DETAILED ACTION
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 October 7, 2025 was filed before the mailing date of this non-final rejection. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Objections
Claim 1 recites “apply reinforcement learning…to weight the retrieved context” and should recite “to weigh the retrieved context.” Claims 12 and 22 recite similar language and should be corrected similarly. Claims 2-11, 13-21 and 23-31 are objected to by virtue of dependency. Appropriate correction is required.
Claim 9 recites “the trainer profile tool” in line 1. The term lacks antecedent basis. For examination purposes, the claim will be interpreted as reciting “the trainer tools”, which has antecedent support from “trainer tools” in claim 1 (as is similarly recited in claims 19 and 29). Appropriate correction is required.
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 1-31 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1
Claims 1-11 recite a system comprising: at least one processor (i.e., machine), claims 12-21 recite a method (i.e., a process), and claims 21-31 recite a computer program embodied on a non-transitory computer-readable medium (i.e., a machine). Therefore, the claims all fall within one of the four statutory categories of invention.
Step 2A, Prong One
Claims 1, 12, and 22 recite providing knowledge-based responses to queries related to human capital management: receiving a user query and enhancing the query through iterative rewriting; storing embedded knowledge; rewriting the user query to separate intent and data content; retrieving context relevant to the rewritten query; applying learning from human feedback to weight the retrieved context; generating a response using retrieval-augmented generation; refining the response; gathering additional data, and providing data retrieval instructions and feedback; and processing the user query, and outputting a final response.
The claim as a whole recites a certain method of organizing human activity. The limitations recited above of providing knowledge-based responses to queries related to human capital management, under broadest reasonable interpretation, recite the abstract idea of a certain method of organizing human activity, e.g., fundamental economic practices. The claims as a whole also and a mental process - concepts performed in the human mind (e.g., including observation, evaluation, judgment, opinion). Therefore, the claim recites an abstract idea.
Step 2A, Prong Two
The judicial exception is not integrated into a practical application. The judicial exception is not integrated into a practical application. Claims 1, 12, and 22 as a whole amount to: 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.
The claims recite the additional elements of: (i) a user interface (claims 1, 12, and 22); (ii) a data store (claims 1, 12, and 22); (iii) at least one processor operatively coupled to the user interface and the data store (claims 1 and 12); (iv) reinforcement learning (claims 1, 12, and 22); (v) chain of thought and tree of thought processing (claims 1, 12, and 22); (vi) a tool execution component configured to invoke/invoking one or more knowledge tools, the tool execution component further comprising trainer tools (claims 1, 12, and 22); and (vii) a computer program embodied on a non-transitory computer-readable medium, the computer program comprising instructions which, when executed by at least one processor, cause the processor to perform a method (claim 22).
The above additional elements of: (i) –(vii) are recited at a high-level of generality such that, when viewed individually as whole/ordered combination, the additional elements amount to no more than 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 (See MPEP 2106.05(f)).
Accordingly, these additional elements, when viewed as a whole/ordered combination (See Fig. 1) do not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. Thus, the claims are directed to an abstract idea.
Step 2B
As discussed above with respect to Step 2A Prong Two, the additional elements amount to no more than: 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. The same analysis applies here in 2B, i.e., 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, does not integrate the abstract idea into a practical application at Step 2A or provide an inventive concept at Step 2B.
Therefore, the additional elements discussed above do not integrate the abstract idea into a practical application at Step 2A or provide an inventive concept at Step 2B. Thus, even when viewed individually and as a whole/ordered combination, nothing in the claims adds significantly more (i.e., an inventive concept) to the abstract idea. Thus, the claims are ineligible.
Dependent claims 5-7, 9, 15-17, 19, 25-27, and 29 further recite details which merely narrow the previously recited abstract idea limitiaitions. For these reasons, as described above with respect to claims 1, 12 and 22, these judicial exceptions are not meaningfully integrated into a practical application or significantly more than the abstract idea. Thus, claims 5-7, 9, 15-17, 19, 25-27, and 29 are also ineligible.
Claim 2 recites the additional elements of a trainer profile tool, a trainer history tool, and a trainer scoring tool. The abstract idea is not integrated into a practical application because the additional elements amount to no more than: 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. (See MPEP 2106.05(f)). The claim does not include limitations sufficient, either alone or in combination, to amount to significantly more than the claimed abstract idea because the aforementioned additional element merely amount to no more than: 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.
Claims 3, 13, and 23 recite: iteratively refining the user query by soliciting clarifying information from the user - which merely narrows the previously recited abstract idea.
The claim adds the additional element of a prompt enhancer. The abstract idea is not integrated into a practical application because the prompt enhancer amounta to no more than: 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. (See MPEP 2106.05(f)). The claim does not include limitations sufficient, either alone or in combination, to amount to significantly more than the claimed abstract idea because the aforementioned additional element merely amounts to no more than: 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.
Claims 4, 14, and 14 recite: separating user intent from data content - which merely narrows the previously recited abstract idea.
The claim adds the additional element of using a fine-tuned large language model. The abstract idea is not integrated into a practical application because the large language model amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (e.g., LLMs, NLP). (See MPEP 2106.05(f)). The claim does not include limitations sufficient, either alone or in combination, to amount to significantly more than the claimed abstract idea because the aforementioned additional element amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use (e.g., LLMs, NLP).
Claims 8, 18 and 28 recite the additional elements at least one knowledge tool selected from the group consisting of an employee information tool, a payroll knowledge tool, a compliance knowledge tool, an HR knowledge tool, and a tax knowledge tool. The abstract idea is not integrated into a practical application because the additional elements amount to no more than: 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. (See MPEP 2106.05(f)). The claim does not include limitations sufficient, either alone or in combination, to amount to significantly more than the claimed abstract idea because the aforementioned additional elements merely amount to no more than: 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.
Claims 10, 20, and 30 recite: storing multidimensional vector embeddings of chunked knowledge; and drift metrics for triggering retraining of embedding models - which merely narrows the previously recited abstract idea.
The abstract idea is not integrated into a practical application because the additional element amounts to no more than: 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. (See MPEP 2106.05(f)). The claim does not include limitations sufficient, either alone or in combination, to amount to significantly more than the claimed abstract idea because the aforementioned additional element merely amounts to no more than: 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.
Claims 11, 21, and 31 recite the additional element of a network interface operatively coupled to the user interface and the at least one processor.
The abstract idea is not integrated into a practical application because the additional element amounts to no more than: 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. (See MPEP 2106.05(f)). The claim does not include limitations sufficient, either alone or in combination, to amount to significantly more than the claimed abstract idea because the aforementioned additional element merely amounts to no more than: 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.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied for establishing a background for determining obviousness under pre-AIA 35 U.S.C. 103(a) 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.
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, 3-4, 6-7, 11-14, 16-17, 21-24, 26-27 and 31 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication No. 2023/0123581 to Lei et al. (hereinafter “Lei”) in view of U.S. Patent Application Publication No. 2026/0220131 to Schaeck et al. (hereinafter “Schaeck”), and even further in view of U.S. Patent Application Publication No. 2025/0061307 to Tran et al. (hereinafter “Tran”).
In regard to claim 1 (and claims 12 and 22), Lei discloses a user interface configured to receive a user query and enhance the query through rewriting (Abst.; Paras. 32-34, 193) (…a client computer having a graphical user interface or a web browser through which a user can (i.e., a user interface configured to) interact with implementations of the systems and techniques described herein… a user inputs a query (i.e., receive a user query)… a machine is detected to not understand a query input by a user… the new query by a machine is greater than the understanding degree to the original query by the machine (i.e., enhance the query through rewriting).)
Lei discloses a store embedded knowledge (Para. 48) (The machine vocabulary collection refers to a collection of words that a machine can understand).
Lei discloses at least one processor operatively coupled to the user interface, the processor configured to perform a method (Fig. 11; Paras. 184-186, 193) (The processor 1101 executes non-transitory software programs, instructions and modules stored in the memory 1102 to execute the various function applications and data processing of a server, that is, implement the query rewriting method provided in any preceding method embodiment….a client computer having a graphical user interface…through which a user can interact with (i.e., coupled to the user interface).)
Lei discloses to rewrite the user query to separate intent and data content (Abst.; Para. 30) (…in response to a query rewriting request (i.e., to rewrite the user query), at least one of context information of an original query (i.e., data content) and intention information of the original query (I.e., separate intent) is extracted).
Lei discloses to retrieve context relevant to the rewritten query from the embedded knowledge (Paras. 50-52) (A target vocabulary (i.e., context relevant) is determined from the machine vocabulary (i.e., from the embedded knowledge) collection based on at least one of the context information of the original query and the intention information of the original query. The target vocabulary is organized according to grammar so that the new query (i.e., relevant to the rewritten query) is obtained.)
Lei discloses to apply learning from human feedback to weigh the retrieved context (Paras. 132, 121-123, 179-181, and 195) (The query feedback module is configured to, after the new query is determined based on the machine vocabulary collection and at least one of semantic information of the original query and/or the intention information of the original query, feed back the new query to a user in the form of a text prompt or two rounds of inquiries. The retrieval module is configured to, after receiving a positive answer (i.e., weigh) from the user (i.e., from human feedback) conduct retrieval according to the new query (i.e., apply learning) and feed back the retrieval result to the user (i.e., to weight the retrieved context)).
Lei discloses to generate a response (Paras. 41 and 45) (…the new query (i.e., a response) is determined (i.e., generate) based on the machine vocabulary collection and at least one of the context information of the original query and the intention information of the original query…The new query refers to a query rewritten from the original query and better understandable by a machine than the original query).
Lei discloses a tool execution component configured to invoke one or more knowledge tools to gather additional data, the tool execution component further comprising trainer tools configured to provide data retrieval instructions and feedback (Paras. 29, 49, 73, 180) (The method may be performed by a query rewriting apparatus. The apparatus may be implemented by software and/or hardware (i.e., a tool execution component)…the new query is determined based on the machine vocabulary collection (i.e., one or more knowledge tools) and at least one of the context information of the original query and the intention information of the original query in the manner below… The sequence conversion network learns the machine vocabulary collection by pre-training (i.e., gather additional data)…The query feedback module (i.e., trainer tools) is configured to, after the new query is determined based on the machine vocabulary collection and at least one of semantic information of the original query and/or the intention information of the original query, feed back the new query to a user in the form of a text prompt or two rounds of inquiries (i.e., provide data retrieval instructions and feedback).)
Lei discloses wherein the system is configured to process the user query through the processor and the tool execution component to output a final response (Abst.; Paras. 29, 49, 73, 180-181) (The method may be performed by a query rewriting apparatus. The apparatus may be implemented by software and/or hardware (i.e., a wherein the system is configured to process the user query through the processor and the tool execution component)…the after the new query is determined based on the machine vocabulary collection and at least one of semantic information of the original query and/or the intention information of the original query, feed back the new query to a user in the form of a text prompt or two rounds of inquiries. The retrieval module is configured to, after receiving a positive answer from the user, conduct retrieval according to the new query and feed back the retrieval result to the user (i.e., output a final response).).
Lei discloses the machine vocabulary (i.e., embedded knowledge) but does not explicitly disclose or teach, however, Schaeck teaches that the knowledge is stored in a data store; that the processor is operatively coupled to the data store; and that the response is generated using retrieval-augmented generation (Paras. 14-15, 30-31 and 37) (Retrieval-Augmented Generation systems utilize an advanced hybrid architecture that can combine information retrieval techniques with generative language models to deliver contextually relevant and accurate responses to user queries (i.e., response is generated using retrieval-augmented generation). These systems can operate by leveraging a retrieval module and a generation module. The retrieval module can be responsible for searching external knowledge sources, such as structured or unstructured databases, document repositories, or curated knowledge bases, to extract information relevant to a user's query (i.e., a data store configured to store…knowledge)…System 100 can comprise a retrieval-augmented generation component 102…retrieval-augmented generation (RAG) component 102 can be linked to an at least one knowledge base…(i.e., that the processor is operatively coupled to the data store).)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the retrieval-augmented generation of Schaeck with the query system of Lei to enhance accuracy and relevance of retrieved results (See Para. 14 of Schaeck).
As discussed above, Lei discloses the rewriting. Lei in view of Schaeck does not explicitly disclose or teach, however, Tran teaches that rewriting is iterative (Para. 25, 31, 96) (To enhance the system's capabilities, iterative refinement is implemented, allowing the system to ask clarifying questions when needed and enabling backtracking and exploration of alternative solutions…Iteratively evaluating and expanding until a solution is found or a termination condition is met).
As discussed above, Lei discloses applying learning from feedback. Lei in view of Schaeck does not explicitly disclose or teach, however, Tran teaches that the learning is reinforcement learning (Para. 130) (These improvements are achieved through the employment of a variety of machine learning methodologies, such as … reinforcement learning. Each of these learning paradigms contributes to the system's ability to autonomously adjust its response patterns and optimization techniques. For instance, through reinforcement learning, the system is capable of identifying successful thought paths and decision outcomes that yield positive results and, conversely, recognising and deprioritising pathways that prove less effective or erroneous).
As discussed above, Lei discloses generating a response. Lei in view of Schaeck does not explicitly disclose or teach, however, Tran teaches that the response is refined using chain of thought and tree of thought processing (Para. 25, 29-31, 66, 69) (To enhance the system's capabilities, iterative refinement is implemented, allowing the system to ask clarifying questions when needed and enabling backtracking and exploration of alternative solutions (i.e., refine the response)…. dynamically refine its approach as it processes information and gradually narrows down on potential solutions. It continues to revise and extend its thought paths, leveraging both the LLM and the specialized reasoning models until a satisfactory solution is reached… For complex problems, the system might employ a collaborative verification approach, combining Chain-of-Thought (CoT)….to verify the solution's correctness (i.e., using chain of though…processing)…. Tree of Thoughts (ToT) framework comprises several key components that work in concert to enable systematic exploration of multiple solution paths, self-evaluation of progress, and flexible search strategies… FIG. 2 illustrates an exemplary tree-of-thought method…Receiving a problem input…Generating thought paths for solving the problem using a large language model…Evaluating the thought paths to determine their likelihood of leading to a solution…Expanding the selected paths by generating additional thoughts…Iteratively evaluating and expanding until a solution is found or a termination condition is met (i.e., using … tree of thought processing).)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the several AI features (iterative refinement, chain of thought, and tree of thought) of Tran with the query rewriting system of Lei in view of Schaeck to provide enhanced problem-solving capabilities (See Para. 5 of Tran).
In regard to claims 3, 13, and 23 as discussed above, Lei discloses the user query. Lei in view of Schaeck does not explicitly disclose or teach, however, Tran teaches a prompt enhancer configured to iteratively refine the user query by soliciting clarifying information from the user (Para. 25) (To enhance the system's capabilities, iterative refinement is implemented, allowing the system to ask clarifying questions when needed and enabling backtracking and exploration of alternative solutions. A feedback loop ensures continuous improvement…).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the clarifying questions of Tran with the query rewriting system of Lei in view of Schaeck to provide continuous improvement (See Para. 25 of Tran).
In regard to claims 4, 14, and 24 as discussed above, Lei discloses the separating user intent data from data content. Lei in view of Schaeck does not explicitly disclose or teach, however, Tran teaches a using a fine-tuned large language model to separate user intent from data content (Paras. 187) (Upon receiving a user query, the meta-reasoning model begins its operation by analyzing the language and context of the query. This includes understanding the semantics, the intent behind the query, and any domain-specific terminology that may be used. The model employs natural language processing algorithms to discern subtle nuances in the phrasing of the query that might indicate the need for a particular kind of expertise or knowledge base (i.e., separate user intent from data content)…The fine-tuned general-purpose LLM (i.e., using a fine-tuned large language model) becomes adept at generating thought paths that are not only relevant to the specific problem at hand but are also reflective of the specialized knowledge encapsulated within the expert reasoning models. It is through this fine-tuning process that the system gains the capability to address the nuances of domain-specific problems while preserving the broad applicability afforded by the general-purpose foundation).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the fine tuned LLM of Tran with the query rewriting system of Lei in view of Schaeck to provide a more efficient query system.
In regard to claims 6, 16, and 26, as discussed above, Schaeck teaches that the knowledge is stored in a data store; and using retrieval-augmented generation. Lei discloses retrieving relevant context by performing a vector similarity search against the embedded knowledge (Paras. 131-145) (A semantic similarity between the at least one first query and the at least one second query is greater than a set similarity threshold…and training logic of the query rewriting model… The obtained vector representation corresponding to each query is input to an initial conversion network and an initial prediction network separately….The context vector of each query is determined to be ci to cn based on the LSTM network of the encoder part and the word relationship information between words in each query. The relevance between the word vector x of a word in a second query and the context vector of each query is calculated by using the attention mechanism and denoted by a. Then context vectors of these queries are added up according to relevance weights to form an overall context vector…The word vector x and cattention are spliced and input to the LSTM network of the decoder part so that a candidate word denoted by x′ is generated).
In regard to claims 7, 17, and 27 as discussed above, Tran teaches refining the response using chain of thought and tree of thought processing. Lei in view of Schaeck does not explicitly disclose or teach, however, Tran teaches refining that the response using chain-of-thought and tree-of-thought processing comprises: decomposing the response into branching thought sequences; and recombining the sequences to produce an enhanced response (Paras. 29-43) (…a Tree of Thoughts (ToT) framework comprises several key components that work in concert to enable systematic exploration of multiple solution paths, self-evaluation of progress, and flexible search strategies based on the nature of the problem. At its core, the framework involves thought decomposition, breaking down complex problems into manageable sub-problems or thought steps. (i.e., decomposing the response into branching thought sequences)… A thought generator creates multiple potential thoughts or solution paths at each decision point, mimicking the human cognitive process of considering various possibilities… Iteratively evaluating and expanding until a solution is found or a termination condition is met (S2410). Outputting a solution or a promising partial solution (S2412) (i.e., recombining the sequences to produce an enhanced response).)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the processing of Tran with the query rewriting system of Lei in view of Schaeck to provide continuous improvement (See Para. 25 of Tran).
In regard to claims 11, 21, and 31, Lei discloses a network interface operatively coupled to the user interface and the at least one processor to transmit the user query to the processor and return the final response to the user (Para. 29, 32, 123, 193-194; Fig. 11) (The systems and techniques described herein may be implemented in a computing system including …a front-end component (for example, a client computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein)…The components of the system may be interconnected by any form or medium of digital data communication (for example, a communication network) (i.e., a network interface operatively coupled to the user interface)… a user inputs a query… after receiving a positive answer from the user, retrieval is conducted according to the new query and a retrieval result is fed back to the user (i.e., return the final response to the user).)
Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over Lei in view of Schaeck and further in view of Tran, as applied to claim 1, and even further in view of U.S. Patent Application Publication No. 2018/0239830 to Dialani et al. (hereinafter “Dialani”).
In regard to claim 2, as discussed above in regard to claim 1, Lei discloses the trainer tools. Lei in view of Schaeck and further in view of Tran, does not explicitly disclose or teach, however, Dialani teaches wherein the trainer tools comprise a trainer profile tool, a trainer history tool, and a trainer scoring tool (Paras. 25, 27, 30, and 90-91) (…creating training…datasets for machine-learned models for automated sourcing…Embodiments can use log data from past years….to train ranking algorithms used by an automated sourcing tool or an intelligent matches tool (i.e., the trainer tools comprise a trainer history tool)… training apparatus …may use maximum-likelihood estimation (MLE) and/or another estimation technique to estimate the parameters of a logistic regression model for calculating expertise scores (i.e., a trainer scoring tool).)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the various training tools of Dialani with the query system of Lei in view of Schaeck and further in view of Tran to solve technical challenges in search queries to data sources (See Para. 2 of Dialani).
Claims 5, 8-9, 15, 18-19, 25, and 28-29 are rejected under 35 U.S.C. 103 as being unpatentable over Lei in view of Schaeck and further in view of Tran, as applied to claims 1, 12, and 22 respectively, and even further in view of U.S. Patent Application Publication No. 2026/0023842 to Neystadt et al. (hereinafter “Neystadt”).
In regard to claims 5, 15, and 25, as discussed above in regard to claims 1,12, and 22, Lei in view of Schaeck and further in view of Tran teaches applying reinforcement learning from human feedback. Lei in view of Schaeck and further in view of Tran does not explicitly disclose or teach, however, Nyestadt teaches calculating a boost factor based on trainer persona data, the trainer persona data including trainer job title and trainer department (Paras. 4 and 19) (The LLM-generated output is dynamically tailored (i.e., a boost factor) to the role of the inquiring user and/or to one or more characteristics of the inquiring user (i.e., based on trainer persona) and/or to characteristics of the organization data to which the original query pertains… information/output/results/insights that are returned to the querying user, based on the user's role/position (i.e., trainer job title), and/or based on data that indicates which department or team is involved and/or which particular project or task are involved (i.e., trainer department).)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the dynamic tailoring of Nyestadt with the query system of Lei in view of Schaeck and further in view of Tran to provide a more accurate output of the rewriting system.
In regard to claims 8, 18, and 28, as discussed above in regard to claims 1,12 and 22, Lei discloses invoking, via a took execution component, one or more knowledge tools to gather additional data. Lei in view of Schaeck and further in view of Tran does not explicitly disclose or teach, however, Nyestadt teaches selecting at least one knowledge tool from the group consisting of an employee information tool (Paras. 4, 19, 36-39) (The LLM-generated output is dynamically tailored to the role of the inquiring user and/or to one or more characteristics of the inquiring user and/or to characteristics of the organization data to which the original query pertains… information/output/results/insights that are returned to the querying user, based on the user's role/position (i.e., employee information tool)….system may include one or more Data Sources 151, such as …. data about organizational structure (e.g., employee directory, organizational chart), and/or other data-items or information that is available to at least some team-members of the organization).)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the dynamic tailoring based on employee information of Nyestadt with the query system of Lei in view of Schaeck and further in view of Tran to provide a more accurate output of the rewriting system.
In regard to claims 9, 19, and 29, as discussed above in regard to claims 1, 12, and 22, Lei discloses the trainer tools; and weighing the retrieved context during learning from human feedback; and Lei in view of Schaeck and further in view of Tran teaches that the weighing is using reinforcement learning.
Lei in view of Schaeck and further in view of Tran does not explicitly disclose or teach, however, Nyestadt teaches wherein the tools are configured to: retrieve trainer persona data from the data store to personalize the … retrieved context (Paras. 4 and 19) (The LLM-generated output is dynamically tailored to the role of the inquiring user and/or to one or more characteristics of the inquiring user (i.e., personalize the retrieved context) and/or to characteristics of the organization data to which the original query pertains… Large Language Model (LLM) that is operably connected to organizational data sources (i.e., persona data from the data store) of an organization.)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the dynamic tailoring of Nyestadt with the query system of Lei in view of Schaeck and further in view of Tran to provide a more accurate output of the rewriting system.
Claims 10, 20 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Lei in view of Schaeck and further in view of Tran, as applied to claims 1, 12, and 22 respectively, and even further in view of U.S. Patent Application Publication No. 2026/0057016 to Vouitsis et al. (hereinafter “Vouitsis”) and even further in view of U.S. Patent Application Publication No. 2025/0013641 to Kamali et al. (hereinafter “Kamali”).
In regard to claims 10, 20, and 30, Lei in view of Schaeck and further in view of Tran does not explicitly disclose or teach, however, Voutsis teaches wherein the data store comprises: a vector database storing multidimensional vector embeddings of chunked knowledge; and embedding models (Para. 73) (…index a set of documents (i.e., data store comprises a vector database) is vectorization. In vectorization each document (or each chunk of a document) is converted or transformed, by an embedding model, into a plurality of embeddings which are stored as a multi-dimensional vector. The multi-dimensional vector is an array of (floating point) numbers that captures the semantic meaning of the document (or the chunk of a document) (i.e., multidimensional vector embeddings of chunked knowledge).). In other words, the multi-dimensional vector is a numeric representation of the content of a document.)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the vectorization of the chunked data of Voutsis with the query system of Lei in view of Schaeck and further in view of Tran to provide a more efficient rewriting system.
As discussed above, Voutsis teaches that the models are embedding models. Lei in view of Schaeck and further in view of Tran, and even further in view of Voutsis does not explicitly disclose or teach, however, Kamali teaches drift metrics for triggering retraining of …models (Paras. 9, 93) (…additional training data are nominated for retraining the model according to the model uncertainties. Retraining can be triggered by model and/or data drift to retain model accuracy).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the data drift of Kamali the query system of Lei in view of Schaeck and further in view of Tran, and even further in view of Voutsis to provide more efficient model retraining in the rewriting system.
Prior Art
The following prior art, made of record and not relied upon, is considered pertinent to Applicant’s disclosure:
U.S. Patent Application Publication No. 2021/0311939 to Shtilkind et al. (hereinafter “Shtilkind”). Shtilkind discloses processing a search query to determine whether any of the one or more query response modules are applicable to the search query. The selected query response module is executed to determine a module answer content. The search query is processed to identify links to content responsive to the search query. A response to the search query is provided including by embedding the module answer content in a user interface providing at least a portion of the identified links.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Rupangini Singh whose telephone number is 571-270-0192. The examiner can normally be reached on Monday – Friday, 9:30 AM – 6:30 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, Shannon Campbell can be reached on Monday – Friday at (571) 272-5587. 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.
/RUPANGINI SINGH/
Examiner, Art Unit 3628
