Detailed Action
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
Applicant’s arguments with respect to claim(s) 1-20 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims [1, 2, 5, 6, 7, 8, 9, 10, 11, 14, 15, 16, 17, 18, 19, 20] are rejected under 35 U.S.C. 103 as
being unpatentable over Renard (US Patent No US20190215249) in view of Hongbin (CN 114861680)
in view of Jones (US. Patent No. US 20220028378 A1) and in further view of Wang (US 12597420
B2).
Regarding claim 1, Renard teaches a method comprising:
generating through execution of a first algorithm in
[0046 "In some embodiments, the AI engine 322 uses automatic speech recognition and natural language processing to determine a customer's intent (i.e. question) and uses an algorithm derived using machine learning to determine an appropriate response"]; [0053, 0054 “In one embodiment, the conversation ranking engine 324 ranks a session based on one or more of the quality of the AI engine’s 322 answers and the sentiment of the session. In one embodiment, the “quality” of the conversation may be determined by the conversation ranking engine 324 using one or more of the following, which may be referred to herein as “qualitative criteria”: The confidence of the AI engine 322 in the user’s (e.g. customer’s) intent. For example, when the AI engine 322 is not confident in the accuracy of what it believes the customer is asking, the conversation ranking engine 324 may affect the ranking in favor of intervention by a user (e.g. a customer service agent). This may be based on an average over the course of the session (e.g. a whole-session average confidence) or a portion of the session (e.g. there has been a series of low- confidence intents, which satisfies a threshold number of low-confidence intents, and argues in favor of human agent intervention, so the conversation ranking engine 324 adjusts the ranking accordingly). Based on different types of algorithm, as mentioned above to compute the confidence, the system 100 may provide the confidence as a percentage or other numerical value. Based on the final confidence score, it may decide to activate the intent expected (e.g. when the confidence is over 80%) or to create a dialog between the user and the machine to ask more precisions (i.e. follow-up questions to better discern the expected intent). When the confidence satisfies a first threshold (e.g. when between 60 and 79.9% accuracy for confidence), the system 100 may automatically activate the session handling by a human (e.g. an agent), or, when another threshold is satisfied (e.g. when under 60% of accuracy for the confidence), the system 100 enters the session in a waiting list to be taken over by a human (e.g. an agent). It should be noted that the thresholds provided are merely examples and others may be used without departing from the disclosure herein. In one embodiment, the thresholds may be parameters in the system 100 that may be defined”].
Generating, through execution of the second algorithm of the
[0060 " In one embodiment, the “sentiment” of the conversation may be evaluated by the conversation ranking engine 324 using machine-based sentiment analysis of each interaction. In one embodiment, sentiment analysis refers to the use of natural language processing, text analysis, computational linguistics and biometrics (e.g. voice) to systematically identify, extract and study affective states and subjective information, e.g., to determine the attitude of a speaker, writer or other subject. Depending on the embodiment, the attitude may be one or more of a judgment or evaluation (as in appraisal theory), affective state (i.e., the emotional state of the author or speaker), or the intended emotional communication (i.e., the emotional effect intended by the author or interlocutor). Depending on the embodiment, the conversation ranking engine 324 uses one or more of the following, which may be referred to as “sentiment criteria,” to evaluate the “sentiment” of the session and rank the session"]; [0074 "In some embodiments, one or more of the foregoing sentiment criteria are used to generate a sentiment metric, which is used by the conversation ranking engine 324 alone, or in combination with a sentiment metric (depending on the embodiment), to rank the conversation"]. This implies generating an emotion parameter based on sentiment analysis.
Generating, through execution of the third algorithm of the
[0055” The confidence of the AI engine’s 322 answer(s). For example, if the AI engine 322 is not confident in the accuracy of an answer, it may affect the ranking in favor of intervention by a user. Based on different types of algorithm, the system 100 may provide the confidence as a percentage or other numerical value. In some embodiments, the confidence in the AI engine’s answers is based on an average over the course of the session. Alternatively, the confidence in the AI engine’s answers may be based on a portion of the session (e.g. there has been a series of low-confidence answers, which satisfies a threshold number of low- confidence answers, and argues in favor of human agent intervention, so the conversation ranking engine 324 adjusts the ranking accordingly”].
determining a service mode for replying to the query content based on the intent parameter,
[0053 “In one embodiment, the conversation ranking engine 324 ranks a session based on one or more of the quality of the AI engine’s 322 answers and the sentiment of the session. In one embodiment, the “quality” of the conversation may be determined by the conversation ranking engine 324 using one or more of the following, which may be referred to herein as “qualitative criteria”: The confidence of the AI engine 322 in the user’s (e.g. customer’s) intent. For example, when the AI engine 322 is not confident in the accuracy of what it believes the customer is asking, the conversation ranking engine 324 may affect the ranking in favor of intervention by a user (e.g. a customer service agent). This may be based on an average over the course of the session (e.g. a whole-session average confidence) or a portion of the session (e.g. there has been a series of low- confidence intents, which satisfies a threshold number of low-confidence intents, and argues in favor of human agent intervention, so the conversation ranking engine 324 adjusts the ranking accordingly). Based on different types of algorithm, as mentioned above to compute the confidence, the system 100 may provide the confidence as a percentage or other numerical value. Based on the final confidence score, it may decide to activate the intent expected (e.g. when the confidence is over 80%) or to create a dialog between the user and the machine to ask more precisions (i.e. follow-up questions to better discern the expected intent). When the confidence satisfies a first threshold (e.g. when between 60 and 79.9% accuracy for confidence), the system 100 may automatically activate the session handling by a human (e.g. an agent), or, when another threshold is satisfied (e.g. when under 60% of accuracy for the confidence), the system 100 enters the session in a waiting list to be taken over by a human (e.g. an agent). It should be noted that the thresholds provided are merely examples and others may be used without departing from the disclosure herein. In one embodiment, the thresholds may be parameters in the system 100 that may be defined”];
[0059 “In some embodiments, one or more of the foregoing qualitative criteria are used to generate a quality metric, which is used by the conversation ranking engine 324 alone, or in combination with a sentiment metric (depending on the embodiment), to rank the conversation”];
[0074 -0075 “In some embodiments, one or more of the foregoing sentiment criteria are used to generate a sentiment metric, which is used by the conversation ranking engine 324 alone, or in combination with a sentiment metric (depending on the embodiment), to rank the conversation. The conversation ranking engine 324 uses one or more of the qualitative criteria and the sentiment criteria (e.g. via sentiment analysis) to determine a rank for a session (e.g. based on the determined metrics). For example, the conversation ranking engine 324 uses the qualitative criteria and the sentiment criteria to determine a value used to determine a session’s rank. The value used to determine the session’s rank may vary depending on the embodiment. For example, the value may be a weighted average calculated using the qualitative criteria and sentiment criteria. In some embodiments, the weight each criterion is assigned may be dynamic (e.g. may be set by a team of users, may vary over time and be assigned using machine learning, etc.”]
But Renard does not teach
parallel processing component of a processor-based machine learning system;
the parallel processing component comprising the first algorithm executing in parallel with at least second and third algorithms and having an input configured to receive the query content, the processor-based machine learning system further comprising a serial processing component having an input coupled to an output of the parallel processing component and including a predefined strategy model and an adaptive strategy model, the predefined strategy model having an input configured to receive the query content and an output coupled to an input of the adaptive strategy model;
wherein responsive to the determined service mode being a particular service mode, a reply to the query content is generated by serial processing of the query content through the predefined strategy model and the adaptive strategy model of the serial processing component of the processor-based machine learning system
However, Hongbin teaches
parallel processing component of a processor-based machine learning system;
[n0054 “Furthermore, after obtaining the dialogue text input by the user in the target service, the dialogue text can also be fed into the Natural Language Understanding (NLU) part. Natural Language Understanding includes four sub-parts: intent recognition, sentiment classification, authenticity detection, and entity recognition. That is, the dialogue text is input into the intent recognition, sentiment classification, authenticity detection, and entity recognition sub-parts in parallel, and the dialogue text outputs the dialogue intent, sentiment polarity, authenticity category, and entity feature information of the dialogue text, respectively. Through an integrated implementation, the dialogue processing efficiency in the dialogue interaction process is improved” where parallel processing is taking place to determine intent and emotion];
the parallel processing component comprising the first algorithm executing in parallel with at least second and third algorithms and having an input configured to receive the query content, the processor-based machine learning system further comprising a serial processing component having an input coupled to an output of the parallel processing component and
[n0054 “Furthermore, after obtaining the dialogue text input by the user in the target service, the dialogue text can also be fed into the Natural Language Understanding (NLU) part. Natural Language Understanding includes four sub-parts: intent recognition, sentiment classification, authenticity detection, and entity recognition. That is, the dialogue text is input into the intent recognition, sentiment classification, authenticity detection, and entity recognition sub-parts in parallel, and the dialogue text outputs the dialogue intent, sentiment polarity, authenticity category, and entity feature information of the dialogue text, respectively. Through an integrated implementation, the dialogue processing efficiency in the dialogue interaction process is improved. In addition, the above four sub-parts can all be in the form of models, namely, intent recognition model, sentiment classification model, authenticity detection model, and entity recognition model” where intent recognition algorithm is running in parallel with sentiment classification algorithm, etc.];
[n0111 “As shown in Figure 2, the dialogue processing in the loan service scenario is divided into three stages: Natural Language Understanding (NLU), Dialogue Management (DM), and Natural Language Generation (NLG). The dialogue text input by user u in the loan service goes through lie detection, entity recognition, intent recognition, and sentiment recognition in parallel, sequentially outputting four pieces of information: lie category, entity feature information, dialogue intent, and sentiment polarity. These four pieces of information are then fed into a neural network model to determine the dialogue state. A decision model then ranks candidate default influencing factors and calculates the user's confidence level in the loan service, thereby generating a response action for the dialogue text. The types of response actions include questioning, confirmation, clarification, information gathering, initiation, and termination. In addition, response actions can also be other types of actions. Based on the response actions and historical dialogue information, and using a text generation model or text generation template, the response text is generated to facilitate user understanding and improve the user experience” where parallel processing includes intention detection and emotional classification and the output of those are feed downstream into serial processing components including neural network/decision model processing”]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Renaud into the teachings of Hongbin because running intention recognition, emotional and confidence processing in parallel would enhance in determining diverse user data simultaneously and would reduce analysis time and increase the accuracy of dialog understanding. Additionally, the input of the results being feed into the serial processing would then allow the process to sequentially and accurately determine dialogue states and generate proper responses. This would thereby reduce processing latency and avoid automated handling of queries and would provide improved responses.
However, Renard in view of Hongbin do not teach
generating a confidence parameter by analyzing a similarity between the query content and training data for training an adaptive strategy.
including a predefined strategy model and an adaptive strategy model, the predefined strategy model having an input configured to receive the query content and an output coupled to an input of the adaptive strategy model;
wherein responsive to the determined service mode being a particular service mode, a reply to the query content is generated by serial processing of the query content through the predefined strategy model and the adaptive strategy model of the serial processing component of the processor-based machine learning system
However, Jones teaches
generating a confidence parameter by analyzing a similarity between the query content and training data for training an adaptive strategy.
generating a confidence parameter based on percentage match and then the behavior changes (adaptive) depending on the computed value - [ 0046 " In operation 430, each intent classifier generates a confidence score for the utterance. The confidence score may be based on comparing the query entities with the one or more entities associated with the intent classifier performing the comparison. In some embodiments, the confidence score is generated based on a number of entities, for the intent classifier generating the score, that match the query entities. In some instances, the confidence score is generated based on a percentage match between query entities and specified entities of the intent classifier"]; [0041" In operation 314, the decision component 150 determines whether a resource or bot exists for the on-topic entity and intent. The resource for the on-topic entity and intent may be a resource or chat bot trained with relevant knowledge on the topic and intent of the query. Where the decision component 150 matches the topic and intent with a resource or bot, the decision component 150 proceeds to operation 316. In operation 316, the decision component 150 connects to the resource or bot and transfers the query or portions thereof to the resource or bot. In some embodiments, in operation 304, where an on-topic intent is identified by the decision component 150 may enact an intent override. In such embodiments, the decision component 150 overrides an utterance topic with a page-based topic, based on the intent. The decision component 150 then proceeds to operation 314"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Renaud in view of Hongbin into the teachings of Jones because executing this independent confidence analysis alongside Renaud in view of Hongbin so that intent, emotion and confidence parameters would be available concurrently for processing and for determining the appropriate responses or service mode accurately.
But Renaud in view of Hongbin and in view of Jones do not teach
including a predefined strategy model and an adaptive strategy model, the predefined strategy model having an input configured to receive the query content and an output coupled to an input of the adaptive strategy model;
wherein responsive to the determined service mode being a particular service mode, a reply to the query content is generated by serial processing of the query content through the predefined strategy model and the adaptive strategy model of the serial processing component of the processor-based machine learning system
However, Wang teaches
including a predefined strategy model and an adaptive strategy model, the predefined strategy model having an input configured to receive the query content and an output coupled to an input of the adaptive strategy model;
wherein responsive to the determined service mode being a particular service mode, a reply to the query content is generated by serial processing of the query content through the predefined strategy model and the adaptive strategy model of the serial processing component of the processor-based machine learning system
[Column 65, 66 lines 52-67, 1-3 “The process begins when the user provides feedback to the AI agent, such as correcting a response or indicating dissatisfaction with a recommendation 2301Next, the AI system searches for responses in the 0KB 2303. The AI system determines whether a predefined response is available in the 0KB 2304. If there are no predefined responses in the 0KB, the AI agent interacts with the user and generates understanding and responses using NLU and NLG 2305. Using advanced reasoning algorithms, the AI agent can determine the root cause of the feedback and adjust its responses accordingly 2306. This process involves analyzing the user's input and context to determine the most likely cause of the feedback and selecting an appropriate response based on this analysis. If a predefined response is in the 0KB, the AI agent interacts with the user using the response identified by the AI system 2307. The AI system analyzes the interaction data and determines whether any adjustments are necessary 2308.”];
[Column 66, lines 13-17 “The flow chart emphasizes the iterative nature of this process, with the AI agent continually learning and adapting based on user feedback to improve its overall performance and provide more contextually relevant and personalized responses.” where serial processing is taking place where predefined responses are searched and if one is not available, then invokes adaptive and advanced reasoning.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Renaud in view of Hongbin in view of Jones into the teachings of Wang because using predefined and adaptive strategies into the systems of Renaud in view of Hongbin and in view of Jones would help generate refined responses to the queries. Using predefined answers first ensures accuracy for common questions and reduces system work and speed and then using adaptive responses if predefined responses are not available ensures the system learns continuously and provides more accurate answers.
Regarding claim 2, Renard in view of Jones in view of Hongbin do not teach in response to receiving the query content input by the user, generating a reply content based on the query content by the predefined strategy model; based on the reply content, determining whether the predefined strategy model has provided a complete reply; and in response to the predefined strategy model having not provided a complete reply, generating a reply content based on the query content by the adaptive strategy model.
However, Wang teaches based
in response to receiving the query content input by the user, generating a reply content based on the query content by the predefined strategy model; based on the reply content, determining whether the predefined strategy model has provided a complete reply; and in response to the predefined strategy model having not provided a complete reply, generating a reply content based on the query content by the adaptive strategy model.
[Column 65, 66 lines 52-67, 1-3 “The process begins when the user provides feedback to the AI agent, such as correcting a response or indicating dissatisfaction with a recommendation 2301. Next, the AI system searches for responses in the 0KB 2303. The AI system determines whether a predefined response is available in the 0KB 2304. If there are no predefined responses in the 0KB, the AI agent interacts with the user and generates understanding and responses using NLU and NLG 2305. Using advanced reasoning algorithms, the AI agent can determine the root cause of the feedback and adjust its responses accordingly 2306. This process involves analyzing the user's input and context to determine the most likely cause of the feedback and selecting an appropriate response based on this analysis. If a predefined response is in the 0KB, the AI agent interacts with the user using the response identified by the AI system 2307. The AI system analyzes the interaction data and determines whether any adjustments are necessary 2308.”];
[Column 66, lines 13-17 “The flow chart emphasizes the iterative nature of this process, with the AI agent continually learning and adapting based on user feedback to improve its overall performance and provide more contextually relevant and personalized responses.” where serial processing is taking place where predefined responses are searched and if one is not unavailable, then invokes adaptive and advanced reasoning.]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Renaud in view of Hongbin in view of Jones into the teachings of Wang because using predefined and adaptive strategies into the systems of Renaud in view of Hongbin and in view of Jones would help generate refined responses to the queries. Such a combination would improve system flexibility and response strength by allowing efficient use of predefined responses when suitable while enabling adaptive generation when the predefined responses are insufficient. This would improve conversational system robustness by enabling adaptive generative strategy when predefined processing fails, thereby enhancing response completeness, system reliability and user experience.
Regarding claim 5, Renard teaches the method according to claim 1, wherein determining a service mode for replying to the query content comprises: determining a decision parameter based on the intent parameter, the emotion parameter, and the confidence parameter;
Renard teaches making a decision based on intent, emotion and confidence parameter –
[0053 “In one embodiment, the conversation ranking engine 324 ranks a session based on one or more of the quality of the AI engine's 322 answers and the sentiment of the session. In one embodiment, the “quality” of the conversation may be determined by the conversation ranking engine 324 using one or more of the following, which may be referred to herein as “qualitative criteria”]; [0059 “In some embodiments, one or more of the foregoing qualitative criteria are used to generate a quality metric, which is used by the conversation ranking engine 324 alone, or in combination with a sentiment metric (depending on the embodiment), to rank the conversation”]; [0074 -0075 “In some embodiments, one or more of the foregoing sentiment criteria are used to generate a sentiment metric, which is used by the conversation ranking engine 324 alone, or in combination with a sentiment metric (depending on the embodiment), to rank the conversation. The conversation ranking engine 324 uses one or more of the qualitative criteria and the sentiment criteria (e.g. via sentiment analysis) to determine a rank for a session (e.g. based on the determined metrics). For example, the conversation ranking engine 324 uses the qualitative criteria and the sentiment criteria to determine a value used to determine a session’s rank. The value used to determine the session’s rank may vary depending on the embodiment. For example, the value may be a weighted average calculated using the qualitative criteria and sentiment criteria. In some embodiments, the weight each criterion is assigned may be dynamic (e.g. may be set by a team of users, may vary over time and be assigned using machine learning, etc.”]
However, Renard in view of Hongbin do not teach determining whether the service mode is a model service mode using the adaptive strategy model or a direct service mode based on the decision parameter and a preset value.
However, Jones teaches determining whether the behavior changes (adaptive) or a direct mode depending on the computed value
[0041" In operation 314, the decision component 150 determines whether a resource or bot exists for the on-topic entity and intent. The resource for the on-topic entity and intent may be a resource or chat bot trained with relevant knowledge on the topic and intent of the query. Where the decision component 150 matches the topic and intent with a resource or bot, the decision component 150 proceeds to operation 316. In operation 316, the decision component 150 connects to the resource or bot and transfers the query or portions thereof to the resource or bot. In some embodiments, in operation 304, where an on-topic intent is identified by the decision component 150 may enact an intent override. In such embodiments, the decision component 150 overrides an utterance topic with a page-based topic, based on the intent. The decision component 150 then proceeds to operation 314"]; [0042 “In embodiments where the decision component 150 does not identify a resource or bot matching the topic and intent, in operation 314, the decision component 150 proceeds to operation 318. In operation 318, the decision component 150 determines whether a suitable or relevant human agent exists to respond to the query based on the intent and topic. The decision component 150 may determine the human agent exists by comparing the entity name and value or intent and topic with information, such as a profile, for the human agent. When the decision component 150 identifies a relevant human agent, the decision component 150 passes the query to the human agent at operation 320. Where the decision component 150 determines no relevant human agent exists, the decision component 150 proceeds to operation 322. In operation 322, the decision component 150 determines whether a topic or intent resource exists. The topic or intent resource may be a network resource, such as a database, a webpage, or other suitable data repository accessible to the decision component 150. The decision component 150 may identify a relevant topic or intent resource by comparing name and value pairs or components of the query with information within or metadata for the resource. Where the decision component 150 identifies a topic or intent resource, the decision component 150 provides the resource as a response to the query in operation 324. Where the decision component 150 identifies no topic or intent resource relevant to the query, the decision component cooperates with one or more other components of the query routing system 102 to generate and present one or more clarification questions to a user within a user interface at operation 310”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Renaud in view of Hongbin into the teachings of Jones because these references address improving automated interactions systems by evaluating user content, emotions and system confidence to determine appropriate handling of user requests and combining known decision parameters to improve routing accuracy whether to route to direct service mode or use an adaptive strategy represents a predictable optimization.
Regarding claim 6 Renard in view of Hongbin in view Jones do teach
the method according to claim 5, wherein determining the decision parameter comprises: determining evaluation factors corresponding to the intent parameter, the emotion parameter, and the confidence parameter, the evaluation factors being one of a level or a weight; and determining the decision parameter based on the evaluation factors corresponding to the intent parameter, the emotion parameter, and the confidence parameter
Renard discloses
[0075 “The conversation ranking engine 324 uses one or more of the qualitative criteria and the sentiment criteria (e.g. via sentiment analysis) to determine a rank for a session (e.g. based on the determined metrics). For example, the conversation ranking engine 324 uses the qualitative criteria and the sentiment criteria to determine a value used to determine a session's rank. The value used to determine the session's rank may vary depending on the embodiment. For example, the value may be a weighted average calculated using the qualitative criteria and sentiment criteria. In some embodiments, the weight each criterion is assigned may be dynamic (e.g. may be set by a team of users, may vary over time and be assigned using machine learning, etc.”]; [0087 “In one embodiment, the rank uses a weighted average. For example, in one embodiment, an accuracy for each parameter, e.g., the sentiment analysis, distribution of words per sentence and/or conversations, number of utterances in the conversation regarding the whole conversation of the same agent, etc. Depending on the embodiment, the computing of accuracy can be through an artificial neural network or simple linear algebra calculation, e.g., vectors distances. In one embodiment, one or more of the parameter is then normalized as a percentage. In one embodiment, a parameter receives a weight to calculate the average accuracy of the conversation. In some embodiments, the weight can be defined as a parameter in the system 100 or another type of artificial neural network is trained to define, through regression, what the best weight distribution for parameters is based on a small number of weights defined by a team in charge to train the system 100”]
Regarding claim 7, Renard teaches the method according to claim 5, wherein determining the
decision parameter comprises: determining a service mode corresponding to the intent parameter, the emotion parameter, and the confidence parameter. Renaud does teach handling based on multiple user state parameters, including sentiment metric representing an emotional state and confidence values associated with the systems determination of user’s intent. It also teaches generating a numeric metric derived from such parameters and arranging or prioritizing sessions based on their relative ranking or proximity to decision criteria
[0053 “In one embodiment, the conversation ranking engine 324 ranks a session based on one or more of the quality of the AI engine's 322 answers and the sentiment of the session. In one embodiment, the “quality” of the conversation may be determined by the conversation ranking engine 324 using one or more of the following, which may be referred to herein as “qualitative criteria”]; [ 0059 “In some embodiments, one or more of the foregoing qualitative criteria are used to generate a quality metric, which is used by the conversation ranking engine 324 alone, or in combination with a sentiment metric (depending on the embodiment), to rank the conversation”].
determining proximity between the service mode and a preset mode; and determining the decision parameter based on the proximity
[0091 “FIGS. 8a-o are example user interfaces presented to a human agent according to one embodiment of the system described above in reference to FIGS. 1-3. In FIG. 8A, the “All” tab 802, which indicates that 140 sessions are in progress is selected. In one embodiment, the grid 804 includes a visual indicator for each of those sessions. A visual indicator for a session may be arranged within the grid relative to other visual indicators associated with other sessions based on one or more criteria. For example, the indicators in the grid may be arranged based on ranking (e.g. so that the conversations that are determined to be going poorly are located in proximity, such as near the top of the UI). In another example, the indicators in the grid may be arranged based on age (e.g. so that the sessions are ordered oldest to newest in the bar 810). In another example, the indicators in the grid may be arranged based on whether an agent has intervened or is intervening (e.g. so that indicators associated with such sessions are located within proximity to one another within bar 810). In another example, the indicators in the grid may be arranged based on channel (e.g. so that sessions associated with SMS are visually grouped, sessions associated with phone calls are visually grouped, and sessions associated with e-mail are visually grouped”].
Regarding claim 8, Renaud in view of Hongbin do not teach
the method according to claim 1, comprising determining whether the query content comprises a preset content; and determining that the service mode is a direct service mode in response to the query content comprising the preset content.
But Jones teaches comparing content to preset content and then determining the service mode is a direct service
[0040 “In some embodiments, if the decision component 150 determines the entity is on-topic, in operation 304, the decision component 150 proceeds to operation 312. In operation 312, the decision component 150 determines if the entity matches a topic and intent of a current URL of the browser. The decision component 150 may compare the one or more of the name and value pair to metadata for or keywords associated with the current URL. Where the decision component 150 determines the topic and intent match between the entity and the current URL, the decision component 150 may proceed to operation 314. In some embodiments, where the decision component 150 determines no topic and intent match occurred between the entity and the current URL, in operation 316, the decision component 150 may redirect the browser to a subsequent URL. The subsequent URL may be a URL which matches one or more of the topic and intent of the query. Once the decision component 150 redirects the browser to the subsequent URL, the decision component 150 may proceed to operation 314”]; [0041 “In operation 314, the decision component 150 determines whether a resource or bot exists for the on-topic entity and intent. The resource for the on-topic entity and intent may be a resource or chat bot trained with relevant knowledge on the topic and intent of the query. Where the decision component 150 matches the topic and intent with a resource or bot, the decision component 150 proceeds to operation 316. In operation 316, the decision component 150 connects to the resource or bot and transfers the query or portions thereof to the resource or bot. In some embodiments, in operation 304, where an on-topic intent is identified by the decision component 150 may enact an intent override. In such embodiments, the decision component 150 overrides an utterance topic with a page-based topic, based on the intent. The decision component 150 then proceeds to operation 314”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Renard in view of Hongbin to incorporate the routing framework of Jones in order to apply Renaud’s in view of Hongbin computed user state parameters within a known bot-versus-human architecture. Doing so would have predictably enabled automated escalation to either a bot or human, thereby improving response reliability and overall user experience.
Regarding claim 9, Renaud in view of Hongbin do not teach
the method according to claim 1, compromising determining whether the adaptive strategy model is in a preset scenario; and determining that the service mode is a direct service mode in response to the adaptive strategy model being in the preset scenario.
But Jones teaches determining the service mode is a direct service model in response to an adaptive strategy
[0042 “In embodiments where the decision component 150 does not identify a resource or bot matching the topic and intent, in operation 314, the decision component 150 proceeds to operation 318. In operation 318, the decision component 150 determines whether a suitable or relevant human agent exists to respond to the query based on the intent and topic. The decision component 150 may determine the human agent exists by comparing the entity name and value or intent and topic with information, such as a profile, for the human agent. When the decision component 150 identifies a relevant human agent, the decision component 150 passes the query to the human agent at operation 320. Where the decision component 150 determines no relevant human agent exists, the decision component 150 proceeds to operation 322. In operation 322, the decision component 150 determines whether a topic or intent resource exists. The topic or intent resource may be a network resource, such as a database, a webpage, or other suitable data repository accessible to the decision component 150. The decision component 150 may identify a relevant topic or intent resource by comparing name and value pairs or components of the query with information within or metadata for the resource. Where the decision component 150 identifies a topic or intent resource, the decision component 150 provides the resource as a response to the query in operation 324. Where the decision component 150 identifies no topic or intent resource relevant to the query, the decision component cooperates with one or more other components of the query routing system 102 to generate and present one or more clarification questions to a user within a user interface at operation 310”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Renard in view of Hongbin to incorporate the routing framework of Jones in order to apply Renaud in view of Hongbin’s computed user state parameters within a known bot-versus-human architecture. Doing so would have predictably enabled automated escalation to a human agent when system confidence is not satisfied, thereby improving response reliability and overall user experience.
Regarding claim 10, Renard et al. recites an electronic device comprising at least one processor and a memory coupled to the at least one processor and having instructions stored therein, wherein the instructions, when executed by the at least one processor, cause the electronic device to perform actions corresponding to the limitations described in claim 1 (see the rejection of claim 1).
Additionally, regarding the processor and memory, Renard discloses
[0110 “A data processing system suitable for storing and/or executing program code may include at least one processor coupled directly or indirectly to memory elements through a system bus. The memory elements can include local memory employed during actual execution of the program code, bulk storage, and cache memories that provide temporary storage of at least some program code in order to reduce the number of times code may be retrieved from bulk storage during execution. Input/output or I/O devices (including but not limited to keyboards, displays, pointing devices, etc.) can be coupled to the system either directly or through intervening I/O controllers”].
As discussed above with respect to claim 1, these steps are rendered obvious in view of Renard in view of Hongbin in combination with Jones and Wang.
Regarding claim 11, the electronic device according to claim 10 where the actions perform the methods of claim 2. Claim 11 is rejected for the same reasons as claim 2.
Regarding claim 14, recites an electronic device according to claim 10, wherein determining a service mode for replying to the query content to perform the method of claim 5. Renard in view of Hongbin and in view of Jones and in further view of Wang do disclose an electronic device as indicated in claim 10. As discussed above with respect to claim 5, these steps are rendered obvious in view of Renard in view of Hongbin and in view of Jones and in further view of Wang.
Regarding Claim 15, it recites the electronic device according to claim 14 wherein determining the decision parameter consists of the methods outlined in claim 6. Claim 15 is rejected for the same reasons as claim 6.
Regarding claim 16, recites an electronic device according to claim 14, wherein determining the decision parameter to perform the method of claim 7. As discussed above with respect to claim 7, these steps are rendered obvious in view of Renard in combination with Jones.
Regarding claim 17, recites an electronic device according to claim 10, wherein the actions perform the method of claim 8. As discussed above with respect to claim 8, these steps are rendered obvious in view of Renard in view of Hongbin and in view of Jones and in further view of Wang.
Regarding claim 18, recites an electronic device according to claim 10, wherein the actions perform the method of claim 9. As discussed above with respect to claim 9, these steps are rendered obvious in view of Renard in view of Hongbin and in view of Jones and in further view of Wang.
Regarding claim 19, recites a computer program product comprising a non-transitory computer-readable medium having machine-executable instructions stored therein, the machine-executable instructions, when executed by a machine, causing the machine to perform actions comprising the method steps of claim 1 (see the rejection of claim 1):
Renard in view of Hongbin in view Jones and in further view of Wang do discloses a computer program tangibly stored on a non-transitory computer-readable medium and comprising machine-executable instructions.
Renaud discloses
[0091 “Furthermore, the technology can take the form of a computer program product accessible from a computer-usable or computer-readable medium providing program code for use by or in connection with a computer or any instruction execution system. For the purposes of this description, a computer-usable or computer readable medium can be any non-transitory storage apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device”].
Claim 19 is rejected for the same reasons as claim 1.
Regarding claim 20, recites a computer program according to claim 19 where the actions perform the methods of claim 2. Claim 20 is rejected for the same reasons as claim 2.
Claim [3, 12] are rejected under 35 U.S.C. 103 as being unpatentable over Renard (US Patent No
US-20190215249-A1) in view of Hongbin (CN 114861680) in view of Jones (US. Patent No. US
20220028378 A1) and in further view of Wang (US 12597420 B2) and in further view of Hasan
(Us. Patent No. US 20230350929).
Regarding claim 3 Renard in view of Hongbin and in further view of Jones and in further view of Want teach the method according to claim 1, wherein generating the intent parameter comprises: generating an evaluated intent value based on the user intent
Renard teaches - [0046 "In some embodiments, the AI engine 322 uses automatic speech recognition and natural language processing to determine a customer's intent (i.e. question) and uses an algorithm derived using machine learning to determine an appropriate response"].
However, Renard in view of Hongbin in view of Jones and in further view of Wang do not teach determining an intent based on both user input and preset intent information.
But Hasan teaches determining an intent associated with a user query by analyzing the user input and evaluating it using predefined intent categories and stored knowledge information - [0129] Further, the method 500, at step 510, may include generating a response corresponding to the intent through the virtual agent based on analyzation and the method 500 terminates at 512. A process of generating the response based on analyzing the knowledgebase is explained in greater detail in conjunction with FIG. 6. Here are some common techniques for generating the response:"]; [0130] Rule-based Systems: Rule-based systems may be utilized to generate responses based on predefined rules and patterns. These systems may have a set of predefined templates or patterns that match specific intents, allowing the virtual agent to select and populate the appropriate response based on the analyzed intent and relevant information from the knowledgebase”]; [0131] Natural Language Generation (NLG): NLG techniques may be employed to automatically generate human-like responses. NLG models may learn patterns and structures from the analyzed knowledgebase and use that knowledge to generate coherent and contextually relevant responses. These models may be trained on large amounts of text data to improve the quality and fluency of the generated responses”]; [0158] The method 700 illustrated by the flow diagram of FIG. 7 for generating a response through a virtual agent starts at step 702. The method 700 may include, at step 704, receiving, by the virtual agent, a query from a user”]; [0159] The method 700, at step 706, may include determining, by the virtual agent, the intent associated with the query. In this step, the virtual agent analyzes the user's query to determine the intent behind it. The intent represents the purpose or goal of the user's query. This may be achieved by the following techniques”]; [0160] Intent Classification: The virtual agent may utilize machine learning techniques, such as supervised learning algorithms, to classify the user's query into predefined intent categories. A training dataset consisting of labeled queries and their corresponding intents is used to train a classifier. The virtual agent then applies this trained model to predict the intent of the user's query “]; [0161] Natural Language Understanding (NLU): NLU techniques may be employed to extract the intent from the user's query. NLU models may analyze the syntactic and semantic structure of the query, identify key phrases or keywords, and map them to predefined intents. Techniques like named entity recognition, part-of-speech tagging, and dependency parsing can be applied to assist in intent determination”]; [0162] Keyword Matching: The virtual agent may use a rule-based approach to match the user's query against a set of predefined keywords or patterns associated with specific intents. If the query contains keywords or phrases that match these predefined patterns, the intent may be determined accordingly”]; [0193 “Utilizing Learned Patterns: The trained LLM has learned patterns from the training data, which includes understanding grammar, syntax, and semantic structures. The LLM can identify patterns in the input query or request and utilize this knowledge to generate a response. For example, if the input query is in the form of a question, the LLM can identify question patterns and provide an appropriate response.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the intent determination framework of Hasan into the primary reference Renard in view of Hongbin in view Jones and in further view of Wang in order to provide a structured intent-based decision mechanism. Such a combination would improve the system ability to accurately interpret user queries by incorporating predefined intent classification into the content-based response framework thereby enabling more structured and reliable decision making.
Regarding claim 12, the electronic device according to claim 10 where the actions perform the methods of claim 3. Claim 12 is rejected for the same reasons as claim 3.
Claim [4, 13] are rejected under 35 U.S.C. 103 as being unpatentable over Renard (US Patent No
US-20190215249-A1) in view of Hongbin (CN 114861680) in view of Jones (US. Patent No. US
20220028378 A1) and in further view of Wang (US 12597420 B2)and in further view of
Moudy (US Patent No. US-10546235-B2).
Regarding claim 4 Renard in view of Hongbin in view Jones and in further view of Wang teach the method according to claim 1, wherein generating the emotion parameter comprises: determining an average emotional value based on scores,
Renard teaches
[0060 " In one embodiment, the “sentiment” of the conversation may be evaluated by the conversation ranking engine 324 using machine-based sentiment analysis of each interaction. In one embodiment, sentiment analysis refers to the use of natural language processing, text analysis, computational linguistics and biometrics (e.g. voice) to systematically identify, extract and study affective states and subjective information, e.g., to determine the attitude of a speaker, writer or other subject. Depending on the embodiment, the attitude may be one or more of a judgment or evaluation (as in appraisal theory), affective state (i.e., the emotional state of the author or speaker), or the intended emotional communication (i.e., the emotional effect intended by the author or interlocutor). Depending on the embodiment, the conversation ranking engine 324 uses one or more of the following, which may be referred to as “sentiment criteria,” to evaluate the “sentiment” of the session and rank the session"]; [0074 "In some embodiments, one or more of the foregoing sentiment criteria are used to generate a sentiment metric, which is used by the conversation ranking engine 324 alone, or in combination with a sentiment metric (depending on the embodiment), to rank the conversation"]. This implies generating an emotion parameter based on sentiment analysis.
[0061- Average (e.g. arithmetic or harmonic) sentiment analysis over all interactions in the conversation/session, which may serve as a forecast of the conversation or as a dialogue atmosphere”]; [0062 ”Sentiment analysis trend of the conversation, e.g., positive to negative, neutral to negative, the opposite, etc.”]; The average sentiment analysis and trend values are interpreted as scores.
[0063 “Offensive language usage, e.g., using keyword detection based on offenses n-gram dictionary.”]; [0064 “Detection of emojis, which may balance the trend of conversation (e.g. a winking, tongue sticking out or smiling emoji may indicate use of an otherwise offensive term is being used playfully or in jest”];
However, Renard in view of Hongbin in view of Jones and in further view of Wang do not teach generating the emotion parameter based on the average emotional value and a preset value.
Moudy discloses using thresholds (preset values) for sentiment analysis – [ Column 25, lines 45-60 “In certain embodiments, the outputs of the sentiment analyzer may be determined in accordance with preprogrammed or dynamically implemented rules and/or criteria. For example, sentiment analyzer output rules may be based on sentiment score thresholds, whereby the feedback analytics server 610 may transmit notifications to predetermined recipients only if a sentiment score calculated in step 804 is greater than a specified threshold, less than a specified threshold, etc. In some cases, certain recipient devices may receive notifications for certain sentiment score thresholds or ranges, while other recipients may receive notifications for other sentiment score thresholds or ranges. In other examples, sentiment analyzer output rules may be based on changes in sentiment scores over time, or identifications of outliers within multiple related sentiment scores. For instance, in professional training or educational system 600, a content provider server 640, authorized client device 630, or other recipient device may receive a notification if the sentiment score for a user (e.g., employee or student), group of users (e.g., class or grade), an instructor or presenter, or one or more content items (e.g., a course, module assignment, test, etc.) falls below a certain threshold. Similarly, in a sentiment analyzer system 600 used with an eCommerce or media distribution system, a content provider 640 or client device 630 may receive a notification in response to unanticipated high or low sentiment scores for a product, product line, or media content, recent changes in sentiment scores for a product or media content, and the like.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the sentiment scoring of Moudy into Renard in view of Hongbin and in view of Jones and in further view of Wang to provide a structured emotional parameter output thereby improving emotional evaluation accuracy withing a decision framework. Additionally, this provides a more predictable and more reliable method of quantifying emotional state.
Regarding claim 13, the electronic device according to claim 10 wherein generating the emotion parameter based on the methods of claim 4 - Claim 13 is rejected for the same reasons as claim 4.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office
action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the
extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from
the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date
of this final action and the advisory action is not mailed until after the end of the THREE-MONTH
shortened statutory period, then the shortened statutory period will expire on the date the advisory
action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will
be calculated from the mailing date of the advisory action. In no event, however, will the statutory
period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHEZA ABDUL AZIZ whose telephone number is (571)272-9610. The examiner can normally be reached Monday-Friday 7:30am-5pm Alternate Fridays off.
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, Daniel Washburn can be reached at (571) 272-5551. 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.
/SHEZA ABDUL AZIZ/Examiner, Art Unit 2657
/DANIEL C WASHBURN/Supervisory Patent Examiner, Art Unit 2657
Prosecution Insights