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 .
This action is responsive to application filed on 18 August 2025. Claims 1-15 are pending in the case. Claims 1, 14, and 15 are the independent claims. This action is non-final.
Priority
Acknowledgment is made of applicant's claim for foreign priority based on an application filed in Republic of Korea on August 20th, 2024. It is noted, however, that applicant has not filed a certified copy of the KR10-2024-0111139 application as required by 37 CFR 1.55.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claim 15 is rejected under 35 U.S.C. § 101 because the claimed invention is directed to non-statutory subject matter. During examination, the claims must be interpreted as broadly as their terms reasonably allow. In re American Academy of Science Tech Center, 367 F.3d 1359, 1369, 70 U.S.P.Q.2d 1827, 1834 (Fed. Cir. 2004). Independent 1 recites in part, “A computer-readable recording medium having recorded thereon a program,” which is not comprehensively defined by the specification. The broadest reasonable interpretation of a claim thus drawn to a computer-readable recording medium (computing device) encompasses software per se in view of the ordinary and customary meaning of the claimed elements as described in the disclosure. Software per se is not a “process,” a “machine,” a “manufacture,” or a “composition of matter” as defined in 35 U.S.C. § 101.
Applicant is respectfully requested to positively recite specific structural components.
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 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 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or 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, 2, 4, 5, 8, and 10-13 are rejected under 35 U.S.C. 103 as being unpatentable over Cao et al. (US 11,984,115 B2) in view of Deutsch et al. (US 2025/0272513 A1).
Regarding claim 1, Cao teaches a method of providing an interactive agent by using a call sequence, the method comprising:
generating a call sequence based on an input text of a vehicle passenger by using a first language model that is pre-trained (see Cao, [Columns 5-6, Lines 44-67 & 1-46], “FIG. 4 is a block diagram of an example dialog system 400 in accordance with one aspect. The system 400 includes an interface 402 to communicate with a user (e.g. by voice, text, etc.). A speech recognizer 404 converts user's speech to a text string. A user's request may be received by text (e.g. a text message, an email, etc.) A semantics extractor 406 extracts semantics of the user request (e.g. intent, context, entities of the request, etc.) from the text string. … An API mapper 408 (e.g. a recurrent neural networks (RNN) module, a reinforcement learning module, or any machine learning module) may be used to model the dialog policy selection based on the human semantics of the user's request. The output of the API mapper 408 is action feature sequences (e.g. the feature sequence in FIG. 3 : (User identity)→(To learnt destination API)→(User identity, From learnt destination API)→(To personal route API)→(category, From personal route API)→(To place API)→(Vehicle Identity, From place API)→(To vehicle communication API)).” [The dialog system may generate an action feature sequence (i.e., call sequence) using an API mapper (i.e., first language model) based on a communication with a user (i.e., based on input text of a vehicle passenger).]);
However, Cao does not explicitly teach:
generating a call sequence based on an input text of a vehicle passenger by using a first language model that is pre-trained;
Deutsh teaches:
generating a call sequence based on an input text of a vehicle passenger by using a first language model that is pre-trained (see Deutsh¸ Paragraph [0058], “the language model may be trained or fine-tuned to recognize context and relevance patterns in conversations with users and the system may integrate such learned behaviors in determining the relevancy of the custom instructions with the prompt in response generation.” [The language model may be trained or fine-tuned (i.e., pre-trained).]);
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined Cao (teaching dialog system capable of semantic-understanding mapping between user intents and machine services) in view of Deutsch (teaching custom model instructions with language models) and arrived at a method that incorporates a pre-trained language model. One of ordinary skill in the art would have been motivated to make such a combination for the purposes of improving existing language model systems (see Deutsh, Paragraph [0005]). In addition, both the references (Cao and Deutsh) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as application programming interface. The close relation between both the references highly suggests an expectation of success.
The combination of Cao, and Deutsch further teaches:
obtaining information of interest by executing the call sequence (see Cao, [Columns 5-6, Lines 44-67 & 1-46], “The API call module 412 then executes a sequence of APIs to generate a response 414 to the user.” [The API call module may execute a sequence of APIs to generate a response to the user (i.e., obtaining information of interest by executing the call sequence).]);
and generating an output text corresponding to the input text, based on the information of interest, by using a second language model that is pre-trained, wherein the call sequence comprises a plurality of calls (see Cao, [Columns 5-6, Lines 44-67 & 1-46], “The output of the API mapper 408 is action feature sequences (e.g. the feature sequence in FIG. 3 : (User identity)→(To learnt destination API)→(User identity, From learnt destination API)→(To personal route API)→(category, From personal route API)→(To place API)→(Vehicle Identity, From place API)→(To vehicle communication API)). … The API call module 412 then executes a sequence of APIs to generate a response 414 to the user.” [The API call module may execute a sequence of APIs to generate a response to the user (i.e., generating an output text corresponding to the input text based on the information of interest, wherein the call sequence comprises a plurality of calls).] Also, see Deutsch¸ Paragraphs [0051], “a user switching between different models using the same user profile (e.g., the system reloads the custom instructions), a user selects a fine-tuned or custom-trained model (e.g., upon starting a session with a custom model, the system may retrieve and add user-specific instructions that optimize interactions with that model),” [0058], “the language model may be trained or fine-tuned to recognize context and relevance patterns in conversations with users and the system may integrate such learned behaviors in determining the relevancy of the custom instructions with the prompt in response generation.” [Different trained language models may be used (i.e., by using a second language model that is pre-trained).]).
Regarding claim 2, Cao in view of Deutsch teaches all the limitations of claim 1. Cao further teaches:
wherein the call sequence comprises all calls for generating the output text (see Cao, [Columns 5-6, Lines 44-67 & 1-46], “The API call module 412 then executes a sequence of APIs to generate a response 414 to the user.” [The API call module may execute a sequence of APIs to generate a response to the user (i.e., wherein the call sequence comprises all calls for generating the output text).]).
Regarding claim 4, Cao in view of Deutsch teaches all the limitations of claim 1. Cao further teaches:
wherein the generating of the call sequence further comprises preprocessing the input text based on at least one of entity search, dialogue example search, and prompt template application (see Cao, [Columns 3-4, Lines 64-67 & 1-7], “FIG. 2 shows an example process of semantic parsing from a user's voice input. A user makes a voice request (e.g. “find KFC”). A speech recognition module (e.g. Cognitive Service) converts the user's voice input to a text string. A semantics understanding module (e.g. Microsoft Language Understanding Intelligent Service (LUIS)) may parse the text string into human semantics, such as entity, intent, or the like. An API (e.g. HERE maps API, etc.) may be then called for generating an output. In this example, the latitude, longitude, distance, or other parameters of the location of the KFC may be output.” [The user’s input may be parsed into human semantics, such as entity to generate an output (i.e., entity search).]).
Regarding claim 5, Cao in view of Deutsch teaches all the limitations of claim 4. Deutsch further teaches:
generating a first search result including an entity corresponding to at least one string constituting the input text, by using a previously generated entity database (see Deutsch¸ Paragraphs [0041]-[0042], “the system implements session-to-session persistence of custom instructions by storing the custom instructions in persistent data storage, such as a user profile database, or a configuration repository, which is indexed by user profiles or assistant ID's. … when a new session is initiated, the system queries this data storage to retrieve the most recent set of custom instructions associated with the user profile or assistant and adds them to the system message before the language model is engaged.… the system may analyze syntactic and lexical similarities by performing keyword extraction and entity recognition to determine whether key concepts in the prompt overlap with those in the custom instructions” [The system may perform entity extraction on the prompt.]);
generating a second search result including at least one dialogue example having, with respect to the input text, a similarity higher than or equal to a threshold, by using a previously generated dialogue example database (see Deutsch¸ Paragraph [0042], “To determine whether the custom instructions are relevant to the prompt, the system may generate a relevancy or matching score and then compare the score with a threshold.” [The system may generate a relevancy or matching score and then compare with a threshold (i.e., a similarity higher than or equal to a threshold).]);
determining at least one call used in at least one dialogue example included in the second search result, as a target call (see Deutsch¸ Paragraph [0055], “context operations may be performed to determine whether the custom instructions are relevant to the prompt. In some embodiments, context operations may involve calling an external tool or subroutine designed to assess relevance, rather than the language model determining relevancy. In some embodiments, the system may invoke a pattern matching module or embeddings-based similarity engine to compare prompts against stored custom instructions. The pattern matching module can identify specific patterns or keywords in the prompts, while the embeddings-based similarity engine can measure the semantic similarity between the prompts and the stored instructions. This combination can allow the system to match user queries with the relevant instructions or responses.” [Determining whether stored custom instructions are relevant to the prompt may involve calling an external tool or subroutine to match user queries with the relevant instructions or responses (i.e., determining a target call included in the second search result).]);
and generating a first input prompt for the first language model by applying, to the input text, the first search result, and the target call, a first prompt template that is previously generated (see Deutsch¸ Paragraph [0007], “a prompt may include batched queries, wherein multiple requests are processed together for efficiency, or contextually enriched inputs, wherein prior conversation history is added to maintain continuity in responses.” Paragraph [0094], “sending the prompt and custom instructions to the language model may include processing the prompt and custom instructions and transmitting them to the language model for response generation. In some embodiments, this process may involve assembling an input structure that includes multiple components, such as a system message, a prompt, custom instructions, and/or conversation history.” [The prompt may be generated for the language model based on the information determined about the prompt.]).
Regarding claim 8, Cao in view of Deutsch teaches all the limitations of claim 1. Cao further teaches:
wherein the obtaining of the information of interest comprises obtaining the information of interest by sequentially executing at least some of the plurality of calls in a preset order or by executing at least some of the plurality of calls in parallel (see Cao, [Columns 5-6, Lines 44-67 & 1-46], “An API mapper 408 (e.g. a recurrent neural networks (RNN) module, a reinforcement learning module, or any machine learning module) may be used to model the dialog policy selection based on the human semantics of the user's request. The output of the API mapper 408 is action feature sequences (e.g. the feature sequence in FIG. 3 … Unlike feedforward neural networks, RNNs can use their internal state (memory) to process sequences of inputs that is suitable for learning objective, in this case the semantic mapping toward finding the optimal service sequence, with a variable length, to address user's need. … The API call module 412 then executes a sequence of APIs to generate a response 414 to the user.” [The API mapper may determine the most optimal service sequence (i.e., obtaining the information of interest by sequentially executing at least some of the plurality of calls in a preset order).]).
Regarding claim 10, Cao in view of Deutsch teaches all the limitations of claim 1. Cao further teaches:
wherein the generating of the call sequence further comprises generating the call sequence based on the input text and a dialogue history of the vehicle passenger (see Cao, [Column 1, Lines 15-36], “The spoken dialog system 100 requires a component 104 that can track what has happened in a dialog, incorporating system outputs, user speech, context from previous turns, and other external information. The output of the dialog state module 104 is then used by the dialog policy module 106 to decide what action the system should take next.”[Columns 5-6, Lines 44-67 & 1-46], “An API mapper 408 (e.g. a recurrent neural networks (RNN) module, a reinforcement learning module, or any machine learning module) may be used to model the dialog policy selection based on the human semantics of the user's request. The output of the API mapper 408 is action feature sequences (e.g. the feature sequence in FIG. 3” [The call sequence may be generated based on the user’s request, and information tracked from dialog, including system outputs, user speech, and context from previous turns (i.e., generating the call sequence based on the input text and a dialogue history of the vehicle passenger).]).
Regarding claim 11, Cao in view of Deutsch teaches all the limitations of claim 10. Cao further teaches:
wherein the generating of the output text further comprises generating the output text based on the input text and the information of interest (see Cao, [Columns 5-6, Lines 44-67 & 1-46], “The system 400 includes an interface 402 to communicate with a user (e.g. by voice, text, etc.). … An API mapper 408 (e.g. a recurrent neural networks (RNN) module, a reinforcement learning module, or any machine learning module) may be used to model the dialog policy selection based on the human semantics of the user's request. ... The API call module 412 then executes a sequence of APIs to generate a response 414 to the user.” [The response may be generated based on the user’s request and human semantics of the user’s request (i.e., generating the output text based on the input text and the information of interest).]).
Regarding claim 12, Cao in view of Deutsch teaches all the limitations of claim 11. Deutsch further teaches:
wherein the generating of the call sequence further comprises updating the dialogue history by adding the input text to the dialogue history, and the generating of the output text comprises generating the output text based on the updated dialogue history and the information of interest (see Deutsch¸ Paragraph [0090], “the system queries data storage for instructions 706, which may include the process by which Assistant APIs retrieve stored user-specific or system-defined instructions that modify the behavior of the language model in response to prompts.” [The prompt may be updated based on stored information.]).
Regarding claim 13, Cao in view of Deutsch teaches all the limitations of claim 12. Deutsch further teaches:
wherein the generating of the output text comprises generating a second input prompt for the second language model by applying, to the updated dialogue history and the information of interest, a second prompt template that is previously generated (see Deutsch¸ Paragraph [0086], “Another input prompt 604 may state “How would you like the language model to respond?”,” Paragraph [0090], “the system queries data storage for instructions 706, which may include the process by which Assistant APIs retrieve stored user-specific or system-defined instructions that modify the behavior of the language model in response to prompts.” Paragraph [0094], “sending the prompt and custom instructions to the language model may include processing the prompt and custom instructions and transmitting them to the language model for response generation. In some embodiments, this process may involve assembling an input structure that includes multiple components, such as a system message, a prompt, custom instructions, and/or conversation history.” [A second prompt may be generated based on the updated information.]).
Regarding claims 14-15, Cao in view of Deutsch teaches all the limitations of claims 1, 2, 4, 5, 8, and 10-13, in method form rather than in apparatus and computer-readable medium form. Cao also discloses an apparatus [Column 7, Lines 34-43], and computer-readable medium [Column 7, Lines 34-43]. Therefore, the supporting rationale of the rejection to claims 1, 2, 4, 5, 8, and 10-13, applies equally as well to those limitations of claims 14-15.
Claims 3, and 6-7 are rejected under 35 U.S.C. 103 as being unpatentable over Cao in view of Deutsch, further in view of Mishra et al. (US 2025/0245319 A1).
Regarding claim 3, Cao in view of Deutsch teaches all the limitations of claim 1. However, the combination of Cao, and Deutsch do not explicitly teach:
wherein the call sequence comprises a plurality of calls having a nested structure.
Mishra teaches:
wherein the call sequence comprises a plurality of calls having a nested structure (see Mishra, Paragraph [0056], “the sequencer module 209 determines the sequence of the event triggering the second event, which, in turn, triggers the third event to create a chain of API call sequences (i.e., the data propagation chain).” [The sequence module determines the sequence of the event triggering the second event, which triggers the third event to create a chain of API call sequences (i.e., wherein the call sequence comprises a plurality of calls having a nested structure).]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined Cao (teaching dialog system capable of semantic-understanding mapping between user intents and machine services) in view of Deutsch (teaching custom model instructions with language models), further in view of Mishra (teaching system/subsystem anomaly detection via simulating end-to-end data propagation) and arrived at a method that incorporates a nested structure. One of ordinary skill in the art would have been motivated to make such a combination for the purposes of reducing redundancies and organizing data efficiently (see Mishra, Paragraph [0018]). In addition, the references (Cao, Deutsh, and Mishra) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as application programming interface. The close relation between the references highly suggests an expectation of success.
Regarding claim 6, Cao in view of Deutsch teaches all the limitations of claim 1. However, the combination of Cao, and Deutsch do not explicitly teach:
wherein the generating of the call sequence further comprises postprocessing an output of the first language model based on at least one of parsing and slot normalization.
Mishra teaches:
wherein the generating of the call sequence further comprises postprocessing an output of the first language model based on at least one of parsing and slot normalization (see Mishra, Paragraph [0056], “the sequencer module 209 normalizes the API call sequences between web application sub-systems, removing redundancies, and categorizes the data based on web applications.” [The sequencer module may normalize the API call sequences (i.e., postprocessing an output based on slot normalization).]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined Cao (teaching dialog system capable of semantic-understanding mapping between user intents and machine services) in view of Deutsch (teaching custom model instructions with language models), further in view of Mishra (teaching system/subsystem anomaly detection via simulating end-to-end data propagation) and arrived at a method that incorporates normalization. One of ordinary skill in the art would have been motivated to make such a combination for the purposes of reducing redundancies and organizing data efficiently (see Mishra, Paragraph [0018]). In addition, the references (Cao, Deutsh, and Mishra) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as application programming interface. The close relation between the references highly suggests an expectation of success.
Regarding claim 7, Cao in view of Deutsch, further in view of Mishra teaches all the limitations of claim 6. Cao, and Mishra further teaches:
generating a structured output representing a structure of the string by parsing the output of the first language model, which is a string representation; and generating the call sequence by converting at least one string included in the structured output, into a normalized expression by using a slot normalization database that is previously generated (see Cao, [Columns 3-4, Lines 64-67 & 1-7], “A semantics understanding module (e.g. Microsoft Language Understanding Intelligent Service (LUIS)) may parse the text string into human semantics, such as entity, intent, or the like.” [Columns 5-6, Lines 44-67 & 1-46], “The output of the API mapper 408 is action feature sequences” [The text string may be parsed, and a call sequence may be generated.] Also, see Mishra, Paragraph [0056], “the sequencer module 209 normalizes the API call sequences between web application sub-systems, removing redundancies, and categorizes the data based on web applications.” Paragraph [0057], “the analyzer module 211 holds the data and metadata in real time, and continuously monitors the data flow. In an example embodiment, the analyzer module 211 removes redundant transactions, and prepares the data for further analysis. In an example embodiment, the analyzer module 211 removes redundant transactions for every categorized web application sub-system. In an example embodiment, the analyzer module 211 maintains the data and/or metadata of all the transactions happening at any level in real-time. In an example embodiment, the analyzer module 211 maintains the transactions for further analysis.” [The API call sequence may be normalized, and the data may be maintained (i.e., slot normalization database).]).
Claim 9 are rejected under 35 U.S.C. 103 as being unpatentable over Cao in view of Deutsch, further in view of Aninakwa (US 2026/0044682 A1).
Regarding claim 9, Cao in view of Deutsch teaches all the limitations of claim 1. However, the combination of Cao, and Deutsch do not explicitly teach:
wherein the obtaining of the information of interest comprises obtaining the information of interest by sequentially executing the plurality of calls based on a depth-first search technique.
Aninakwa teaches:
wherein the obtaining of the information of interest comprises obtaining the information of interest by sequentially executing the plurality of calls based on a depth-first search technique (see Aninakwa, Paragraph [0109], “The prompt extractor 308 may further support handling nested or hierarchical prompts by applying a recursive or depth-first parsing logic to ensure proper delimitation of embedded custom prompts within standard prompts.” Paragraph [0115], “Upon receipt of a prompt, the agent may parse the input, determine a sequence of required actions (e.g., summarization, classification, prompt expansion, figure interpretation), and orchestrate calls to the appropriate modules or LLMs accordingly.” [The information of interest may be obtained by determining a sequence of calls, and orchestrating the calls based on a depth-first parsing logic (i.e., obtaining the information of interest by sequentially executing the plurality of calls based on a depth-first search technique).]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined Cao (teaching dialog system capable of semantic-understanding mapping between user intents and machine services) in view of Deutsch (teaching custom model instructions with language models), further in view of Aninakwa (teaching automated technical draft generation) and arrived at a method that incorporates a depth-first search technique. One of ordinary skill in the art would have been motivated to make such a combination for the purposes of improving performance and accuracy of understanding natural language (see Aninakwa, Paragraph [0105]). In addition, the references (Cao, Deutsh, and Aninakwa) teach features that are directed to analogous art and they are directed to the same field of endeavor, such as application programming interface. The close relation between the references highly suggests an expectation of success.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
1.) Lu et al. (US 2025/0291559 A1), which discloses a language model-based interface for simulation systems and applications.
2.) Trufinescu et al. (US 2018/0131642 A1), which discloses a method for relating to a conversation run time for automating transitions of conversational user interfaces.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HUSAM TURKI SAMARA whose telephone number is (571)272-6803. The examiner can normally be reached on Monday - Thursday, Alternate Fridays.
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, Apu Mofiz can be reached on (571)-272-4080. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
HUSAM TURKI SAMARA/Examiner, Art Unit 2161
/APU M MOFIZ/Supervisory Patent Examiner, Art Unit 2161
