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 .
Priority
Acknowledgment is made of applicant’s claim for foreign priority under 35 U.S.C. 119 (a)-(d). The certified copy has been filed in parent Application No. KR10-2023-0164108, filed on 11/23/2023.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 11/22/2024 and 03/04/2025 is considered by the examiner.
Drawings
The information disclosure statement (IDS) submitted on 11/22/2024 is 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.
Claim1 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim(s) recite(s), “generating, by a knowledge grounded conversation (KGC) model, utterance corresponding to knowledge and dialogue history; determining whether to refine the utterance based on source-faithfulness score indicating a degree to which the utterance reflects the knowledge; extracting a named entity from the knowledge; and regenerating the utterance based on the knowledge, the utterance, and the named entity.”. The limitation, as drafted is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for recitation of generic computer components. That is, other than reciting "grounded conversation (KGC) model” nothing in the claim element precludes the step from practically being formed in the mind. For example, but for the
"grounded conversation (KGC) model” language "generating" and "determining", “extracting” and “regenerating” in the context of this claims encompasses a first person generate an answer or reply (utterance) in a conversation with another person who asked a question to the first person related to a topic. The first person generate the answer based on determining correctness of the answer whether to present the answer differently based on subject matter knowledge related to the question and recalling past conversation with other persons regarding the topics. If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it fall with the "Mental Process" grouping of abstract ideas. Accordingly, the claim recites an abstract idea.
The judicial exception is not integrated into a practical application. In particular, the claim recites addition elements - generating, by a knowledge grounded conversation (KGC) model, utterance corresponding to knowledge and dialogue history. The use of "knowledge grounded conversation (KGC) model" to generate utterance corresponding to knowledge and dialogue history, is recited at a high-level of generality (i.e. generating by a computing device a response/result to a text input/natural language request) such that it amounts no more than mere instructions to apply the exception using generic computer components.
Accordingly, this additional elements does not integrate the abstract into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea.
The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of “generating, by a knowledge grounded conversation (KGC) model, utterance corresponding to knowledge and dialogue history, steps amounts to no more than mere instructions to apply the exception using a generic computer component. The use of a knowledge grounded conversation (KGC) model, as the claim recites, provide only the idea of a solution or outcome and fails to recite details of how a solution to a problem is accomplished. Without any description of the knowledge grounded conversation (KGC) model for accomplishing the result using generic processor, does not integrate a judicial exception into a practical application or provide significantly more because this type of recitation is equivalent to the words "apply it" (See MPEP 2106.05(f), "The recitation of claim limitations that attempt to cover any solution to an identified problem with no restriction on how the result is accomplished and no description of the mechanism for accomplishing the result, does not integrate a judicial exception into a practical application or provide significantly more because this type of recitation is equivalent to the words "apply it". See Electric Power Group, LLC V. Alstom, S.A., 830 F.3d 1350, 1356, 119 USPQ2d 1739, 1743-44 (Fed. Cir. 2016); Intellectual Ventures I V. Symantec, 838 F.3d 1307, 1327, 120 USPQ2d 1353, 1366 (Fed. Cir. 2016): Internet Patents Corp. V. Active Network, Inc., 790 F.3d 1343, 1348, 115 USPQ2d 1414, 1417 (Fed. Cir. 2015). In contrast, claiming a particular solution to a problem or a particular way to achieve a desired outcome may integrate the judicial exception into a practical application or provide significantly more. See Electric Power, 830 F.3d at 1356, 119 USPQ2d at 1743."). Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
Even when considered in combination, the additional elements represent mere instruction to apply an exception and insignificant extra-solution activity which cannot provide an invention concept. Claim 1, is thus patent ineligible.
With respect to Claim 2, which depends on claim 1, and include all the limitation of claim 1. The limitation of claim 2, “wherein the source-faithfulness score is computed by utilizing Dependency Arc Entailment (DAE) as a scoring function”, similarly a mathematical grouping of abstract idea and considered data gathering and insignificant extra solution activity (See MPEP 2106.05(g)). Accordingly, this additional elements does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
With respect to Claim 3, which depends on claim 2, and include all the limitation of claim 2. The limitation of claim 3, “wherein the determining whether to refine the utterance comprises determining to refine the utterance when the source-faithfulness score is lower than a threshold.”, similarly a mental grouping of abstract idea, such as based on a score the person determines to generate the answer differently and considered data gathering and insignificant extra solution activity (See MPEP 2106.05(g)). Accordingly, this additional elements does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
With respect to Claim 4, which depends on claim 2, and include all the limitation of claim 2. The limitation of claim 4, “wherein the regenerating comprises regenerating the utterance using an utterance regeneration model that is trained to receive at least one of the knowledge, the utterance, and the entity name included in the knowledge and to output the refined utterance.”, similarly a mental grouping of abstract idea but for the use of “utterance generation model”, such as based on recalling a past conversation and knowledge about the topic person determines to generate the answer differently and considered data gathering and insignificant extra solution activity (See MPEP 2106.05(g)). Accordingly, this additional elements does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
Even when considered in combination, the additional elements in claims 2-4, represent mere instruction to apply an exception and insignificant extra-solution activity which cannot provide an invention concept. Claims 2-4, are thus patent ineligible.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 1-4, are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Khosla et al.(Us 20205/0005058 A1).
Regarding Claim 1, Khosla et al teach: A post-hoc utterance refining method, performed by a computing device comprising at least one processor, the method comprising ([0009] Aspects of the present disclosure relate to systems and methods for providing a network-based service for natural language question processing. More specifically, one or more aspects of the present application can include a network-based service for processing natural language queries (e.g., questions, prompts, commands, etc.) provided by a computing device to supplement, optimize, or otherwise modify the natural language query. One or more aspects of the present application can include a network-based service for further processing natural language queries utilizing LLM-based processing resources based on processed queries and processing result validation.): generating, by a knowledge grounded conversation (KGC) model (LLM), utterance (answer) corresponding to knowledge and dialogue history (previous asked questions (or prompts) and previously provided answers) ([0063] The LLM component 106 may also be configured to process multiple questions (or prompts) from a user (e.g., customer of a network-based service or services) of a customer computing device 122 such that the LLM component 106 may utilize previous asked questions (or prompts) and previously provided answers to form an evidence pool. The evidence pool may be used to provide an answer to a current question. The LLM component 106 may store the multiple previous questions as conversational context. [0064] The LLM component 106 may gain access to the user's credentials (e.g., which services they are subscribed to, usage history, knowledge graphs of the customer, etc.) to generate an answer which can include API commands as an answer.); determining whether to refine the utterance (determines if the answer was generated in error) based on source-faithfulness (confirm faithfulness) score (if the score is above a threshold ) indicating a degree to which the utterance reflects the knowledge (determine whether there is a contradiction between the answers and the retrieved passages) ([0045] The aggregator component 104 may also use a similarity score to determine which passages retrieved are relevant (e.g., not out of scope). The retrieved passages may be sent through a dense encoder and to get their dense embeddings. Scores may be generated by the aggregator component 104 for each passage in relativity to the natural language question (e.g., how well the passage is related to the question). [0056] The verifier component 108 may combine the four techniques by the modules above (textual overlap module 234, NLI module 236, relational NLI module 238, membership inference attack module 240) to determine whether an answer will be shown. For example, the verifier component 108 may use the outputs and scores of the above four techniques to create a threshold score and provide a generated answer if the score is above a threshold (e.g., greater than 0.5). [0066] At (7), the LLM component 106 sends the generated answer and retrieved passages to the verifier component 108. At (8), the verifier component 108 determines if the answer was generated in error (e.g., hallucinated). As stated above, the verifier component 108 may look for textual overlap between an answer and retrieved passages, determine whether there is a contradiction between the answers and the retrieved passages, use head/tail/relational triples to confirm faithfulness, use membership inference attacks techniques to confirm whether a question (e.g., or similar) is in a dataset, and/or a score of any of the four combined. At (9), if the answer was not hallucinated, the verifier component 108 sends the answer and retrieved passages to the attribution component 109. ); extracting a named entity (retrieves passages) from the knowledge ([0072] At block 406, aggregator component 104 retrieves passages (e.g., documents, links, API calls, multimedia, etc.) related to the answer from the search systems 124. The aggregator component 104 may retrieve whole documents of from the search systems 124 or retrieve certain text (e.g., inline text) from documents but not the whole document. ); and regenerating the utterance (customer may respond to the LLM component 106 (e.g., or alternatively the natural language question answering service 102) and indicate to the LLM component 106 that the API command should be executed) based on the knowledge, the utterance, and the named entity ([0064] The LLM component 106 may then generate an answer which provides the API command (e.g., to perform the API command on the network-based AI service). After determining the type and kind of API command to generate on behalf of the customer, the LLM component 106 may then send the API command to the customer with a request or prompt as to whether the API command should be executed against the network-based service. The customer may respond to the LLM component 106 (e.g., or alternatively the natural language question answering service 102) and indicate to the LLM component 106 that the API command should be executed. After receiving a request to execute the API command, the LLM component 106 may execute the API command against the network-based service and send the customer a confirmation that the API command finished successfully.).
Regarding Claim 2, Khosla et al teach: The method of claim 1, wherein the source-faithfulness score is computed by utilizing Dependency Arc Entailment (DAE) (metrics techniques such as a recall calculation) as a scoring function (See rejection of claim 1 and [0052] The textual overlap module 234 may determine the textual overlap between an answer and the retrieved passages by using metrics techniques such as a recall calculation (e.g., true positive\(true positive+false negative)), an F1 score (e.g., 2×(precision*recall\precision+recall)), and the like. ).
Regarding Claim 3, Khosla et al teach: The method of claim 2, wherein the determining whether to refine the utterance comprises determining to refine the utterance when the source-faithfulness score is lower than a threshold (See rejection of claim 1 and [0056] The verifier component 108 may combine the four techniques by the modules above (textual overlap module 234, NLI module 236, relational NLI module 238, membership inference attack module 240) to determine whether an answer will be shown. For example, the verifier component 108 may use the outputs and scores of the above four techniques to create a threshold score and provide a generated answer if the score is above a threshold (e.g., greater than 0.5). [0066] At (7), the LLM component 106 sends the generated answer and retrieved passages to the verifier component 108. At (8), the verifier component 108 determines if the answer was generated in error (e.g., hallucinated). As stated above, the verifier component 108 may look for textual overlap between an answer and retrieved passages, determine whether there is a contradiction between the answers and the retrieved passages, use head/tail/relational triples to confirm faithfulness, use membership inference attacks techniques to confirm whether a question (e.g., or similar) is in a dataset, and/or a score of any of the four combined. At (9), if the answer was not hallucinated, the verifier component 108 sends the answer and retrieved passages to the attribution component 109.).
Regarding Claim 4, Khosla et al teach: The method of claim 3, wherein the regenerating comprises regenerating the utterance using an utterance regeneration model (aggregator component 104) that is trained to receive at least one of the knowledge, the utterance, and the entity name included in the knowledge and to output the refined utterance (See rejection of claim 3 and [0060] At (2), the natural language question answering service 102 utilizes the aggregator component 104 to determine relevant passages to retrieve from search systems 124. As described herein, the aggregator component 104 may utilize partial string techniques and DPR techniques to determine the meaning of the natural language question and also determine which network-based services or computing domains, passages should be retrieved from. At (3), the aggregator component 104 retrieves the passages from the search systems 124. Moreover, the aggregator component 104 may modify and supplement the question with the retrieved passages to form a prompt. The prompt may comprise selected passages and QA pairs for the LLM component 106 to provide an answer to. [0064] The LLM component 106 may then generate an answer which provides the API command (e.g., to perform the API command on the network-based AI service). After determining the type and kind of API command to generate on behalf of the customer, the LLM component 106 may then send the API command to the customer with a request or prompt as to whether the API command should be executed against the network-based service. The customer may respond to the LLM component 106 (e.g., or alternatively the natural language question answering service 102) and indicate to the LLM component 106 that the API command should be executed. After receiving a request to execute the API command, the LLM component 106 may execute the API command against the network-based service and send the customer a confirmation that the API command finished successfully. [0066] At (7), the LLM component 106 sends the generated answer and retrieved passages to the verifier component 108. At (8), the verifier component 108 determines if the answer was generated in error (e.g., hallucinated). As stated above, the verifier component 108 may look for textual overlap between an answer and retrieved passages, determine whether there is a contradiction between the answers and the retrieved passages, use head/tail/relational triples to confirm faithfulness, use membership inference attacks techniques to confirm whether a question (e.g., or similar) is in a dataset, and/or a score of any of the four combined. At (9), if the answer was not hallucinated, the verifier component 108 sends the answer and retrieved passages to the attribution component 109.).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. The prior art of record Bayless et al.(US 2025/0111192 A1), GENERATING KNOWLEDGE GRAPHS USING LARGE LANGUAGE MODELS, teach : Techniques for a knowledge-graph system to use large language models (LLMs) to build knowledge graphs to answer queries submitted to a chatbot by users. The knowledge-graph system builds the knowledge graph using answers produced by an LLM for novel queries. The chatbot will continue to use the LLM to answer novel queries, but the chatbot may harness the knowledge graph to answer repeat questions to gain various efficiencies over LLM-backed chatbots. For example, the knowledge-graph system may easily debug or otherwise improve the answers in knowledge graphs, store provenance information in knowledge graphs, and augment the knowledge graphs using other data sources. Thus, the reliability and correctness of chatbots will be improved as the bugs and inaccuracies in answers provided by the LLM will be corrected in the knowledge graphs, but the chatbots can still harness the abilities of LLMs to provide answers across various subject-matter domains.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MOHAMMAD K ISLAM whose telephone number is (571)270-5878. The examiner can normally be reached Monday -Friday, EST (IFP).
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, Paras Shah can be reached at 571-270-1650. 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.
/MOHAMMAD K ISLAM/Primary Examiner, Art Unit 2653
Prosecution Insights