CTNF 18/830,344 CTNF 98511 Notice of Pre-AIA or AIA Status 07-03-aia AIA 15-10-aia 1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA. Claim Objections 07-29-01 AIA 2. Claim s 11-12 are objected to because of the following informalities: in claim 11, please correct “based on which the the first generative artificial intelligence model predicts” in line 13 to “based on which the the first generative artificial intelligence model predicts” . Appropriate correction is required. Claim Rejections - 35 USC § 103 07-06 AIA 15-10-15 3. 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. 07-20-aia AIA 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. 07-23-aia AIA The factual inquiries 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. 07-21-aia AIA 4. Claim s 1-30 are rejected under 35 U.S.C. 103 as being unpatentable over Rahmfeld (US 2021/0390099) in view of Tomkins (US 2021/0294828) . Regarding Claim 1: Rahmfeld discloses a computer-implemented method comprising: receiving, at a digital assistant, a natural language utterance from a user during a session between the user and the digital assistant (Rahmfeld: ¶18 discloses receiving a natural language utterance from a user because the system receives user input utterances, i.e., natural language queries, through a user interface. The utterance is received as part of a data conversation, which corresponds to a session between the user and the virtual assistant) ; obtaining a topic context instance for the natural language utterance (Rahmfeld: ¶21 discloses the topic view may comprise a listing of one or more topics where each topic comprises one or more columns and their respective tables, join conditions and other metadata, ¶35 discloses that the effective schema identification component may identify a topic and/or one or more references steps of a data conversation that should serve as the target utterance, this teaches obtaining a topic context instance by identifying a topic or referenced conversation step that serves as the target context for the utterance, the columns, tables and metadata are all topic specific context used for interpretation of the utterance) , wherein the obtaining comprises: executing, based on the natural language utterance, a search on a current session context instance, a data store, or both (Rahmfeld: ¶18 discloses new utterances may leverage previous conversation steps (including e.g., utterances, structured query language query and result set data conversation) and ¶52 discloses the ability to determine whether the utterance includes one or more references to previous steps, including in the data conversation and determining a topic and/or a result set from a previous step. This teaches searching current conversational context because it determines whether the new utterance references previous steps, including prior utterances, SQL statements, result sets and metadata. It also teaches using a data store because the selected topic maps to tables and columns of an underlying schema in data store(s) and the translated utterance is executed against the relevant data sets) , based on the search, determining whether the natural language utterance satisfies a threshold level of similarity with one or more topics represented in the current session context instance, the data store, or both (Rahmfeld: ¶84 discloses the effective schema identification layer 134 may determine the effective schema using a machine learning prediction model, the selection model may generate N utterance schema pairs based on N number of effective schema options. ¶92 discloses the SoftMax layer 280 may generate the probabilities for each topic/step to hold the answer for the utterance 114. The topic/step with the largest probability may be chosen. This altogether teaches comparing the utterance to candidate topics or steps by forming utterance schema pairs for each possible topic or referenced step and using a machine learning model to generate probabilities. Selecting the topic/step having the largest probability teaches determining that the utterance corresponds to a topic or previous step) , responsive to determining the natural language utterance satisfies the threshold level of similarity with the one or more topics, identifying the topic context instance associated with the one or more topics (Rahmfeld: ¶18 and ¶82 discloses the model determines the topic or previous step most likely to contain the answer, Rahmfeld identifies the corresponding effective schema. That effective schema, including the topic’s columns and metadata or the previous result set columns, corresponds to the claimed topic context instance associated with the identified topic) , and associating the natural language utterance with the topic context instance (Rahmfeld: ¶50-52 associates the utterance with the selected topic/effective by making the utterance part of a conversation step that includes the generated SQL, result set and metadata. The utterance with the selected topic or previous step provides the effective schema for the utterances, so the utterance is tied to the topic specified context user for translation execution) ; generating, by a first generative artificial intelligence model, a list comprising one or more executable actions based on one or more candidate actions associated with the topic context instance (Rahmfeld: ¶22 discloses the query conversations may occur via one or more machine learning models that may determine which (subset of a) schema should be included in a translation task for a given utterance, identify whether and how elements from previous data conversation steps may be incorporated into the query, and translate the utterance into Structured Query Language (SQL). ¶117 discloses one or more intermediate SQL output tokens that form an intermediate SQL statement and ¶123 discloses intermediate SQL may encode one or more actions where an action may not directly correspond to standard SQL function of keyword. Altogether this teaches a machine-learning encoder/decoder model that generates intermediate SQL tokens forming an intermediate SQL statement. This intermediate SQL can encode one or more actions, including higher-level actions. These actions are generated based on the selected effective schema topic and the utterance. Therefore, the encoder/decoder translation model corresponds to the claimed generative AI model and the generated intermediate SQL actions correspond to the claimed executable actions) ; creating, based on the list, an execution plan comprising the one or more executable actions (Rahmfeld: ¶36-37 and ¶124 teaches creating an execution plan because the intermediate SQL identifies the operations to be performed and the post processing layer converts those operations into executable SQL statements. The executable SQL specifies the concrete operations to be carried out against the data store or previous result set) ; generating an updated topic context instance by updating the topic context instance to include the execution plan (Rahmfeld: ¶47, discloses a conversation history window, ¶50 discloses a result set from query execution and ¶57 discloses the result may be included with the utterance of the present step. This altogether discloses updating the conversation/topic context because, after the utterance is translated and executed the present conversation step is updated to include the generated SQL query, result set and metadata. That updated conversation step becomes part of the conversation history and can be used by later utterances. Therefore, Rahmfeld updates the topic/conversation context to include the generated executable SQL execution plan and it is associated results and metadata) ; executing the execution plan based on the updated topic context instance, wherein the executing comprises executing the executable action using an asset to obtain an output (Rahmfeld: ¶56 discloses executing the execution plan because the post processed executable SQL statement is executed against the underlying database schema or datastores. The data query engine is the claimed asset and the result set produced from executing the SQL statement is the claimed output) ; and sending the output or a communication derived from the output to the user (Rahmfeld: ¶57 teaches sending the output to the user because the result set and metadata from executing the statement are communicated and displayed through the user interface) . Rahmfeld does not explicitly disclose a threshold . However, Tomkins discloses an express threshold (Tomkins: ¶39 teaches the threshold concept because it selects an ontology graph vertex only when the distance between embedding vectors satisfies a distance threshold) . Rahmfeld and Tomkins are combinable because they are from the same field of endeavor, specifically natural language processing, e.g., both disclose methods for processing natural language and generating model output for question answering. Rahmfeld teaches determining the relevant topic and step for a natural language utterance by generating utterance schema pairs for candidate topics and producing probabilities for each topic step and selecting the most likely answer, however it does not explicitly state comparing to a threshold, but it does compare similarity to known topics. Tomkins does teach selecting a concept based on an embedding vector distance satisfying a distance threshold. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to disclose Tomkin’s threshold based similarity determination into Rahmfeld’s probability based topic selection process. The suggestion/motivation for doing so is “some embodiments may improve the speed and accuracy of responses to queries” as disclosed in ¶41. Regarding Claim 2: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1, wherein generating the list comprises selecting the one or more executable actions from the one or more candidate actions based on each of the one or more executable actions satisfying a threshold level of similarity with the natural language utterance and context within the topic context instance (Rahmfeld: ¶35 discloses using the natural language utterance and the conversation context to identify the relevant or referenced prior step, this selected topic step is the context used for generating the later action, ¶82 discloses the columns configured for the topic and corresponding meta-data, or the columns of the result set of the referenced step respectively, may constitute the effective schema which maps to the claimed context within the topic context instance, ¶94 discloses generating an intermediate SQL statement based on both the utterance and the identified schema/topic context with a SoftMax layer for finding similarity. ¶123 then goes on to explain the intermediate SQL actions which correspond to the claimed executable actions. The candidate actions are the available intermediate SQL higher level actions supported by the system; Tomkins ¶39 discloses threshold level of similarity) . Rahmfeld and Tomkins are combinable because they are from the same field of endeavor, specifically natural language processing, e.g., both disclose methods for processing natural language and generating model output for question answering. Rahmfeld teaches determining the relevant topic and step for a natural language utterance by generating utterance schema pairs for candidate topics and producing probabilities for each topic step and selecting the most likely answer, however it does not explicitly state comparing to a threshold, but it does compare similarity to known topics. Tomkins does teach selecting a concept based on an embedding vector distance satisfying a distance threshold. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to disclose Tomkin’s threshold based similarity determination into Rahmfeld’s probability based topic selection process. The suggestion/motivation for doing so is “some embodiments may improve the speed and accuracy of responses to queries” as disclosed in ¶41. Regarding Claim 3: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1 further comprising: responsive to a user logging into an application associated with the digital assistant, creating the current session context instance for the session (Tomkins: ¶51 and ¶96 disclose a login session between a client device and server and that the session and account parameters such as login identifier, username, session identifier, account identifier and domain indicators are available during the session) , wherein the current session context instance comprises prior natural language utterances from the user during the session between the user and the digital assistant (Rahmfeld: ¶18 a data conversation may contain a series of one or more data conversation steps that operate within a conversational context, a data conversation step may contain one or more utterances, new utterances etc. ¶50 discloses a data conversation 110 may include one or more conversation steps 111 where each step 111 may include an utterance a structured query language query, a result set from query execution and associated step metadata) , and wherein each of the prior natural language utterances (a) is resolved and associated with the topic context instance or other topic context instance associated with the current session context instance (Rahmfeld: ¶35 discloses the schema identifying a topic or steps of data conversation that may serve as a target for an utterance and the schema may use a model to make a prediction from the received utterance and the metadata, ¶52 discloses the schema layer may determine a topic and the result sets from previous steps, Altogether this teaches the topic or previous step result set that should server as the target for the utterance. Once the utterance is translated and executed it becomes a conversation step including the utterance, SQL, result set and metadata therefore the utterances are resolved) or (b) is unresolved and associated with a tentative topic context instance (Rahmfeld¶19 discloses feedback may highlight errors allowing a user to understand if an utterance was interpreted as intended and ¶48 discloses retraining based on this indication) . Rahmfeld and Tomkins are combinable because they are from the same field of endeavor, specifically natural language processing, e.g., both disclose methods for processing natural language and generating model output for question answering. Rahmfeld teaches determining the relevant topic and step for a natural language utterance in a session with a user, however it does not specifically discuss the user logging into an application associated with the digital assistant. Tomkins does teach this user login configuration explicitly. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to disclose Tomkin’s threshold based similarity determination into Rahmfeld’s probability based topic selection process. The suggestion/motivation for doing so is to “increase the speed of document retrieval and increase the relevance of retrieved information by providing knowledge within the scope of a user's indicated expertise” by providing user specific information as suggested in ¶40. Regarding Claim 4: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1, wherein: the digital assistant is configured to handle a plurality of actions associated with a plurality of topics including the one or more topics (Rahmfeld: ¶21 discloses the topic view may list one or more topics, where each topic comprises on or more columns and their respective tables, respective join conditions and other metadata, ¶123 discloses intermediate SQL may encode one or more actions, where an action may not directly correspond to a standard SQL function or keyword and ¶125 discloses table 1 includes example keywords, actions example utterances and example intermediate SQL encoding generated by the translation layer 136. Altogether teaching a system that handles multiple topics because its topic view includes a plurality of topics and a plurality of actions because the intermediates SQL may encode multiple actions) ; the topic context instance is specific to the one or more topics and is associated one or more actions of the plurality of actions (Rahmfeld: ¶82 discloses the columns configured for the topic and corresponding meta-data which may constitute the effective schema, ¶94 discloses translating an utterance in the schema context to an intermediate SQL statement the translation layer may use an encoder decoder like framework, ¶124 discloses the intermediate SQL keywords may include ‘combine’ ‘exclude’ which could translate to executable SQL containing joins of various types of result sets) ; and determining whether the natural language utterance satisfies the threshold level of similarity with the one or more topics is a function of similarity between the natural language utterance and the associated one or more actions (Rahmfeld: ¶84, ¶92 and ¶125 discloses generating schema pairs, probabilities for topics and steps and comparison to example actions, keywords and utterances; Tomkins: ¶39 discloses a distance threshold. In total Rahmfeld teaches selecting the topic most likely to answer the utterance based on candidate topic options and teaches generation actions from the utterance in that selected context, Tomkins supplies the ability to demand it meet a certain threshold) . Rahmfeld and Tomkins are combinable because they are from the same field of endeavor, specifically natural language processing, e.g., both disclose methods for processing natural language and generating model output for question answering. Rahmfeld teaches determining the relevant topic and step for a natural language utterance in a session with a user, however it does not specifically discuss the user logging into an application associated with the digital assistant. Tomkins does teach this user login configuration explicitly. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to disclose Tomkin’s threshold based similarity determination into Rahmfeld’s probability based topic selection process. The suggestion/motivation for doing so is to “increase the speed of document retrieval and increase the relevance of retrieved information by providing knowledge within the scope of a user's indicated expertise” by providing user specific information as suggested in ¶40. Regarding Claim 5: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1, further comprising: receiving, at the digital assistant, a subsequent natural language utterance from the user during the session between the user and the digital assistant (Rahmfeld: ¶51 discloses receiving additional utterances during the same data conversation because a data conversation includes multiple conversation steps and later utterances may be input through the same user interface) ; obtaining a tentative topic context instance for the subsequent natural language utterance (Rahmfeld: ¶38 teaches handling utterances whose interpretation may be uncertain by providing feedback and allowing the user to validate whether the interpretation is correct, this reasonably supports a tentative context for an utterance that is not yet confidently resolved) , wherein the obtaining comprises: executing, based on the subsequent natural language utterance, a search on the current session context instance, the data store, or both (Rahmfeld: ¶52 searches/uses both current conversation context and data store topic information by checking previous steps and determining the relevant topic or prior result set) , based on the search, determining the subsequent natural language utterance does not satisfy the threshold level of similarity with one or more topics represented in the current session context instance, the data store, or both (Rahmfeld: ¶82 discloses confidence based interpretation. Tomkins and ¶148 discloses threshold concepts. If the utterance does not meet the threshold, the system would not finally resolve it to a topic) , responsive to determining the subsequent natural language utterance does not satisfy the threshold level of similarity with the one or more topics, creating the tentative topic context instance associated with the current session context instance (Rahmfeld: ¶19 discloses highlighting ambiguities and/or errors in utterances, and or facilitate the user in providing another utterance; Tomkins: ¶39 and ¶148 supplies the ability to do this in the context of thresholds) , and associating the subsequent natural language utterance with the tentative topic context instance (Rahmfeld: ¶51 discloses associating the utterance with a present conversation step in the conversation history, for an unresolved utterance, that present step corresponds to the tentative context) . Rahmfeld and Tomkins are combinable because they are from the same field of endeavor, specifically natural language processing, e.g., both disclose methods for processing natural language and generating model output for question answering. Rahmfeld teaches determining the relevant topic and step for a natural language utterance in a session with a user, however it does not specifically discuss the user logging into an application associated with the digital assistant. Tomkins does teach this user login configuration explicitly. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to disclose Tomkin’s threshold based similarity determination into Rahmfeld’s probability based topic selection process. The suggestion/motivation for doing so is to “increase the speed of document retrieval and increase the relevance of retrieved information by providing knowledge within the scope of a user's indicated expertise” by providing user specific information as suggested in ¶40. Regarding Claim 6: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 5, further comprising: receiving, at the digital assistant, another subsequent natural language utterance from the user during the session between the user and the digital assistant (Rahmfeld: ¶18 discloses receiving multiple utterances as multiple steps in a data conversation) ; obtaining the topic context instance for another subsequent natural language utterance, (Rahmfeld: ¶52 teaches finding the topic from prior and data store information) wherein the obtaining comprises: executing, based on the another subsequent natural language utterance, a search on the current session context instance, the data store, or both (Rahmfeld: ¶52 teaches searching prior conversation steps and topic data store information for later/new utterances) , based on the search, determining the another subsequent natural language utterance satisfies the threshold level of similarity with the same or different one or more topics represented in the current session context instance, the data store, or both (Rahmfeld ¶92 generates probabilities for each topic/step to hold the answer, Tomkins¶39 discloses distance satisfying a distance threshold) , responsive to determining the another subsequent natural language utterance satisfies the threshold level of similarity with the same or different one or more topics, identifying the same or different topic context instance associated with the same or different one or more topics (Rahmfeld: ¶52 teaches identifying the topic of previous step/result set that serves as the target context for the utterance) , and associating the another subsequent natural language utterance with the same or different topic context instance (Rahmfeld: ¶52 teaches associating the utterance with a conversation step that includes the selected topic/result set context, SQL, result and meta data) . Rahmfeld and Tomkins are combinable because they are from the same field of endeavor, specifically natural language processing, e.g., both disclose methods for processing natural language and generating model output for question answering. Rahmfeld teaches determining the relevant topic and step for a natural language utterance in a session with a user, however it does not specifically discuss the user logging into an application associated with the digital assistant. Tomkins does teach this user login configuration explicitly. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to disclose Tomkin’s threshold based similarity determination into Rahmfeld’s probability based topic selection process. The suggestion/motivation for doing so is to “increase the speed of document retrieval and increase the relevance of retrieved information by providing knowledge within the scope of a user's indicated expertise” by providing user specific information as suggested in ¶40. Regarding Claim 7: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 6, further comprising: responsive to receiving the another subsequent natural language utterance from the user, reevaluating the subsequent natural language utterance associated with the tentative topic context instance (Rahmfeld: ¶18-19 discloses using a later utterance to continue or clarify the conversation after ambiguity/error feedback in order to reevaluate, specifically stating “new utterances may leverage previous conversation steps…” and “highlight ambiguities and/or errors… and or facilitate the user in providing another utterance” meaning the system may reevaluate the utterance by recognizing errors, using the context from the previous erred attempt and receive new input from the user) , wherein the reevaluating comprises: executing, based on the subsequent natural language utterance, a search on the current session context instance, the data store, or both (Rahmfeld: ¶52 teaches searching prior conversation steps and topic data store information for later/new utterances) , based on the search, determining the subsequent natural language utterance satisfies the threshold level of similarity with the same or different one or more topics represented in the current session context instance, the data store, or both (Rahmfeld ¶92 generates probabilities for each topic/step to hold the answer, Tomkins¶39 discloses distance satisfying a distance threshold) , responsive to determining the subsequent natural language utterance satisfies the threshold level of similarity with the same or different one or more topics, identifying the same or different topic context instance associated with the same or different one or more topics (Rahmfeld: ¶52 teaches identifying the topic of previous step/result set that serves as the target context for the utterance) , and associating the subsequent natural language utterance with the same or different topic context instance (Rahmfeld: ¶52 teaches associating the utterance with a conversation step that includes the selected topic/result set context, SQL, result and meta data) . Rahmfeld and Tomkins are combinable because they are from the same field of endeavor, specifically natural language processing, e.g., both disclose methods for processing natural language and generating model output for question answering. Rahmfeld teaches determining the relevant topic and step for a natural language utterance in a session with a user, however it does not specifically discuss the user logging into an application associated with the digital assistant. Tomkins does teach this user login configuration explicitly. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to disclose Tomkin’s threshold based similarity determination into Rahmfeld’s probability based topic selection process. The suggestion/motivation for doing so is to “increase the speed of document retrieval and increase the relevance of retrieved information by providing knowledge within the scope of a user's indicated expertise” by providing user specific information as suggested in ¶40. Examiner Interpretation: Rahmfeld teaches the flow of claims 5-7 because Rahmfeld’s data conversion is bult around conversation steps and each step can include the utterance, generated SQL, result set and metadata. New utterances can leverage previous conversation steps, including previous utterances, SQL queries and result set data. Rahmfeld discloses when the current utterance is ambiguous or erroneous, it does not treat the conversation as over, the virtual assistance can highlight ambiguities and/or errors, help the user understand whether the utterance was interpreted correctly and facilitate the user providing another utterance as in claim 5. For claim 6 Rahmfeld also teaches the next iteration directly. A later utterance is processed using the conversation context. The system determines whether the utterance references previous steps, identifies the topic or result set from the previous steps on which to operate and then translates the utterance. For claim 7, Rahmfeld’s use of historical input is key. Because new utterances may leverage previous conversation steps and because Rahmfeld specifically handles ambiguity/error feedback by letting the user provide another utterance, the later utterance because additional context for resolving what was previously ambiguous. So, the earlier unresolved present step can be reevaluated using the now expanded conversation history. Regarding Claim 8: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1, wherein: the natural language utterance references a prior conversation between the user and the digital assistant (Rahmfeld: ¶20 discloses later use of saved/replayed prior data conversations) ; obtaining the topic context instance for the natural language utterance (Rahmfeld: ¶20 and 35 discloses obtaining a topic context instance by identifying a topic or referenced conversation step that serves as the target context for the utterance) further comprises: based on the reference to the prior conversation and the search, identifying a past topic context instance associated with the same or different one or more topics (Rahmfeld: ¶49 identifies a past topic by teaching accessing prior saved conversations and using them as a basis for further analysis, which corresponds to identifying a past topic/conversation context) , and linking, using a virtual pointer, the topic context instance with the past topic context instance (Rahmfeld: ¶47 discloses the conversation window may include a hyperlink associated with each conversation step for fast navigation to any step of a data conversation, this functions as a virtual pointer linking the current context to the past context) ; and the one or more executable actions are selected from the one or more candidate actions based on each of the one or more executable actions satisfying the threshold level of similarity with the natural language utterance, the context within the topic context instance, and additional context within the past topic context instance (Rahmfeld: ¶36 teaches selecting/generating actions using the current utterance plus previous conversation queries/result sets. That maps to using additional past topic context when selecting executable actions. Tomkins: ¶39 and ¶148 discloses a distance threshold) . Rahmfeld and Tomkins are combinable because they are from the same field of endeavor, specifically natural language processing, e.g., both disclose methods for processing natural language and generating model output for question answering. Rahmfeld teaches determining the relevant topic and step for a natural language utterance in a session with a user, however it does not specifically discuss the user logging into an application associated with the digital assistant. Tomkins does teach this user login configuration explicitly. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to disclose Tomkin’s threshold based similarity determination into Rahmfeld’s probability based topic selection process. The suggestion/motivation for doing so is to “increase the speed of document retrieval and increase the relevance of retrieved information by providing knowledge within the scope of a user's indicated expertise” by providing user specific information as suggested in ¶40. Regarding Claim 9: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1, wherein: responsive to determining the natural language utterance satisfies the threshold level of similarity with the one or more topics, identifying multiple topic context instances associated with the current session context instance and associated with the one or more topics (Rahmfeld: ¶32 teaches that multiple topics and result sets can be identified as relevant to an utterance) ; and obtaining the topic context instance for the natural language utterance further comprises merging the multiple topic context instances to create the topic context instance as a composite of the multiple topic context instance (Rahmfeld: ¶33 discloses combining multiple prior result sets and topics for the current utterance. This corresponds to merging multiple topic contexts into a composite context for analysis) . Regarding Claim 10: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 9, wherein: the context within the topic context instance includes a conversation history between the user and the digital assistant (Rahmfeld: ¶47 discloses a conversation history as part of the data conversation context) ; context within each of the multiple topic context instances includes additional conversation history between the user and the digital assistant (Rahmfeld ¶50 discloses each prior step and result set context includes its own utterance, query, result and associated step metadata) ; and merging the multiple topic context instances includes concatenating the conversation history with each of the additional conversation histories (Rahmfeld: ¶58 discloses using multiple prior steps together in a later query. Therefore, whether system builds later queries on multiple previous steps, the conversation history for the current step is combined with the additional histories of the referenced prior steps, thereby merging/concatenation the relevant conversation histories into the composite context) . Regarding Claim 11: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1, wherein: the one or more candidate actions are identified as being associated with the topic context instance by executing, using the natural language utterance, a semantic search of potential actions represented in the data store that are associated with the digital assistant (Rahmfeld: ¶82, ¶96 and ¶125 teaches identifying actions associated with the selected topic context because the model input includes the utterance, selected schema context and SQL keywords. The table 1 higher level actions are the potential actions associated with the system and the translation layer selects/generates the appropriate action encoding based on the utterance and context) ; the potential actions have a zero confidence level for satisfying the threshold level of similarity with the natural language utterance and the context within the topic context instance (Rahmfeld: ¶53 discloses the translation layer may determine an estimated confidence level of the translation, which can be based on the aggregated and/or individual token probabilities. If potential actions that do not correspond to the utterance or context arise for lack of similarity, they would have no selected probability or effectively ‘zero’ confidence for the generated action sequence as recited in the claim. Tomkins: ¶39 and ¶148 further supports treating nonmatching items as failing a threshold similarity test, i.e., having zero confidence for satisfying the threshold) ; the one or more candidate actions have a positive confidence level for satisfying the threshold level of similarity with the natural language utterance and the context within the topic context instance (Rahmfeld: ¶122 the decoder chooses among possible intermediate SQL and keywords. Candidate actions correspond to action keywords having positive model probability/confidence for the utterance and selected context) , and the one or more executable actions have a positive confidence level and do satisfy the threshold level of similarity with the natural language utterance and the context within the topic context instance (Rahmfeld: ¶55 discloses the positive confidence for generated intermediate SQL action tokens and an aggregated confidence measure for the translation. Tomkins ¶39, ¶77, ¶86, ¶148 disclose threshold requirements. Together the generated executable action is one having positive confidence and satisfying a threshold similarity standard) , based on which the first generative artificial intelligence model predicts that the one or more executable actions are relevant for responding to the natural language utterance with a high confidence level (Rahmfeld: ¶55 teaches that the generated translation action has an associated confidence measure. A high confidence generated intermediate SQL action is predicted as relevant for responding to the utterance) . Rahmfeld and Tomkins are combinable because they are from the same field of endeavor, specifically natural language processing, e.g., both disclose methods for processing natural language and generating model output for question answering. Rahmfeld teaches determining the relevant topic and step for a natural language utterance in a session with a user, however it does not specifically discuss the user logging into an application associated with the digital assistant. Tomkins does teach this user login configuration explicitly. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to disclose Tomkin’s threshold based similarity determination into Rahmfeld’s probability based topic selection process. The suggestion/motivation for doing so is to “increase the speed of document retrieval and increase the relevance of retrieved information by providing knowledge within the scope of a user's indicated expertise” by providing user specific information as suggested in ¶40. Regarding Claim 12: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 11, further comprising: constructing, based on the topic context instance, an input prompt comprising the one or more candidate actions (Rahmfeld: ¶96 the model input includes the utterance, topic/effective schema context and SQL actions/keywords, these SQL and action keywords correspond to candidate actions) , at least a portion of the context associated with the topic context instance, and the natural language utterance (Rahmfeld: ¶94 provides the model input to the encoder decoder translation model) ; and providing the input prompt to the first generative artificial intelligence model, wherein the first generative artificial intelligence model generates the list comprising the executable action based on the input prompt (Rahmfeld: ¶117 and ¶123 disclose the encoder/decoder generates intermediate SQL tokens encoding the executable actions from the utterance, schema context and action keywords) . Regarding Claim 13: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1, further comprising: executing, based on the one or more candidate actions, a search on user-preferences in the data store (Rahmfeld: ¶22 disclose query conversion may determine which schema should be included and identify whether and how elements from previous data conversation steps may be incorporated and translate the utterance, ¶50 discloses each step 111 may include an utterance… a result set from query execution and associated step metadata (e.g., a step name, a step number, virtual assistant feedback, user annotations, etc.). In total this teaches using prior conversation steps, result sets and step metadata when generating the executable query action, user preferences may be stored in metadata in the data conversation session. Rahmfeld’s use of step metadata and context supports searching relevant contextual metadata for generating the executable action related to the user) to identify one or more user-preferences that are relevant to the one or more candidate actions (Rahmfeld: ¶49 and ¶52 identifies prior conversation steps, result sets and associated relevant to the present utterance action. Because the metadata can include user annotations, the system identifies user provided preferences or instructions relevant to the candidate action query) , wherein the creating the execution plan comprises embedding the one or more user-preferences into the execution plan (Rahmfeld: ¶49 and 54 teaches creating the executable SQL/action plan using prior steps and their associated context. Because those prior steps can include user annotations, edits and intended result information, the generated executable SQL embeds the relevant user preference metadata into the execution plan) . Regarding Claim 14: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1, wherein: the context within the topic context instance includes a conversation history between the user and the digital assistant (Rahmfeld: ¶47, ¶50 teaches that the topic conversation context includes a conversation is made up of steps and each step includes the user utterance, generated query, result set and metadata. The conversation history window stores and displays that history) ; and the current session context instance is associated with the topic context instance and one or more other topic context instances, each of the one or more other topic context instances includes additional conversation history between the user and the digital assistant (Rahmfeld: ¶33 and ¶50 teach that different topic/result sets and metadata. Those prior steps are additional conversation history associated with the other topic contexts) . Regarding Claim 15: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 14, further comprising: generating a summary of the conversation history and the additional conversation history between the user and the digital assistant (¶47 discloses a conversation history window for one or more data conversation, ¶50 discloses a virtual assistant receives an explanation of the translated executable SQL statement in understandable terms, altogether teaching storing conversation history and additional conversation history through conversation steps that include utterances, SQL queries, result sets and metadata. Rahmfeld also teaches that the virtual assistant generates understandable explanations of the translated executable query. That explanation is a summary of what occurred in the conversation step) ; revising the current session context instance to include the summary of the conversation history (Rahmfeld: ¶47 discloses a conversation history window for one or more data conversation, ¶50 discloses a virtual assistant receives an explanation of the translated executable SQL statement in understandable terms, altogether teaching storing conversation history and additional conversation history through conversation steps that include utterances, SQL queries, result sets and metadata. Rahmfeld also teaches that the virtual assistant generates understandable explanations of the translated executable query. That explanation is a summary of what occurred in the conversation step execution) ; and computing performance metrics for the digital assistant based on the revised current session context instance (Rahmfeld: ¶48 the retraining module may enable a translation model retraining workflow based on a user indicating that a specific utterance was not translated by the data conversation system as intended by the user ¶53 the translation layer may determine an estimated confidence level of translation and ¶55 the post processing layer may create an aggregated confidence measure for the translation. Altogether this teaches computing performance metrics because the system determines estimated and aggregated confidence levels for the translation and supports retraining based on mistranslated utterances. These confidence and mistranslation retraining signals are computed from the revised conversation context and reflect the digital assistant’s performance in interpreting and responding to user utterances) . Regarding Claim 16: The proposed combination of Rahmfeld in view of Tomkins further discloses the computer-implemented method of claim 1, further comprising: constructing, based on the output and the topic context instance, an input prompt comprising the output, at least a portion of the context associated with the topic context instance, and the natural language utterance (Rahmfeld: ¶38 which discloses the virtual assistant receiving or communicating explanation of the translated executable structured query language statement, the result sets produced by the execution of the structured query language statement and the data filters used in the execution of the query, ¶57 the data conversion system 101 may display the virtual assistant feedback generated by the post processing layer 138. This teaches using the executed query output, including result sets, filters and post-processing information, together with the utterance step context to generate virtual assistant feedback. This corresponds to constructing an input for generating a user facing response based on the output topic/effective schema context and natural language utterance) ; providing the input prompt to a second generative artificial intelligence model (Rahmfeld: ¶38, 55 discloses a second response generation stage after the executable SQL action is generated and executed, the virtual assistant receives the result information and produces a human understandable explanation. To the extent the claim requires a separate ‘second generative AI model,” Rahmfeld’s virtual assistant explanation/summarization stage corresponds to the second model or response generation component that generates the user facing communication after execution) ; and generating, by the second generative artificial intelligence model, a response to the natural language utterance based on the input prompt (Rahmfeld: ¶55-57 discloses generating a user facing response because the virtual assistant provides understandable feedback based on the utterance, executable query, result set, filters and post processing information) , wherein the response is the communication derived from the output (Rahmfeld: ¶7 and ¶38 teaches that the response/virtual assistant feedback is derived from the execution output because the system communicates result sets, metadata, filters, confidence, explanations and post processing information to the user after executing the SQL action) . Regarding Claim 17: Claim 17 has been analyzed with regard to claims 1 (see rejection above) and is rejected for the same reasons of obviousness as used above. It is noted Rahmfeld discloses a system comprising: one or more processing systems; and one or more computer-readable media storing instructions which, when executed by the one or more processing systems, cause the system to perform the operations of claim 17 and further dependent claims at least at ¶158. Regarding Claim 18: Claim 18 has been analyzed with regard to claims 2 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 19: Claim 19 has been analyzed with regard to claims 3 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 20: Claim 20 has been analyzed with regard to claims 4 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 21: Claim 21 has been analyzed with regard to claims 5 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 22: Claim 22 has been analyzed with regard to claims 6 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 23: Claim 23 has been analyzed with regard to claims 7 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 24: Claim 24 has been analyzed with regard to claims 8 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 25: Claim 25 has been analyzed with regard to claims 9 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 26: Claim 26 has been analyzed with regard to claims 10 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 27: Claim 27 has been analyzed with regard to claims 1 (see rejection above) and is rejected for the same reasons of obviousness as used above. It is noted Rahmfeld discloses one or more non-transitory computer-readable media storing instructions which, when executed by one or more processors, cause a system to perform operations of claim 27 and further dependent claims at least at ¶156. Regarding Claim 28: Claim 28 has been analyzed with regard to claims 2 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 29: Claim 29 has been analyzed with regard to claims 3 (see rejection above) and is rejected for the same reasons of obviousness as used above. Regarding Claim 30: Claim 30 has been analyzed with regard to claims 4 (see rejection above) and is rejected for the same reasons of obviousness as used above. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to IAN SCOTT MCLEAN whose telephone number is (703)756-4599. The examiner can normally be reached "Monday - Friday 8:00-5:00 EST, off Every 2nd Friday". 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, Hai Phan can be reached at (571) 272-6338. 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. /IAN SCOTT MCLEAN/Examiner, Art Unit 2654 /HAI PHAN/Supervisory Patent Examiner, Art Unit 2654 Application/Control Number: 18/830,344 Page 2 Art Unit: 2654 Application/Control Number: 18/830,344 Page 3 Art Unit: 2654 Application/Control Number: 18/830,344 Page 4 Art Unit: 2654 Application/Control Number: 18/830,344 Page 5 Art Unit: 2654 Application/Control Number: 18/830,344 Page 6 Art Unit: 2654 Application/Control Number: 18/830,344 Page 7 Art Unit: 2654 Application/Control Number: 18/830,344 Page 8 Art Unit: 2654 Application/Control Number: 18/830,344 Page 9 Art Unit: 2654 Application/Control Number: 18/830,344 Page 10 Art Unit: 2654 Application/Control Number: 18/830,344 Page 11 Art Unit: 2654 Application/Control Number: 18/830,344 Page 12 Art Unit: 2654 Application/Control Number: 18/830,344 Page 13 Art Unit: 2654 Application/Control Number: 18/830,344 Page 14 Art Unit: 2654 Application/Control Number: 18/830,344 Page 15 Art Unit: 2654 Application/Control Number: 18/830,344 Page 16 Art Unit: 2654 Application/Control Number: 18/830,344 Page 17 Art Unit: 2654 Application/Control Number: 18/830,344 Page 18 Art Unit: 2654 Application/Control Number: 18/830,344 Page 19 Art Unit: 2654 Application/Control Number: 18/830,344 Page 20 Art Unit: 2654 Application/Control Number: 18/830,344 Page 21 Art Unit: 2654 Application/Control Number: 18/830,344 Page 22 Art Unit: 2654 Application/Control Number: 18/830,344 Page 23 Art Unit: 2654 Application/Control Number: 18/830,344 Page 24 Art Unit: 2654 Application/Control Number: 18/830,344 Page 25 Art Unit: 2654 Application/Control Number: 18/830,344 Page 26 Art Unit: 2654 Application/Control Number: 18/830,344 Page 27 Art Unit: 2654 Application/Control Number: 18/830,344 Page 28 Art Unit: 2654 Application/Control Number: 18/830,344 Page 29 Art Unit: 2654
Prosecution Insights