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 Office Action is in response to RCE filed on June 15, 2026.
Claim 20 is canceled.
Claim 21 has been added.
The objections and rejections from the prior correspondence that are not restated herein are withdrawn.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on June 15, 2026 has been entered.
Response to Arguments
Applicant's arguments filed on June 15, 2026 with regards to the 101 rejection have been fully considered and are persuasive. Therefore, the 101 rejections have been withdrawn.
Applicant’s arguments with regards to the 103 rejections have been fully considered but are not persuasive. Applicant argues that the combination of cited prior art does not teach generating machine learning encodings of the client features from the value metric of the digital account and the previous client device interactions because Peng's citation are to encoding a representation of a user that "describes or indicates the user's behavior over a period of time," which does not appear to take into account a value metric.
In response to Applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). Specifically, Sivasankar was relied upon to teach generating […] the client features from the value metric of the digital account and the previous client device interactions ([28] teaches determining a responsive action corresponding to the identified intent, where the responsive action may be identified by applying a set of rules to the identified intent and to certain query metadata items (such as the user identifier), where the identified intent may be “account balance inquiry,” and the parameters of the query may further identify one or more accounts to which the intended action should be applied; [29] teaches performing the identified responsive action, which may involve performing one or more operations on one or more accounts associated with the user (e.g., executing one or more database queries to compute the account balance for one or more accounts associated with the user); see [14] as taught above for importing an automated assistant transcript, which may include a set of records reflecting user interactions with an automated assistant, where each record may include the user's query, its classification by the automated assistant (e.g., by assigning a topic and a subtopic to each query), the user's intent inferred by the automated assistant from the query based on the assigned topic and subtopic, and the responsive action inferred by the automated assistant from the query based on the identified user's intent), and Peng was relied upon to teach machine learning encodings of the client features ([41] teaches the first representation of the user is encoded into a first feature representation that is representative of the user's behavior (see Fig. 2 206), and [54-55] teach a predicted user behavior model is generated by applying the deep recurrent neural network to input data, which includes the feature representations of a user generated or encoded by the unsupervised deep neural networks (see Fig. 2 210)), where the combination of Sivasankar and Peng teaches the entire limitation.
Applicant also argues that Sivasankar [0029] describing “executing one or more database queries to compute the account balance” fails to teach accessing a value metric because Sivasankar describes computing the account balance as a "responsive action" that is performed after the model has already identified the user's intent.
In response to Applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., the responsive action is performed before the model has identified the user's intent) are not recited in the rejected claim(s). 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).
Applicant also argues that the combination of cited prior art does not teach generating an automated interaction response utilizing the predicted client disposition classification, the disposition classification probability, and a disposition classification threshold because this limitation plainly recites an inference-time operation where a probability of a predicted classification is evaluated against a threshold to determine how the system will respond to a client.
In response to Applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., an inference-time operation where a probability of a predicted classification is evaluated against a threshold to determine how the system will respond to a client) are not recited in the rejected claim(s). 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).
Furthermore, Sivasankar teaches the limitation generating an automated interaction response utilizing the predicted client disposition classification ([13] teaches an automated assistant (e.g., a chat bot or an interactive voice response system) may analyze a received query in order to associate it with a topic and a subtopic and identify the user's intent by machine learning-based models, where the automated assistant may then identify (e.g., by applying a set of rules to the identified intent and to certain query metadata items) and perform a responsive action corresponding to the identified intent (i.e. generating an automated interaction response), and upon performing the responsive action, the automated assistant may produce a response and return it to the requesting client; [14] teaches initiating a model training session by importing an automated assistant transcript, which may include a set of records reflecting user interactions with an automated assistant. Each record may include the user's query, its classification (i.e. utilizing the predicted client disposition classification) by the automated assistant (e.g., by assigning a topic and a subtopic to each query), the user's intent inferred by the automated assistant from the query based on the assigned topic and subtopic, and the responsive action inferred by the automated assistant from the query based on the identified user's intent), the disposition classification probability, and a disposition classification threshold ([16] teaches supervised model training may involve running the model on a data sample from a training data set, comparing the actual model output with the desired model output (e.g., the subtopic and the corresponding confidence score, the topic and the corresponding confidence score, or the intent and the corresponding confidence score), and adjusting values of one or more model parameters responsive to determining that the difference of the actual and desired model output exceeds a specified threshold classification error; [18] teaches responsive to detecting significant model imbalances, the model training platform may retrain one or more models using additional training data sets, where the model retraining operations may be iteratively repeated until the detected model imbalance would satisfy a specified model imbalance threshold; [19] teaches responsive to successfully evaluating the model, the model training platform may publish the model to a model deployment environment, which may involve storing the model in a designated depository and notifying the deployment workflows associated with the respective deployment environments of the model identifiers in the repository).
Applicant also argues that Sivasankar's threshold is used exclusively during supervised model training to compute a classification error and adjust model weights, where Applicant does not understand the relevance to any citation to training a model when a clause deals with inference, and thus, Sivasankar does not teach or suggest using a threshold to evaluate a prediction probability at runtime in order to generate an automated interaction response.
Similar to the responses above, in response to Applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., a threshold to evaluate a prediction probability is used at runtime in order to generate an automated interaction response) are not recited in the rejected claim(s). 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). Examiner suggests incorporating such limitations into the claims (and provide support in Applicant’s specification) in order to distinguish the claimed invention from Examiner’s broadest reasonable interpretation of the claims. Furthermore, Sivasankar teaches the limitation generating an automated interaction response utilizing the predicted client disposition classification, the disposition classification probability, and a disposition classification threshold (see Sivasankar [13-14], [16], and [18-19] above, where a machine-learning-based models are trained by using imported an automated assistant transcript and adjusting model parameters to achieve the desired model output, where the automated assistant may produce a response and return it to the requesting client).
Applicant further argues that the combination of cited prior art fails to teach extracting client features corresponding to the client of the automated client interaction system by accessing, via a digital database, a value metric of a digital account and previous client device interactions and then generating, utilizing a machine learning model to analyze the machine learning encodings of the extracted client features, a predicted client disposition classification, where Sivasankar explicitly describes computing the account balance as a "responsive action" that is performed after the model has already identified the user's intent, i.e. as Sivasankar explains at paragraphs [0028] and [0029], the system first identifies the intent (e.g., "account balance inquiry"), and then performs the responsive action (e.g., executing database queries to compute the account balance), and thus, Sivasankar does not extract an account balance to use as an input feature for its machine learning model; rather, Sivasankar's model analyzes the query text to predict the intent, and then subsequently retrieves the account balance to fulfill the request.
The Examiner agrees that Sivasankar [28] teaches determining a responsive action corresponding to the identified intent. It is unclear to the Examiner whether Applicant is stating that the claimed invention determines a responsive action prior to identifying the user’s intent, which is not recited in the claims. As stated above, Sivasankar teaches machine learning-based models employed by the automated assistant is trained using the automated assistant transcript ([14]) and adjust model parameters to achieve the desired model output ([16], [18]), where the model is published/deployed after successful evaluation ([19]). The automated assistant, which employs the machine learning-based models, may then identify and perform a responsive action corresponding to the identified intent and return the response to the requesting client ([13]). Therefore, the Examiner has not “fundamentally misconstrue[d] the temporal and functional operation of Sivasankar by treating the post-prediction output and fulfillment action as a preprediction input feature”, but rather Sivasankar teaches the limitations of claim 1 as outlined in the rejection below.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries 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.
Claims 1-2, 4-12, and 14-18 are rejected under 35 U.S.C. 103 as being unpatentable over Sivasankar (Pub. No.: US 20210365834 A1 Sivasankar), hereafter Sivasankar, in view of Peng (Pub. No.: US 20190147356 A1), hereafter Peng.
Regarding claim 1, Sivasankar teaches:
in response to a client contacting an automated client interaction system, utilizing client credentials to extract client features corresponding to the client of the automated client interaction system by accessing, via a digital database, a value metric of a digital account and previous client device interactions; ([14] teaches a model training platform may initiate a model training session by importing an automated assistant transcript, which may include a set of records reflecting user interactions with an automated assistant, where each record may include the user's query, its classification by the automated assistant (e.g., by assigning a topic and a subtopic to each query), the user's intent inferred by the automated assistant from the query based on the assigned topic and subtopic, and the responsive action inferred by the automated assistant from the query based on the identified user's intent; [29] teaches performing one or more operations on one or more accounts associated with the user (e.g., executing one or more database queries to compute the account balance for one or more accounts associated with the user); see Fig. 5 #520A-N Database server.)
generating […] the client features from the value metric of the digital account and the previous client device interactions ([28] teaches determining a responsive action corresponding to the identified intent, where the responsive action may be identified by applying a set of rules to the identified intent and to certain query metadata items (such as the user identifier), where the identified intent may be “account balance inquiry,” and the parameters of the query may further identify one or more accounts to which the intended action should be applied; [29] teaches performing the identified responsive action, which may involve performing one or more operations on one or more accounts associated with the user (e.g., executing one or more database queries to compute the account balance for one or more accounts associated with the user); see [14] as taught above for importing an automated assistant transcript, which may include a set of records reflecting user interactions with an automated assistant, where each record may include the user's query, its classification by the automated assistant (e.g., by assigning a topic and a subtopic to each query), the user's intent inferred by the automated assistant from the query based on the assigned topic and subtopic, and the responsive action inferred by the automated assistant from the query based on the identified user's intent);
generating, utilizing a machine learning model to analyze the […] extracted client features ([14] teaches a model training platform may initiate a model training session by importing an automated assistant transcript), a predicted client disposition classification ([14] teaches a set of records reflecting user interactions with an automated assistant. Each record may include the user's query, its classification by the automated assistant (e.g., by assigning a topic and a subtopic to each query), the user's intent inferred by the automated assistant from the query based on the assigned topic and subtopic, and the responsive action inferred by the automated assistant from the query based on the identified user's intent.) and a disposition classification probability from the […] client features. ([16] teaches comparing the actual model output with the desired model output (e.g., the subtopic and the corresponding confidence score, the topic and the corresponding confidence score, or the intent and the corresponding confidence score))
generating an automated interaction response utilizing the predicted client disposition classification ([13] teaches an automated assistant (e.g., a chat bot or an interactive voice response system) may analyze a received query in order to associate it with a topic and a subtopic and identify the user's intent by machine learning-based models, where the automated assistant may then identify (e.g., by applying a set of rules to the identified intent and to certain query metadata items) and perform a responsive action corresponding to the identified intent (i.e. generating an automated interaction response), and upon performing the responsive action, the automated assistant may produce a response and return it to the requesting client; [14] teaches initiating a model training session by importing an automated assistant transcript, which may include a set of records reflecting user interactions with an automated assistant. Each record may include the user's query, its classification (i.e. utilizing the predicted client disposition classification) by the automated assistant (e.g., by assigning a topic and a subtopic to each query), the user's intent inferred by the automated assistant from the query based on the assigned topic and subtopic, and the responsive action inferred by the automated assistant from the query based on the identified user's intent), the disposition classification probability, and a disposition classification threshold ([16] teaches supervised model training may involve running the model on a data sample from a training data set, comparing the actual model output with the desired model output (e.g., the subtopic and the corresponding confidence score, the topic and the corresponding confidence score, or the intent and the corresponding confidence score), and adjusting values of one or more model parameters responsive to determining that the difference of the actual and desired model output exceeds a specified threshold classification error; [18] teaches responsive to detecting significant model imbalances, the model training platform may retrain one or more models using additional training data sets, where the model retraining operations may be iteratively repeated until the detected model imbalance would satisfy a specified model imbalance threshold; [19] teaches responsive to successfully evaluating the model, the model training platform may publish the model to a model deployment environment, which may involve storing the model in a designated depository and notifying the deployment workflows associated with the respective deployment environments of the model identifiers in the repository); and
providing the automated interaction response to the client via the automated client interaction system ([13] teaches the automated assistant may then identify (e.g., by applying a set of rules to the identified intent and to certain query metadata items) and perform a responsive action corresponding to the identified intent. Upon performing the responsive action, the automated assistant may produce a response and return it to the requesting client).
Sivasankar does not appear to explicitly teach machine learning encodings of the client features.
However, Peng teaches the limitation ([41] teaches the first representation of the user is encoded into a first feature representation that is representative of the user's behavior (see Fig. 2 206), and [54-55] teach a predicted user behavior model is generated by applying the deep recurrent neural network to input data, which includes the feature representations of a user generated or encoded by the unsupervised deep neural networks (see Fig. 2 210)).
Accordingly, it would have been obvious to a person having ordinary skill in the art at the time of the effective filing of the invention, having the teachings of Sivasankar and Peng before them, to include Peng’s feature learning in Sivasankar’s system that performs supervised learning of an automated assistant. One would have been motivated to make such a combination in order to avoid inaccurately predicting user behavior and also account for temporal user behavior or time dependent user behavior as taught by Peng [3].
Regarding claim 2, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above. Sivasankar also teaches:
identifying a client query via the automated client interaction system; ([0013] teaches an automated assistant (e.g., a chat bot or an interactive voice response system) may analyze a received query in order to associate it with a topic and a subtopic and identify the user's intent.)
in response to identifying the client query, providing the automated interaction response, ([Fig 1] teaches receiving and processing query, identifying intent, and performing responsive action.)
PNG
media_image1.png
819
424
media_image1.png
Greyscale
wherein the automated interaction response comprises an indicator of the predicted client disposition classification. ([40] teaches an identifier of the intent that the automated assistant has associated with the query.)
Regarding claim 4, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above. Sivasankar also teaches:
monitoring client interaction with the automated client interaction system to determine a ground truth client disposition; ([40] teaches adjusting values of one or more model parameters responsive to determining that the difference of the actual and desired model output exceeds a specified threshold classification error)
training the machine learning model by comparing the predicted client disposition classification and the ground truth client disposition ([0016] teaches a Supervised model training may involve running the model on a data sample from a training data set, comparing the actual model output with the desired model output (e.g., the subtopic and the corresponding confidence score, the topic and the corresponding confidence score, or the intent and the corresponding confidence score))
Regarding claim 5, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above. Sivasankar also teaches:
utilizing the machine learning model comprises generating the predicted client disposition classification ([14] teaches a set of records reflecting user interactions with an automated assistant. Each record may include the user's query, its classification by the automated assistant (e.g., by assigning a topic and a subtopic to each query), the user's intent inferred by the automated assistant from the query based on the assigned topic and subtopic, and the responsive action inferred by the automated assistant from the query based on the identified user's intent.)
and the disposition classification probability ([16] teaches comparing the actual model output with the desired model output (e.g., the subtopic and the corresponding confidence score, the topic and the corresponding confidence score, or the intent and the corresponding confidence score))
utilizing one or more of a random forest model or gradient boosted decision tree model ([0032] teaches the above-referenced models employed to identify the topic, subtopic, intent, and responsive action may utilize a variety of automatic classification methodologies, such as Bayesian classifiers, support vector machines (SVMs), random forest classifiers, gradient boosting classifiers, neural networks, etc. Supervised training of a model may involve adjusting, based on example input-output pairs, one or more parameters of a model that maps an input (e.g., a vector of feature values characterizing an object) to an output (e.g., a category of a predetermined set of categories).)
Regarding claim 6, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above. Sivasankar also teaches:
determining a previous disposition from a previous interaction by the client with the automated client interaction system; generating the predicted client disposition classification and the disposition classification probability from the previous disposition utilizing the machine learning model. ([Fig. 3A] teaches storing previous user queries, probabilities, and responses in a table to refer to for future queries)
PNG
media_image2.png
596
419
media_image2.png
Greyscale
Regarding claim 7, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above. Sivasankar also teaches:
Extracting client features comprises extracting a digital account duration, a direct deposit status of a digital account, or application device activity on the digital account. ([0028] teaches At block 150, the computer system may determine a responsive action corresponding to the identified intent. In certain implementations, the responsive action may be identified by applying a set of rules to the identified intent and to certain query metadata items (such as the user identifier). In an illustrative example, the identified intent may be “account balance inquiry,” and the parameters of the query may further identify one or more accounts to which the intended action should be applied.)
Regarding claim 8, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above. Sivasankar also teaches:
determining that the disposition classification probability satisfies the disposition classification threshold and generating the automated interaction response comprising an indicator of the predicted client disposition classification. ([40] teaches comparing the actual model output with the desired model output (e.g., the subtopic and the corresponding confidence score, the topic and the corresponding confidence score, or the intent and the corresponding confidence score), and adjusting values of one or more model parameters responsive to determining that the difference of the actual and desired model output exceeds a specified threshold classification error)
Regarding claim 9, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above. Sivasankar also teaches:
generating, utilizing the machine learning model, an additional predicted client disposition classification and an additional disposition classification probability from additional client features; withholding an additional automated interactive response corresponding to the additional predicted client disposition classification based on comparing the additional disposition classification probability and a disposition classification threshold. ([0033] teaches supervised model training may utilize one or more training data sets. Each training data set includes a plurality of data items, such that each data item specifies a set of classification feature values for an object (e.g., represented by a vector, each element of which represents the number of occurrences in the query of the word identified by the index of the element) and a corresponding classification of the object (e.g., represented by the confidence score associated with a topic of the query). Supervised model training may involve running the model on the data items from the training data set, comparing the actual model output with the desired model output (i.e., the category and the corresponding confidence score associated with the data item by the training data set), and adjusting values of one or more model parameters responsive to determining that the value of a predetermined quality metric exceeds a specified threshold value.) ([13] teaches the automated assistant may then identify (e.g., by applying a set of rules to the identified intent and to certain query metadata items) and perform a responsive action corresponding to the identified intent. Upon performing the responsive action, the automated assistant may produce a response and return it to the requesting client)
Regarding claim 10, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above. Sivasankar also teaches:
providing an interactive voice response indicating the predicted client disposition classification or providing an automated text response indicating the predicted client disposition classification in a digital message thread. ([0013] teaches an automated assistant (e.g., a chat bot or an interactive voice response system) may analyze a received query in order to associate it with a topic and a subtopic and identify the user's intent. (Where a chatbot produces an automated text response))
Regarding claim 11, the claim recites similar limitation as corresponding to claim 1 and is rejected for similar reasons as claim 1 using similar teachings and rationale. Sivasankar also teaches:
A non-transitory computer-readable medium storing instructions that, when executed by at least one processor, cause a computer system. ([0069] The data storage device 618 may include a computer-readable storage medium 628 on which may be stored one or more sets of instructions (e.g., instructions of the methods 100, 200, and/or 400 of supervised machine learning for automated assistants, in accordance with one or more aspects of the present disclosure) implementing any one or more of the methods or functions described herein. The may also reside, completely or at least partially, within main memory 604 and/or within processing device 602 during execution thereof by computer system 600, main memory 604 and processing device 602 also constituting computer-readable media. The instructions may further be transmitted or received over a network 620 via network interface device 608.)
Regarding claim 12, the claim recites similar limitation as corresponding claim 2 and is rejected for similar reasons as claim 2 using similar teachings and rationale.
Regarding claim 14, the claim recites similar limitation as corresponding claim 4 and is rejected for similar reasons as claim 2 using similar teachings and rationale.
Regarding claim 15, the claim recites similar limitation as corresponding claim 5 and is rejected for similar reasons as claim 2 using similar teachings and rationale.
Regarding claim 16, the claim recites similar limitation as corresponding claim 6 and is rejected for similar reasons as claim 6 using similar teachings and rationale.
Regarding claim 17, the claim recites similar limitation as corresponding to claim 1 and is rejected for similar reasons as claim 1 using similar teachings and rationale. Sivasankar also teaches:
A system comprising: at least one processor; and at least one non-transitory computer-readable storage medium storing instructions that, when executed by the at least one processor, cause the system ([0068] The computer system 600 may further include a network interface device 608, which may communicate with a network 620. The computer system 600 also may include a video display unit 610 (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT)), an alphanumeric input device 612 (e.g., a keyboard), a cursor control device 614 (e.g., a mouse) and/or an acoustic signal generation device 616 (e.g., a speaker). In one embodiment, video display unit 610, alphanumeric input device 612, and cursor control device 614 may be combined into a single component or device (e.g., an LCD touch screen).)
Regarding claim 18, the claim recites similar limitation as corresponding claim 2 and is rejected for similar reasons as claim 2 using similar teachings and rationale.
Claims 3, 13, 19 are rejected under 35 U.S.C. 103 as being unpatentable over Sivasankar in view of Peng as applied to claims 1, 11, and 17 above, and further in view of Nair (Pub. No.: US 2021/ 0105246 A1), hereafter Nair.
Regarding claim 3, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above.
Sivasankar in view of Peng does not appear to explicitly teach in response to a user interaction with the automated interaction response, initiating a client-agent response session between the client and an agent device; providing the predicted client disposition classification for display via the agent device.
Nair teaches:
in response to a user interaction with the automated interaction response, initiating a client-agent response session between the client and an agent device; providing the predicted client disposition classification for display via the agent device. ( [0016] teaches conventionally, messaging centers may be accessed on a user device via an interactive interface using at least a messaging application available on the interface. FIG. 1 presents an exemplary messaging center application interface and solution 100. In particular, FIG. 1 illustrates a user device 102 with messaging center interface 104. The user device 102 maybe a tablet, iPad, cell phone or the like. For exemplary purposes, user device 104 can be a smart phone or laptop. The user device 102 may be equipped with various applications for performing various tasks. For example, the user device 102 may be used for web browsing, video streaming, bill payments, and online purchases. Additionally, the user device 10 be equipped with applications that enable the user to make purchases and transfers using a payment provider application and/or a digital wallet, and/or access application with the payment provider, merchant, messaging center, etc. Further, the user device 102 may be capable of making phone calls and communicating with one or more other communications devices using a cellular network, Wi-Fi, Bluetooth, BLE, NFC, WLAN, etc. For example, in the communication the user may communicate via the user device 102 with a service agent, bot, or other at an application dashboard of a messaging center 104. In the exemplary message center dashboard 104 of FIG. 1, for example, a user can review account activity associated with a payment processing service. The account activity can include information about a user's account balance, invoicing, and other recent activity 106. At this messaging center dashboard 104, the user may also message, chat or otherwise communicate with customer service agent regarding their account. As illustrated FIG. 1, the user may also be flagged on pending notifications108 regarding previous communications with the customer service center representative. Note that the term customer service representative is being broadly used to represent a bot, agent, or other entity which may be used in communicating with a user/customer. In some instances, the term customer service agent may be interchanged with the term customer service user or simply agent.)
Accordingly, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention, having the teachings of Sivasankar, Peng, and Nair before them, to include Nair’s feature of initiating session with an agent using an automated assistant in Sivasankar and Peng’s system that performs supervised learning of an automated assistant. One would have been motivated to make such a combination in order to improve efficiency by initiating a live chat if the model cannot understand a query as taught by Nair [0015])
Regarding claim 13, the claim recites similar limitation as corresponding claim 3 and is rejected for similar reasons as claim 3 using similar teachings and rationale.
Regarding claim 19, the claim recites similar limitation as corresponding claim 3 and is rejected for similar reasons as claim 3 using similar teachings and rationale.
Claim 21 is rejected under 35 U.S.C. 103 as being unpatentable over Sivasankar in view of Peng as applied to claim 1 above, and further in view of Lara Maldonado (Pub. No.: US 20230019856 A1), hereafter Maldonado.
Regarding claim 21, Sivasankar in view of Peng teaches the elements of claim 1 as outlined above.
Sivasankar in view of Peng does not appear to explicitly teach wherein generating, utilizing a machine learning model to analyze the machine learning encodings of the extracted client features, a predicted client disposition classification and a disposition classification probability from the machine encodings of the client features comprises utilizing a heuristic model in tandem with the machine learning model.
However, Sivasankar in view of Peng and Maldonado teaches the limitation (Maldonado [0065] teaches allocation platform 508 may be implemented as another conjoined machine learning model and/or a heuristic model, where Sivasankar [13-14], [16], and [18-19] teach the claimed “predicted client disposition classification and a disposition classification probability” (see claim 1 above)).
Accordingly, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention, having the teachings of Sivasankar, Peng, and Maldonado before them, to include Maldonado’s conjoining machine learning model and heuristic model in Sivasankar and Peng’s system that performs supervised learning of an automated assistant. One would have been motivated to make such a combination in order to implement machine learning outputs using objective training data and prioritize for outputs that a human cannot compute as taught by Maldonado [0067]).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ANDREW JUNG whose telephone number is (571)270-3779. The examiner can normally be reached Monday through Friday from 9am to 5pm.
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, DAVID WILEY can be reached on 571-272-4150. 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.
/ANDREW J JUNG/Supervisory Patent Examiner, Art Unit 2175
Prosecution Insights