DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This Office Action is in response to the claim amendment filed on May 14, 2026 and wherein claims 1-3, 5-9, 11-14, 16-17, 19 amended.
In virtue of this communication, claims 1-20 are currently pending in this Office Action.
With respect to the rejection of claims 1-20 under 35 USC §101, as set forth in the previous Office Action, the Applicant’s amendment, and argument, see paragraphs 3-4 of page 15, paragraphs 1-2 of page 16, and paragraphs 1-2 of page 17 in Remarks filed on May 14, 2026, have been fully considered and wherein applicant has referred claimed limitations to the specification as cited in the paragraphs above, and thus, the argument is persuasive. Therefore, the rejection of claims 1-20 under 35 USC §101, as set forth in the previous Office Action, has been withdrawn.
The Office appreciates the explanation of the amendment and analyses of the prior arts, and however, although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993) and MPEP 2145.
Examiner Comment
Claim 14 is submitted with claim status “(original)”, and applicant Remarks stated “Claims 1-3, 5-9, 11-13, 16-17, and 19 have been amended herein …” which has excluded any amendment of claim 14, and however, claim 14 added limitations underscored and crossed original limitations, which is confusing because it is unclear whether amendment claim 14 should be considered or original claim 14 should be considered in the prosecution and it appears to have “original” that should be replaced with “Currently amended” and Remarks should include claim 14 that is amended “herein”. A correction is required accordingly.
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 of this title, 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.
Claims 1-2, 4-7, 10-12, 14-20 are rejected under 35 U.S.C. 103 as being unpatentable over Sen et al. (US 20200073983 A1, hereinafter Sen) and in view of reference Nouri et al. (US 20240038226 A1, hereinafter Nouri).
Claim 11: Sen teaches a system (title and abstract, ln 1-9, a system in fig. 1) comprising one or more processors (one or more processors 16 in fig. 3, para 112) to
generate, based at least on a logical reasoning engine (including domain reasoning engine 147, etc., in fig. 1) processing a translated representation of one or more natural language statements representative of content of a query (including an annotated parse tree outputted from nested query detection module 143, corresponding to the query, para 21), a representation of one or more logical assessments of the one or more natural language statements expressed in a logic specification language (the nested query class type determined to have logic operators, such as NOT as class 1, number in class 2, whether a condition is satisfied for an entity retrieval in class 3, etc., in the set of token words, phrases, or clauses, para 19, 23-30, as expressed in logic specification language for assessing domain intelligence of the set of tokens in the tree and identifying and determining subqueries and a joint condition for the subqueries, para 36); and
apply, to modules (query interpretation module 148, query execution module 151, database 152, etc.), one or more prompts (generated an outer portion of query 160-1 and an inner portion of query 160-2 combined with a joint condition 160-3, outputted from the domain reasoning engine 147, to the query interpreter module 148) to cause the modules to generate a natural language response to the query (the query results 154 generated and stored through a logical construct of the nested query generated by the query interpreter module 148, and followed by the query execution module 151) based at least on a representation of the one or more logical assessments of the one or more natural language statements (based on at least the outer portion of query 160-1 and inner portion of query 160-2 as a representation of natural language query 160 for generating the final results 154, para 37).
However, Sen does not explicitly teach the modules are one or more language models LMs.
Nouri teaches an analogous field of endeavor by disclosing a method (title and abstract, ln 1-17 and method steps in fig. 2) and wherein one or more language models LMs are disclosed (including general language models practiced, para 52) to have applied one or more prompts to (applying additional questions with user feedback, as one or more prompts, to the language model 200, steps 222-226, para 45-46) for generating a natural language response to a query (revised task specific output at 226, as response to received task specific request at 202) based on a representative of one or more logical assessments of one or more natural language statements (the further questions to and answers from the user or instructions in the feedback as the representative of evaluation of question-answer pairs as revised question-answer pairs, i.e., logical assessment) for benefits of improving efficacy by using the machine learning model (improving accuracy of the generative tasks by matching one or more keywords, para 35, enhancing text representation based on given answers, para 62, and improving efficacy of user interface, para 2-3).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have applied the one or more language models, as taught by Nouri, to the modules in the system, as taught by Sen, for the benefits discussed above.
Claim 19 recited a method comprising essential limitations of claim 11 and thus, claim 19 rejected according to claim 11 above and the combination of Sen and Nouri further teaches,
obtaining, based at least on a logical reasoning engine (Sen, including domain reasoning engine 147, etc., in fig. 1 and Nouri, part of language model for generating question-answer pairs) processing a translated representation that expresses content of a query in a particular language (Sen, the annotated parse tree or the annotated token sets processed by domain reasoning engine 147 and the annotated parse tree or token sets as translated representation of the parse tree expressing the natural language query 160 in natural language with tree structure and Nouri, a list of answer choices is generated by processing the generated question-answer pairs at 206, para 37-38), a representation of one or more logical assessments of the content of the query (Sen, the nested query class type, including NOT as class 1, whether a condition is satisfied for an entity retrieval in class 3, etc., para 23-30, i.e., representation of logic assessments of content of the natural language query 160, para 36, and Nouri, generated task specific output for further processing at 216, para 42); and
obtaining, based at least on one or more language models LMs (Sen, modules, and Nouri, the general language models, para 52 and discussed in claim 11 above) processing one or more prompts (Sen, the generated outer portion of query 160-1, inner portion of query 160-2 combined with the joint condition 160-3 from the domain reasoning engine 147 are processed and Nouri, the further questions and feedback further processed by the language model 200, para 45-46) that cause the one or more LMs to evaluate the representation of the one or more logical assessments of the content of the query (Sen, forming a logic construct of the nested query based on the elements 160-1, 160-2, and 160-3, para 37 and Nouri, evaluating by revising the task-specific output at 226), a response to the query in the particular language or at least one other language (Sen, the query results 154 obtained for storage through the formed logical construct, para 37 and Nouri, the revised task-specific output 226, para 46 and discussed in claim 11 above).
Claim 14: the combination of Sen and Nouri further teaches, wherein the one or more processors are further to generate the translated representation of the one or more natural language statements (Sen, an annotated parse tree or the tokens sets are generated by semantic parsing module 145, etc., para 43-46 and Nouri, the generated task specific output at 216) based at least on applying, to the one or more LMs, one or more initial prompts (Sen, a parse tree outputted from NLQ parser module 141, para 19 and Nouri, by applying the revised question-answer pairs to the language model 200 at step 214-216) to cause the one or more LMs to convert the one or more natural language statements representative of the content of the query into the translated representation of one or more natural language statements (Sen, the annotated parse tree as translated form of the parse tree and containing the content of the natural language query, e.g., based on morphological analyses and syntactic analysis, etc., para 19 and Nouri, translating the task specific request to the task specific output 216).
Claim 1 has been analyzed and rejected according to claims 11, 14, 19 and the combination of Sen and Nouri further teaches, one or more processors comprising circuitry (Sen, circuitry in fig. 5 and processors or processing units 16 in fig. 3, para 112 and Nouri, task processor 104 in fig. 1 and with circuitry in figs. 5-6) to:
apply, to one or more language models LMs (Sen, included in nested query detection module 143 and as one of combination of modules 141, 143, 148, 151, etc., in fig. 1 and Nouri, language models discussed in claim 1 above), one or more first prompt (Sen, formed trees prior to annotation and output from the NLQ parser Module 141 and Nouri, generated question-answer pairs by replacing generated answers with received answers 214 in fig. 2) to cause the one or more LMs to convert content of a query into one or more logical statements that express the content of the query in a logic specification language (annotated tree with semantic labels on their determined semantic type from the tree structure, para 39, 43-44 and Nouri, the task specific output based on the question-answer pairs 216);
generate, using a logical reasoning engine corresponding to the logic specification language (Sen, including domain reasoning engine 147 and receiving the nested query class label and parsed token sets, para 36), a representation of one or more logical assessments of the one or more logical statements that express the content of the query (Sen, generating outer portion 160-1, inner portions 160-2, and join clause to join subqueries, etc., as representation of the nested query class of natural language query 160, para 36 and as discussed in claim 11 above);
apply, to the one or more language model LMs (Sen, including query execution module 151, database 152, etc. and Nouri, the language models 200 in fig. 2 and 130 in fig. 1), one or more second prompts (Sen, the logic construct of a nested query based on generated 160-1, 160-2, with 160-3 from the domain reasoning engine 147 and discussed in claim 11 above, and Nouri, the additional questions with user feedback, as one or more prompts, applied to the language model 200, steps 222-226, para 45-46) to cause the one or more LMs to generate a natural language response comprising at least one of an explanation of, or an answer to a question about, the content of the query (Sen, the query results 154 through a logical construct of the nested query by the element 148, 151, etc., and discussed in claim 11 above and Nouri, the revised task specific output at 226, as response to received task specific request at 202) based at least on the representation of the one or more logical assessments (Sen, based on at least the elements 160-1, 160-2 as a representation of natural language query 160 for generating the final results 154, para 37 and discussed in claim 11 above, and Nouri, based on the further questions to and answers from the user and instructions in the feedback of generated revised question-answer pairs, i.e., logical assessment, as discussed in claim 11 above); and
provide the natural language response as an output (Sen, providing the results to a datastore 154 and Nouri, and outputting the revised task-specific output 226 to the ender user, e.g., through a display for the output 220).
Claim 2: the combination of Sen and Nouri further teaches, according to claim 1 above, wherein the processing circuitry is further to generate the one or more first prompts (Sen, formed trees prior to annotation and output from the NLQ parser Module 141 and Nouri, generated question-answer pairs by replacing generated answers with received answers 214 and discussed in claim 1 above) based at least on inserting the one or more input statements into one or more template prompts (Sen, parsed tree with tokens as filled in tree structure with the words, phrase, and clause of the natural language query, para 19, and Nouri, the request task with the contextual information from the task processor to the language model in fig. 2, question plus contextual as template prompts) that instruct the one or more LLMs to convert the one or more input statements into the logic specification language (Sen, the annotated tree with phrases, clauses, para 43-44, and Nouri, language model is instructed and caused to perform converting the revised question-answer pairs to the task specific output and/or additional questions, from the task request with parameters via 204-206 in fig. 2).
Claim 4: the combination of Sen and Nouri further teaches, according to claim 1 above, wherein the logical reasoning engine implements one or more solvers (Sen, semantic reasoning is performed on the set of tokens to associate the tokens with tokens by join action, and formulating data structure for the following query interpreter module 148, para 36-37).
Claim 5: the combination of Sen and Nouri further teaches, according to claim 1 above, wherein the one or more logical assessments of the one or more logical statements represent at least one of: a proof or refutation of one or more statements representing the content of the query, a deduced fact based at least on the one or more statements, or a consistency check of the one or more statements (Markush rule applied, see MPEP 2117, Sen, parsing the set of tokens into the outer and inner portions of the query, i.e., deduced fact, and determining the join correlation of the portions by associating the join condition, para 55, including semantic type “NOT”, “NotEqual”, “Numeric”, “Instance”, and other semantic types, para 44).
Claim 6: the combination of Sen and Nouri further teaches, according to claim 1 above, wherein the processing circuitry is further to generate the one or more second prompts using retrieval augmented generation to augment the one or more second prompts with representation of retrieved content explaining the logic specification language (Sen, the logic construct from the outer and the inner portions of the query and augmented by the join for logic correlation of the content of the subqueries, para 37, and according to the semantic type, para 44, and Nouri, the additional questions and augmented with instructions or answers in the user feedback, para 45-46, and further answers to additional questions, para 57-60).
Claim 7: the combination of Sen and Nouri further teaches, according to claim 1 above, wherein the natural language response comprises an explanation of the one or more logical assessments, or an answer to a question about the content of the query based at least on the one or more logical assessments (Sen, an answer to query clause of class 2, e.g., the answer to “what is the number of companies generating average revenues more than 1 billion in 2017?”, or answer to the request “Show me stocks having highest traded value in 2018” in class 3 type, etc., para 22-30, and results from executing the converted SQL query, para 37 and Nouri, the task specific output in fig. 4C, including date/time, location, style, members, etc., from the query keywords “wedding ceremony”, “reception” in the query, para 35).
Claim 10: the combination of Sen and Nouri further teaches, according to claim 1 above, wherein the processing circuitry is comprised in at least one of: a control system for an autonomous or semi-autonomous machine; a perception system for an autonomous or semi-autonomous machine; a system for performing simulation operations; a system for performing digital twin operations; a system for performing light transport simulation; a system for performing collaborative content creation for 3D assets; a system for performing deep learning operations; a system for performing remote operations; a system for performing real-time streaming; a system for generating or presenting one or more of augmented reality content, virtual reality content, or mixed reality content; a system implemented using an edge device; a system implemented using a robot; a system for performing conversational AI operations; a system for generating synthetic data; a system for generating synthetic data using AI; a system incorporating one or more virtual machines VMs; a system implemented at least partially in a data center; or a system implemented at least partially using cloud computing resources ( Markush rule applied, MPEP 2117, Sen, service providers, business intelligence system, para 16, and Nouri, smartphone for performing autonomous or semi-autonomous machine for smartphone, para 69, perception system for an autonomous or semi-autonomous machine such as temperature control for the model, maximize outputs size for the model, etc., para 56).
Claim 12: the combination of Sen and Nouri further teaches, according to claim 11 above, wherein the processing circuitry is further to generate the translated representation of the one or more natural language statements (Sen, including an annotated parse tree outputted from nested query detection module 143, corresponding to the query, para 21 and Nouri, the task request with the keywords and then generating a list of the question-answer at steps 204-206 in fig. 2) based at least on inserting the one or more natural language statements into one or more template prompts (Sen, using semantic type label placed to the parsed tree to generate annotated parse tree, and identified nested query by nested query detection module 143, and Nouri, the request task with the contextual information from the task processor to the language model in fig. 2, question plus contextual as template prompts) that instruct the one or more LMs to convert the one or more input statements into the logic specification language (Sen, by using identified various types of joins conditioned to the subqueries, para 31-32, and Nouri, language model is instructed and caused to perform converting or generating a list of relevant question-answer from the task request with parameters via 204-206 in fig. 2).
Claim 15 has been analyzed and rejected according to claims 11, 5 above.
Claim 16 has been analyzed and rejected according to claims 11, 6 above.
Claim 17 has been analyzed and rejected according to claims 11, 7 above.
Claim 18 has been analyzed and rejected according to claims 11, 10 above.
Claim 20 has been analyzed and rejected according to claims 19, 10 above.
Claims 3, 13 are rejected under 35 U.S.C. 103 as being unpatentable over Sen (above) and in view of reference Nouri (above) and Yang et al. (CN 117076607 A, hereinafter Yang, its translation and original attached herein and its translation applied in paragraph citation under Li below).
Claim 3: the combination of Sen and Nouri further teaches, according to claim 1 above, wherein the processing circuitry is further to generate the one or more logical statements using a first large language model of the one or more LMs (Sen, annotated tree with semantic labels on their determined semantic type from the tree structure, para 39, 43-44 and Nouri, the task specific output based on the question-answer pairs 216, and the language processing to generating question-answer list which is logically related to the task request with parameters via steps 204-206 through the language model in fig. 2, and the language model is large language model, para 24), except explicitly teaching wherein the first large language model tuned to the logic specification language.
Yang teaches an analogous field of endeavor by disclosing a method (title and abstract, ln 1-13 and method steps in figs. 1-2) and wherein the first large language model is disclosed (a large language model, abstract) to be tuned to the logic specification language (when the large language model is not continuously trained, S103, the large language model is provided with an example of conversion from natural language text to logical expression, and the natural language text in order to obtain logic expression from the large language model, the last paragraph of page 7,and the 1st paragraph of page 8, i.e., the large language model is trained to accept a conversion from the natural language to a language representing the logic expression, and output the logic expression based on the input of the natural language text corresponding to the logic expression) for benefits of improving the efficiency of the large language model (by saving computation power, and adapted for diversified requirements, the last paragraph of page 7, with more accurate output of the logic expression corresponding to the natural language text from the large language model in an automatic, efficient, and convenient manner, para 3 of page 8).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have applied wherein the first large language model tuned to the logic specification language, as taught by Yang, to the first large language model implemented by the one or more processors, as taught by the combination of Sen and Nouri, for the benefits discussed above.
Claim 13 has been analyzed and rejected according to claims 11, 3 above (the translated representation of the one or more natural language statements in claim 13 mapped to the one or more logical statements in claim 3).
Claims 8-9 are rejected under 35 U.S.C. 103 as being unpatentable over Sen (above) and in view of references Nouri (above) and Suenbuel et al. (US 20150261744 A1, hereinafter Suenbuel).
Claim 8: the combination of Sen and Nouri further teaches, according to claim 1 above, the content of the query (Sen, nested query “show me stocks …”, “what is …?”, etc., para 23-29, and Nouri, the content is about travel plans, user biography, generating a dialog, written works, event summaries, etc., para 34), except explicitly teaching wherein the content of the query comprises a sequence of image data of a video.
Suenbuel teaches an analogous field of endeavor by disclosing one or more processors comprising processing circuitry (title and abstract, ln 1-11, DSP 320, general-purpose processor 310 in fig. 3 and comprising processing circuitry including connected GNSS receiver 360, wireless transceiver 330, connected by bus 301, para 22) and wherein a content of a query is disclosed (user input including a query or question, and containing multiple requests for information, para 25) and the content of the query comprising a sequence of image data of a video (including image input component by one or more cameras, para 66) for benefits of improving performance of the query system (by not just matching of terms, but based on proof search during reasoning processing, para 14, 33-34, and by deducing logic relationship from preset attributes without explicitly wording, para 42-44 and handing multiple questions in the same query, para 25, 47).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have applied the content of the query that comprises the sequence of image data of video, as taught by Suenbuel, to the content of the query processed by the processing circuitry of the one or more processors, as taught by the combination of Sen and Nouri, for the benefits discussed above.
Claim 9: the combination of Sen, Nouri, and Suenbuel further teaches, according to claim 1 above, wherein the content of the query comprises a sequence of time-series data (Sen, the content of the query, discussed in claim 8 above, and Nouri, the query involved in user’s first task specific request and then later, selection of answer in the question-answer pairs provided by the language model, etc., and discussed in claim 1 above, and Suenbuel, multiple questions in series in the query, e.g., “Show a company with a debt of more than 100 million dollars. What is the nationality of the client?” for different databases to be accessed, para 25), wherein the natural language response comprises at least one of: an explanation of, or an answer to a question about, at least a portion of the time-series data (Suebuel, incorporating answer to previous questions in one query to answer additional questions in the same query in fig. 7B, para 52).
Response to Arguments
Applicant's arguments filed on May 14, 2026 have been fully considered and but are moot in view of the new ground(s) of rejection necessitated by the applicant amendment. Although a new ground of rejection has been used to address additional limitations that have been added to claims 1-3, 5-9, 11-14, 16-17, 19, a response is considered necessary for several of applicant’s arguments since reference Nouri will continue to be used to meet several claimed limitations.
With respect to the prior art rejection of independent claim 1, under 35 USC §103(a), as set forth in the Office Action, applicant argued: “None of the cited passages appears to involve the claimed one or more first prompts” and neither “converting content of a query into one or more logical statements that express the content of the query in a logic specification language”, as asserted in paragraph 2 of page 13 in Remarks filed on May 14, 2026.
In response to the argument cited above, the Office respectfully disagrees because (1) claim 1 broadly claimed “one or more first prompts” that merely “cause the one or more LMs to convert content of a query into one or more logical statemen” and thus, its BRI is applied (MPEP 2111) and thus, (2) any inputs to the language models LMs would be interpreted as “prompt” or “prompts” because any inputs would cause the LMs to perform functions (Nouri, e.g., converting task specific request in the query into question-answer pairs 202-204 and generating the list of answer choices in the question-answer pairs) and the function is to “convert content of a query into one or more logical statements …” (Nouri, questions-answers pairs self are logical statements reflecting the content of the query in a logic form or specification language), which is essentially consistent with the argued feature about “first one or more prompts” and “cause”, but applicant is in silence. Applicant appears to look at exact wording of “prompt” and “converting”, etc., in the prior art, but however, no exact wording to the claimed term does not mean no teaching of claimed term by the prior art.
Applicant further argued “one or more second prompt” applied to “language model” to cause the one or more LMs to generate natural language response”, etc., as asserted in paragraph 3 of page 14 in Remarks filed on May 14, 2026.
In response to argument above, the Office further disagrees because (1) claim 1 broadly recited “one or more second prompt” with no recitation of what it is and but merely “cause” the LMs to perform function, and therefore, similarly, its BRI would be applied and under its BRI, Nouri clearly teaches receiving inputs (user’s further answers with feedback 224 in fig. 2) that cause the LMs(200) to perform revise the task specific output (226), and forming a final task specific output to be displayed on the task processor of client (220), which is also essentially consistent with the argued features about “one or more second prompt” and “cause” above, but applicant is also silence. In order to progress the prosecution, prior art Sen is applied to further meet the argued features above.
In the response to this office action, the Office respectfully requests that support be shown for language added to any original claims on amendment and any new claims. That is, indicate support for newly added claim language by specifically pointing to page(s) and line numbers in the specification and/or drawing figure(s). This will assist the Office in prosecuting this application.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LESHUI ZHANG whose telephone number is (571)270-5589. The examiner can normally be reached Monday-Friday 6:30amp-4:00pm EST.
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, Vivian Chin can be reached at 571-272-7848. 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.
/LESHUI ZHANG/
Primary Examiner,
Art Unit 2695
Prosecution Insights