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 .
Continued Examination Under 37 CFR 1.114
2. 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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 4/21/2026 has been entered.
Accordingly, claims 1-20 are pending in this application. Claims 1 and 11 are currently amended.
Response to Arguments
3. Applicant’s arguments with respect to amended pending claims filed on 4/21/2026 have been fully considered. In view of the claim amendment filed, the rejection has been withdrawn. However, upon further consideration, a new ground(s) of rejection is made.
Further, regarding the new limitations recited in claims 1 and 11, it is submitted that they are properly addressed by the new ground of rejection.
Furthermore, it is also submitted that all limitations in pending claims, including those not specifically argued, are properly addressed. The reason is set forth in the rejections. See claim analysis below for detail.
Claim Rejections - 35 USC § 103
4. In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
5. 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.
6. Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Vasylyev (US 20240412720 A1) in view of Venkateshwaran et al. (US 20240232539 A1).
7. Regarding Claim 1, Vasylyev discloses a computer-implemented method ([Abstract]: a method for providing a contextualized response to a user) executed on data processing hardware that causes the data processing hardware to perform operations comprising (Fig. 1; assistant system 2; [0020] FIG. 1 is a schematic diagram showing an exemplary implementation of an AI assistant system):
receiving a query and context associated with the query (Fig. 2; [0012]: The processor may be further configured to execute instructions to proactively generate a plurality of candidate conversational responses based on the conversational context data prior to receiving a subsequent user query; [0194]: When the user issues a query or command, assistant system 2 retrieves the relevant audio data and contextual information from the active windows);
determining, from the context, two or more context components ([0050]: The Assistant may be configured to access external storage… and fulfill user commands. Useful examples of such data include but are not limited to weather updates, stock quotes, road conditions, names or other attributes of people in an address book, latest news on a specific subject, various documents stored locally or on a cloud, and so on);
for each respective context component of the two or more context components:
determining a corresponding priority score based on a relevance of the respective context component to the query ([0107] The dynamic adjustment of the conversation memory time window may be configured to utilize quantifying and combining several key factors which can be normalized and combined into an overall score).
However, Vasylyev does not explicitly teach “the corresponding priority score determined for the respective context component indicating relevance of the respective context component compared to each other context component of the two or more context components ; and biasing the respective context component by weighting the respective context component based on a value of the corresponding priority score; and generating, using a neural network model, a response based on the query and the biased two or more context components”.
On the other hand, in the same field of endeavor, Venkateshwaran teaches
the corresponding priority score determined for the respective context component indicating relevance of the respective context component compared to each other context component of the two or more context components (Fig. 2; [0087]: Next, the “context extractor” scores each of the extracted contexts… and keeps only the ones with highest correlation to the question being addressed, (3) using a machine learning model that has been previously trained using supervised learning towards the end goal to score the contexts. The content extractor then combines the scores using use-case and question specific priority rules and comes up with the final score, keeping the contexts that have the highest final score); and
biasing the respective context component by weighting the respective context component based on a value of the corresponding priority score (Fig. 1; [0087]: These and other aspects of context tuning are disclosed herein, including use of weights to bias the outputs and various aspects of machine learning; [0093]-[0094]: Tuning weights may be used to either bias or un-bias the learnings and suggestions, as appropriate; Here, tuning also includes adjusting the weights in the scoring rules for scoring candidate context sin coming up with the final context… Through this process, reinforcement training may add a “good” bias); and
generating, using a neural network model, a response based on the query and the biased two or more context components (Figs. 2-3; [0087]-[0088]: Systems/processes 300-1400 may include machine-learning modules/routines… that may be used to generate various recommendations and suggestions, and to ask and answer questions; [0097]: FIG. 3 illustrates an example system 300 for generating suggestions/alerts based on an expert system approach, with score and context… configuring words/phrases and NLP models).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the teaching of Vasylyev to incorporate the teachings of Venkateshwaran to determine a score indicating relevance and performing the biasing by weighting the component based on the score.
The motivation for doing so would be to enhance a customer's business process, as recognized by Venkateshwaran ([0114]: Thus, the reinforcement machine learning provides a positive bias that enhances a customer's business process).
Regarding Claim 2, the combined teachings of Vasylyev and Venkateshwaran disclose the method of claim 1.
Vasylyev further teaches wherein the neural network model comprises an automatic speech recognition model or a large language model (Fig. 1; [0033]: Large Language Models (LLMs), and/or other known forms or combinations of generative AI technology. The LLMs may be trained on a large corpus of text and utilize a neural network; [0093]: According to one embodiment, processor 122 incorporates processors inspired by the structure and function of biological neural networks… These processors can be particularly useful for tasks that require real-time, low-power processing, such as always-on speech recognition).
Regarding Claim 3, the combined teachings of Vasylyev and Venkateshwaran disclose the method of claim 1.
Vasylyev further teaches wherein the neural network model resides at a user device (Fig. 1; [0066]-[0067]: Useful examples of mobile devices that can incorporate assistant 2 include but are not limited to: smartphones, tablets, smartwatches, wearable devices, laptops, e-readers, portable gaming devices, personal digital assistants; [0080]: For processing images captured by the camera or image sensor, the AI Assistant may use convolutional neural networks (CNNs) to learn features).
Regarding Claim 4, the combined teachings of Vasylyev and Venkateshwaran disclose the method of claim 1.
Vasylyev further teaches wherein the context comprises contextual data elements (Fig. 1; [0102]: audio processing unit 125 may be configured to perform user identification based on the unique vocal characteristics of each user… This user identification functionality may be integrated with the other components of assistant system 2, such as the contextual memory unit 116, which can maintain separate conversation histories and contextual data for each identified user),
each respective contextual data element associated with a corresponding context modality ([0015]: The system may also include a wireless communication device for accessing external databases and internet resources, and a multi-modal input processing unit for processing various inputs such as speech, visual, text, and gesture).
Regarding Claim 5, the combined teachings of Vasylyev and Venkateshwaran disclose the method of claim 4.
Vasylyev further teaches wherein each respective context component of the one or more context components comprises one or more of the contextual data elements each associated with the same corresponding context modality (Fig. 1; [0401]: Assistant system 2 may be configured to include the content of its output to the users (or its representation in any form, such as audio, text, tokens, encoded data, contextual data, vectors, etc.) into the information it stores in its memory in conjunction with the ongoing conversation between the users. In other words, the stored conversation may include responses from the systems as a part of it).
Regarding Claim 6, the combined teachings of Vasylyev and Venkateshwaran disclose the method of claim 1.
Vasylyev further teaches wherein the operations further comprise, for each respective context component of the one or more context components:
for each respective context model of a plurality of context models, determining a corresponding intermediate weight based on a respective relevance of the respective context component to the query ([0107]: Additionally, the system can monitor user behavior during conversations, such as scrolling back to review older context or requesting information from earlier in the dialogue. These actions can be tracked and analyzed using a weighted moving average algorithm); and
determining a corresponding final weight based on each corresponding intermediate weight determined for the respective context component ([0112]: In step 2, assistant system 2 may assign weights to each normalized factor based on their relative importance), wherein the corresponding priority score for the respective context component corresponds to the corresponding final weight ([0117]-[0122]: In step 2, assistant system 2 applies weights, e.g., default weights from the above-described examples. In step 3, assistant system 2 calculates utility score).
Regarding Claim 7, the combined teachings of Vasylyev and Venkateshwaran disclose the method of claim 6.
Vasylyev further teaches wherein determining the corresponding final weight comprises selecting the greatest corresponding intermediate weight determined for the respective context component as the corresponding final weight ([0107]: These actions can be tracked and analyzed using a weighted moving average algorithm to calculate an implicit user preference score (e.g., implicit_window_preference), in the range, e.g., [0.0, 1.0], where higher values indicate a desire for longer context retention; [0110]: Assistant system 2 may employ various algorithms and mathematical models to weight and combine the above-described factors into a unified metric; [0113]: In step 3, assistant system 2 may calculate the weighted average of the normalized factors to obtain the overall utility score).
Regarding Claim 8, the combined teachings of Vasylyev and Venkateshwaran disclose the method of claim 6.
Vasylyev further teaches wherein each respective context model is configured to process a particular type of context modality ([0205]: Assistant system 2 may utilize its multi-modal input processing capabilities, including speech recognition, natural language processing, reasoning, and sentiment analysis, to extract key information and patterns from the conversation data; [0282]: Similarly, as assistant system 2 continues its engagement in a conversation with the user(s) it may continuously update the conversational context data stored in contextual memory unit 116 with a suitable representation the generated conversational response, preferably matching the type of representation of the input data stored in this memory unit).
Regarding Claim 9, the combined teachings of Vasylyev and Venkateshwaran disclose the method of claim 1.
Vasylyev further teaches wherein the operations further comprise:
selecting, from the biased one or more context components, a subset of biased context components based on the corresponding priority score of each respective biased context component, wherein the generating the response is further based on the subset of biased context components (Fig. 1; [0244] Assistant system 2 may also be configured to employ prioritized memory retention where it assigns priority scores to different parts of the conversational context based on their estimated relevance and informativeness. When the context window needs to be reduced, the system may preferentially retain the highest-priority segments while discarding or offloading the lower-priority ones; [0307]: The GPT model may be configured to detect when the conversation history exceeds the model's input window or context window capacity and generate a summary for a subset of this data and then iteratively update this summary with information from the next subset).
Regarding Claim 10, the combined teachings of Vasylyev and Venkateshwaran disclose the method of claim 9.
Vasylyev further teaches wherein each respective biased context component in the subset of biased context components is associated with a corresponding priority score that satisfies a priority score threshold ([0244]: Assistant system 2 may also be configured to employ prioritized memory retention where it assigns priority scores to different parts of the conversational context based on their estimated relevance and informativeness).
Regarding Claim 11, Vasylyev discloses a system comprising: data processing hardware; and memory hardware in communication with the data processing hardware (Fig. 1; assistant system 2; [0020] FIG. 1 is a schematic diagram showing an exemplary implementation of an AI assistant system) the memory hardware storing instructions that when executed on the data processing hardware cause the data processing hardware to perform operations comprising:
receiving a query and context associated with the query (Fig. 2; [0012]: The processor may be further configured to execute instructions to proactively generate a plurality of candidate conversational responses based on the conversational context data prior to receiving a subsequent user query; [0194]: When the user issues a query or command, assistant system 2 retrieves the relevant audio data and contextual information from the active windows);
determining, from the context, two or more context components ([0050]: The Assistant may be configured to access external storage… and fulfill user commands. Useful examples of such data include but are not limited to weather updates, stock quotes, road conditions, names or other attributes of people in an address book, latest news on a specific subject, various documents stored locally or on a cloud, and so on);
for each respective context component of the two or more context components
determining a corresponding priority score based on a relevance of the respective context component to the query ([0107]: The dynamic adjustment of the conversation memory time window may be configured to utilize quantifying and combining several key factors which can be normalized and combined into an overall score).
However, Vasylyev does not explicitly teach “the corresponding priority score determined for the respective context component indicating relevance of the respective context component compared to each other context component of the two or more context components ; and biasing the respective context component by weighting the respective context component based on a value of the corresponding priority score; and generating, using a neural network model, a response based on the query and the biased two or more context components”.
On the other hand, in the same field of endeavor, Venkateshwaran teaches
the corresponding priority score determined for the respective context component indicating relevance of the respective context component compared to each other context component of the two or more context components (Fig. 2; [0087]: Next, the “context extractor” scores each of the extracted contexts… and keeps only the ones with highest correlation to the question being addressed, (3) using a machine learning model that has been previously trained using supervised learning towards the end goal to score the contexts. The content extractor then combines the scores using use-case and question specific priority rules and comes up with the final score, keeping the contexts that have the highest final score); and
biasing the respective context component by weighting the respective context component based on a value of the corresponding priority score (Fig. 1; [0087]: These and other aspects of context tuning are disclosed herein, including use of weights to bias the outputs and various aspects of machine learning; [0093]-[0094]: Tuning weights may be used to either bias or un-bias the learnings and suggestions, as appropriate; Here, tuning also includes adjusting the weights in the scoring rules for scoring candidate context sin coming up with the final context… Through this process, reinforcement training may add a “good” bias); and
generating, using a neural network model, a response based on the query and the biased two or more context components (Figs. 2-3; [0087]-[0088]: Systems/processes 300-1400 may include machine-learning modules/routines… that may be used to generate various recommendations and suggestions, and to ask and answer questions; [0097]: FIG. 3 illustrates an example system 300 for generating suggestions/alerts based on an expert system approach, with score and context… configuring words/phrases and NLP models).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the teaching of Vasylyev to incorporate the teachings of Venkateshwaran to determine a score indicating relevance and performing the biasing by weighting the component based on the score.
The motivation for doing so would be to enhance a customer's business process, as recognized by Venkateshwaran ([0114]: Thus, the reinforcement machine learning provides a positive bias that enhances a customer's business process).
Regarding Claim 12, the combined teachings of Vasylyev and Venkateshwaran disclose the system of claim 11.
Vasylyev further teaches wherein the neural network model comprises an automatic speech recognition model or a large language model. (Fig. 1; [0033]: Large Language Models (LLMs), and/or other known forms or combinations of generative AI technology. The LLMs may be trained on a large corpus of text and utilize a neural network; [0093]: According to one embodiment, processor 122 incorporates processors inspired by the structure and function of biological neural networks… These processors can be particularly useful for tasks that require real-time, low-power processing, such as always-on speech recognition).
Regarding Claim 13, the combined teachings of Vasylyev and Venkateshwaran disclose the system of claim 11.
Vasylyev further teaches wherein the neural network model resides at a user device (Fig. 1; [0066]-[0067]: Useful examples of mobile devices that can incorporate assistant 2 include but are not limited to: smartphones, tablets, smartwatches, wearable devices, laptops, e-readers, portable gaming devices, personal digital assistants; [0080]: For processing images captured by the camera or image sensor, the AI Assistant may use convolutional neural networks (CNNs) to learn features)..
Regarding Claim 14, the combined teachings of Vasylyev and Venkateshwaran disclose the system of claim 11.
Vasylyev further teaches wherein the context comprises contextual data elements (Fig. 1; [0102]: audio processing unit 125 may be configured to perform user identification based on the unique vocal characteristics of each user… This user identification functionality may be integrated with the other components of assistant system 2, such as the contextual memory unit 116, which can maintain separate conversation histories and contextual data for each identified user), each respective contextual data element associated with a corresponding context modality ([0015]: The system may also include a wireless communication device for accessing external databases and internet resources, and a multi-modal input processing unit for processing various inputs such as speech, visual, text, and gesture).
Regarding Claim 15, the combined teachings of Vasylyev and Venkateshwaran disclose the system of claim 14.
Vasylyev further teaches wherein each respective context component of the one or more context components comprises one or more of the contextual data elements each associated with the same corresponding context modality (Fig. 1; [0401]: Assistant system 2 may be configured to include the content of its output to the users (or its representation in any form, such as audio, text, tokens, encoded data, contextual data, vectors, etc.) into the information it stores in its memory in conjunction with the ongoing conversation between the users. In other words, the stored conversation may include responses from the systems as a part of it).
Regarding Claim 16, the combined teachings of Vasylyev and Venkateshwaran disclose the system of claim 11.
Vasylyev further teaches wherein the operations further comprise, for each respective context component of the one or more context components: for each respective context model of a plurality of context models, determining a corresponding intermediate weight based on a respective relevance of the respective context component to the query ([0107]: Additionally, the system can monitor user behavior during conversations, such as scrolling back to review older context or requesting information from earlier in the dialogue. These actions can be tracked and analyzed using a weighted moving average algorithm); and
determining a corresponding final weight based on each corresponding intermediate weight determined for the respective context component, wherein the corresponding priority score for the respective context component corresponds to the corresponding final weight ([0112]: In step 2, assistant system 2 may assign weights to each normalized factor based on their relative importance), wherein the corresponding priority score for the respective context component corresponds to the corresponding final weight ([0117]-[0122]: In step 2, assistant system 2 applies weights, e.g., default weights from the above-described examples. In step 3, assistant system 2 calculates utility score).
Regarding Claim 17, the combined teachings of Vasylyev and Venkateshwaran disclose the system of claim 16.
Vasylyev further teaches wherein determining the corresponding final weight comprises selecting the greatest corresponding intermediate weight determined for the respective context component as the corresponding final weight ([0107]: These actions can be tracked and analyzed using a weighted moving average algorithm to calculate an implicit user preference score (e.g., implicit_window_preference), in the range, e.g., [0.0, 1.0], where higher values indicate a desire for longer context retention; [0110]: Assistant system 2 may employ various algorithms and mathematical models to weight and combine the above-described factors into a unified metric; [0113]: In step 3, assistant system 2 may calculate the weighted average of the normalized factors to obtain the overall utility score).
Regarding Claim 18, the combined teachings of Vasylyev and Venkateshwaran disclose the system of claim 16.
Vasylyev further teaches wherein each respective context model is configured to process a particular type of context modality ([0205]: Assistant system 2 may utilize its multi-modal input processing capabilities, including speech recognition, natural language processing, reasoning, and sentiment analysis, to extract key information and patterns from the conversation data; [0282]: Similarly, as assistant system 2 continues its engagement in a conversation with the user(s) it may continuously update the conversational context data stored in contextual memory unit 116 with a suitable representation the generated conversational response, preferably matching the type of representation of the input data stored in this memory unit).
Regarding Claim 19, the combined teachings of Vasylyev and Venkateshwaran disclose the system of claim 11.
Vasylyev further teaches wherein the operations further comprise: selecting, from the biased one or more context components, a subset of biased context components based on the corresponding priority score of each respective biased context component, wherein the generating the response is further based on the subset of biased context components (Fig. 1; [0244] Assistant system 2 may also be configured to employ prioritized memory retention where it assigns priority scores to different parts of the conversational context based on their estimated relevance and informativeness. When the context window needs to be reduced, the system may preferentially retain the highest-priority segments while discarding or offloading the lower-priority ones; [0307]: The GPT model may be configured to detect when the conversation history exceeds the model's input window or context window capacity and generate a summary for a subset of this data and then iteratively update this summary with information from the next subset).
Regarding Claim 20, the combined teachings of Vasylyev and Venkateshwaran disclose the system of claim 19.
Vasylyev further teaches wherein each respective biased context component in the subset of biased context components is associated with a corresponding priority score that satisfies a priority score threshold ([0244]: Assistant system 2 may also be configured to employ prioritized memory retention where it assigns priority scores to different parts of the conversational context based on their estimated relevance and informativeness).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHIRLEY D. HICKS whose telephone number is (571)272-3304. The examiner can normally be reached Mon - Fri 7:30 - 4:00.
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, Charles Rones can be reached on (571) 272-4085. 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.
/S.D.H./Examiner, Art Unit 2168
/CHARLES RONES/Supervisory Patent Examiner, Art Unit 2168
Prosecution Insights