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 .
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.
Priority
Acknowledgment is made of applicant's claim for domestic priority based on US application 17/668968 filed on 02/10/2022.
Claim Rejections - 35 USC § 103
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 103 that form the basis for the rejections under this section made 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, 3-10, and 12-20 are rejected under 35 USC 103(a) as being unpatentable over Somech et al. (US 11699039 B2) in view of Wassertblat et al. (US 2012/0215535 A1).
Regarding Claims 1, 10, and 19, Somech discloses a system (Fig. 6, computing device 600) comprising:
one or more hardware processors (Col 39, Row 41, one or more processors 614); and
memory / One or more non-transitory computer readable media storing instructions that, when executed by the one or more hardware processors, cause the system (Col 39, Row 60 - Col 40, Row 27, memory 612 includes computer storage media storing computer readable instructions; Col 40, Rows 30-32, processors 614 read data from memory 612) to perform operations comprising:
identifying a first conversation conducted over a first communication channel type of a plurality of communication channel types (Col 20, Rows 25-28, Communication Session “CS” monitor and analyzer 290 (Fig. 2, CS Monitor & Analyzer 290 includes CS-Content Analyzer 292 and CS-Context Analyzer 294) monitors and analyzes content and context of one or more Communication Sessions “CS” that the user is currently participating in or has previously participated in; i.e., monitoring and analyzing a first communication session that the user is participating in or has participated in; per Col 8, Rows 1-25, communication sessions include email exchanges, textual chat session, video chat session, telephone / video call; per Col 10, Rows 55-63 and Col 22, Rows 1-5, CS-context analyzer 294 determines contextual features of the communication session, the contextual feature includes an indication or link for materials referenced and/or exchanged in the conversation such as emails, audible, visual, and/or textual content);
applying a semantic recognition model to the first conversation to encode the first conversation based on semantic content of the first conversation (Col 20, Rows 59-67, CS-content analyzer 292 analyzes content line of each CS (i.e., a first communication session) that a user is or has participated in using content-substance model; Col 21, Rows 10-18 and Rows 48-53, content substance model is a combination of topic models and keyword models where content substance model is employed to determine the substance of the content comprising content substance features indicating the semantics or meanings of the natural language utterances or expressions embedded in the conversations, topics being conversed about, and intention / sentiments of the conversation);
identifying a second conversation conducted over a second communication channel type of the plurality of communication channel types (Col 20, Rows 25-28, Communication Session “CS” monitor and analyzer 290 monitors and analyzes content and context of one or more Communication Sessions “CS” that the user is currently participating in or has previously participated in; i.e., monitoring and analyzing a second communication session that the user is currently participating in or has previously participated in);
applying the semantic recognition model to the second conversation to encode the second conversation based on the semantic content of the second conversation (Col 20, Rows 59-67, CS-content analyzer 292 analyzes content line of each CS (i.e., a second communication session) that a user is or has participated in using content-substance model; Col 21, Rows 10-18 and Rows 48-53, content substance model is a combination of topic models and keyword models where content substance model is employed to determine the substance of the content comprising content substance features indicating the semantics or meanings of the natural language utterances or expressions embedded in the conversations, topics being conversed about, and intention / sentiments of the conversation);
based on the encoding of the first conversation and the second conversation by the semantic recognition model:
identifying a first segment of the first conversation conducted over the first communication channel type and a second segment of the second conversation conducted over the second communication channel type as corresponding to a first transaction (Col 22, Rows 1-5 and Rows 44-47, CS-context analyzer 294 determines contextual features of the communication session indicating or encoding meeting titles, meeting subjects, meeting agendas, or other information indicating the structure and topics of a conversation of the communication session; Col 22, Rows 15-21 and Rows 52-64, content relevance determine 280 receives content and determined communication session features (e.g., CS content features, contextual features, and content-substance) from the CS monitor and analyzer 290 to determine the relevance of the content of one or more communication sessions for the user and notify the user of identified highly relevant content);
extracting the first segment and the second segment (Col 24, Rows 25-34, content relevance analyzer 280 implements content relevance logic 234 to determine the relevance of each portion of the CS content from the one or more communication sessions, identifying the portions of the content that are likely relevant to the user, and provide notifications of content that are identified as highly relevant); and
populating a database with a summary of the first segment and the second segment to the first transaction (Col 25, Row 61 – Col 26, Row 11, CS summary engine 260 generates summarized versions of the likely relevant content portions; in view of Col 6, Rows 45-47, summaries may be stored and indexed for use in record keeping and querying).
Somech does not disclose populating the database with a first mapping of the first segment and the second segment to the first transaction.
Wassertblat discloses apparatus for multi-channel categorization (Abstract; Fig. 1) comprising:
identifying a first conversation conducted over a first communication channel type of a plurality of communication channel types (Figs. 1-2, Interaction Analytics 136 with ¶¶20-21, capturing interactions from various sources and channels; ¶¶28-29 and ¶¶45-46, normalize interactions include telephone / voice over IP 112 (i.e., first conversation over first communication channel), video conference, emails, chat sessions etc. into unified format so that interactions captured from different channel types are later handled in the same manner; ¶47, store captured interactions in interaction database 324 with meta data such as date and time of an interaction, identifying data of the interaction);
applying a semantic recognition processing to the first conversation to encode the first conversation based on semantic content of the first conversation (¶¶48-49, perform natural language processing, key phrases detection and scoring, and sentiment analysis and index the result into indexed database 340; ¶¶53-54, input content from interaction database 324 into categorization process to categorize interactions into multi-channel main categories);
identifying a second conversation conducted over a second communication channel type of the plurality of communication channel types (Figs. 1-2, Interaction Analytics 136 with ¶¶20-21, capturing interactions from various sources and channels; ¶¶28-29 and ¶¶45-46, normalize interactions include telephone / voice over IP 112, video conference (i.e., second conversation over second communication channel), emails, chat sessions etc. into unified format so that interactions captured from different channel types are later handled in the same manner; ¶47, store captured interactions in interaction database 324 with meta data such as date and time of an interaction, identifying data of the interaction);
applying the semantic recognition processing to the second conversation to encode the second conversation based on the semantic content of the second conversation (¶¶48-49, perform natural language processing, key phrases detection and scoring, and sentiment analysis and index the result into indexed database 340; ¶¶53-54, input content from interaction database 324 into categorization process to categorize interactions into multi-channel main categories);
based on the encoding of the first conversation and the second conversation by the semantic recognition model:
identifying a first segment of the first conversation conducted over the first communication channel type and a second segment of the second conversation conducted over the second communication channel type as corresponding to a first transaction (¶54, categorization process categorizes interactions into multi-channel main categories which include related interactions captured from different channel types; ¶63, assign calls into categories by clustering; ¶67, in clustering, interactions are divided into clusters);
extracting the first segment and the second segment (¶68, on multi-channel categorization 440, related clusters or sub-categories are unified and correlated where clusters related to the same subject but to two or more different interaction types are unified; i.e., interactions from two or more different channels / interaction types are extracted and unified); and
populating a database with a first mapping of the first segment and the second segment to the first transaction (¶70, using semantic inference engines to combine categories having the same semantic context or information in which categories from different channels are explicitly combined; ¶71, assigned interactions and their clusters are stored in a categorized database 448).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to populate the database with a first mapping of the first segment and the second segment to the first transaction so that if several customers from different channels contacted the call center regarding the same topic / transaction, the combination provides a system to group the different interactions from all channels into one category and generate reports or raise an alert addressing the business issue / transaction (Wassertblat, ¶23).
Regarding Claims 3, 12, and 20, Somech discloses wherein identifying the first segment and the second segment as corresponding to the first transaction comprises applying a trained classification-type machine learning model to the first conversation encoded by the semantic recognition model and the second conversation encoded by the semantic recognition model (Col 11, Rows 39-45, employ deep learning methodologies to train one or more topical models, keyword models, and semantic models to determine the content substance features of content lines; i.e., Col 21, Rows 10-18 and Rows 48-53, content substance model being a combination of the topic models and keywords models means it is a trained classification type machine learning model to classify semantics or meanings of content sessions), and wherein the trained classification-type machine learning model classifies the first segment and the second segment as corresponding to the first transaction (Col 21, Rows 50-53, content substance features from content substance model indicate topics being conversed about; Col 22, Rows 1-5 and Rows 44-47, CS-context analyzer 294 determines contextual features of the communication session indicating or encoding meeting titles, meeting subjects, meeting agendas, or other information indicating the structure and topics of a conversation of the communication session; Col 24, Rows 43-55, content relevance analyzer 282 uses contextual features and content-substance features to determine a relevance of portions of received content; i.e., using the machine learning content substance model’s classification results (i.e., topics) to determine relevance of portions of communication sessions to a particular topic user is interested in).
Regarding Claims 4 and 13, Somech discloses training a classification-type machine learning model to classify encoded segments of conversations (Col 11, Rows 48-50, train content-substance models based on historical communication session content data) at least by:
obtaining a first training data set of historical conversation data for a plurality of conversations (Col 19, Rows 52-69, train content substance natural language model using content from a communication session log and communication session metadata),
the first training data set of historical conversation data comprising:
a third conversation (Col 19, Row 60, content from a communication session “CS” log; in view of Col 8, Rows 1-14, “CS content” or “content” includes exchange of emails between two or more users, real-time chats / IM session, exchange of textual, audible, and visual content);
a communication channel type, of a plurality of communication channel types, over which the third conversation was conducted (Col 19, Rows 52-69, train the model using communication session metadata; per Col 10, Rows 55-63 and Col 22, Rows 1-5, determining contextual feature based on metadata associated with the communication session indicating emails, audible, visual, and / or textual content being exchanged); and
a label classifying a first portion of the third conversation (Col 19, Rows 4-6, customizing machine learning data model to a particular user based on training the machine learning model with data that is specific to the user (i.e., data specifically labeled to a particular user));
applying the first training data set to the classification-type machine learning model to generate the trained classification-type machine learning model (Col 19, Rows 52-69, train content substance natural language model using content from a communication session log and communication session metadata).
Regarding Claims 5 and 14, Somech discloses identifying a third segment of the first conversation conducted over the first communication channel type as corresponding to a second transaction (Col 22, Rows 1-5 and Rows 44-47, CS-context analyzer 294 determines contextual features of the communication session indicating or encoding meeting titles, meeting subjects, meeting agendas, or other information indicating the structure and topics of a conversation of the communication session; i.e., identifying portions of communication session corresponding to a second meeting title, meeting subject, meeting agenda, or second topic of a conversation; Col 22, Rows 15-21 and Rows 52-64, content relevance determine 280 receives content and determined communication session features (e.g., CS content features, contextual features, and content-substance) from the CS monitor and analyzer 290 to determine the relevance of the content of one or more communication sessions for the user and notify the user of identified highly relevant content; i.e., determine relevance of content corresponding to the second meeting title, meeting subject, meeting agenda, or second topic of a conversation);
extracting the third segment (Col 24, Rows 25-34, content relevance analyzer 280 implements content relevance logic 234 to determine the relevance of each portion of the CS content from the one or more communication sessions, identifying the portions of the content that are likely relevant to the user, and provide notifications of content that are identified as highly relevant; i.e., identifying portions of the content that are highly relevant to the second meeting title, meeting subject, meeting agenda, or second topic of a conversation); and
As modified by Wassertblat, populating the database with a second mapping of the third segment to the second transaction (Somech, Col 25, Row 61 – Col 26, Row 11, CS summary engine 260 generates summarized versions of the likely relevant content portions; in view of Col 6, Rows 45-47, summaries may be stored and indexed for use in record keeping and querying; Wassertblat, ¶68, on multi-channel categorization 440, related clusters or sub-categories (i.e., clustered content corresponding to the second meeting title, meeting subject, meeting agenda, or second topic of a conversation) are unified and correlated where clusters related to the same subject but to two or more different interaction types are unified; i.e., interactions from two or more different channels / interaction types corresponding to the second meeting title, meeting subject, meeting agenda, or second topic of a conversation are extracted and unified).
Regarding Claims 6 and 15, Somech as modified by Wassertblat discloses wherein the operations further comprise: identifying a third segment of the first conversation conducted over the first communication channel type (Wassertblat, ¶56, filtering voice interactions (i.e., voice channel interaction) into one or more initial categories; ¶57, filtering textual interaction (i.e., text channel interaction) categorized into one or more initial categories; ¶58, interactions categorized into one or more initial categories are filtered and one or more of them are passed to further processing); and
refraining from populating the database with the third segment (Wassertblat, ¶7 and ¶77, filtering component filters interactions in order to select the interactions that should be further analyzed and categorized where interactions not suitable for such analysis are filtered out; i.e., interactions filtered out are not stored in categorized database 448 per Wassertblat, ¶71).
Regarding Claims 7 and 16, Somech as modified by Wassertblat discloses wherein refraining from populating the database with the third segment is based at least on a classification-type machine learning model identifying the third segment as not corresponding to any particular transaction (Somech, Col 21, Rows 10-18 and Rows 48-53, content substance model / deep learning ML model is employed to determine substance of the content comprising content substance features indicating the intention / sentiments of the conversation; Wassertblat, ¶62, sentimental analysis filtering 420 in which words indicating emotions are searched for and such interactions passes the filtering; i.e., too short interactions (Wassertblat, ¶77) not containing sentimental / emotional / intentional words are filtered out).
Regarding Claims 8 and 17, Somech as modified by Wassertblat discloses wherein the operations further comprise:
identifying a third segment of a third conversation conducted over a third communication channel type (Somech, Col 22, Rows 1-5, CS-context analyzer 294 analyzes the context of the communication session to determine contextual features based on metadata associated with the communication session; per Col 10, Rows 55-63, the contextual feature indicates the communication session corresponds to emails, audible, visual, and/or textual content (e.g., a third communication session comprising audio content in an audio communication channel));
identifying first semantic content in the first segment (Somech, Col 20, Rows 59-67, CS-content analyzer 292 analyzes content line of each CS (i.e., the first communication session comprising textual email communication over email channel) that a user is or has participated in using content-substance model to determine the substance of the content comprising content substance features indicating the semantics or meanings of the natural language utterances or expressions embedded in the conversations, topics being conversed about, and intention / sentiments of the conversation per Col 21, Rows 10-18 and Rows 48-53);
identifying the first semantic content in the third segment (Somech, Col 20, Rows 59-67, CS-content analyzer 292 analyzes content line of each CS (i.e., audio content in the audio communication channel) that a user is or has participated in using content-substance model to determine the substance of the content comprising content substance features indicating the semantics or meanings of the natural language utterances or expressions embedded in the conversations, topics being conversed about, and intention / sentiments of the conversation per Col 21, Rows 10-18 and Rows 48-53; Col 22, Rows 15-21 and Rows 52-64, content relevance determine 280 receives content and determined communication session features (e.g., CS content features, contextual features, and content-substance) from the CS monitor and analyzer 290 to determine the relevance of the content of one or more communication sessions for the user and notify the user of identified highly relevant content; i.e., the email content over email channel in the first session and the audio content over the audio channel in the third session having content-substance highly relevant to a user topic); and as modified by Wassertblat
refraining from extracting the third segment to populate the database with the third segment (Wassertblat, ¶60; further filter vocal interactions using audio filtering; ¶77, filtering interactions to select interactions that should be further analyzed and categorized),
wherein extracting the first segment and populating the database with the first segment are based at least on the first semantic content and the first communication channel type (¶70, using semantic inference engines to combine categories having the same semantic context or information in which categories from different channels are explicitly combined; ¶71, assigned interactions and their clusters are stored in a categorized database 448 (e.g., email content over email channel in the first session having a particular semantic topic / sentiment is stored in database 448)), and
wherein refraining from extracting the third segment to populate the database with the third segment are based at least on the first semantic content and the third communication channel type (Wassertblat, ¶60, vocal interactions are further filtered using emotion analysis; e.g., in a situation where vocal interactions is filtered out because its emotion analysis indicated certain sentiment).
Regarding Claims 9 and 18, Somech discloses wherein extracting the first segment comprises:
extracting a first portion of a first length of the first conversation based on the first communication channel type (Col 10, Rows 35-36 and Row 55-63, contextual features indicate (1) initiating and terminating time stamps of the communication session and (2) the communication session content being email, audio, video, and text), and
wherein extracting the second segment comprises:
extracting a second portion of a second length of the second conversation based on the second communication channel type (Col 10, Rows 35-36 and Row 55-63, contextual features indicate (1) initiating and terminating time stamps of the communication session and (2) the communication session content being email, audio, video, and text).
While Somech does not suggest wherein the second length is different from the first length, Wassertblat suggests a first conversation having a first portion first length and a second conversation having a second portion of a second length different from the first length (Wassertblat, ¶77, some interactions are short and some interactions are long).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to have a first conversation having a first portion first length and a second conversation having a second portion of a second length different from the first length because interactions / conversation lengths can be long or short.
Claims 2 and 11 are rejected under 35 USC 103(a) as being unpatentable over Somech et al. (US 11699039 B2) in view of Wassertblat et al. (US 2012/0215535 A1) as applied to claims 1 and 10, in further view of Hosseinisianaki et al. (US 2020/0344194 A1).
Regarding Claims 2 and 11, Somech discloses wherein the semantic recognition model is an encoding-type deep learning machine learning model that transforms text content into symbolic representations (Col 9, Rows 35-38, natural language models include deep learning audio encoders and decoders to analyze and generate natural language expressions including symbol expressions).
Somech does not disclose wherein the semantic recognition model is an encoding-type deep learning machine learning model that transforms text content into numerical representations.
Hosseinisianaki discloses encoding type deep learning machine learning model as a semantic recognition model to transform text content into numerical representations (¶¶62-63, deep learning model comprising content embedder 402 and content encoder 406 as part of content encoder 302; ¶65, content embedder 402 can project content to a high dimensional space by transforming each word in content 310 to a space in which semantically related words are closer to each other where words are expressed as vectors).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to implement the semantic recognition model is an encoding-type deep learning machine learning model that transforms text content into numerical / vector representations in order to represent semantically related words / topics in semantically related space (Hosseinisianaki, ¶65).
Conclusion
Prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
US 2023/0177269 A1 discloses conversation topic extraction technique that divides received text of a communication channel into conversation documents based on conversation threads of the communication channel where phrases of the text of the conversation documents are tokenized and importance scores are assigned to the tokenized phrases using unsupervised topic extraction to determine topic phrases for the conversation documents.
US 2019/0245974 A1 discloses receiving a first communication associated with a contact on a communication channel of a plurality of communication channels, receiving at least one additional communication from the contact on a different communication channel of the plurality of communication channels, and linking the first communication and the additional communication by the contact by storing a transcript of the first communication and the additional communication, a category, and a score to enable a user to navigate between the stored first communication and the additional communication.
US 9715494 B1 discloses computing a past period for a channel on a messaging platform, extract a set of channel characteristics from a past period data posted in the channel during the past period, and construct a recommendation to include a channel identifier of the channel to suggest posting the message in the channel and a change to the message according to corresponding characteristics in the set of channel characteristics.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to examiner Richard Z. Zhu whose telephone number is 571-270-1587 or examiner’s supervisor Hai Phan whose telephone number is 571-272-6338. Examiner Richard Zhu can normally be reached on M-Th, 0730:1700.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/RICHARD Z ZHU/Primary Examiner, Art Unit 2654 03/27/2026