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 .
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.
Claim(s) 1-2, 5-20 are rejected under 35 U.S.C. 103 as being unpatentable over Cai (CA 3,042,921) in view of Zengliang (NPL “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks”).
With respect to claim 17 (similarly claim 1), Cai teaches a system (e.g. an example agent platform 100 Fig 1 [0042]) for generating a support summary from a natural language chat record of a support interaction with a support user (e.g. the agent platform 100 can process the data to generate summary reports, risk profiles, regulatory issues, and so on, in response to inquiries received at virtual agent 180 using natural language processor 120 [0044]), the support summary comprising natural language text identifying a support issue and information germane to the support issue (e.g. the natural language processor 120 can computationally summarize an article/inputted text to extract the main concepts/ideas [0125], Fig 5 [0204]-[0207]), the system comprising:
a tokenization module (e.g. tokenization function [0055]) configured to tokenize the natural language chat record (e.g. tokenize natural language data including a data representation of user-inputted text [0126]);
a sentiment identification module configured to produce sentiment data from the tokenized natural language chat record (e.g. the agent platform 100 uses natural language processor 120 for sentiment analysis [0128]), the sentiment data reflecting sentiment of the support user during the support interaction (e.g. Sentiment analysis is the process of computationally identifying the attitude/sentiment of a given text. The natural language processor 120 can classify sentences as either very negative, negative, neutral, positive, or very positive [0128]);
a semantic analysis module (e.g. semantic process [0108]) configured to extract support-relevant features from the tokenized natural language chat record (e.g. determine concepts, select sentences that represent the concepts, and generate the report summary using the selected sentences [0108], see also [0168]); and
(e.g. acoustic model [0103]-[0105]) configured to generate the support summary from the support-relevant features and the sentiment data (e.g. The virtual agent 180 can integrate with natural language processor 120 for text analysis and summary report generation. The virtual agent 180 can integrate with cognitive search to enable processing of search requests and retrieval of search results [0163], see also [0205]-[0209] where a summary report is generated from the support-relevant features and the sentiment data).
However, Cai fails to teach the acoustic model of [0103]-[0105] is a language model to generate support summary
Zengliang teaches a language model KNN-LM in page 8-9 to generate support summary as suggested in pp21-26.
Cai and Zengliang are analogous art because they all pertain to generating summaries. Therefore it would have been obvious to people having ordinary skill in the art before the effective filing date of the claimed invention to modify Cai with the language model of Zengliang to include: a language model to generate support summary, as suggested by Zengliang in pp21-26. The benefit of the modification would be to improve the accuracy of the whole system.
With respect to claim 18, Cai teaches the system of claim 17, wherein the support-relevant features include at least one support issue, and the support summary identifies the support issue (e.g. [0044], [0062]-[0067], [0090]-[0097] where the incident ticket includes at least one support issue and the summary report identifies the support issue).
With respect to claim 19 (similarly claim 13), Cai in view of Zengliang teaches the system of claim 17, wherein the language model is a large language model configured to generate the support summary from the natural language chat record using the sentiment data and the support-relevant features for context injection (e.g. [0044], [0162]-[0164] suggest the language model as modified by Zengliang is a large language model configured to generate the support summary from the natural language chat record using the sentiment data and the support-relevant features for context injection, as suggested in [0205]-[0209] where a summary report is generated from the support-relevant features and the sentiment data for context injection).
With respect to claim 20 (similarly claim 2), Cai teaches the system of claim 17, further comprising an audiovisual capture device configured to capture audio or video data corresponding to the support interaction with the support user (e.g. [0099]-[0103] describes an audio/video input device to capture audio or video data corresponding to the support interaction with the support user), wherein the sentiment identification is configured to produce the sentiment data at least in part from the captured audio or video data (e.g. [0111], [0128], [0152], [0195]-[0196] disclose the sentiment data produced from at least in part from the captured audio or video data).
With respect to claim 5, Cai teaches the method of claim 1, wherein the sentiment data comprises: identification of a plurality of user sentiments expressed by the support user during the support chat (e.g. the sentiment comprises identification of a plurality of user sentiments very negative, negative, neutral, positive, or very positive [0128]); and identification of a sentiment transition between the plurality of user sentiments (e.g. [0152] identifies a sentiment transition between the plurality of user sentiments i.e. +1 for each positive word and -1 for each negative word). The natural language processor 120 can classify the sentences based on aggregated score (for example: +4 or greater = Very Positive, +2 or greater = Positive, 0 = Neutral, -2 or less = Negative, -4 or less = Very Negative).
With respect to claim 6, Cai teaches the method of claim 1, wherein the support summary includes an identification of user sentiment (e.g. The interface 500 includes visual elements for distribution of sentence sentiments based on the input data 508 [0205]-[0206], [0209]).
With respect to claim 7, Cai teaches the method of claim 1, wherein producing the support summary using the language model comprises correlating the user sentiment and a change in the user sentiment with at least one chat string included among the chat information (e.g. +1 for each positive word and -1 for each negative word [0152] suggest correlating the user sentiment and a change in the user sentiment with at least one chat string included among the chat information).
With respect to claim 8, Cai teaches the method of claim 1, further comprising performing feature extraction on natural language technical notes recorded by a support technician (e.g. [0087] performs feature extraction on an incident/ natural language technical notes recorded by a support technician), and the language model produces the support summary at least in part based on features extracted from the natural language technical notes (e.g. Fig 5 [0205]-[0209] the language model produces the summary report based on features extracted from the natural language technical notes).
With respect to claim 9, Cai teaches the method of claim 1, wherein the language model produces the support summary at least in part from the extracted support-relevant features, the sentiment data, and the natural language data (e.g. Fig 5 [0205]-[0209] the language model produces the summary report at least in part from the extracted support-relevant features, the sentiment data, and the natural language data).
With respect to claim 10, Cai teaches the method of claim 9, wherein the language model receives the extracted support-relevant features and the sentiment data in the form of context injection for the generative production of the support summary (e.g. Fig 5 [0205]-[0209] the language model receives the extracted support-relevant features and the sentiment data in the form of context injection for the generative production of the support summary).
With respect to claim 11, Cai teaches the method of claim 1, wherein performing semantic analysis on the tokenized language comprises classifying the tokenized language according to semantic features (e.g. [0095]-[0096], [0108]-[0109], [0128], [0193], [0206] classify the tokenized language according to semantic features).
With respect to claim 12, Cai teaches the method of claim 11, further comprising performing feature extraction on the chat information to flag the semantic features, wherein the feature extraction includes at least one of bag-of-words (BOW) analysis, bag-of-n-grams analysis, term frequency-inverse document frequency (TF-IDF) vectorization analysis, and One Hot encoding (e.g. [0129] and claim 11 disclose the feature extraction includes term frequency-inverse document frequency (TF-IDF) vectorization analysis).
With respect to claim 14, Cai teaches the method of claim 13 including the generation of the support summary using at least one of the sentiment data and the support-relevant features.
However, Cai fails to teach wherein the generation of the support summary is constrained by retrieval augmented generation (RAG) using at least one of the sentiment data and the support-relevant features.
Zengliang teaches a generation of tokens/sequence pp29-31 constrained by retrieval augmented generation (RAG) using at least one of the tokens and sequence.
Cai and Zengliang are analogous art because they all pertain to generating summaries/tokens/sequence. Therefore it would have been obvious to people having ordinary skill in the art before the effective filing date of the claimed invention to modify Cai with the RAG models of Zengliang to include: wherein the generation of the support summary is constrained by retrieval augmented generation (RAG) using at least one of the sentiment data and the support-relevant features, as suggested by Zengliang in pp29-31. The benefit of the modification would be to improve the accuracy of the whole system.
With respect to claim 15, Cai teaches a method of providing technical support to a user (e.g. a method for providing summary reports, risk profiles, regulatory issues, and so on, in response to inquiries received at virtual agent 180 using natural language processor 120 [0044]), the method comprising: generating a support summary according to the method of claim 1 (e.g. generating a report summary [0044]);
reviewing the support summary (reviewing the summary report as suggested in [0205]-[0209]); providing technical support related to the support issue identified in the support summary, to the user (e.g. The platform 100 can implement a virtual agent 180 which can be an automated chatbot agent with a focus on providing IT production support based on predictive/prescriptive models 126, natural language processor 120 and machine learning [0165]).
With respect to claim 16, Cai in view of Zengliang teaches the method of claim 15, wherein the language model produces the support summary at least in part base on features extracted from natural language technical notes (e.g. the language model as modified by Zengliang produces the summary report of Fig 5 based on features extracted from IT incident tickets [0004], [0049], [0087]) the method further comprising updating the technical notes while or after providing the technical support to the user (e.g. process text fields of IT incident tickets to update a knowledge base for a natural language processor and machine learning [0004], see also [0060]), the technical notes including at least one of open questions, actions taken or pending, and possible or excluded diagnoses (e.g. [0062]-[0066] disclose the IT incident tickets include at least one of open questions, actions taken or pending, and possible or excluded diagnoses).
Claim(s) 3 is rejected under 35 U.S.C. 103 as being unpatentable over Cai (CA 3,042,921) in view of Zengliang (NPL “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks”) and further in view of Yoon (US 2019/0005957).
With respect to claim 3, Cai in view of Zengliang teaches the method of claim 2 including performing sentiment analysis on the AV data and generating the sentiment identification.
However, Cai fails to teach wherein performing sentiment analysis on the AV data comprises: extracting audio waveform data from the AV data; and generating the sentiment identification based in part on the extracted audio waveform data.
Yoon teaches extracting audio waveform data from the AV data (e.g. receiving a sound signal composed of sound waveforms as input data [0004]); and generating the sentiment identification based in part on the extracted audio waveform data (e.g. identifying words or word sequences, and extracting the meaning thereof [0004]).
Cai and Yoon are analogous art because they all pertain to processing audio sound signals. Therefore it would have been obvious to people having ordinary skill in the art before the effective filing date of the claimed invention to modify Cai with the teachings of Yoon to include: wherein performing sentiment analysis on the AV data comprises: extracting audio waveform data from the AV data; and generating the sentiment identification based in part on the extracted audio waveform data, as suggested by Yoon in [0004]. The benefit of the modification would be for enabling a user to accurately control a smart terminal using a set of voice commands only under preset conditions, in which the set of voice commands differs from condition to condition Yoon [0008].
Claim(s) 4 is rejected under 35 U.S.C. 103 as being unpatentable over Cai (CA 3,042,921) in view of Zengliang (NPL “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks”) and further in view of Saxena (US 2023/0206692).
With respect to claim 4, Cai in view of Zengliang teaches the method of claim 2 including performing sentiment analysis on the AV data and generating the sentiment identification.
However, Cai fails to teach wherein performing sentiment analysis on the AV data comprises: extracting facial expression information from the AV data; and generating the sentiment identification based in part on the facial expression information.
Saxena teaches extracting facial expression information from the AV data (e.g. extracting facial expression from the video data [0061], claim 1, see also [0019]); and generating the sentiment identification based in part on the facial expression information (e.g. step 808, the method 800 determines a sentiment score for the customer for the facial expression information Fig 8 [0061], claim 1, [0019]).
Cai and Saxena are analogous art because they all pertain to extracting features from AV/video data. Therefore, it would have been obvious to people having ordinary skill before the effective filing date of the claimed invention to modify Cai with the teachings of Saxena to include: wherein performing sentiment analysis on the AV data comprises: extracting facial expression information from the AV data; and generating the sentiment identification based in part on the facial expression information, as suggested by Saxena in [0061], claim 1. The benefit of the modification would be such performance scores determined over several such sales calls of the salesperson with same or different customers or potential customers are combined to compute a cumulative performance score (CPS) for the salesperson, Saxena [0019].
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to IBRAHIM SIDDO whose telephone number is (571)272-4508. The examiner can normally be reached 9:00-5:30PM.
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, Akwasi Sarpong can be reached at 5712703438. 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.
/IBRAHIM SIDDO/Primary Examiner, Art Unit 2681