generate the customized reply” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Regarding Claim 7
Step 2A, Prong 2
Regarding the “wherein the parametric data includes one or more large language model weights and one or more fine-tuned weight adjustments” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of a LLM that has been trained and fine-tuned. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (an LLM that has been trained and fine-tuned). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Step 2B
Regarding the “wherein the parametric data includes one or more large language model weights and one or more fine-tuned weight adjustments” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Regarding Claim 8
Step 2A, Prong 1
wherein the customized reply to the request is to be received within a predefined amount of time. (under the broadest reasonable interpretation, a customer service representative can provide a reply within a predefined amount of time, such as if there is a company policy to respond to a caller’s request within 1 minute of putting the caller on hold)
Regarding Step 2A, Prong 2, the claim does not include any additional elements that integrate the judicial exception into a practical application and regarding Step 2B, there are no additional elements recited that amount to significantly more than the judicial exception.
Regarding Claim 9
Step 2A, Prong 2
Regarding the “wherein the receiving the request comprises receiving a plurality of requests for a plurality of customized replies at a particular request rate, the plurality of requests including the request” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)).
Step 2B
Regarding the “wherein the receiving the request comprises receiving a plurality of requests for a plurality of customized replies at a particular request rate, the plurality of requests including the request” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Regarding Claim 10
Step 2A, Prong 2
Regarding the “wherein the one or more templates are generated using a large language model, the large language model trained using a reduced set of computing resources” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of an LLM trained using a reduced set of computing resources. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (an LLM trained using a reduced set of computing resources). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Step 2B
Regarding the “wherein the one or more templates are generated using a large language model, the large language model trained using a reduced set of computing resources” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Regarding Claim 11
Step 2A, Prong 1
Claim 11 recites a computer system that corresponds to the method of claim 1, and therefore the analysis under Step 2A, Prong 1 with respect to claim 1 also applies to this claim 11. While claim 11 recites additional generic computing components (“computer-readable storage medium”, “processor set”), such additional generic computing components do not change the analysis under Step 2A, Prong 1.
Step 2A, Prong 2
Claim 11 recites a computer system that corresponds to the method of claim 1, and therefore the analysis under Step 2A, Prong 2 with respect to claim 1 also applies to this claim 11. While claim 11 recites additional generic computing components (“computer-readable storage medium”, “processor set”), such additional generic computing components do not change the analysis under Step 2A, Prong 2. Such limitations (“computer-readable storage medium”, “processor set”) are recited at a high-level of generality and amount to no more than adding the words “apply it” (or an equivalent) with the judicial exception. These additional elements are recited at a high-level of generality and amount to no more than mere instructions to apply the exception using generic computer components. Accordingly, these additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Step 2B
Claim 11 recites a computer system that corresponds to the method of claim 1, and therefore the analysis under Step 2B with respect to claim 1 also applies to this claim 11. While claim 11 recites additional generic computing components (“computer-readable storage medium”, “processor set”), such additional generic computing components do not change the analysis under Step 2B. Such limitations are recited at a high-level of generality and amount to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, these additional elements do not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Claims 12-15 recite a computer system that depend from claim 11 and correspond to the methods of claims 3-5 and 9, and are therefore rejected for the same reasons explained above with respect to claims 12 and 3-5, and 9, respectively.
Regarding Claim 16
Step 2A, Prong 1
Claim 16 recites a computer program product that corresponds to the method of claim 1, and therefore the analysis under Step 2A, Prong 1 with respect to claim 1 also applies to this claim 16. While claim 16 recites additional generic computing components (“computer-readable storage medium”, “program instructions”), such additional generic computing components do not change the analysis under Step 2A, Prong 1.
Step 2A, Prong 2
Claim 16 recites a computer program product that corresponds to the method of claim 1, and therefore the analysis under Step 2A, Prong 2 with respect to claim 1 also applies to this claim 16. While claim 16 recites additional generic computing components (“computer-readable storage medium”, “program instructions”), such additional generic computing components do not change the analysis under Step 2A, Prong 2. Such limitations (“computer-readable storage medium”, “processor set”) are recited at a high-level of generality and amount to no more than adding the words “apply it” (or an equivalent) with the judicial exception. These additional elements are recited at a high-level of generality and amount to no more than mere instructions to apply the exception using generic computer components. Accordingly, these additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Step 2B
Claim 16 recites a computer program product that corresponds to the method of claim 1, and therefore the analysis under Step 2B with respect to claim 1 also applies to this claim 16. While claim 16 recites additional generic computing components (“computer-readable storage medium”, “program instructions”), such additional generic computing components do not change the analysis under Step 2B. Such limitations are recited at a high-level of generality and amount to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, these additional elements do not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Claims 17-20 recite a computer program product that depend from claim 16 and correspond to the methods of claims 3-5 and 9, and are therefore rejected for the same reasons explained above with respect to claims 16 and 3-5, and 9, respectively.
Regarding Claim 21
Step 2A, prong 1 (Is the claim directed to a law of nature, a natural phenomenon or an abstract idea).
Claim 21 recites the following mental processes, that in each case under the broadest reasonable interpretation, covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components (e.g., “computer-implemented”).
inputting information in the one or more templates to provide one or more populated templates (under the broadest reasonable interpretation, a human such as a customer service agent, can take a template written on paper and populate the blanks when reading a script to a customer (a requestor), such as populating the blanks for the customer’s name, a product/service identifier, etc.)
Claim 21 further recites the following limitations that fall under the “managing personal behavior or relationships or interactions between people” judicial exception category. See MPEP 2106.04(a)(2) II.C.
generating a customized reply to the request based on the one or more populated templates (under the broadest reasonable interpretation, a human such as a customer service agent, can take a template written on paper and populate the blanks when reading a script to a customer (a requestor), and then based on such populated template, the human can read the populated template (which is a customized reply for the customer), where such conversation is merely providing information to another human being)
providing to the requester the customized reply that is generated based on the one or more populated templates. (under the broadest reasonable interpretation, a human such as a customer service agent, can take a template written on paper and populate the blanks when reading a script to a customer (a requestor), and then based on such populated template, the human can read the populated template to the customer (which is a customized reply), where such conversation is merely providing information to another human being)
Step 2A, prong 2 (Does the claim recite additional elements that integrate the judicial exception into a practical application?).
The judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements (e.g., “computer-implemented”) which are recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f)).
Regarding the “A computer-implemented method of facilitating processing within a computing environment, the computer-implemented method comprising” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of a generic computing environment. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (a generic computing environment). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Regarding the “receiving from a requester a request over one or more networks of the computing environment, the request related to a selected event” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)).
Regarding the “retrieving one or more templates stored in a selected location, the one or more templates retrieved based, at least in part, on one or more ontologies constructed for the selected event” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)).
Regarding the “the one or more templates created using a large language model trained to create the one or more templates based on the selected event” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of a LLM creating a template. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (a LLM creating a template). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Regarding the “the large language model shielded from direct request traffic” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of a LLM being shielded from direct request traffic. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (a LLM being shielded from direct request traffic). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Accordingly, at Step 2A, prong two, after considering all claim elements individually and as an ordered combination, it is determined that the claims do not integrate the judicial exception into a practical application.
Step 2B (Does the claim recite additional elements that amount to significantly more than the judicial exception?)
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception. As discussed above, the additional elements (e.g., “computer-implemented”) are recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f)).
Regarding the “A computer-implemented method of facilitating processing within a computing environment, the computer-implemented method comprising” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Regarding the “receiving from a requester a request over one or more networks of the computing environment, the request related to a selected event” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Regarding the “retrieving one or more templates stored in a selected location, the one or more templates retrieved based, at least in part, on one or more ontologies constructed for the selected event” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Regarding the “the one or more templates created using a large language model trained to create the one or more templates based on the selected event” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Regarding the “the large language model shielded from direct request traffic” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Accordingly, at Step 2B after considering all claim elements individually and as an ordered combination, it is determined that the claims do not integrate the judicial exception into a practical application.
Claim 22 recites a computer-implemented method that depends from claim 21 and correspond to the method of claim 8, and is therefore rejected for the same reasons explained above with respect to claims 8 and 21.
Claim 23 recites a computer-implemented method that depends from claim 21 and correspond to the method of claim 10, and is therefore rejected for the same reasons explained above with respect to claims 10 and 21.
Claim 24 recites a computer-implemented method that depends from claim 21 and correspond to the method of claim 9, and is therefore rejected for the same reasons explained above with respect to claims 9 and 21.
Regarding Claim 25
Step 2A, Prong 2
Regarding the “wherein the large language model that is trained is tuned using one or more tuning techniques” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of a tuned LLM. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (a tuned LLM). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Step 2B
Regarding the “wherein the large language model that is trained is tuned using one or more tuning techniques” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-2, 11, and 16 are rejected under 35 U.S.C. 103 as being unpatentable over US 20250133042 A1, hereinafter referenced as TSVETKOV, in view of US 20230114066 A1, hereinafter referenced as SOBLE.
Regarding Claim 1
TSVETKOV teaches:
A computer-implemented method of facilitating processing within a computing environment, the computer-implemented method comprising: (TSVETKOV, para. 0003: “According to one implementation, a method is disclosed for using a large language model (LLM) to generate a customized email according to a methodology that reduces the likelihood of the customized email including hallucinated facts or undesired (e.g., irrelevant) personal identifying information (PII).”;
TSVETKOV, para. 0023: “FIG. 1 illustrates an example system 100 that uses generative AI to create customized email content according to a methodology that lowers the likelihood of the model outputs including hallucinated facts or sensitive personal identity information (PII) that may pose a legal risk to the creating entity. Within the system 100, a user provides input to an email application 102 to initiate generation of a customized email. The email application 102 may include software components that execute locally on a user device, remotely in a cloud platform, or some combination thereof.”;
TSVETKO, para. 0036: “In some implementations, the user's compute environment includes tool(s) that collect and store meeting information 130, such as a transcripts or summaries of different meetings. For example, Microsoft's Sales Copilot® is an add-on to Microsoft Teams®, which business users frequently use to conduct web-based calls.”;
Examiner’s Note”: Fig. 1 illustrates the computing environment for processing emails)
receiving from a requester a request for a customized reply, the receiving the request using one or more networks of the computing environment; (TSVTETKOV, para. 0023: “Within the system 100, a user provides input to an email application 102 to initiate generation of a customized email”;
TSVETKOV, para. 0024: “In various implementations, aspects of the LLM context confiner 108 reside locally on the device of the end user, remotely in the cloud, or some combination thereof.”
TSVETKOV, para. 0034: “Within the email application 102, the user also identifies a recipient of the customized email that is to be drafted, such as by selecting a contact card for the recipient, typing the recipient's name, email address, or other form of recipient identifier 116. A context mining tool 114 uses the recipient identifier 116 along with an identifier of the user composing the message (e.g., sender ID 118) and the PII-filtered template 132 to mine contextual data 128 from various data sources that is potentially relevant to the draft email.”;
TSVETKOV, para. 0076: “One or more applications 540 (e.g., the LLM context confiner 108 of FIG. 1 or any of its subcomponents) are loaded in the memory 504 and executed on the operating system 510 by the processing system 502. In some implementations, aspects of the LLM context confiner 108 of FIG. 1 are loaded into memory of different processing devices connected across a network.”;
TSVETKOV, para. 0077: “Additionally, the applications 540 may receive input from one or more remote devices, such as remotely-located servers or smart devices, by communicating with such devices over a wired or wireless network using more communication transceivers 530 and an antenna 532 to provide network connectivity (e.g., a mobile phone network, Wi-Fi®, Bluetooth®). The processing device 500 may also include one or more storage devices 528 (e.g., non-volatile storage) for storing email templates, emails, user profile data, meeting summaries, etc. Other configurations may also be employed.”;
Examiner’s Note: the user (corresponding to recited “requestor”) requests a customized email (corresponding to recited “customized reply”, where the customized email is a reply in response to the request by the user), and the various components of system 100 may be spread across a computer network such that the request for an email can come from a user using a remote device and communicating with a server hosting the email application)
retrieving one or more templates stored in a selected location based ... (TSVETKOV, para. 0028: “For example, the template 104 is selected from a library of pre-generated templates that are each tailored for a different purpose, product, and/or target customer (e.g., based on specific characteristics of the customer or customer's business). The types of templates present in the template library and content of each template may vary considerably depending upon the nature of the business entity using the system 100, such as based on the types of products or services offered by the business and/or the types of customer communications that the business regularly conducts.”;
TSVETKOV, para. 0070: “In some implementations, the email template is an email previously composed by the first user. In other implementations, the email template is selected from a template library comprising templates crafted to serve as templates, such as that provide information on different products or services, target customers with different characteristics, etc.”
Examiner’s Note: templates are retrieved for use from the template library (corresponding to recited “selected location”)
inputting into the one or more templates information relating to the requester to provide one or more populated templates; (TSVETKOV, para. 0042: “ The relevant contextual information 124 is provided as input to an LLM prompt generator 134 along with “a template body,” which refers to either the PII-filtered template 132 or the template 104 (e.g., in use cases that do not include PII filtering) along with metadata of the template, if any exists. Based on the relevant contextual information 124 and the template body, the LLM prompt generator 134 generates the prompt 110. The prompt 110 includes both the template body and the relevant contextual information 124 and further includes an instruction asking the LLM 112 to draft an email based on the relevant contextual information 124 following the template 104.”;
TSVETKOV, para. 0062: “Here, the customized email 230 includes an intro portion 236 that includes stylized language from the template 204. This intro portion is followed by a product description portion 238 that includes detail from the template with product names populated from the relevant contextual information 224.”;
Examiner’s Note: as shown in the example customized email of 230, the LLM inputs contextual information into the template in order to populate the template)
generating the customized reply to the request based on the one or more populated templates; and (TSVETKOV, para. 0061: “ FIG. 2C illustrates a customized email 230 that is output by an LLM 232 in response to receiving the prompt 212, as described above, which is generated per the operations described with respect to FIGS. 2A and 2B.”;
TSVETKOV, para. 0062: “Here, the customized email 230 includes an intro portion 236 that includes stylized language from the template 204. This intro portion is followed by a product description portion 238 that includes detail from the template with product names populated from the relevant contextual information 224. A third paragraph 240 following the product description portion 238 includes additional context from the relevant contextual information 224 that references the recent meeting between the two parties and also summarizes the “follow-ups” identified in the relevant contextual information 224 as affirmative actions that are being taken. The email includes links 242 from the template 204, and the remaining portion 244 of the email following the links 242 also includes text that is taken, verbatim, from the template 204. In scenarios where the template 204 was an email previously drafted by the user, the user's own writing style is therefore captured in the customed email 230 generated by the LLM 232.”;
Examiner’s Note: the LLM uses the populated template to customize the generated email as shown in Fig. 2C)
providing the customized reply to the requester. (TSVETKOV, para. 0074: “A returning operation 412 returns the customized email to the email application for display in an email composition window presented on a user interface.”)
However, TSVETKOV fails to teach:
... based, at least in part, on at least one ontology constructed for one or more domains associated with the request for the customized reply;
However, in a related field of endeavor (a clinical ontology with respect to medical report templates, see para. 0002), SOBLE teaches and makes obvious:
retrieving one or more templates stored in a selected location based, at least in part, on at least one ontology constructed for one or more domains associated with the request for the customized reply; (SOBLE, para. 0020: “The system and methods of the present invention also permits a user to produce a report template according to a clinical ontology chosen by the user. By the selection of a topic, heading, and subheading of the template, a user may obtain access to the relevant information organized by the system and thereby be able to prepare a medical report and complete a clinical study for a subject more efficiently.”;
SOBLE, para. 0021: “ By organizing the template and the information according to the same chosen clinical ontology, an efficient display of the information may be formed. The efficient display of information formed through the use of the present invention is termed also a “categorical display”. The categorical display may be distributed through a network to one or more display outputs to which are connected one or more displays on which the categorical display may be shown to one or more users. A “categorical display” for purposes of this application is one in which at least certain or all the information that has been obtained for a subject is organized and made accessible according to the same clinical ontology on which the template report is based so that a user may be provided with the corresponding appropriate information simply by selecting the heading or subheading of the template report through the configurable display. Through the use of the template report and the categorical display tool, the user can conduct and report the results for a clinical study more efficiently.”;
SOBLE, para. 0029: “Embodiments of the present invention may include an information resource that facilitates such processing by a chosen organizational ontology and in which at least the ontology-defined information may be stored and made accessible such as in one or more information retention elements such that a user—by entering a selection, for example, of a heading or subheading of the template report—may easily obtain the corresponding relevant information.”;
Examiner’s Note: the TSVETKOV-SOBLE combination now modifies the email response system of TSVETKOV to organize data using an ontology system as in SOBLE, where the ontology can relate to a domain like the medical field; for example, the TSVETKOV-SOBLE combination now uses the email response system of TSVETKOV in a hospital setting, where the templates of TSVETKOV are populated using the ontology of a medical domain as in SOBLE)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV and SOBLE as explained above. As disclosed by SOBLE, one of ordinary skill would have been motivated to do so because storing information with respect to an ontology enables the efficient display of information. (para. 0021).
Regarding Claim 2
TSVETKOV and SOBLE teach the method of claim 1 as explained above. TSVETKOV further teaches:
wherein the one or more templates are one or more artificial intelligence templates. (TSVETKOV, para. 0028: “For example, the template 104 is selected from a library of pre-generated templates that are each tailored for a different purpose, product, and/or target customer (e.g., based on specific characteristics of the customer or customer's business). The types of templates present in the template library and content of each template may vary considerably depending upon the nature of the business entity using the system 100, such as based on the types of products or services offered by the business and/or the types of customer communications that the business regularly conducts.”;
TSVETKOV, para. 0070: “In some implementations, the email template is an email previously composed by the first user. In other implementations, the email template is selected from a template library comprising templates crafted to serve as templates, such as that provide information on different products or services, target customers with different characteristics, etc.”
TSVETKOV, para. 0041: “This relevant contextual information 124 is ultimately inserted into the prompt 110 that is provided to the LLM 112.”;
Examiner’s Note: templates are retrieved for use from the template library and provided to the LLM, and thus are considered to be “artificial intelligence templates” because the templates are utilized by an LLM, which is a form of artificial intelligence)
Regarding Claim 11
TSVETKOV teaches:
A computer system for facilitating processing within a computing environment, the computer system comprising: a processor set; a set of at least one computer-readable storage medium; and program instructions, collectively stored in the set of at least one computer-readable storage medium for causing the processor set to perform the following computer operations including: (TSVETKOV, para. 0076: “ One or more applications 540 (e.g., the LLM context confiner 108 of FIG. 1 or any of its subcomponents) are loaded in the memory 504 and executed on the operating system 510 by the processing system 502. In some implementations, aspects of the LLM context confiner 108 of FIG. 1 are loaded into memory of different processing devices connected across a network.”)
The remaining limitations correspond to the method of claim 1, and therefore this claim is rejected for the same reasons explained above with respect to claim 1.
Regarding Claim 16
TSVETKOV teaches:
A computer program product for facilitating processing within a computing environment, the computer program product comprising: a set of at least one computer-readable storage medium; and program instructions, collectively stored in the set of at least one computer-readable storage medium for causing a processor set to perform the following computer operations including: (TSVETKOV, para. 0076: “ One or more applications 540 (e.g., the LLM context confiner 108 of FIG. 1 or any of its subcomponents) are loaded in the memory 504 and executed on the operating system 510 by the processing system 502. In some implementations, aspects of the LLM context confiner 108 of FIG. 1 are loaded into memory of different processing devices connected across a network.”)
The remaining limitations correspond to the method of claim 1, and therefore this claim is rejected for the same reasons explained above with respect to claim 1.
Claim 3, 12, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBLE and further in view of US 20190251165, hereinafter referenced as BACHRACH, and further in view of US 20240265269 A1, hereinafter referenced as CHEN.
Regarding Claim 3
TSVETKOV and SOBLE teach the method of claim 1 as explained above. However, TSVETKOV and SOBLE fail to explicitly teach:
wherein the one or more templates are created using a large language model that is tuned to isolate one or more chosen parts of the large language model and to prune one or more other parts from the large language model that are not chosen.
However, in a related field of endeavor (training global language models, see para. 0003), BACHRACH teaches and makes obvious:
wherein the one or more templates are created using a large language model (BACHRACH, para. 0085: “This training data may be used to train a predictive model, wherein the conversational agent is configured to apply the predictive model to messages within a text dialogue to predict a response template to use to respond to the messages.”;
Examiner’s Note: The TSVETKOV-SOBLE-BACHRACH combination now modifies TSVETKOV so that the LLM of TSVETKOV is used to actually create a response template as in BACHRACH)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, and BACHRACH as explained above. As disclosed by BACHRACH, one of ordinary skill would have been motivated to do so because a predictive model that uses templates “offers advantages over comparative neural ‘chatbot’ architectures, for example, “greater control may be applied and unintelligible outputs may be avoided.” (para. 0044).
However, TSVETKOV, SOBLE, and BACHRACH fail to explicitly teach:
...that is tuned to isolate one or more chosen parts of the large language model and to prune one or more other parts from the large language model that are not chosen.
However, in a related field of endeavor (training global language models, see para. 0003), CHEN teaches and makes obvious:
... that is tuned to isolate one or more chosen parts of the large language model and to prune one or more other parts from the large language model that are not chosen. (CHEN, para. 0068: “In some implementations, applying the global embedding mask to the global embedding matrix to generate the sparsified global language model can cause a subset of global weights, of the global language model, to be pruned.”;
CHEN, para. 0069: “In some versions of those implementations, the corresponding update generated locally at each of the computing devices of the population can be for only an additional subset of global weights, of the global language model, that are not pruned in applying the global embedding mask to the global embedding matrix.”;
Examiner’s Note: CHEN discloses that, for a language model, there is a subset of global weights that are pruned, and that there is a subset of global weights that is now pruned; the TSVETKOV-SOBLE-CHEN combination now prunes the language models of TSVETKOV using the teachings of CHEN)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, BACHRACH, and CHEN as explained above. As disclosed by CHEN, one of ordinary skill would have been motivated to do so in order to prune a global language model so that it includes fewer parameters than the original model, while still matching the precision and/or recall of the original global language model. (para. 0026). One of ordinary skill in the art would understand that a pruned language model would require fewer computing resources at inference time.
Claim 12 depends from claim 11 and claims a computer system that corresponds to the method of claim 3, and is therefore rejected for the same reasons explained with respect to claims 3 and 11.
Claim 17 depends from claim 16 and claims a computer program product that corresponds to the method of claim 3, and is therefore rejected for the same reasons explained with respect to claims 3 and 16.
Claims 4, 13, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBLE and further in view of US 20250117421 A1, hereinafter referenced as MCCURDY.
Regarding Claim 4
TSVETKOV and SOBLE teach the method of claim 1 as explained above. However, TSVETKOV and SOBLE fail to explicitly teach:
wherein the retrieving the one or more templates comprises: using a key generated based on a prompt of the request to select for retrieval at least one template of the one or more templates stored in the selected location.
However, in a related field of endeavor (language models, see paras. 0004-0005), MCCURDY teaches and makes obvious:
wherein the retrieving the one or more templates comprises: using a key generated based on a prompt of the request to select for retrieval at least one template of the one or more templates stored in the selected location. (MCCURDY, para. 0038: “As an additional example, query module 130 can extract one or more key terms from a user prompt and use those key terms to query vector database 180 and/or database 170 to retrieve one or more templates, forms, diagnostic checklists, etc.”;
Examiner’s Note: the TSVETKOV-SOBLE-MCCURDY combination now generates a key from key terms in the prompt as in MCCURDY, and uses such key to retrieve a template as in MCCURDY)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, and MCCURDY as explained above. As disclosed by MCCURDY, one of ordinary skill would have been motivated to do so in order to because MCCURDY teaches methods that “reduce the total human support technician time required to solve user technical problems.” (para. 0059).
Claim 13 depends from claim 11 and claims a computer system that corresponds to the method of claim 4, and is therefore rejected for the same reasons explained with respect to claims 4 and 11.
Claim 18 depends from claim 16 and claims a computer program product that corresponds to the method of claim 4, and is therefore rejected for the same reasons explained with respect to claims 4 and 16.
Claims 5, 14, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBLE and further in view of US 20250037340 A1, hereinafter referenced as SRINIVASA.
Regarding Claim 5
TSVETKOV and SOBLE teach the method of claim 1 as explained above. However, TSVETKOV and SOBLE fail to explicitly teach:
receiving requester feedback regarding the customized reply provided to the requester; and modifying one or more weights of the one or more templates based on the requester feedback, wherein the one or more weights are modified based on performing reinforcement learning context generation.
However, in a related field of endeavor (generative language models, see para. 0025), SRINIVASA teaches and makes obvious:
receiving requester feedback regarding the customized reply provided to the requester; and modifying one or more weights of the one or more templates based on the requester feedback, wherein the one or more weights are modified based on performing reinforcement learning context generation. (SRINIVASA, para. 0025: “In various embodiments, the facility uses various kinds of generative language models, invoked in various ways. In various such embodiments, the facility uses a stock or standard generative language model; uses a generative language model trained from scratch on materials generated or selected by the designers or operators of the facility and/or mental health experts, in some cases including the script templates or example scripts; uses a stock or standard generative language model that is further trained with materials generated or selected by designers or operators of the facility, in some cases including script templates or example scripts; uses standard or stock generative language models that are fine-tuned using materials generated or selected by the designers and/or operators of the facility. In various embodiments, this fine tuning includes such approaches as updating the model's embedding layers; updating the model's language modeling head; updating parameters of the model; prompt engineering; prompt-tuning or optimization; reinforcement learning from human feedback, such as pre-training a language model, gathering data and training a reward model, or fine-tuning the language model with reinforcement learning. In some embodiments, the facility submits materials generated or selected by the designers and/or operators of the facility as part of the prompt that the facility submits to the generative language model.”;
Examiner’s Note: the TSVETKOV-SOBLE-SRINIVASA combination now trains the LLMs of TSVETKOV using human feedback and reinforcement learning as in SRINISVASA to update the weights of the LLM of TSVETKOV)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, and SRINIVASA as explained above. As disclosed by SRINIVASA, one of ordinary skill would have been motivated to do so in order to because SRINIVASA teaches methods that utilize the feedback of mental health professionals to create and refine scripts using a language model to “address range of behavioral issues or scenarios.” (para. 0011).
Claim 14 depends from claim 11 and claims a computer system that corresponds to the method of claim 5, and is therefore rejected for the same reasons explained with respect to claims 5 and 11.
Claim 19 depends from claim 16 and claims a computer program product that corresponds to the method of claim 5, and is therefore rejected for the same reasons explained with respect to claims 5 and 16.
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBLE and further in view of US 20240086647, hereinafter referenced as RAMANUJAM.
Regarding Claim 6
TSVETKOV and SOBLE teach the method of claim 1 as explained above. TSVETKOV further teaches:
performing retrieval augmented generation to establish one or more context vectors that provide supporting information for generation of the customized reply to the request; and (TSVETKOV, para. 0050: “Upon receiving inputs including the recipient ID 216, the template 204, and/or the sender ID 218, the context mining tool 214 accesses stored CRM data 220 and mines contextual data that is associated with one or more of the inputs. In this case, the context mining tool 214 searches for customer profiles within the CRM data 220 that include the recipient ID 216.”;
TSVETKOV, para. 0055: “ By applying the above-described scoring and ranking logic to information items identified within the CRM data 220, the context mining tool 214 assembles and outputs relevant contextual information 224. As shown, this information includes a summary of a recent meeting including the parties that attended the meeting, the date of the meeting, a product discussed, highlights of the meeting, and action items that were identified during the meeting. This contextual information is provided as input to an LLM prompt generator 234, along with the template 204.”;
Examiner’s Note: context mining tool 214 searches CRM data 220 for contextual data (corresponding to recited “one or more context vectors” to provide additional information to the LLM)
establishing the one or more context vectors, the establishing comprising: retrieving selected data from one or more external data sources, ... , the knowledge data to be used to generate the customized reply; and (TSVETKOV, para. 0050: “Upon receiving inputs including the recipient ID 216, the template 204, and/or the sender ID 218, the context mining tool 214 accesses stored CRM data 220 and mines contextual data that is associated with one or more of the inputs. In this case, the context mining tool 214 searches for customer profiles within the CRM data 220 that include the recipient ID 216.”;
TSVETKOV, para. 0055: “ By applying the above-described scoring and ranking logic to information items identified within the CRM data 220, the context mining tool 214 assembles and outputs relevant contextual information 224. As shown, this information includes a summary of a recent meeting including the parties that attended the meeting, the date of the meeting, a product discussed, highlights of the meeting, and action items that were identified during the meeting. This contextual information is provided as input to an LLM prompt generator 234, along with the template 204.”;
Examiner’s Note: context mining tool 214 searches CRM data 220 (corresponding to recited “one or more external data sources”) for contextual data (corresponding to recited “one or more context vectors” to provide additional information to the LLM)
combining the selected data that is retrieved with parametric data encoded by the large language model to establish the one or more context vectors. (TSVETKOV, para. 0050: “Upon receiving inputs including the recipient ID 216, the template 204, and/or the sender ID 218, the context mining tool 214 accesses stored CRM data 220 and mines contextual data that is associated with one or more of the inputs. In this case, the context mining tool 214 searches for customer profiles within the CRM data 220 that include the recipient ID 216.”;
TSVETKOV, para. 0055: “ By applying the above-described scoring and ranking logic to information items identified within the CRM data 220, the context mining tool 214 assembles and outputs relevant contextual information 224. As shown, this information includes a summary of a recent meeting including the parties that attended the meeting, the date of the meeting, a product discussed, highlights of the meeting, and action items that were identified during the meeting. This contextual information is provided as input to an LLM prompt generator 234, along with the template 204.”;
Examiner’s Note: context mining tool 214 searches CRM data 220 (corresponding to recited “selected data” from “one or more external data sources”) for contextual data (corresponding to recited “one or more context vectors” to provide additional information to the LLM (e.g., the selected data is combined with the LLM)
However, TSVETKOV and SOBLE fail to explicitly teach:
... the selected data being knowledge data missing from a large language model used to create the one or more templates
However, in a related field of endeavor (language models, see para. 0035), RAMANUJAM teaches and makes obvious: (RAMANUJAM, para. 0042: “The section mapper 206 of the automated authoring engine exemplarily illustrated in FIG. 2, executes the section mapping algorithm for mapping the sections configured in the predefined CSR template with the content from the source documents as disclosed in the description of FIG. 5. In the execution of the section mapping algorithm, the section mapper 206 matches the sections defined in the predefined CSR template with the sections in the source documents uploaded by a user and utilizes the machine learning model 313 to obtain predictions on new data, for example, the content from the uploaded source documents.”;
Examiner’s Note: RAMANUJAM teaches that new data that has not been considered by the machine learning model 313 (corresponding to recited “knowledge data missing from a large language model”) can be used at inference; the TSVETKOV-SOBLE-RAMANUJAM combination now utilizes the new data of RAMANUJAM at inference with respect to the LLM of TSVETKOV such that new data is utilized with respect to the context data)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, and RAMANUJAM as explained above. As disclosed by RAMANUJAM, one of ordinary skill would have been motivated to do so in order to utilize new data to retrain machine learning models. (para. 0042).
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBLE and RAMANUJAM and further in view of US 20230245651 A1, hereinafter referenced as WANG.
Regarding Claim 7
TSVETKOV and SOBLE and RAMANUJAM teach the method of claim 6 as explained above. However, TSVETKOV and SOBLE and RAMANUJAM fail to explicitly teach:
wherein the parametric data includes one or more large language model weights and one or more fine-tuned weight adjustments.
However, in a related field of endeavor (language modeling, see para. 0136), WANG teaches and makes obvious:
wherein the parametric data includes one or more large language model weights and one or more fine-tuned weight adjustments. (WANG, para. 0244: “Fine-tuning: Language models can be fine-tuned for specific domains or topics to improve its ability to generate contextually relevant responses for those domains. This involves training the model on a dataset of conversational data in the target domain, which allows it to learn the specific language and patterns of conversation in that domain.”;
Examiner’s Note: the TSVETKOV-SOBLE-RAMANUJAM-WANG combination now fine-tunes the LLM of TSVETKOV as taught by WANG in order to both train and fine-tune weights)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, RAMANUJAM, and WANG as explained above. As disclosed by WANG, one of ordinary skill would have been motivated to do so in order to “save time and resources by leveraging pre-existing knowledge and models.” (para. 0081).
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBLE and WANG.
Regarding Claim 8
TSVETKOV and SOBLE teach the method of claim 1 as explained above. However, TSVETKOV and SOBLE fail to explicitly teach:
wherein the customized reply to the request is to be received within a predefined amount of time.
However, in a related field of endeavor (language modeling, see para. 0136), WANG teaches and makes obvious:
wherein the customized reply to the request is to be received within a predefined amount of time. (WANG, para. 0342: “If no response is detected within the time limit, the AI system iteratively searches for another entity with conversational capability, ensuring a suitable conversational partner is found 1404.”;
WANG, para. 0345: “The AI system waits for a response from the selected entity within a limited period of time. If a positive response indicating conversational capability is received, the AI system proceeds with further interaction to establish the user’s identity and their role in the healthcare setting. (6) Iterative search: If no response is received within the time limit, the AI system iteratively searches for another entity with potential conversational capability in the environment, repeating the validation process until a suitable conversational partner is found.”;
Examiner’s Note: the TSVETKOV-SOBLE-WANG combination now has a time limit with respect to when a response can be transmitted and received as in WANG)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, and WANG as explained above. As disclosed by WANG, one of ordinary skill would have been motivated to do so in order to “save time and resources by leveraging pre-existing knowledge and models.” (para. 0081).
Claims 9, 15, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBEL and further in view of US 20240275730 A1, hereinafter referenced as BLACK.
Regarding Claim 9
TSVETKOV and SOBLE teach the method of claim 1 as explained above. However, TSVETKOV and SOBLE fail to explicitly teach:
wherein the receiving the request comprises receiving a plurality of requests for a plurality of customized replies at a particular request rate, the plurality of requests including the request.
However, in a related field of endeavor (distributed computing, see para. 0087), BLACK teaches and makes obvious:
wherein the receiving the request comprises receiving a plurality of requests for a plurality of customized replies at a particular request rate, the plurality of requests including the request. (BLACK, para. 0027: “For example, when rate limiting server 140 receives requests at a rate lower than the rate limit (e.g., for a given path), rate limiting server 140 is able to direct its client to fulfill all requests and does not have the process of rate limiting in place.”;
Examiner’s Note: the TSVETKOV-SOBLE-BLACK combination now uses a rate limit (as in BLACK) to monitor the recited particular request rate with respect to TSVETKOV)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, and BLACK as explained above. As disclosed by BLACK, one of ordinary skill would have been motivated to do so in order to utilize rate limiting to “limit the number of requests sent or received in a network for a given time frame” and to “reduce strain on the network or prevent malicious attacks.” (para. 0002).
Claim 15 depends from claim 11 and claims a computer system that corresponds to the method of claim 9, and is therefore rejected for the same reasons explained with respect to claims 9 and 11.
Claim 20 depends from claim 16 and claims a computer program product that corresponds to the method of claim 9, and is therefore rejected for the same reasons explained with respect to claims 9 and 16.
Claims 10 is rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBEL and BACHRACH and further in view of US 20240427810 A1, hereinafter referenced as TISHBI.
Regarding Claim 10
TSVETKOV and SOBLE teach the method of claim 1 as explained above. However, TSVETKOV and SOBLE fail to explicitly teach:
wherein the one or more templates are generated using a large language model, the large language model trained using a reduced set of computing resources.
However, in a related field of endeavor (training global language models, see para. 0003), BACHRACH teaches and makes obvious:
wherein the one or more templates are generated using a large language model (BACHRACH, para. 0085: “This training data may be used to train a predictive model, wherein the conversational agent is configured to apply the predictive model to messages within a text dialogue to predict a response template to use to respond to the messages.”;
Examiner’s Note: The TSVETKOV-SOBLE-BACHRACH combination now modifies TSVETKOV so that the LLM of TSVETKOV is used to actually create a response template as in BACHRACH)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, and BACHRACH as explained above. As disclosed by BACHRACH, one of ordinary skill would have been motivated to do so because a predictive model that uses templates “offers advantages over comparative neural ‘chatbot’ architectures, for example, “greater control may be applied and unintelligible outputs may be avoided.” (para. 0044).
However, TSVETKOV, SOBLE, and BACHRACH fail to explicitly teach:
the large language model trained using a reduced set of computing resources.
However, in a related field of endeavor (large language models, see para. 0001), TISHBI teaches and makes obvious:
the large language model trained using a reduced set of computing resources. (TISHBI, para. 0036: “On advantage of the present disclosure is providing a method and system which include a check and balance for an answer received as an output from an LLM which is trained for a cybersecurity solution. For example, by fine-tuning a pretrained LLM to generate answers based on a semantic data layer, token usage is reduced, thereby decreasing resources required to train and process the neural network of the LLM.”;
Examiner’s Note: the TSVETKOV-SOBLE-BACHRACH-TISHBI combination now trains the LLM of TSVETKOV using reduced resources as in TISHBI)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, BACHRACH, and TISHBI as explained above. As disclosed by TISHBI, one of ordinary skill would have been motivated to do so because TISHBI teaches techniques for reducing tokenization in the LLM. (para. 0035).
Claim 21 is rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBLE and BACHRACH.
Regarding Claim 21
TSVETKOV teaches:
A computer-implemented method of facilitating processing within a computing environment, the computer-implemented method comprising: (TSVETKOV, para. 0003: “According to one implementation, a method is disclosed for using a large language model (LLM) to generate a customized email according to a methodology that reduces the likelihood of the customized email including hallucinated facts or undesired (e.g., irrelevant) personal identifying information (PII).”;
TSVETKOV, para. 0023: “FIG. 1 illustrates an example system 100 that uses generative AI to create customized email content according to a methodology that lowers the likelihood of the model outputs including hallucinated facts or sensitive personal identity information (PII) that may pose a legal risk to the creating entity. Within the system 100, a user provides input to an email application 102 to initiate generation of a customized email. The email application 102 may include software components that execute locally on a user device, remotely in a cloud platform, or some combination thereof.”;
TSVETKO, para. 0036: “In some implementations, the user's compute environment includes tool(s) that collect and store meeting information 130, such as a transcripts or summaries of different meetings. For example, Microsoft's Sales Copilot® is an add-on to Microsoft Teams®, which business users frequently use to conduct web-based calls.”;
Examiner’s Note”: Fig. 1 illustrates the computing environment for processing emails)
receiving from a requester a request over one or more networks of the computing environment, the request related to a selected event; receiving from a requester a request for a customized reply, the receiving the request using one or more networks of the computing environment; (TSVETKOV, para. 0014: “For example, a natural language processing (NLP) model can be trained on customer data and outgoing customer emails pertaining to different products and services and then asked to generate an email for a specific customer (or for a customer with a specific characteristics) about a particular product or service.”;
TSVTETKOV, para. 0023: “Within the system 100, a user provides input to an email application 102 to initiate generation of a customized email”;
TSVETKOV, para. 0024: “In various implementations, aspects of the LLM context confiner 108 reside locally on the device of the end user, remotely in the cloud, or some combination thereof.”
TSVETKOV, para. 0034: “Within the email application 102, the user also identifies a recipient of the customized email that is to be drafted, such as by selecting a contact card for the recipient, typing the recipient's name, email address, or other form of recipient identifier 116. A context mining tool 114 uses the recipient identifier 116 along with an identifier of the user composing the message (e.g., sender ID 118) and the PII-filtered template 132 to mine contextual data 128 from various data sources that is potentially relevant to the draft email.”;
TSVETKOV, para. 0076: “One or more applications 540 (e.g., the LLM context confiner 108 of FIG. 1 or any of its subcomponents) are loaded in the memory 504 and executed on the operating system 510 by the processing system 502. In some implementations, aspects of the LLM context confiner 108 of FIG. 1 are loaded into memory of different processing devices connected across a network.”;
TSVETKOV, para. 0077: “Additionally, the applications 540 may receive input from one or more remote devices, such as remotely-located servers or smart devices, by communicating with such devices over a wired or wireless network using more communication transceivers 530 and an antenna 532 to provide network connectivity (e.g., a mobile phone network, Wi-Fi®, Bluetooth®). The processing device 500 may also include one or more storage devices 528 (e.g., non-volatile storage) for storing email templates, emails, user profile data, meeting summaries, etc. Other configurations may also be employed.”;
Examiner’s Note: the user (corresponding to recited “requestor”) requests a customized email (corresponding to recited “customized reply”, where the customized email is a reply in response to the request by the user), and the various components of system 100 may be spread across a computer network such that the request for an email can come from a user using a remote device and communicating with a server hosting the email application, and where the product/service selected by the user (and the user’s use thereof) corresponds to the recited “selected event”)
retrieving one or more templates stored in a selected location, ... (TSVETKOV, para. 0028: “For example, the template 104 is selected from a library of pre-generated templates that are each tailored for a different purpose, product, and/or target customer (e.g., based on specific characteristics of the customer or customer's business). The types of templates present in the template library and content of each template may vary considerably depending upon the nature of the business entity using the system 100, such as based on the types of products or services offered by the business and/or the types of customer communications that the business regularly conducts.”;
TSVETKOV, para. 0070: “In some implementations, the email template is an email previously composed by the first user. In other implementations, the email template is selected from a template library comprising templates crafted to serve as templates, such as that provide information on different products or services, target customers with different characteristics, etc.”
Examiner’s Note: templates are retrieved for use from the template library (corresponding to recited “selected location”)
... the large language model shielded from direct request traffic; (TSVTETKOV, para. 0023: “Within the system 100, a user provides input to an email application 102 to initiate generation of a customized email”;
Examiner’s Note: As shown in Fig. 1, the LLM 112 does not directly receive requests from a user, because the user initiates requests through e-mail application 102 and does not have direct access to the LLM)
inputting information in the one or more templates to provide one or more populated templates; (TSVETKOV, para. 0042: “ The relevant contextual information 124 is provided as input to an LLM prompt generator 134 along with “a template body,” which refers to either the PII-filtered template 132 or the template 104 (e.g., in use cases that do not include PII filtering) along with metadata of the template, if any exists. Based on the relevant contextual information 124 and the template body, the LLM prompt generator 134 generates the prompt 110. The prompt 110 includes both the template body and the relevant contextual information 124 and further includes an instruction asking the LLM 112 to draft an email based on the relevant contextual information 124 following the template 104.”;
TSVETKOV, para. 0062: “Here, the customized email 230 includes an intro portion 236 that includes stylized language from the template 204. This intro portion is followed by a product description portion 238 that includes detail from the template with product names populated from the relevant contextual information 224.”;
Examiner’s Note: as shown in the example customized email of 230, the LLM inputs contextual information into the template in order to populate the template)
generating a customized reply to the request based on the one or more populated templates; and (TSVETKOV, para. 0061: “ FIG. 2C illustrates a customized email 230 that is output by an LLM 232 in response to receiving the prompt 212, as described above, which is generated per the operations described with respect to FIGS. 2A and 2B.”;
TSVETKOV, para. 0062: “Here, the customized email 230 includes an intro portion 236 that includes stylized language from the template 204. This intro portion is followed by a product description portion 238 that includes detail from the template with product names populated from the relevant contextual information 224. A third paragraph 240 following the product description portion 238 includes additional context from the relevant contextual information 224 that references the recent meeting between the two parties and also summarizes the “follow-ups” identified in the relevant contextual information 224 as affirmative actions that are being taken. The email includes links 242 from the template 204, and the remaining portion 244 of the email following the links 242 also includes text that is taken, verbatim, from the template 204. In scenarios where the template 204 was an email previously drafted by the user, the user's own writing style is therefore captured in the customed email 230 generated by the LLM 232.”;
Examiner’s Note: the LLM uses the populated template to customize the generated email as shown in Fig. 2C)
providing to the requester the customized reply that is generated based on the one or more populated templates. (TSVETKOV, para. 0074: “A returning operation 412 returns the customized email to the email application for display in an email composition window presented on a user interface.”)
However, TSVETKOV fails to explicitly teach:
the one or more templates retrieved based, at least in part, on one or more ontologies constructed for the selected event, the one or more templates created using a large language model trained to create the one or more templates based on the selected event,
However, in a related field of endeavor (a clinical ontology with respect to medical report templates, see para. 0002), SOBLE teaches and makes obvious:
the one or more templates retrieved based, at least in part, on one or more ontologies constructed for the selected event (SOBLE, para. 0020: “The system and methods of the present invention also permits a user to produce a report template according to a clinical ontology chosen by the user. By the selection of a topic, heading, and subheading of the template, a user may obtain access to the relevant information organized by the system and thereby be able to prepare a medical report and complete a clinical study for a subject more efficiently.”;
SOBLE, para. 0021: “ By organizing the template and the information according to the same chosen clinical ontology, an efficient display of the information may be formed. The efficient display of information formed through the use of the present invention is termed also a “categorical display”. The categorical display may be distributed through a network to one or more display outputs to which are connected one or more displays on which the categorical display may be shown to one or more users. A “categorical display” for purposes of this application is one in which at least certain or all the information that has been obtained for a subject is organized and made accessible according to the same clinical ontology on which the template report is based so that a user may be provided with the corresponding appropriate information simply by selecting the heading or subheading of the template report through the configurable display. Through the use of the template report and the categorical display tool, the user can conduct and report the results for a clinical study more efficiently.”;
SOBLE, para. 0029: “Embodiments of the present invention may include an information resource that facilitates such processing by a chosen organizational ontology and in which at least the ontology-defined information may be stored and made accessible such as in one or more information retention elements such that a user—by entering a selection, for example, of a heading or subheading of the template report—may easily obtain the corresponding relevant information.”;
Examiner’s Note: the TSVETKOV-SOBLE combination now modifies the email response system of TSVETKOV to organize data using an ontology system as in SOBLE, where the ontology can relate to a domain like the medical field; for example, the TSVETKOV-SOBLE combination now uses the email response system of TSVETKOV in a hospital setting, where the templates of TSVETKOV are populated using the ontology of a medical domain as in SOBLE)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV and SOBLE as explained above. As disclosed by SOBLE, one of ordinary skill would have been motivated to do so because storing information with respect to an ontology enables the efficient display of information. (para. 0021).
However, TSVETKOV and SOBLE fail to explicitly teach:
the one or more templates created using a large language model trained to create the one or more templates based on the selected event,
However, in a related field of endeavor (training global language models, see para. 0003), BACHRACH teaches and makes obvious:
the one or more templates created using a large language model trained to create the one or more templates based on the selected event (BACHRACH, para. 0085: “This training data may be used to train a predictive model, wherein the conversational agent is configured to apply the predictive model to messages within a text dialogue to predict a response template to use to respond to the messages.”;
Examiner’s Note: The TSVETKOV-SOBLE-BACHRACH combination now modifies TSVETKOV so that the LLM of TSVETKOV is used to actually create a response template as in BACHRACH)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, and BACHRACH as explained above. As disclosed by BACHRACH, one of ordinary skill would have been motivated to do so because a predictive model that uses templates “offers advantages over comparative neural ‘chatbot’ architectures, for example, “greater control may be applied and unintelligible outputs may be avoided.” (para. 0044).
Claim 22 is rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBLE and BACHRACH and further in view of WANG.
Regarding Claim 22
TSVETKOV, SOBLE, and BACHRACH teach the method of claim 21 as explained above. However, TSVETKOV, SOBLE, and BACHRACH fail to explicitly teach:
wherein the providing the customized reply is performed within a selected amount of time
However, in a related field of endeavor (language modeling, see para. 0136), WANG teaches and makes obvious:
wherein the providing the customized reply is performed within a selected amount of time. (WANG, para. 0342: “If no response is detected within the time limit, the AI system iteratively searches for another entity with conversational capability, ensuring a suitable conversational partner is found 1404.”;
WANG, para. 0345: “The AI system waits for a response from the selected entity within a limited period of time. If a positive response indicating conversational capability is received, the AI system proceeds with further interaction to establish the user’s identity and their role in the healthcare setting. (6) Iterative search: If no response is received within the time limit, the AI system iteratively searches for another entity with potential conversational capability in the environment, repeating the validation process until a suitable conversational partner is found.”;
Examiner’s Note: the TSVETKOV-SOBLE-BACHRACH-WANG combination now has a time limit with respect to when a response can be transmitted and received as in WANG, where such time limit is “selected” by the designer of the AI system)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, BACHRACH, and WANG as explained above. As disclosed by WANG, one of ordinary skill would have been motivated to do so in order to “save time and resources by leveraging pre-existing knowledge and models.” (para. 0081).
Claim 23 is rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBEL and BACHRACH and further in view of TISHBI
Regarding Claim 23
TSVETKOV, SOBLE, and BACHRACH teach the method of claim 21 as explained above. However, TSVETKOV, SOBLE, and BACHRACH fail to explicitly teach:
wherein the large language model is trained using a reduced set of computing resources.
However, in a related field of endeavor (large language models, see para. 0001), TISHBI teaches and makes obvious:
the large language model trained using a reduced set of computing resources. (TISHBI, para. 0036: “On advantage of the present disclosure is providing a method and system which include a check and balance for an answer received as an output from an LLM which is trained for a cybersecurity solution. For example, by fine-tuning a pretrained LLM to generate answers based on a semantic data layer, token usage is reduced, thereby decreasing resources required to train and process the neural network of the LLM.”;
Examiner’s Note: the TSVETKOV-SOBLE-BACHRACH-TISHBI combination now trains the LLM of TSVETKOV using reduced resources as in TISHBI)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, BACHRACH, and TISHBI as explained above. As disclosed by TISHBI, one of ordinary skill would have been motivated to do so because TISHBI teaches techniques for reducing tokenization in the LLM. (para. 0035).
Claim 24 is rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBEL and BACHRACH and further in view of BLACK.
Regarding Claim 24
TSVETKOV, SOBLE, and BACHRACH teach the method of claim 1 as explained above. However, TSVETKOV, SOBLE, and BACHRACH fail to explicitly teach:
wherein the receiving the request comprises receiving a plurality of requests at a particular request rate, the plurality of requests including the request.
However, in a related field of endeavor (distributed computing, see para. 0087), BLACK teaches and makes obvious:
wherein the receiving the request comprises receiving a plurality of requests at a particular request rate, the plurality of requests including the request. (BLACK, para. 0027: “For example, when rate limiting server 140 receives requests at a rate lower than the rate limit (e.g., for a given path), rate limiting server 140 is able to direct its client to fulfill all requests and does not have the process of rate limiting in place.”;
Examiner’s Note: the TSVETKOV-SOBLE-BACHRACH-BLACK combination now uses a rate limit (as in BLACK) to monitor the recited particular request rate with respect to TSVETKOV)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, BACHRACH, and BLACK as explained above. As disclosed by BLACK, one of ordinary skill would have been motivated to do so in order to utilize rate limiting to “limit the number of requests sent or received in a network for a given time frame” and to “reduce strain on the network or prevent malicious attacks.” (para. 0002).
Claim 25 is rejected under 35 U.S.C. 103 as being unpatentable over TSVETKOV in view of SOBLE and BACHRACH and further in view of WANG.
Regarding Claim 25
TSVETKOV and SOBLE and BACHRACH teach the method of claim 21 as explained above. However, TSVETKOV and SOBLE and BACHRACH fail to explicitly teach:
wherein the large language model that is trained is tuned using one or more tuning techniques.
However, in a related field of endeavor (language modeling, see para. 0136), WANG teaches and makes obvious:
wherein the large language model that is trained is tuned using one or more tuning techniques. (WANG, para. 0244: “Fine-tuning: Language models can be fine-tuned for specific domains or topics to improve its ability to generate contextually relevant responses for those domains. This involves training the model on a dataset of conversational data in the target domain, which allows it to learn the specific language and patterns of conversation in that domain.”;
Examiner’s Note: the TSVETKOV-SOBLE-BACHRACH-WANG combination now fine-tunes the LLM of TSVETKOV as taught by WANG)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of TSVETKOV, SOBLE, BACHRACH, and WANG as explained above. As disclosed by WANG, one of ordinary skill would have been motivated to do so in order to “save time and resources by leveraging pre-existing knowledge and models.” (para. 0081).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 20240061866 A1 (Sehon). “The output of the model (e.g., model 406) may be used to determine relevant ontologies based on the user action. Furthermore, the ontologies determined by the model (e.g., model 406) may be included in a template presented to a user wherein the user may select data types for a new data asset type.” (para. 0048).
US 12405945 B2 (Zhang). “The object-centric data model is centered on the notion of data objects and properties of the data objects. Furthermore, the object-centric data model is based on an ontology that defines hierarchical object types and property types. For example, a data object in the body of data may have the hierarchical object type “Employee” and have a “Name” property, a “Title” property,” and a “Salary” property. The ontology may define the “Employee” object type as a child object type of the “Person” object type. Thus, the “Employee” object is also a “Person” object according to the object-centric data model.” (col. 2, lines 44-55). “Selecting a particular search template may cause generating a user interface 300 that includes a plurality of search fields. For example, FIG. 5A depicts an example plurality of search fields 500A-N provided to the user in response to selecting a particular search template. In an embodiment, a search field may accept a keyword and/or a property value 122 as input.” (col. 6, lines 24-30).
US 20250158940 A1 (Jin). “With the media and user insight, the reply system selects appropriate prompt templates and populates them with the extracted details to prepare structured prompts tailored to the context. These prompts are sent to large language models optimized for conversational text generation. The AI systems analyze the prompts and return numerous suggested replies relevant to the media. The reply system applies quality checks and filtering to select and display the best responses to a user of the client device 106.” (para. 0053).
US 20200133691 A1 (Goenka). “Alternatively and/or additionally, a second plurality of requests received from the first client device during the third activity session may be monitored and/or analyzed. For example, a second quantity of requests of the second plurality of requests, a second request-rate at which requests of the second plurality of requests are received from the first client device, etc. may be determined based upon the second activity. For example, a second measure of requesting activity may be generated based upon the second quantity of requests of the second plurality of requests and/or the second request-rate at which requests of the second plurality of requests are received from the first client device. In some examples, during the third activity session, the second measure of requesting activity may be updated periodically and/or responsive to detecting activity associated with the second activity.” (para. 0080).
Thenmalar, S., and T. V. Geetha. "Automatic Generation of Templates using Ontology." Proceedings of the Third International Symposium on Women in Computing and Informatics. 2015. “Our goal is to generate templates automatically from the domain corpus using domain ontology.” (p. 668, section 1).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MICHAEL C LEE whose telephone number is (571)272-4933. The examiner can normally be reached M-F 12:00 pm - 8:00 pm ET.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Omar Fernandez Rivas can be reached at 571-272-2589. 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.
/MICHAEL C. LEE/Examiner, Art Unit 2128
Prosecution Insights