DETAILED ACTION
This examination is in response to the communication filed on 11/07/2024. Claims 1-28 are currently pending, where claims 1 and 15 are independent.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 02/20/2025 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
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 1-11 and 15-25 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. Independent claims 1 and 15 recite “receiving a task prompt…”, “identifying, based on the task prompt, a context of the user input”, “determining…a user correction prompt… ”, “providing…the task prompt…” and “providing the personalized response to the user input….”
The limitations of “receiving…”, “identifying…”, “determining…”, “providing…” and “providing…” as drafted, are a process that, under a broadest reasonable interpretation, covers the abstract idea of “mental processes” because they cover concepts performed in the human mind, including observation, evaluation, judgement and opinion. See MPEP 2106.04(a)(2). That is, other than reciting “an LLM”, “a user device”, a “data processing processor” (claim 15), and “memory” (claim 15), nothing in the claimed elements preclude the steps from practically being performed by a person
receiving a task prompt representative of a user input from a user, the task prompt specifying a task for a large language model (LLM) to perform responsive to the user input (e.g., by the person receiving a written or audible request from a user);
identifying, based on the task prompt, a context of the user input (e.g., by the person determining what task the user is requesting );
determining, based on the context of the user input, a user correction prompt comprising one or more user changes made by the user to one or more prior outputs of the LLM (e.g., by the person reviewing a log or transcript of previous requests to determine user feedback/corrections );
providing, as input to the LLM, the task prompt conditioned on the user correction prompt to cause the LLM to generate a personalized response to the user input (e.g., by the person providing the correction prompt, e.g., in written or audible form, for processing by an LLM ); and
providing the personalized response to the user input for output associated with the user (e.g., by the user providing a written response to the user request).
This judicial exception is not integrated into a practical application because the additional elements of “an LLM”, “a user device”, a “data processing processor” (claim 15), and “memory” (claim 15), are all recited at a high-level of generality, and ¶[0036] of the Specification describes the use of a general-purpose processor, e.g., CPU. 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. In addition, the LLM is recited as an intended use or source of the user feedback/correction and not as an active step in the claimed process. Thus, the claims as a whole are directed to an abstract idea (Step 2A, prong two).
Claims 1 and 15 do not include any additional elements that are sufficient to amount to significantly more than the judicial exception because, as discussed above with respect to integration of the abstract idea into a practical application, the additional elements of “an LLM”, “a user device”, a “data processing processor” (claim 15), and “memory” (claim 15) amount to no more than mere instructions to apply the exception using generic computer components. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept (Step 2B).
With respect to dependent claims 2-4, 8, 16-18 and 22, these claims are directed the criteria for determining the context of the user input. These limitations also relate to the abstract idea of “mental processes.” That is nothing in the claimed elements preclude the steps from practically being performed by a person using the recited criteria when identifying the task being requested by the user request. No additional elements are present.
With respect to dependent claims 5-7, 19-21 and 10, these claims are directed the user request being in a audio form and the task being speech recognition. These limitations also relate to the abstract idea of “mental processes.” That is nothing in the claimed elements preclude the steps from practically being performed by a person receiving an audible user request and transcribing it into a written request. No additional elements are present.
With respect to dependent claims 9-11 and 23-25, these claims are directed the additional step of applying weights to the user’s previous changes and the criteria for applying said weights. These limitations also relate to the abstract idea of “mental processes.” That is nothing in the claimed elements preclude the steps from practically being performed by a person ranking the transcript/log of previous user changes using the recited criteria. No additional elements are present.
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 non-obviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 5-8, 12-15, 19-22 and 26-28 are rejected under 35 U.S.C. 103 as being unpatentable over Madaan et al. “MemPrompt: Memory-assisted Prompt Editing with User Feedback” arXiv:2201.06009v7 [cs.CL] 18 Feb 2023; herein “Madaan” further in view of Kaufman et al. (WO 2022/192797 A1; herein “Kaufman”).
Regarding claim 1 and 15, Madaan teaches a computer-implemented method executed on data processing hardware and a system comprising data processing hardware and memory hardware in communication with the data processing hardware, that causes the data processing hardware to perform operations comprising:
receiving a task prompt representative of a user input from a user, the task prompt specifying a task for a large language model (LLM) to perform responsive to the user input (Figure 2, user input x and page 3, second column teaches “…given an input x, we prompt the model to generate an output y…”);
identifying, based on the task prompt, a context of the user input (Fig. 2, output u and Pag 3, first column, section 3.1 teaches “…given an input x, a model generates an output y and a sentence u expressing its understanding of the task.” u is interpreted as the context of the user input x, i.e., what task the user is requesting);
determining, based on the context of the user input, a user correction prompt comprising one or more user changes made by the user to one or more prior outputs of the LLM (Figure 2, M(x) and Page 3, second column teaches “Given a new query, MemPrompt uses fb from similar, prior queries to enrich the(few-shot) prompt p. In addition to the input x, MemPrompt retrieves a fb if a question similar to x has been asked before” );
providing, as input to the LLM, the task prompt conditioned on the user correction prompt to cause the LLM to generate a personalized response to the user input (Figure 2, input x + M(x) input the GPT-3 model and page 6, second column teaches “…If a similar question (e.g., what is akin to pretty?) is asked later by the same or a different user, the corresponding feedback (I wanted a synonym) is attached with the question to generate the answer”); and
providing the personalized response to the user input
Although the system taught by Madaan inherently requires a processing device, Madaan is silent regarding the architecture/structure of the processing system. Therefore, Madaan fails to explicitly disclose that the personalized response is provided for output from a user device associated with the user.
Kaufman teaches a context speech-to-text system which utilizes a machine learning based on user feedback/correction to enhance accuracy (Kaufman ¶[012]). More specifically, Kaufman, ¶¶[035]-[036], teaches “the contextual STT platform 311 interfaces with the knowledge repository 305 and enables users to submit user data 312, which may include recorded or live-streamed audio as well as user feedback information. The STT platform 311 also includes an STT engine 314 that implements the software functions and logic to accomplish the various tasks and operations detailed here…a machine learning (ML) facility 318 may be implemented within the STT platform 3111 which analyzes the knowledge repository 305 as well as the user data 312 to further refine the contextual vocabulary 316. User data 312 includes audio recordings as well as user feedback, such as user feedback concerning errors identified in any transcripts of those audio recordings” In addition, Fig. 2 and ¶[029] teaches “remote server 250 may receive an audio file and one or more keywords from computing device 220. The remote server 250 may identify one or more speech sounds within the audio file…perform the cloud-based speech recognition technique on the audio file, and return one or more words identified within the audio file to computer device 220” Computing device 220 is interpreted as a user device. Therefore, Kaufman teaches the personalized response is provided for output from a user device associated with the user.
Madaan differs from the claimed invention, as defined by claims 1 and 15, in that Madaan fails to specifically disclose that the task input and resulting output is provided via a user device. Machine learning system which utilize client/server type architecture are known in the art as evidenced by Kaufman. Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to have implemented the memory-assisted prompt editing taught by Madaan in a client/server type architecture as taught by Kaufman in order to “improve the accuracy, efficiency and impact of knowledge management for teams of people and individuals” by allowing the “system to evaluate documents and things used by an enterprise, which may have its own industry-specific lexicon, to identify words and phrases that are more prevalently used by that enterprise.” (Kaufman, ¶[005])
Regarding claims 5 and 19, the combination of Madaan and Kaufman teaches all of the elements of claims 1 and 15 (see detailed element mapping above). In addition, Kaufman further teaches the user input comprises audio data characterizing an utterance spoken by the user (Fig. 2 and ¶[029] teaches “remote server 250 may receive an audio file and one or more keywords from computing device 220. The remote server 250 may identify one or more speech sounds within the audio file…”); and
the task prompt representative of the user input comprises a speech recognition representation of the utterance (Fig. 2 and ¶[029] teaches “…The remote server 250 may identify one or more speech sounds within the audio file…perform the cloud-based speech recognition technique on the audio file, and return one or more words identified within the audio file to computer device 220”).
Madaan differs from the claimed invention, as defined by claims 5 and 19, in that Madaan fails to specifically disclose that the task input and resulting output is a speech to text process. Speech to text systems which utilizes user feedback to improve the process are well known in as evidenced by Kaufman. Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to have implemented the memory-assisted prompt editing taught by Madaan in an context speech to text system as taught by Kaufman in order to “improve the accuracy, efficiency and impact of knowledge management for teams of people and individuals” by allowing the “system to evaluate documents and things used by an enterprise, which may have its own industry-specific lexicon, to identify words and phrases that are more prevalently used by that enterprise.” (Kaufman, ¶[005])
Regarding claims 6 and 20, the combination of Madaan and Kaufman teaches all of the elements of claims 5 and 19 (see detailed element mapping above). In addition, Madaan further teaches the one or more user changes comprise corrections made by the user to prior transcriptions generated by the LLM (¶[036] teaches “User data 312 includes audio recordings as well as user feedback, such as user feedback concerning errors identified in any transcripts of those audio recordings”).
Madaan differs from the claimed invention, as defined by claims 6 and 20, in that Madaan fails to specifically disclose that the task input and resulting output is a speech to text process. Speech to text systems which utilizes user feedback to improve the process are well known in as evidenced by Kaufman. Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to have implemented the memory-assisted prompt editing taught by Madaan in an context speech to text system as taught by Kaufman in order to “improve the accuracy, efficiency and impact of knowledge management for teams of people and individuals” by allowing the “system to evaluate documents and things used by an enterprise, which may have its own industry-specific lexicon, to identify words and phrases that are more prevalently used by that enterprise.” (Kaufman, ¶[005])
Regarding claims 7 and 21, the combination of Madaan and Kaufman teaches all of the elements of claims 5 and 19 (see detailed element mapping above). In addition, Madaan further teaches the speech recognition representation comprises at least one of:
an audio encoding of the audio data characterizing the utterance, the audio encoding output by an audio encoder of a speech recognition model (the “or” makes this element optional);
a list of speech recognition hypotheses for the utterance output by the speech recognition model (the “or” makes this element optional); or
a transcription of the utterance output by the speech recognition model (¶[040] teaches “the remote STT service 360 returns one or more proposed transcripts of the audio recording to the STT engine 314”).
Madaan differs from the claimed invention, as defined by claims 7 and 21, in that Madaan fails to specifically disclose that the task input and resulting output is a speech to text process. Speech to text systems which utilizes user feedback to improve the process are well known in as evidenced by Kaufman. Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to have implemented the memory-assisted prompt editing taught by Madaan in an context speech to text system as taught by Kaufman in order to “improve the accuracy, efficiency and impact of knowledge management for teams of people and individuals” by allowing the “system to evaluate documents and things used by an enterprise, which may have its own industry-specific lexicon, to identify words and phrases that are more prevalently used by that enterprise.” (Kaufman, ¶[005])
Regarding claims 8 and 22, the combination of Madaan and Kaufman teaches all of the elements of claims 1 and 15 (see detailed element mapping above). In addition, Madaan further teaches the user correction prompt is configured to guide the LLM to generate the personalized response while parameters of the LLM are held fixed (Page 1, Abstract teaches “Our goal is to allow users to correct such errors directly through interaction – without retraining” and Figure 1 caption teaches “Our approach is simple because only the prompt needs to be updated with the retrieved relevant feedback, and no retraining is necessary”).
Regarding claims 12 and 26, the combination of Madaan and Kaufman teaches all of the elements of claims 1 and 15 (see detailed element mapping above). In addition, Kaufman further teaches the LLM executes on a remote computing system in communication with the data processing hardware via a network (Fig. 3, Remote STT service 360 and ¶[033] teaches “remote STT service 360 is a cloud-based Speech-To-Text service”); and
providing the task prompt conditioned on the user correction prompt as input to the LLM comprises transmitting, from the data processing hardware to the remote computing system via the network, the task prompt conditioned on the user correction prompt (¶[029] teaches “remote server 250 may receive an audio file and one or more keywords from computing device 220. The remote server 250 may identify one or more speech sounds within the audio file…perform the cloud-based speech recognition technique on the audio file, and return one or more words identified within the audio file to computer device 220” ).
Madaan differs from the claimed invention, as defined by claims 12 and 26, in that Madaan fails to specifically disclose that the LLM executes on a remote computing system. Machine learning system which utilize client/server type architecture wherein the cloud-based services are provided via remote computing system are known in the art as evidenced by Kaufman. Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to have implemented the memory-assisted prompt editing taught by Madaan in a client/server type architecture as taught by Kaufman in order to “improve the accuracy, efficiency and impact of knowledge management for teams of people and individuals” by allowing the “system to evaluate documents and things used by an enterprise, which may have its own industry-specific lexicon, to identify words and phrases that are more prevalently used by that enterprise.” (Kaufman, ¶[005])
Regarding claims 13 and 27, the combination of Madaan and Kaufman teaches all of the elements of claims 12 and 26 (see detailed element mapping above). In addition, Madaan further teaches the remote computing system does not retain the one or more user changes (As shown in Fig. 3, the user data 312 is stored and processed at the facility. In addition, ¶[035] teaches “The STT platform 311 also includes an STT engine 314 that implements the software functions and the logic to accomplish the various tasks and operations detailed here” Accordingly, the user data, i.e., user feedback/corrections to transcripts, is not stored at the remote server).
Madaan differs from the claimed invention, as defined by claims 13 and 27, in that Madaan fails to specifically disclose that the LLM executes on a remote computing system. Machine learning system which utilize client/server type architecture wherein the cloud-based services are provided via remote computing system are known in the art as evidenced by Kaufman. Therefore, it would have been obvious to one having ordinary skill in the art before the effective filing date of the invention to have implemented the memory-assisted prompt editing taught by Madaan in a client/server type architecture as taught by Kaufman in order to “improve the accuracy, efficiency and impact of knowledge management for teams of people and individuals” by allowing the “system to evaluate documents and things used by an enterprise, which may have its own industry-specific lexicon, to identify words and phrases that are more prevalently used by that enterprise.” (Kaufman, ¶[005])
Regarding claims 14 and 28, the combination of Madaan and Kaufman teaches all of the elements of claims 1 and 15 (see detailed element mapping above).
In addition, Madaan further teaches providing the task prompt conditioned on the user correction prompt as input to the LLM comprises processing, using the LLM, the task prompt conditioned on the user correction prompt to generate the personalized response to the user input (Figure 2 teaches “…searches for feedback from prior queries with a similar intent as x using a retrieval function…x is then concatenated to the retrieved feedback and appended to the prompt for querying GPT-3”).
In addition, Kaufman further teaches the LLM executes on the data processing hardware (Fig.3, ML facility 318 and ¶[036] teaches “a machine learning (ML) facility 318 may be implemented within the STT platform 311.
Madaan differs from the claimed invention, as defined in claims 14 and 28, in that Madaan fails to disclose that the LLM maybe housed in the same processing hardware as the user feedback selection processing. Systems capable of handling machine learning processes and user feedback selection processing are known in the art as evidenced by Kaufman. Therefore, it would have been obvious to one skilled in the art before the effective filing date of the invention to have implemented the memory-assisted prompt editing system of Madaan such that the prompt editing and LLM are housed in the same processing hardware as taught by Kaufman as it merely constitutes the combination of known architecture to achieve the predictable result of decreasing latency.
Claims 2-3 and 16-17 are rejected under 35 U.S.C. 103 as being unpatentable over the combination of Madaan and Kaufman as applied to claims 1 and 15 above, and further in view of Liu et al. (CN 116843795 A; herein “Liu”).
Regarding claims 2 and 16, the combination of Madaan and Kaufman teaches all of the elements of claims 1 and 15 (see detailed element mapping above).
In addition, Madaan teaches determining the user correction prompt comprises selecting the one or more user changes made by the user to prior outputs of the LLM when performing tasks associated with the task type (Figure 2 teaches “…searches for feedback from prior queries with a similar intent as x using a retrieval function…x is then concatenated to the retrieved feedback and appended to the prompt for querying GPT-3”).
The combination of Madaan and Kaufman fails to explicitly disclose that identifying the context of the user input comprises identifying a task type for the task specified by the task prompt for the LLM to perform.
Liu teaches a natural language process that includes, inter alia, performing prompt optimization based on types of task. More specifically, Liu teaches in-context learning, e.g., example selection for few-shot prompting, based on task type. (Liu page 14, 6th paragraph). Therefore, Liu teaches identifying a task type for the task specified by the task prompt for the LLM to perform.
The combination of Madaan and Kaufman differs from the claimed invention, as defined in claims 1 and 16, in that the combination fails to specifically disclose that utilizing task type when selecting in-context learning for prompt optimization. In-context learning for prompt optimization based on task type is known in the art as evidenced by Liu. Therefore, it would have been obvious to one having ordinary skill in the art to have modified the recall function taught by Madaan to factor in task type as taught by Liu as it merely constitutes the combination of known processes to achieve the predictable result of optimizing LLM prompting while factoring in the type of task being requested in the prompt.
Regarding claims 3 and 17, the combination of Madaan, Kaufman and Liu teaches all of the elements of claims 2 and 16 (see detailed element mapping above). In addition, Madaan further teaches wherein the task type comprises at least one of a speech recognition task (the “or” makes this limitation optional), a text prediction task (the “or” makes this limitation optional), or a text generation task (Figure 1 teaches the LLM input is a text generation task).
Claims 4 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over the combination of Madaan and Kaufman as applied to claims 1 and 15 above, and further in view of Macklin et al. (US 2023/0368043 A1; herein “Macklin”).
Regarding claims 4 and 18, the combination of Madaan and Kaufman teaches all of the elements of claims 1 and 15 (see detailed element mapping above). In addition, Madaan further teaches determining the user correction prompt comprises selecting the one or more user changes made by the user to prior outputs of the LLM responsive to corresponding prior user inputs from the user associated with the topic (Figure 2 teaches “…searches for feedback from prior queries with a similar intent as x using a retrieval function…x is then concatenated to the retrieved feedback and appended to the prompt for querying GPT-3”). However, the combination of Madaan and Kaufman fails to explicitly disclose identifying the context of the user input includes identifying a topic associated with the user input.
Macklin teaches systems and methods for machine learning models for interaction insights. More specifically, Macklin teaches “classifying the acquired data by adding one or more labels to each of the acquired data may further include grouping the one or more labels to a core topic, wherein the core topic indicates the communication intent of the user” (Macklin, ¶[0014]). Therefore, Macklin teaches identifying the context of the user input includes identifying a topic associated with the user input.
The combination of Madaan and Kaufman differs from the claimed invention, as defined by claims 4 and 18, in that the combination fails to disclose that determining the intent of the user includes identifying a topic associated with the user input. Identifying topics are a source of user intent is known in the art as evidenced by Macklin. Therefore, it would have been obvious to one having ordinary skill in the art to have modified the system taught by the combination of Madaan and Kaufman to include using topic as an indication of intent as taught by Macklin as it merely constitutes the combination of known processes to achieve the predictable result of optimizing LLM prompting while factoring in the topic being requested in the prompt.
Claims 9-10 and 23-24 are rejected under 35 U.S.C. 103 as being unpatentable over the combination of Madaan and Kaufman as applied to claims 1 and 15 above, and further in view of Potter (US 2023/0368043 A1; herein “Potter”).
Regarding claims 9 and 23, the combination of Madaan and Kaufman teaches all of the elements of claims 1 and 15 (see detailed element mapping above). However, the combination of Madaan and Kaufman fails to disclose applying a corresponding weight to each of the one or more user changes; and determining the user correction prompt based on the corresponding weight applied to each of the one or more user changes.
Potter teaches a dialog analysis method which analyzes dialog turn information to determine confusing prompts. More specifically, ¶[0205] of Potter teaches obtaining an indication of which prompts in a dialog record are causing confusion based on several factors including, among other user actions, the number of times a recognized response to the prompt was denied or otherwise canceled. Paragraph [0211] further teaches that different weighting factors can be applied to various factors. Therefore, Potter teaches applying a corresponding weight to each of the one or more user changes; and determining the user prompt based on the corresponding weight applied to each of the one or more user changes.
The combination of Madaan and Kaufman differs from the claimed invention, as defined by claims 9 and 23, in that the combination fails to disclose weighting the user feedback/changes. Weighting the criteria, such as user feedback, used to identify prompts needing correction is known in the art as evidenced by Potter. Therefore, it would have been obvious to one having ordinary skill in the art to have modified the system taught by the combination of Madaan and Kaufman to include weighting the user feedback/changes as taught by Potter as it merely constitutes the combination of known processes to achieve the predictable result of optimizing LLM prompting based on user feedback.
Regarding claims 10 and 24, Madaan, Kaufman, and Potter teaches all of the elements of claims 9 and 23 (see detailed element mapping above). In addition, Potter further teaches applying the corresponding weight to each of the one or more user changes comprises, for each particular user change of the one or more user changes:
determining a number of times that the particular user change was made by the user (¶[0210] teaches “…Denial rate: (for turns of Type Ask): the number of times a Semantic Item value (i.e., a recognized response to the prompt) was denied or otherwise cancelled” ); and
determining the corresponding weight to apply to the particular user change based on the number of times that the particular user change was made by the user (¶[0211] teaches “These individual totals are summed over the number of instances of the prompt in the data…different weighting factors can be applied to component (a)-(d). The resulting rating can be used alone…” ).
The combination of Madaan and Kaufman differs from the claimed invention, as defined by claims 10 and 24, in that the combination fails to disclose weighting the user feedback/changes based on the number of times the changes are made. Weighting the criteria, such as user feedback, used to identify prompts needing correction based on the number of times it is made is known in the art as evidenced by Potter. Therefore, it would have been obvious to one having ordinary skill in the art to have modified the system taught by the combination of Madaan and Kaufman to include weighting the user feedback/changes as taught by Potter as it merely constitutes the combination of known processes to achieve the predictable result of optimizing LLM prompting based on user feedback.
Claims 11 and 25 are rejected under 35 U.S.C. 103 as being unpatentable over the combination of Madaan, Kaufman and Potter as applied to claims 9 and 23 above, and further in view of Deng et al. (CN 112000883; herein “Deng”).
Regarding claims 11 and 25, the combination of Madaan, Kaufman, and Potter teaches all of the elements of claims 9 and 23 (see detailed element mapping above). However, the combination fails to disclose determining an elapsed time since when the particular user change was last made by the user; and determining the corresponding weight to apply to the particular user change based on the elapsed time since when the particular user change was last made ( ).
Deng teaches a method for automatically adjusting a weight value. More specifically, Deng teaches adjusting the weight value of parameters based on the elapsed time period of the user feedback notification. (See page 5, sixth paragraph) Therefore, determining an elapsed time since when the particular user change (i.e., user feedback) was last made by the user; and determining the corresponding weight to apply to the particular user change based on the elapsed time since when the particular user change was last made.
The combination of Madaan, Kaufman and Potter differs from the claimed invention, as defined by claims 11 and 25, in that the combination fails to disclose weighting the user feedback/changes based on an elapsed time since receiving the user feedback/change. Weighting the criteria, such as user feedback, used to identify prompts needing correction based on the elapsed time since the change was made is known in the art as evidenced by Deng. Therefore, it would have been obvious to one having ordinary skill in the art to have modified the system taught by the combination of Madaan, Kaufman, and Potter to include weighting the user feedback/changes as taught by Lie as it merely constitutes the combination of known processes to achieve the predictable result of optimizing LLM prompting based on user feedback.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Sheikh et al. (US 2023/0368284 A1) teaches a system and method for enabling application of autonomous agents using a client-server architecture;
Braho et al. (US 2007/0198269 A1) teaches a method and system for assessing and improving the performance of a speech recognition system;
Ljolje et al. (US 2015/0348540 A1) teaches a system and method for optimizing speech recognition and natural language parameters with user feedback; and
Kelkar et al. (US 2023/0274095 A1) teaches an autonomous conversational AI system which utilizes user’s feedback and number of times an intent has been repeated to calculate a user’s dissatisfaction threshold.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PENNY L CAUDLE whose telephone number is (703)756-1432. The examiner can normally be reached M-Th 8:00 am to 5:00 pm eastern.
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, Daniel Washburn can be reached at 571-272-5551. 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.
/PENNY L CAUDLE/Examiner, Art Unit 2657
/DANIEL C WASHBURN/Supervisory Patent Examiner, Art Unit 2657
Prosecution Insights