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 .
This non-final rejection is in response to the application filed on: 12/05/2024 and preliminary amendments filed on: 06/16/2025.
Drawings
The drawings filed on: 12/05/2024 are accepted.
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.
Claim 2-22 are rejected under 35 U.S.C. 101 because the claimed invention is directed to abstract idea without significantly more.
Claim 2 Analysis:
101 Analysis Step 2A, Prong One
Claim 2 recites the following limitations (of which bolded limitations constitute a ‘mental process’ that covers performance of the limitations in the human mind).
A method comprising: performing at least once the following: generating a synthetic chat between a customer and an agent using one or more large language model (LLMs), wherein the generating includes submitting a first prompt to cause generation of a first response to use as a turn of the synthetic chat for the customer, and generating and submitting a second prompt based on the first response to cause generation of a second response to use as a turn of the synthetic chat for the agent; adding, based on the synthetic chat, content to the second prompt to improve performance of the agent for use in a specific domain; receiving, from a user of a user computing device, a first turn of a conversation; submitting to an LLM a third prompt that includes the content and that is based on the first turn of the conversation; and causing communication to the user of the user computing device a third response generated by the LLM responsive to the third prompt.
As a note, steps fall within the mental process groupings of abstract ideas because they
cover concepts performed in the human mind, including observation, evaluation,
judgement and opinion (See MPEP 2106.04(a)(2), subsection III).
With respect to the particular limitations that were bolded above, these steps can be practically performed in the human mind using observation, evaluation, judgment and/or opinion. For example, the particular limitations encompass: 1) making a judgment to produce synthetic chat content for use between a customer and agent, 2) evaluating the first response and making a judgment of a second prompt based on the evaluation, 3) evaluating synthetic chat content and making a judgment to add content to the second prompt, and 4) evaluating a third prompt and making a judgement to produce a third response.
101 Analysis Step 2A, Prong Two
With regards to the following additional elements of:
“generating a synthetic chat between a customer and an agent using one or more large language model (LLMs), wherein the generating … of a first response to use as a turn of the synthetic chat for the customer, and generating ” ; “… generation of a second response …”, “of the user computing device”, “ .. generated by the LLM responsive” , “of the user computing device …”, “..generated by the LLM responsive ….” - these additional elements are considered merely reciting the words "apply it" (or an equivalent) with the judicial exception, or merely including instructions to implement an abstract idea on a computer (apply prompt to LLM , which generates), or merely using a computer as a tool to perform an abstract idea. More specifically, these elements apply a judicial exception to a generic computer that employs generic computer functions (execution of steps and producing output based on the execution). The courts have identified these types of limitations as insufficient to integrate a judicial exception into a practical application.
“includes submitting a first prompt … ”, “submitting a second prompt …”, “receiving, from a user of a user computing device, a first turn of a conversation; submitting to an LLM a third prompt that includes the content and that is based on the first turn of the conversation”… “receiving, from a user of a user computing device, a first turn of a conversation; submitting to an LLM a third prompt that includes the content and that is based on the first turn of the conversation” -these additional elements are considered merely adding insignificant extra-solution activity to the judicial exception as they are considered ‘mere data gathering’ iv. Obtaining information about transactions using the Internet to verify credit card transactions, CyberSource v. Retail Decisions, Inc., 654 F.3d 1366, 1375, 99 USPQ2d 1690, 1694 (Fed. Cir. 2011). The courts have identified these types of limitations as insufficient to integrate a judicial exception into a practical application.
101 Analysis Step 2B
The claim does not amount to significantly more than the recited exception:
With regards to the following additional elements of:
“generating a synthetic chat between a customer and an agent using one or more large language model (LLMs), wherein the generating … of a first response to use as a turn of the synthetic chat for the customer, and generating ” ; “… generation of a second response …”, “of the user computing device”, “ .. generated by the LLM responsive” , “of the user computing device …”, “..generated by the LLM responsive ….” - these additional elements are considered merely reciting the words "apply it" (or an equivalent) with the judicial exception, or merely including instructions to implement an abstract idea on a computer (apply prompt to LLM , which generates), or merely using a computer as a tool to perform an abstract idea. More specifically, these elements apply a judicial exception to a generic computer that employs generic computer functions (execution of steps and producing output based on the execution). The courts have identified these types of limitations as insufficient to integrate a judicial exception into a practical application.
“includes submitting a first prompt … ”, “submitting a second prompt …”, “receiving, from a user of a user computing device, a first turn of a conversation; submitting to an LLM a third prompt that includes the content and that is based on the first turn of the conversation”… “receiving, from a user of a user computing device, a first turn of a conversation; submitting to an LLM a third prompt that includes the content and that is based on the first turn of the conversation” -these additional elements are considered merely adding insignificant extra-solution activity to the judicial exception as they are considered ‘mere data gathering’ iv. Obtaining information about transactions using the Internet to verify credit card transactions, CyberSource v. Retail Decisions, Inc., 654 F.3d 1366, 1375, 99 USPQ2d 1690, 1694 (Fed. Cir. 2011). The courts have identified these types of limitations as insufficient to integrate a judicial exception into a practical application. The courts have identified these types of limitations as insufficient to qualify as ‘significantly more’ when recited in a claim with a judicial exception.
101 Analysis for claims 3-12:
Claims 3-12 do not recite any further limitations that cause the claim to be patent
eligible. Rather, the limitations of dependent claims are directed toward additional
aspects of the judicial exception that do not integrate the judicial exception into a
practical application nor amount to significantly more than the judicial exception (as the types of additional elements present in one or more of these dependent claims have already been addressed in claim 1).
101 Analysis for claim 13
With regards to claim 13, it is rejected under similar rationale as claim 1. It is noted that it additionally recites the additional element of ‘a non-transitory machine-readable storage medium’, which is considered merely reciting the words "apply it" (or an equivalent) with the judicial exception, or merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea,. More specifically, this element applies a judicial exception to a generic computer. The courts have identified these types of limitations as insufficient to integrate the judicial exception into a practical application and insufficient to amount to significantly more than the judicial exception.
101 Analysis for claims 14-22
Claims 14-22 do not recite any further limitations that cause the claim to be patent
eligible. Rather, the limitations of dependent claims are directed toward additional
aspects of the judicial exception that do not integrate the judicial exception into a
practical application nor amount to significantly more than the judicial exception (as the types of additional elements present in one or more of these dependent claims have already been addressed in claim 13).
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
Claims 2 and 13 of the instant application (hereinafter ‘892) are rejected on the ground of nonstatutory double patenting as being unpatentable over claim 1 and 8 of U.S. Patent No. 12277396 (hereinafter ‘396). Although the claims at issue are not identical, they are not patentably distinct from each other because:
With regards to claim 2 of ‘892, claim 1 of ‘396 teaches the limitations of claim 2 of ‘892 since claim 2 of ‘892 is broader than claim 1 of ‘396.
With regards to claim 13 of ‘892, claim 8 of ‘396 teaches the limitations of claim 13 of ‘396 since claim 13 of ‘892 is broader than claim 8 of ‘396.
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.
Claim(s) 2-6, 8, 13-15, 17, and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Shah et al (“Bootstrapping a Neural Conversational Agent with Dialogue Self-Play, Crowdsourcing and On-Line Reinforcement Learning”, published: June 2018, publisher: NAACL-HLT 2018, pages 41-51) in view of Zheng et al (“A Pre-Training Based Personalized Dialogue Generation Model with Persona-Sparse Data”, published: 2020, publisher: The Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI-20), pages 9693) in view of Shen et al (US Application: US 20140317030, published: Oct. 23, 2014, filed: Apr. 22, 2013).
With regards to claim 2, Shah et al teaches a method comprising:
performing at least once the following:
generating a synthetic chat between a customer and an agent using one or more large language model (LLMs), wherein the generating includes submitting a first prompt to cause generation of a first response to use as a turn of the synthetic chat for the customer, and generating and submitting a second prompt based on the first response to cause generation of a second response to use as a turn of the synthetic chat for the agent (a synthetic chat is generated via M2M approach (page 41). As explained in page 43, a framework is implemented to generate iterative prompts/dialog to a user/customer simulator (model) and an agent (model) as dialog self-play, such that the customer and agent each take turns in a dialog exchange (iterative exchange). );
adding, based on the synthetic chat, content to the second prompt to improve performance of the agent LLM for use in a specific domain (page 43: task information and/or annotations are also added to the iterative prompts such as additional information to help an agent ‘perform’ by moving a dialog forward for an Agent associated with the specific task domain (i.e. in this example an agent that is specific to ticket booking domain/scheme) using the task and/or the annotation data);
receiving, from a user of a user computing [entity] , a first turn of a conversation (page 43: a back and forth dialog exchange (turns are implemented between the user and agent);
submitting to an LLM a third prompt that includes the content and that is based on the first turn of the conversation (page 47: a neural conversational model is refined/trained based upon the M2M generated data sets, such that it accepts prompts/conversation from users (the examiner notes that the claim limitations addressed earlier explained that the generated data set includes multiple turn/prompts (such as the claimed first turned dialog) and have been incorporated into training the neural conversational model)); and
causing communication to the user of the user computing [entity] a third response generated by the LLM responsive to the third prompt (page 47: the neural conversational model is designed to interact/respond to users upon deployment).
However Shah et al does not expressly teach a method for generating a synthetic chat such that … the chatbots/customer and agent models are specifically a type of large language model/LLM; … and the user computing entity of Shah et al , is a computing device’.
Yet Zheng et al teaches different models/bots are specifically a type of large language model/LLM (Figure 2, pages 9694, 9695 and 9696: each model/bot having a persona is implemented using large language model type architecture to conduct dialogs having personas/roles reflective of persona-related features ).
It would have been obvious to one of ordinary skill in the art before the effective filing of the invention to have modified Shah et al’s ability to implement customer and agent models to implement and generate dialog/prompts between the models/entities, such that the models could have been implemented as LLM specific models as taught by Zheng et al. The combination would have allowed the models to have been trained to reflect more human like personas such that the models “generate more coherent and persona consistent responses” (Zheng et al, Abstract).
However although Shen et al and Zheng et al teaches a method for generating a synthetic chat using computing entities, they don’t expressly teach computing devices.
Yet Shen teaches a computer implemented method that includes the user computing device (Fig. 1, Fig. 9: a computer implemented method incorporating use if processor(s) and memor(ies) is applied in a chat/conversation system ).
It would have been obvious to one of ordinary skill in the art before the effective filing of the invention to have modified Shah et al and Zheng et al’s ability implement computing entities for generating a synthetic chat, such that the computing entities are specifically computing devices, as taught by Shen. The combination would have allowed conversation to progress in a manner that would yield a better experience by taking into account a user’s communication style (Shen et al, paragraph 0011).
With regards to claim 3. (Currently Amended) The method of claim 2, the combination of Shah et al and Zheng et al teaches the adding… , as similarly explained in the rejection of claim 2, and is rejected under similar rationale.
However the combination of Shah et al and Zheng et al does not take into account wherein the adding is based on linguistic properties comprising one or more of tone, complexity, nuance, domain-specific words, domain-specific phrases, corporate policies, brand, and length of response.
Yet Shen et al teaches wherein the adding is based on linguistic properties comprising one or more of tone, complexity, nuance, domain-specific words, domain-specific phrases, corporate policies, brand, and length of response (paragraphs 0042, 0050 and 0073: additional information (properties such as tone, length of response, statements) are provided back to the system which is then each weighed/scored to help the agent provide an optimal response to the user).
It would have been obvious to one of ordinary skill in the art before the effective filing of the invention to have modified Shah et al and Zheng et al’s ability to add information to help forward/progress a dialog/conversation, such that additional information from a response that is generated can be weighed/scored to help progress the dialog/conversation, as taught by Shen et al. The combination would have allowed conversation to progress in a manner that would yield a better experience by taking into account a user’s communication style (Shen et al, paragraph 0011).
With regards to claim 4. (Previously Presented) The method of claim 3, the combination of Sha et al, Zheng et al and Shen et al teaches wherein the linguistic properties comprise a plurality of properties, and wherein each of the plurality of properties is scored, as similarly explained in the rejection of claim 3 (paragraphs 0042, 0050 and 0073 in Shen et al were explained in the rejection of claim 3 to teach that additional information (properties such as tone, length of response, statements) are provided back to the system which is then each weighed/scored to help the agent provide an optimal response to the user), and is rejected under similar rationale.
With regards to claim 5. (Previously Presented) The method of claim 4, the combination of Sha et al, Zheng et al and Shen et al teaches wherein a score for each synthetic chat is based on a weight assigned to each of the plurality of properties, as similarly explained in the rejection of claim 3 (paragraphs 0042, 0050 and 0073 in Shen et al were explained in the rejection of claim 3 to teach that additional information (properties such as tone, length of response, statements) are provided back to the system which is then each weighed/scored to help the agent provide an optimal response to the user), and is rejected under similar rationale.
With regards to claim 6. (Currently Amended) The method of claim 2, the combination of Sha et al, Zheng et al and Shen et al teaches wherein the one or more LLMs include a customer LLM and an agent LLM, as similarly explained in the rejection of claim 2 above (Shah et al’s models were modified with Zheng et al’ teachings to be LLM models), and is rejected under similar rationale.
With regards to claim 8. (Currently Amended) The method of claim 2, the combination of Sha et al, Zheng et al and Shen et al teaches wherein the generating and the adding is performed iteratively, as similarly explained in the rejection of claim 2 above (Shah et al’s models were modified with Zheng et al’ teachings to be LLM models), and is rejected under similar rationale.
With regards to claim 13, the combination of Sha et al, Zheng et al and Shen et al teaches a non-transitory machine-readable storage medium that provides instructions that, if executed by a system, are configurable to cause said system to perform operations comprising:
performing at least once the following:
generating a synthetic chat between a customer and one or more large language model (LLMs), wherein the generating includes submitting a first prompt to cause generation of a first response to use as a turn of the synthetic chat for the customer, and generating and submitting a second prompt based on the first response to cause generation of a second response to use as a turn of the synthetic chat for the agent;
adding, based on the synthetic chat, content to the second prompt to improve performance of the agent for use in a specific domain;
receiving, from a user of a user computing device, a first turn of a conversation;
submitting to an LLM a third prompt that includes the content and that is based on the first turn of the conversation; and causing communication to the user of the user computing device a third response generated by the LLM responsive to the third prompt, as similarly explained in the rejection of claim 2, and is rejected under similar rationale.
With regards to claim 14. (Currently Amended) The non-transitory machine-readable storage medium of claim 13, the combination of Sha et al, Zheng et al and Shen et al teaches wherein the adding is based on linguistic properties comprising one or more of tone, complexity, nuance, domain-specific words, domain-specific phrases, corporate policies, brand, and length of response, as similarly explained in the rejection of claim 3, and is rejected under similar rationale.
With regards to claim 15. (Currently Amended) The non-transitory machine-readable storage medium of claim 14, the combination of Sha et al, Zheng et al and Shen et al teaches wherein the linguistic properties comprise a plurality of properties, and wherein each of the plurality of properties is scored, wherein a score for each synthetic chat is based on a weight assigned to each of the plurality of properties, as similarly explained in the rejection of claim 5, and is rejected under similar rationale.
With regards to claim 17. (Currently Amended) The non-transitory machine-readable storage medium of claim 13, the combination of Sha et al, Zheng et al and Shen et al teaches wherein the one or more LLMs include a customer LLM and an agent LLM, as similarly explained in the rejection of claim 6, and is rejected under similar rationale
With regards to claim 19. (Currently Amended) The non-transitory machine-readable storage medium of claim 13, the combination of Sha et al, Zheng et al and Shen et al teaches wherein the generating and the adding is performed iteratively, as similarly explained in the rejection of claim 8, and is rejected under similar rationale
Claim(s) 12 and 22 is/are rejected under 35 U.S.C. 103 as being unpatentable over Shah et al (“Bootstrapping a Neural Conversational Agent with Dialogue Self-Play, Crowdsourcing and On-Line Reinforcement Learning”, published: June 2018, publisher: NAACL-HLT 2018, pages 41-51) in view of Zheng et al (“A Pre-Training Based Personalized Dialogue Generation Model with Persona-Sparse Data”, published: 2020, publisher: The Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI-20), pages 9693) in view of Shen et al (US Application: US 20140317030, published: Oct. 23, 2014, filed: Apr. 22, 2013) in view of Czarnecki et al (US Application: US 20190354867, published: Nov. 21, 2019, filed: May 20, 2019).
With regards to claim 12. (Currently Amended) The method of claim 2, the combination of Sha et al, Zheng et al and Shen et al teaches wherein the generating and the adding is performed iteratively to produce synthetic chats, as similarly explained in the rejection of claim 2, and is rejected under similar rationale.
However the combination does not teach … until at least one of the synthetic chats is scored above a threshold.
Yet Czarnecki et al teaches [generating and adding/accumulating training data] … until at least one of the [outputs] is scored above a threshold (paragraph 0089: iterative generation of training data/output data is stopped when amount of training data has been assessed to be more than a threshold amount).
It would have been obvious to one of ordinary skill in the art before the effective filing of the invention to have modified Shah et al, Zheng and Shen et al’s ability to train using generative dialogs/outputs , such that the training would end upon the outputs being considered to exceed a threshold amount of training data, as taught by Czarnecki et al. The combination would have allowed Shah et al and Zheng et al to have been able to conditionally/flexibly apply a desired training policy to a plurality of networks.
With regards to claim 22. (New) The non-transitory machine-readable storage medium of claim 13, the combination of Shah et al, Zheng et al, Shen et al and Czarnecki et al teaches wherein the generating and the adding is performed iteratively to produce synthetic chats until at least one of the synthetic chats is scored above a threshold, as similarly explained in the rejection of claim 12 above, and is rejected under similar rationale.
Claim(s) 7 and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Shah et al (“Bootstrapping a Neural Conversational Agent with Dialogue Self-Play, Crowdsourcing and On-Line Reinforcement Learning”, published: June 2018, publisher: NAACL-HLT 2018, pages 41-51) in view of Zheng et al (“A Pre-Training Based Personalized Dialogue Generation Model with Persona-Sparse Data”, published: 2020, publisher: The Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI-20), pages 9693) in view of Shen et al (US Application: US 20140317030, published: Oct. 23, 2014, filed: Apr. 22, 2013) in view of Foster et al (US Application: US 20190354594, published: Nov. 21, 2019, filed: May 20, 2018).
With regards to claim 7. (Currently Amended) The method of claim 2, the combination of Shah et al, Zheng et al and Shen et al teaches wherein the specific domain, as similarly explained in the rejection of claim 2, and is rejected under similar rationale.
However, although the combination teaches different LLM models are implemented having personas , in a dialog associated with a type of domain, the combination does not teach the domain is medical.
Yet Foster et al teaches … the domain is medical (paragraph 0027: model persona can be specialized to process medical conversations with the user).
It would have been obvious to one of ordinary skill in the art before the effective filing of the invention to have modified Shah et al, Zheng et al and Shen et al’s ability to customize dialog between models based on domain data of a conversation, such that the domain data would have further included medical domain, as taught by Foster et al. The combination would have allowed Shah et al and Zheng et al to have implemented models that have a persona with role and domain specific characteristics.
With regards to claim 18. (Currently Amended) The non-transitory machine-readable storage medium of claim 13, the combination of Shah et al, Zheng et al, Shen et al and Foster et al teaches wherein the specific domain is medical, as similarly explained in the rejection of claim 7, and is rejected under similar rationale.
Claim(s) 9 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Shah et al (“Bootstrapping a Neural Conversational Agent with Dialogue Self-Play, Crowdsourcing and On-Line Reinforcement Learning”, published: June 2018, publisher: NAACL-HLT 2018, pages 41-51) in view of Zheng et al (“A Pre-Training Based Personalized Dialogue Generation Model with Persona-Sparse Data”, published: 2020, publisher: The Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI-20), pages 9693) in view of Shen et al (US Application: US 20140317030, published: Oct. 23, 2014, filed: Apr. 22, 2013) in view of Chandra et al (US Application: US 20210042800, published: Feb. 11, 2021, filed: Aug. 6, 2019).
With regards to claim 9. (Currently Amended) The method of claim 8, Shah et al, Zheng et al and Shen et al teaches the synthetic chat is based on [assessment] … of the synthetic chat, … the second prompt, as similarly explained in rejection of claim 2, and is rejected under similar rationale.
However the combination does not expressly teach …. wherein a score of the synthetic chat is based on a comparison of a summary of the synthetic chat to the second prompt.
Yet Chandra et al teaches …. wherein a score of the synthetic chat is based on a comparison of a summary of the synthetic chat to the [subsequent] prompt (paragraph 0027 and 0050: response from a customer/user is summarized as a score based upon features from user communication and a subsequent (‘second’) prompt by the agent is influenced based upon the scoring (such as when scoring is above a predefined threshold)).
It would have been obvious to one of ordinary skill in the art before the effective filing of the invention to have modified Shah et al, Zheng et al and Shen et al’s ability to train a conversational agent through synthetic chat generation, such that the chat generation is influenced by further taking into account chat summary via scoring, as taught by Chandra et al. The combination would have allowed implementation of an agent that would be capable of dynamically adapting decisions based upon customer interaction.
With regards to claim 20. (Currently Amended) The non-transitory machine-readable storage medium of claim 19, Shah et al, Zheng et al, Shen et al and Chandra et al teaches wherein a score of the synthetic chat is based on a comparison of a summary of the synthetic chat to the second prompt, as similarly explained in the rejection of claim 9, and is rejected under similar rationale.
Claim(s) 10, 11 and 21 is/are rejected under 35 U.S.C. 103 as being unpatentable over Shah et al (“Bootstrapping a Neural Conversational Agent with Dialogue Self-Play, Crowdsourcing and On-Line Reinforcement Learning”, published: June 2018, publisher: NAACL-HLT 2018, pages 41-51) in view of Zheng et al (“A Pre-Training Based Personalized Dialogue Generation Model with Persona-Sparse Data”, published: 2020, publisher: The Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI-20), pages 9693) in view of Shen et al (US Application: US 20140317030, published: Oct. 23, 2014, filed: Apr. 22, 2013) in view of Cheng et al (US Application: US 20200395007, published: Dec. 17, 2020, filed: Apr. 20, 2020).
With regards to claim 10. (Currently Amended) The method of claim 2, the combination of Shah et al, Zheng et al and Shen et al teaches wherein the adding is based on identifying from the synthetic chat, as similarly explained in the rejection of claim 2, and is rejected under similar rationale.
However the combination does not expressly teach … identifying … whether a category of information was shared as part of the synthetic chat.
Yet Cheng et al teaches … identifying … whether a category of information was shared as part of the synthetic chat (Abstract, Fig 2, Fig. 19A, paragraph 0097-0098, 0211, 0258-0259: a plurality of agents/models can each be prompted based domain specific information encountered with user chat/input. Also a category/domain/concept is identified from the chat to assist and invoke a subsequent contributing agent).
It would have been obvious to one of ordinary skill in the art before the effective filing of the invention to have modified Shah et al, Zheng et al, and Shen et al’s ability to implement models as LLM specific models that generate and also add information based upon chat content analysis, such that the analysis would have further included category/concept identification and the LLM specific models would have included a third model that can be selected based upon domain /category specific information encountered with user chat/input, as taught by Cheng et al. The combination would have provided an enhanced conversation experience through an ability to scale a request in a multi model/agent network.
With regards to claim 11. (Previously Presented) The method of claim 10, the combination of Shah, Zheng et al, Shen et al and Cheng et al teaches wherein the identifying, .. the synthetic chat included information with the category of information includes prompting a third LLM regarding whether the synthetic chat included information with the category of information (as similarly explained in the rejection of claim 10 a category/concepts are identified to invoke an additional LLM model/agent), and is rejected under similar rationale.
With regards to claim 21. (Currently Amended) The non-transitory machine-readable storage medium of claim 13, the combination of Shah, Zheng et al, Shen et al and Cheng et al teaches wherein the generating and the adding is performed is based on identifying from the synthetic chat whether a category of information was shared as part of the synthetic chat, as similarly explained in the rejection of claim 10 above, and is rejected under similar rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Baldwin et al (US Application: US 2008/0091406): This reference teaches a cooperative conversation interface based on a particular domain.
Pasupalak et al (US Application: US 2015/0066479): This reference teaches a conversational agent.
Cai et al (US Application: US 20190121853): This reference teaches facilitating automatic detection of relationships between sentences in conversations.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to WILSON W TSUI whose telephone number is (571)272-7596. The examiner can normally be reached Monday - Friday 9 am -6 pm.
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, Adam Queler can be reached at (571) 272-4140. 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.
/WILSON W TSUI/Primary Examiner, Art Unit 2172
Prosecution Insights