WO2024073872A1 - Suggestion d'action pour une session de communication - Google Patents

Suggestion d'action pour une session de communication Download PDF

Info

Publication number
WO2024073872A1
WO2024073872A1 PCT/CN2022/123709 CN2022123709W WO2024073872A1 WO 2024073872 A1 WO2024073872 A1 WO 2024073872A1 CN 2022123709 W CN2022123709 W CN 2022123709W WO 2024073872 A1 WO2024073872 A1 WO 2024073872A1
Authority
WO
WIPO (PCT)
Prior art keywords
action
poll
intent
session
analyzing
Prior art date
Application number
PCT/CN2022/123709
Other languages
English (en)
Inventor
Tianqi Liu
Archana SASEETHARAN
Binbin He
Chenxiao LI
Dongmei Zhang
Yunpeng TANG
Genglin Huang
Huitian JIAO
Qiang Dong
Jing Liu
Ke Wang
Kun Liu
Manpratap SURI
Mengyu ZHOU
Shi Han
Shiraz Cupala
Tao Wu
Tiantian WANG
Le XIA
Walter Hoy Toh WONG
Wenfei Tang
Yan ZHAI
Yao KE
Original Assignee
Microsoft Technology Licensing, Llc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Microsoft Technology Licensing, Llc filed Critical Microsoft Technology Licensing, Llc
Priority to PCT/CN2022/123709 priority Critical patent/WO2024073872A1/fr
Publication of WO2024073872A1 publication Critical patent/WO2024073872A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0282—Rating or review of business operators or products
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06395—Quality analysis or management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0278—Product appraisal
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0281—Customer communication at a business location, e.g. providing product or service information, consulting

Definitions

  • Multi-party communication session is a typical implementation of communication session, which supports multiple users or participants. It is beneficial to automatically provide various types of suggestion or assistance to users in a communication session, especially in a multi-party communication session.
  • Embodiments of the present disclosure propose methods and apparatuses for providing action suggestion for a communication session.
  • Session insight information may be generated based on session data of the communication session.
  • Poll insight information may be generated based on poll data of at least one previous poll associated with the communication session.
  • An action suggestion may be generated based at least on the session insight information and the poll insight information.
  • FIG. 1A to FIG. 1C illustrate exemplary application scenarios according to embodiments of the present disclosure.
  • FIG. 2 illustrates an exemplary process of providing action suggestion for a communication session according to an embodiment.
  • FIG. 3 illustrates an exemplary process of session insight information generation according to an embodiment.
  • FIG. 4 illustrates an exemplary process of poll insight information generation according to an embodiment.
  • FIG. 5 illustrates an exemplary process of action suggestion generation according to an embodiment.
  • FIG. 6 to FIG. 9 illustrate exemplary user interfaces according to embodiments of the present disclosure.
  • FIG. 10 illustrates a flowchart of an exemplary method for providing action suggestion for a communication session according to an embodiment.
  • FIG. 11 illustrates an exemplary apparatus for providing action suggestion for a communication session according to an embodiment.
  • FIG. 12 illustrates an exemplary apparatus for providing action suggestion for a communication session according to an embodiment.
  • An exemplary suggestion or assistance provided to users in a communication session is a suggested poll which is usually supported by a form service.
  • a suggested poll which is usually supported by a form service.
  • a user may easily and quickly obtain information from other users in a communication session.
  • existing approaches for providing poll suggestion in a communication session would trigger a poll suggestion through detecting and analyzing content presented in the communication session or utterances from users in the communication session.
  • Embodiments of the present disclosure propose a mechanism for providing action suggestion for a communication session.
  • the term “communication session” may refer to various types of session occurred among two or more users or participants, e.g., online meeting, email, etc.
  • the communication session may be supported by various communication session applications, e.g., online meeting application, email service, etc.
  • the term “application” may widely refer to various application programs, application clients, software, systems, services, network platforms, etc.
  • action suggestion involved in the embodiments of the present disclosure may not only comprise poll-related suggestion, but also comprise various other types of suggested action that can be taken by a user for a communication session.
  • action suggestion may be provided based on both information related to the communication session, e.g., session data of the communication session, and information related to at least one previous poll associated with the communication session, e.g., poll data of the at least one previous poll.
  • the embodiments of the present disclosure may provide action suggestion at least in consideration of the at least one previous poll.
  • the embodiments of the present disclosure may provide action suggestion at various time points relative to a communication session, e.g., during the communication session, after the communication session, before the communication session, etc.
  • the embodiments of the present disclosure may provide action suggestion in a follow-up or multi-round approach in response to previous polls.
  • the embodiments of the present disclosure may effectively predict what action is most likely to be taken by a user for a communication session, and prompt the user with a generated action suggestion. Accordingly, the embodiments of the present disclosure may provide more valuable and helpful suggestions to a user, assist the user to take actions efficiently, improve user experience, etc.
  • the embodiments of the present disclosure may be applied in various application scenarios.
  • FIG. 1A illustrates an exemplary application scenario 100A according to an embodiment.
  • the application scenario 100A shows a scenario of providing action suggestion before a communication session.
  • a user 102 is planning to have a communication session with at least one another user 104.
  • the user 102 is using a terminal device 106, and the at least one user 104 is using at least one terminal device 108.
  • the user 102 may perform a user operation, e.g., scheduling a communication session, on the terminal device 106.
  • a user operation e.g., scheduling a communication session
  • the user 102 schedules an online meeting through, e.g., a calendar application, an email service, etc., wherein the online meeting is scheduled at 10 am on Thursday right after a holiday.
  • information collecting may be performed so as to collect information related to the user operation, information related to the communication session, etc.
  • the information related to the user operation may comprise, e.g., scheduling an online meeting, etc.
  • the information related to the communication session may comprise various types of session data associated with the communication session, e.g., meeting time information, meeting metadata, etc. associated with the scheduled online meeting.
  • action suggestion may be provided according to the collected information.
  • the action suggestion may comprise a suggestion of launching an ice-breaker poll during the online meeting for collecting participants’ feeling about the holiday.
  • the action suggestion may be presented to the user 102 on the terminal device 106. The user 102 may accept to take the action suggestion.
  • the communication session scheduled by the user 102 is established for the user 102 and the at least one user 104 through a communication session application.
  • the scheduled online meeting is established through an online meeting application on the terminal device 106 and an online meeting application on the at least one terminal device 108.
  • the user 102 may perform an action corresponding to the action suggestion. For example, at 118, the user 102 sends an ice-breaker poll suggested in the action suggestion on the terminal device 106 to the at least one user 104. Then, the at least one user 104 may respond to the action performed by the user 102. For example, at 120, the at least one user 104 may respond to the ice-breaker poll on the at least one terminal device 108. The user 102 may obtain responses from the at least one user 104. For example, at 122, the user 102 may receive, on the terminal device 106, a poll response for the poll sent at 118, wherein the poll response may be a set of responses made by the at least one user 104.
  • FIG. 1B illustrates an exemplary application scenario 100B according to an embodiment.
  • the application scenario 100B shows a scenario of providing action suggestion during a communication session.
  • the same reference numbers in FIG. 1A and FIG. 1B indicate the same elements.
  • a communication session is established through a communication session application on the terminal device 106 and a communication session application on the at least one terminal device 108.
  • a communication session application on the terminal device 106 For example, an online meeting is established through an online meeting application on the terminal device 106 and an online meeting application on the at least one terminal device 108.
  • the user 102 sends a poll on the terminal device 106 to the at least one user 104. Then, at 134, the at least one user 104 may respond to the poll on the at least one terminal device 108. At 136, the user 102 may obtain, on the terminal device 106, a poll response for the poll sent at 132.
  • information collecting may be performed so as to collect information related to the communication session, information related to the poll sent at 132, etc.
  • the information related to the communication session may comprise various types of session data associated with the communication session.
  • the information related to the poll may comprise various types of poll data associated with the poll.
  • action suggestion may be provided according to the collected information.
  • the action suggestion may be presented to the user 102 on the terminal device 106.
  • the user 102 may perform a suggested action on the terminal device 106 according to the action suggestion.
  • the communication session is terminated.
  • the user 102 may also perform a suggested action on the terminal device 106 according to the action suggestion after the communication session, as shown by the dotted block 142’ .
  • FIG. 1C illustrates an exemplary application scenario 100C according to an embodiment.
  • the application scenario 100C shows a scenario of providing action suggestion after a communication session.
  • the same reference numbers in FIG. 1A, FIG. 1B and FIG. 1C indicate the same elements.
  • a communication session is established through a communication session application on the terminal device 106 and a communication session application on the at least one terminal device 108.
  • a communication session application on the terminal device 106 For example, an online meeting is established through an online meeting application on the terminal device 106 and an online meeting application on the at least one terminal device 108.
  • the user 102 sends a poll on the terminal device 106 to the at least one user 104. Then, at 154, the at least one user 104 may respond to the poll on the at least one terminal device 108. At 156, the user 102 may obtain, on the terminal device 106, a poll response for the poll sent at 152.
  • information collecting may be performed so as to collect information related to the communication session, information related to the poll sent at 152 etc.
  • the information related to the communication session may comprise various types of session data associated with the communication session.
  • the information related to the poll may comprise various types of poll data associated with the poll.
  • action suggestion may be provided according to the collected information.
  • the action suggestion may be presented to the user 102 on the terminal device 106.
  • the user 102 may perform a suggested action on the terminal device 106 according to the action suggestion.
  • the at least one user 104 respond to the poll on the at least one terminal device 108 after the communication session, as shown by the dotted block 154’ . Then, the user 102 may obtain, on the terminal device 106, a poll response for the poll sent at 152 after the communication session, as shown by the dotted block 156’ . Accordingly, the information collecting may be performed after the communication session, as shown by the dotted block 158’ .
  • the application scenarios in FIG. 1A to FIG. 1C are exemplary, and these application scenarios may be changed or combined in any approaches.
  • the processes in FIG. 1A to FIG. 1C may be concatenated, thus achieving the providing of a series of follow-up action suggestions in a multi-round approach.
  • the poll sent at 132 in FIG. 1B or at 152 in FIG. 1C may refer to the poll sent at 118 in FIG. 1A.
  • the operation of information collecting may cover various approaches of collecting a poll response, e.g., detecting the poll response from applications on the terminal device 106, obtaining the poll response from a form service that is used by the user 102 for sending a poll, etc.
  • the user 102 is a target user who is provided with action suggestion in FIG. 1A to FIG. 1C
  • anyone of the at least one user 104 may also be provided with respective action suggestion on a corresponding terminal device in a similar way.
  • the processing of providing action suggestion at 114, 140 and 162 may be implemented as an additional function in the communication session application or any other applications interactive with the user 102 on the terminal device 106, or implemented as an independent application that is capable of assessing or interacting with applications on the terminal device 106.
  • FIG. 2 illustrates an exemplary process 200 of providing action suggestion for a communication session according to an embodiment.
  • the process 200 is an exemplary implementation of the operation of providing action suggestion in the application scenarios discussed above.
  • the process 200 may be performed for providing action suggestion for the communication session to a target user.
  • session data 210 of the communication session may be obtained.
  • the session data 210 comprises various types of information item associated with the communication session, e.g., session textual content, session metadata, session time information, user operation information, external information, etc.
  • the session textual content is a collection of text detected with respect to the communication session.
  • the session textual content may comprise at least one of: title or subject of the communication session; textual content presented in or identified from the communication session; text or transcript converted from audio signals in the communication session, e.g., from voice conversation or speech by users; and so on.
  • session textual content of the online meeting may comprise, e.g., meeting title, textual content in a document shared on the screen during the online meeting, transcript converted from voice conversation among participants, etc.
  • session textual content of the email may comprise, e.g., subject of the email, text in the body of the email, etc.
  • the embodiments of the present disclosure are not limited to any specific type of session textual content, and are not limited to any specific approach for obtaining the session textual content.
  • the session metadata comprises various types of information related to attributes of the communication session.
  • the session metadata may comprise at least one of: session size, e.g., the number of participants, etc.; session type, e.g., regular session, occasional session, etc.; participant identification (ID) , e.g., IDs of all the participants of the communication session; role of the target user, who is to be provided with action suggestion, in the communication session, e.g., organizer, presenter, attendee, etc.; session ID for uniquely identifying the communication session; and so on.
  • session metadata of the online meeting may comprise, e.g., meeting size, meeting type, role of the target user in the online meeting, meeting ID, etc.
  • the embodiments of the present disclosure are not limited to any specific type of session metadata, and are not limited to any specific approach for obtaining the session metadata.
  • the session time information comprises information of occurrence time, duration time, etc. associated with the communication session, e.g., starting time point, end time point, etc.
  • the embodiments of the present disclosure are not limited to any specific type of session time information, and are not limited to any specific approach for obtaining the session time information.
  • the user operation information comprises various types of information related to an operation taken by the target user for the communication session.
  • the user operation information may be, e.g., information about an operation of scheduling the communication session, information about an operation of sharing or sending a document in the communication session, etc.
  • the embodiments of the present disclosure are not limited to any specific type of user operation information, and are not limited to any specific approach for obtaining the user operation information.
  • the external information comprises various types of information that are not directly related to the communication session but are still useful for providing action suggestion for the communication session.
  • the external information may comprise: personal information of the target user, e.g., job title, occupation, location, etc.; information about the company or entity in which the target user works; information about content or user operations in applications other than the communication session application; and so on.
  • the embodiments of the present disclosure are not limited to any specific type of external information, and are not limited to any specific approach for obtaining the external information.
  • session data 210 may comprise any one or more of the above exemplary information items associated with the communication session, and may also comprise any other types of information item associated with the communication session.
  • session insight information generation may be performed, in which session insight information 230 is generated based on the session data 210.
  • insight information may widely refer to reference information useful for determining how to take a next action, which is obtained through analyzing original input data.
  • the session insight information 230 contains insight information generated according to the session data 210 and predicted for an action suggestion to be provided.
  • the session insight information 230 may comprise an action 232 and an intent 234 that are predicted for an action suggestion.
  • the session insight information generation at 220 may comprise predicting the action 232 through performing a series of action analyzing to the session data 210, and predicting the intent 234 through performing a series of intent analyzing to the session data 210.
  • the action 232 indicates a type of an action suggested in an action suggestion.
  • the action 232 may comprise various types of action that are possible to be taken next, e.g., creating a poll, launching a poll, sending a document, setting up a new communication session, recording an annotation, sending a message, etc.
  • a plurality of action types may be predefined, and an action type indicated by the action 232 may be determined from the predefined action types according to information items in the session data 210.
  • the embodiments of the present disclosure are not limited to any specific action type indicated by the action 232.
  • the intent 234 is a scenario description of the action 232, which defines details for the performing of the action 232 from various aspects.
  • the intent 234 may also be referred to as, e.g., intent taxonomy, etc.
  • the intent 234 may be represented in a multi-dimension approach. Dimensions in the intent 234 may comprise, e.g., purpose, target, domain, time information, recipients, etc.
  • the dimension of purpose indicates what purpose or objective is desired to achieve through performing an action, e.g., feedback, rating, share information, confirmation, etc.
  • the purpose of the poll may be “feedback” .
  • the dimension of target indicates a type of a communication session, e.g., training, propaganda, discussion, internal session, etc.
  • the target of the online meeting may be “training” .
  • the dimension of domain indicates the industry or field of a communication session, e.g., high technology, education, society, health, etc.
  • the domain of the online meeting may be “high technology” .
  • the dimension of time information indicates a time point or time interval for performing an action.
  • the time information may be denoted in various approaches, e.g., by a specific date and/or time point, by a time interval with a specific starting time and/or end time, by a relative time with respect to a date or time point, etc.
  • the time information of the action may be “December 23”and/or “before Christmas Day” .
  • the dimension of recipients indicates users to whom an action is to perform, which may comprise user IDs, user emails, etc. of the recipients. As an example, if a poll is to send to all the participants of an online meeting, the dimension of recipients may comprise these participants’ IDs. It should be understood that the intent 234 is not limited to any of the above exemplary dimensions, but may also comprise any other dimensions useful for providing scenario description.
  • poll data 240 of at least one previous poll associated with the communication session may be obtained.
  • a poll may have been previously sent by the target user to other users in the communication session, and accordingly, poll data of this poll may be obtained.
  • the poll data 240 comprises various types of information associated with the previous poll, e.g., poll metadata, poll response, etc.
  • the poll metadata comprises various types of information related to the previous poll.
  • the poll metadata may comprise at least one of: poll content, e.g., questions, options, etc. in the previous poll; launching time information, e.g., the sending time of the previous poll; role of the target user in the previous poll, e.g., initiator, recipient, etc.; poll recipients; and so on.
  • poll content e.g., questions, options, etc. in the previous poll
  • launching time information e.g., the sending time of the previous poll
  • role of the target user in the previous poll e.g., initiator, recipient, etc.
  • poll recipients e.g., poll recipients
  • the poll response is a set of responses made by responders among the at least one recipient.
  • the poll response may comprise, e.g., selections made by responders, content input by responders, responder ID of each response, responding time information of each response, etc.
  • the embodiments of the present disclosure are not limited to any specific item contained in the poll response, and are not limited to any specific approach for obtaining the poll response.
  • poll insight information generation may be performed, in which poll insight information 260 is generated based on the poll data 240.
  • the poll insight information 260 contains insight information generated according to the poll data 240 and predicted for an action suggestion to be provided.
  • the poll insight information 240 may comprise an action 262 and an intent 264 that are predicted for an action suggestion.
  • the poll insight information generation at 250 may comprise obtaining the action 262 and the intent 264 through performing action analyzing and intent analyzing to the poll data 240 respectively.
  • At least one intermediate analysis result may be obtained through performing statistical analyzing to the poll response according to a type of the at least one previous poll, and then the action 262 may be obtained through performing action analyzing to the intermediate analysis result, and the intent 264 may be obtained through performing intent analyzing to the intermediate analysis result and poll content in the poll metadata.
  • the poll insight information 240 may further comprise an analysis result for the poll response.
  • the poll insight information generation at 250 may further comprise generating the analysis result according to the intermediate analysis result.
  • the action 262 and the intent 264 may have the same meanings and definitions as the action 232 and the intent 234 respectively.
  • action suggestion generation may be performed, in which an action suggestion 280 is generated based at least on the session insight information 230 and the poll insight information 260.
  • the action suggestion 280 comprises a prompt associated with a suggested action.
  • the action suggestion 280 may inform the target user what action can be taken next. Accordingly, the target user may perform the suggested action according to the action suggestion 280.
  • the action suggestion 280 may inform the target user to create a specific poll, and accordingly the target user may be guided to create the poll through, e.g., a form service.
  • the action suggestion 280 may not only inform the target user what action can be taken next, but also provide assistance to the target user for performing the suggested action.
  • the action suggestion 280 may inform the target user to launch a specific poll and provide a suggested example of the poll, and accordingly the target user may simply accept and launch the suggested poll provided in the action suggestion 280 through performing, e.g., a one-click operation on a button in the action suggestion 280.
  • the action suggestion 280 may be visually presented on a terminal device of the target user.
  • the embodiments of the present disclosure are not limited to any specific content and format of the action suggestion 280, and are not limited to any specific approach for presenting the action suggestion 280.
  • the process 200 may be performed iteratively. For example, after a suggested action in an action suggestion is taken by the target user, the process 200 may be performed again so as to provide further action suggestion. In this way, a series of follow-up action suggestions may be provided in a multi-round approach. Moreover, the process 200 may be triggered to perform periodically or in response to meeting any predefined conditions.
  • FIG. 3 illustrates an exemplary process 300 of session insight information generation according to an embodiment.
  • Session insight information generation 310 in the process 300 is an exemplary implementation of the operation of session insight information generation at 220 in FIG. 2.
  • Session data 302 is an input to the session insight information generation 310, which may comprise at least one of session textual content, session metadata, session time information, user operation information, external information, etc.
  • the session insight information generation 310 may comprise an action analyzing chain 320 for predicting an action 304 through performing a series of action analyzing to the session data 302, and an intent analyzing chain 330 for predicting an intent 306 through performing a series of intent analyzing to the session data 302.
  • the action analyzing chain 320 may comprise multiple sequential action analyzing stages for processing corresponding items in the session data 302 respectively, e.g., a text action analyzing stage 321, a metadata action analyzing stage 323, a time action analyzing stage 325, a user operation action analyzing stage 327, an external information action analyzing stage 329, etc.
  • the text action analyzing stage 321 performs action analyzing to the session textual content in the session data 302 so as to predict an action.
  • Textual content in the communication session may be useful for predicting an action. For example, if an utterance from the target user indicates that the target user desires to know the feelings of other users about the communication session, there is a high probability that the target user would accept a suggested action of launching a rating poll.
  • a neural network model e.g., a BERT model, may be adopted at the text action analyzing stage 321 for predicting an action.
  • the embodiments of the present disclosure are not limited to any specific approach of implementing the text action analyzing stage 321.
  • the metadata action analyzing stage 323 performs action analyzing to the session metadata in the session data 302 so as to predict an action.
  • Session metadata of the communication session may be useful for predicting an action. For example, if a role of the target user is an attendee instead of an organizer or presenter, there is a low probability that the target user would take an action of creating an ice-breaker poll. For example, if a role of the target user is an organizer and a session size indicates that a large number of participants are involved in the communication session, there is a high probability that the target user would accept a suggested action of creating a feedback poll for conveniently collecting feedbacks from all the participants rather than taking other approaches for collecting feedbacks.
  • various heuristic rules may be predefined with respect to the session metadata, and these heuristic rules may be adopted at the metadata action analyzing stage 323 for predicting an action.
  • the embodiments of the present disclosure are not limited to any specific heuristic rules adopted at the metadata action analyzing stage 323.
  • the time action analyzing stage 325 performs action analyzing to the session time information in the session data 302 so as to predict an action. Session time information of the communication session may be useful for predicting an action. For example, if a starting time point of the communication session is right after a holiday, there is a high probability that the target user would accept a suggested action of launching a poll for collecting the participants’ feeling about the holiday.
  • various heuristic rules may be predefined with respect to the session time information, and these heuristic rules may be adopted at the time action analyzing stage 325 for predicting an action. The embodiments of the present disclosure are not limited to any specific heuristic rules adopted at the time action analyzing stage 325.
  • the user operation action analyzing stage 327 performs action analyzing to the user operation information in the session data 302 so as to predict an action.
  • User operation information may be useful for predicting an action. For example, if the target user has performed an operation of sending a document to other users, there is a high probability that the target user would accept a suggested action of creating a poll for confirming whether the document is received.
  • various heuristic rules may be predefined with respect to the user operation information, and these heuristic rules may be adopted at the user operation action analyzing stage 327 for predicting an action.
  • the embodiments of the present disclosure are not limited to any specific heuristic rules adopted at the user operation action analyzing stage 327.
  • the external information action analyzing stage 329 performs action analyzing to the external information in the session data 302 so as to predict an action.
  • External information may be useful for predicting an action. For example, if a job title of the target user is a manager, there is a high probability that the target user would accept a suggested action of sending a message for assigning tasks.
  • various heuristic rules may be predefined with respect to the external information, and these heuristic rules may be adopted at the external information action analyzing stage 329 for predicting an action.
  • the embodiments of the present disclosure are not limited to any specific heuristic rules adopted at the external information action analyzing stage 329.
  • the processing by each action analyzing stage in the action analyzing chain 320 may comply with one or more predetermined strategies.
  • each action analyzing stage may implement basic action prediction.
  • the action analyzing stage may perform action analyzing to a corresponding item in the session data, so as to obtain a current predicted action.
  • the current action analyzing stage may obtain a current predicted action through performing action analyzing to a corresponding item in the session data, and replace the preceding predicted action by the current predicted action.
  • each action analyzing stage may be predefined with a corresponding confidence value in various approaches. If a confidence value of an action analyzing stage is lower than the predetermined confidence threshold, it may be determined that a predicted action output by this action analyzing stage does not meet the predetermined confidence threshold.
  • the current action analyzing stage may maintain the preceding predicted action, without performing action analyzing at the current action analyzing stage.
  • the action predicted by the action analyzing chain may be updated or refined along with the multiple sequential action analyzing stages. It should be understood that the above strategies are merely exemplary, and the embodiments of the present disclosure may also cover any other types of strategy predetermined for the action analyzing stages in the action analyzing chain 320.
  • the intent analyzing chain 330 may comprise multiple sequential intent analyzing stages for processing corresponding items in the session data 302 respectively, e.g., a text intent analyzing stage 331, a metadata intent analyzing stage 333, a time intent analyzing stage 335, a user operation intent analyzing stage 337, an external information intent analyzing stage 339, etc.
  • the text intent analyzing stage 331 performs intent analyzing to the session textual content in the session data 302 so as to predict one or more dimensions in an intent.
  • Textual content in the communication session may be useful for intent prediction. For example, if a meeting title of an online meeting is “Training course for new staff” , the dimension of target in the intent may be predicted as “training” . For example, if the text converted from voice conversation contains a large amount of words such as “AI” , “machine learning” , etc., the dimension of domain in the intent may be predicted as “high technology” .
  • a neural network model e.g., a BERT model, may be adopted at the text intent analyzing stage 331 for intent prediction.
  • the embodiments of the present disclosure are not limited to any specific approach of implementing the text intent analyzing stage 331.
  • the metadata intent analyzing stage 333 performs action analyzing to the session metadata in the session data 302 so as to predict one or more dimensions in an intent.
  • Session metadata of the communication session may be useful for intent prediction. For example, if a session size indicates that there are only two participants in the communication session, the dimension of target in the intent may be predicted as “one-on-one session” instead of, e.g., “training” .
  • the session metadata comprises participants’ IDs
  • the dimension of recipients in the intent may be predicted as comprising IDs of all or a part of the participants.
  • various heuristic rules may be predefined with respect to the session metadata, and these heuristic rules may be adopted at the metadata intent analyzing stage 333 for intent prediction.
  • the embodiments of the present disclosure are not limited to any specific heuristic rules adopted at the metadata intent analyzing stage 333.
  • the time intent analyzing stage 335 performs action analyzing to the session time information in the session data 302 so as to predict one or more dimensions in an intent.
  • Session time information of the communication session may be useful for intent prediction. For example, if the session time information includes a starting time point and an end time point of the communication session, the dimension of time information in the intent may be predicted as “during a time interval” , wherein the time interval is between the starting time point and the end time point.
  • various heuristic rules may be predefined with respect to the session time information, and these heuristic rules may be adopted at the time intent analyzing stage 335 for intent prediction.
  • the embodiments of the present disclosure are not limited to any specific heuristic rules adopted at the time intent analyzing stage 335.
  • the user operation intent analyzing stage 337 performs action analyzing to the user operation information in the session data 302 so as to predict one or more dimensions in an intent.
  • User operation information may be useful for intent prediction. For example, if the target user has performed an operation of sending a document to other users, the target user is very likely to accept a suggested action of launching a poll for confirming whether the document is received, and thus the dimension of purpose in the intent may be predicted as “confirmation” .
  • various heuristic rules may be predefined with respect to the user operation information, and these heuristic rules may be adopted at the user operation intent analyzing stage 337 for intent prediction.
  • the embodiments of the present disclosure are not limited to any specific heuristic rules adopted at the user operation intent analyzing stage 337.
  • the external information intent analyzing stage 339 performs intent analyzing to the external information in the session data 302 so as to predict one or more dimensions in an intent.
  • External information may be useful for intent prediction. For example, if the name of a company in which the target user works includes words such as “education group” , the dimension of domain in the content may be predicted as “education” .
  • various heuristic rules may be predefined with respect to the external information, and these heuristic rules may be adopted at the external information intent analyzing stage 339 for intent prediction.
  • the embodiments of the present disclosure are not limited to any specific heuristic rules adopted at the external information intent analyzing stage 339.
  • the processing by each intent analyzing stage in the intent analyzing chain 330 may comply with one or more predetermined strategies.
  • each intent analyzing stage may implement basic intent prediction.
  • the intent analyzing stage may perform intent analyzing to a corresponding item in the session data, so as to obtain a current predicted intent.
  • each intent analyzing stage may obtain a current predicted intent through performing intent analyzing to a corresponding item in the session data, and update a preceding predicted intent output by a preceding intent analyzing stage with the current predicted intent.
  • one or more dimensions in the current predicted intent may be added into the preceding predicted intent.
  • each intent analyzing stage may be predefined with a corresponding confidence value in various approaches. If a confidence value of the current intent analyzing stage is higher than a confidence value of the preceding intent analyzing stage, one or more dimensions in the current predicted intent may be used for replacing corresponding dimensions in the preceding predicted intent.
  • the current intent analyzing stage may obtain a current predicted intent through performing intent analyzing to a corresponding item in the session data, and replace the preceding predicted intent by the current predicted intent.
  • the current intent analyzing stage may maintain the preceding predicted intent, without performing intent analyzing at the current intent analyzing stage.
  • the intent predicted by the intent analyzing chain may be updated or refined along with the multiple sequential intent analyzing stages. It should be understood that the above strategies are merely exemplary, and the embodiments of the present disclosure may also cover any other types of strategy predetermined for the intent analyzing stages in the intent analyzing chain 330.
  • the orders of the multiple action analyzing stages and the multiple intent analyzing stages may be changed in any approach.
  • the multiple action analyzing stages and the multiple intent analyzing stages may be ordered according to their confidence values.
  • any action analyzing stage or intent analyzing stage may be omitted from the process 300, or any other action analyzing stage or intent analyzing stage may be added into the process 300.
  • FIG. 4 illustrates an exemplary process 400 of poll insight information generation according to an embodiment.
  • Poll insight information generation 410 in the process 400 is an exemplary implementation of the operation of poll insight information generation at 250 in FIG. 2.
  • Poll data 402 of at least one previous poll is an input to the poll insight information generation 410, which may comprise at least one of poll metadata, poll response, etc.
  • At least one intermediate analysis result may be obtained through performing statistical analyzing to the poll response according to a type of the at least one previous poll.
  • poll types e.g., rating poll, ranking poll, word cloud poll, etc.
  • a question in a rating poll may ask a responder to give a rating level for a target event, wherein the rating level may be defined in various approaches, e.g., rating from A to C, rating from score 10 to score 1, etc.
  • a question in a ranking poll may ask a responder to rank a plurality of options in an order of, e.g., from high to low, from low to high, etc.
  • a question in a word cloud poll may ask a responder to input a textual answer to the question.
  • the process 400 may perform different statistical analyzing to poll responses for different types of polls.
  • an intermediate analysis result may refer to a result derived by performing statistical analyzing to a poll response, which is merely a direct statistical result of the poll response.
  • task generation may be performed at 420, in which the type of the previous poll may be identified and a computation processing corresponding to the type of the previous poll may be triggered to perform statistical analyzing to the poll response.
  • rating distribution computation may be performed at 432 so as to obtain an intermediate analysis result according to the poll response.
  • the intermediate analysis result output by the rating distribution computation at 432 may comprise at least one of: all the possible rating levels; for each rating level, user IDs of responders who give the rating level; for each rating level, the number of responders who give the rating level; and so on.
  • ranking distribution computation may be performed at 434 so as to obtain an intermediate analysis result according to the poll response.
  • the intermediate analysis result output by the ranking distribution computation at 434 may comprise at least one of: all the options in the poll; for each ordered position of each option, user IDs of responders who assign the ordered position to the option; for each ordered position of each option, the number of responders who assign the ordered position to the option; and so on.
  • word cloud computation may be performed at 436 so as to obtain an intermediate analysis result according to the poll response.
  • the intermediate analysis result output by the word cloud computation at 436 may comprise at least one of: a textual answer input by each responder; for each textual answer, user IDs of responders who input the textual answer; for each textual answer, the number of responders who input the textual answer; and so on.
  • the process 400 may comprise one or more of the rating distribution computation, the ranking distribution computation and the word cloud computation, and may further comprise any other computation processings for obtaining an intermediate analysis result.
  • post processing may be performed to the intermediate analysis result output by the rating distribution computation 432, the ranking distribution computation 434 or the word cloud computation 436.
  • the post processing at 440 may comprise determining whether the intermediate analysis result meets a predefined statistical significance condition.
  • the statistical significance condition may be represented by a statistical significance threshold. If the statistical significance of the intermediate analysis result is not lower than the statistical significance threshold, the intermediate analysis result may be construed as meeting the statistical significance condition.
  • the embodiments of the present disclosure are not limited to any specific approach for evaluating whether the intermediate analysis result meets the statistical significance condition.
  • the post processing at 440 may further comprise triggering to perform at least one of action analyzing, intent analyzing and result generation.
  • action analyzing may be performed to the intermediate analysis result so as to predict an action 404.
  • Information in the intermediate analysis result may be useful for predicting an action. For example, if the previous poll is a rating poll sent by the target user for collecting other users’ feeling about the communication session, and the intermediate analysis result shows that most of responders give a lowest rating level for the communication session, there is a high probability that the target user would accept a suggested action of launching a poll for further collecting feedbacks about improvement opinions from these responders.
  • various heuristic rules may be predefined with respect to the intermediate analysis result, and these heuristic rules may be adopted by the action analyzing at 450 for predicting an action.
  • the embodiments of the present disclosure are not limited to any specific heuristic rules adopted by the action analyzing at 450.
  • intent analyzing may be performed to the intermediate analysis result and poll content in the poll metadata so as to predict an intent 406.
  • Information in the intermediate analysis result and the poll content may be useful for intent prediction. For example, if the poll content indicates that the previous poll is a rating poll, and the intermediate analysis result shows that most of responders give a lowest rating level for the communication session, the target user is very likely to accept a suggested action of launching a poll for further collecting feedbacks about improvement opinions from these responders, and thus the dimension of purpose in the intent may be predicted as “feedback” , and the dimension of recipients in the intent may be predicted as comprising IDs of those responders who give the lowest rating level.
  • various heuristic rules may be predefined with respect to the intermediate analysis result, and these heuristic rules may be adopted by the intent analyzing at 460 for intent prediction.
  • the embodiments of the present disclosure are not limited to any specific heuristic rules adopted by the intent analyzing at 460.
  • result generation may be performed to the intermediate analysis result so as to generate an analysis result 408.
  • the result generation at 470 may generate the analysis result 408 through expressing the intermediate analysis result in a more readable and understandable way.
  • the analysis result 408 may describe statistical information contained in the intermediate analysis result in natural language.
  • the result generation at 470 may generate the analysis result 408 through further performing, e.g., reasoning operation, inducing operation, summarizing operation, etc. to the intermediate analysis result.
  • the result generation at 470 may reason out that most of the participants are satisfied with the communication session.
  • the result generation at 470 may be implemented through various predefined heuristic rules and/or a machine learning model. The embodiments of the present disclosure are not limited to any specific approach for implementing the result generation at 470.
  • an intermediate analysis result for each previous poll may be obtained.
  • an intermediate analysis result may be obtained through performing statistical analyzing to a poll response of this previous poll according to a type of this previous poll, e.g., the intermediate analysis result of this previous poll may be obtained through performing one of the rating distribution computation 432, the ranking distribution computation 434, the word cloud computation 436, etc. which corresponds to the type of this previous poll. In this way, two or more intermediate analysis results corresponding to the two or more previous polls may be obtained.
  • the post processing at 440 may further comprise ranking the two or more intermediate analysis results according to any predetermined criteria, and selecting the highest-ranked intermediate analysis result. Then, for the highest-ranked intermediate analysis result, the post processing at 440 may trigger to perform at least one of the action analyzing 450, the intent analyzing 460 and the result generation 470.
  • FIG. 5 illustrates an exemplary process 500 of action suggestion generation according to an embodiment.
  • Action suggestion generation 540 in the process 500 is an exemplary implementation of the operation of action suggestion generation at 270 in FIG. 2.
  • Session insight information 510 and poll insight information 520 are inputs to the action suggestion generation 540.
  • the session insight information 510 may comprise an action 512 and an intent 514 that are predicted according to the process 300 in FIG. 3, and the poll insight information 520 may comprise an action 522 and an intent 524 that are predicted according to the process 400 in FIG. 4.
  • session metadata 530 of the communication session may also be input to the action suggestion generation 540.
  • action merging may be performed to merge the action 512 in the session insight information 510 and the action 522 in the poll insight information 520 into at least one merged action. For example, if the action 512 is different from the action 522, the action merging at 550 may output the action 512 and the action 522 directly as two merged action. For example, if the action 512 is the same as the action 522, the action merging at 550 may output a single merged action which is the same as the action 512 and the action 522.
  • intent merging may be performed to merge the intent 514 in the session insight information 510 and the intent 524 in the poll insight information 520 into at least one merged intent.
  • an enumerating approach may be adopted by the intent merging at 560.
  • the at least one merged intent may be formed through combining any one or more dimensions in the intent 514 with any one or more dimensions in the intent 524 in an enumerating approach.
  • the prediction of action and intent according to session data and the prediction of action and intent according to poll data may be effectively complementary with each other.
  • An action suggestion may be further generated based at least on the at least one merged action and the at least one merged intent.
  • suggestion generation may be performed so as to generate at least one action suggestion candidate through searching a predetermined knowledge base 572 according to the at least one merged action and the at least one merged intent.
  • the knowledge base 572 may comprise a plurality of action suggestion candidates, each of which has a corresponding action and a corresponding intent.
  • the suggestion generation at 570 may form a plurality of [action, intent] pairs through combining each of the at least one merged action with each of the at least one merged intent in an enumerating approach, and then retrieve an action suggestion candidate from the knowledge base 572 for each [action, intent] pair. In this way, at least one action suggestion candidate may be output by the suggestion generation at 570.
  • a feedback poll for a training communication session may be retrieved from the knowledge base 572 as an action suggestion candidate.
  • the suggestion generation at 570 may further comprise utilizing the session metadata 530 for introducing more information related to the communication session into an action suggestion candidate.
  • a suggestion of sending a document after communication session may be retrieved from the knowledge base 572, and then a link of the shared document may be obtained or identified from the session metadata 530 and added into the retrieved suggestion to form an action suggestion candidate.
  • the suggestion generation at 570 may be implemented through various predefined heuristic rules. The embodiments of the present disclosure are not limited to any specific approach for implementing the suggestion generation at 570.
  • a single knowledge base 572 is shown in FIG. 5, multiple different types of knowledge base may be pre-established for multiple different types of action respectively.
  • suggestion ranking may be performed so as to rank the at least one action suggestion candidate output by the suggestion generation at 570.
  • the ranking of the at least one action suggestion candidate may be performed according to relevance with the target user or the communication session. For example, a regression model may be adopted for generating a relevance probability for each action suggestion candidate, and the at least one action suggestion candidate may be ranked according to their relevance probabilities.
  • the highest-ranked action suggestion candidate may be output by the action suggestion generation 540 as an action suggestion 590 which is to be provided to the target user.
  • the relevance probability of the highest-ranked action suggestion candidate may be compared with a predefined relevance probability threshold, and only when the relevance probability of the highest-ranked action suggestion candidate is not lower than the relevance probability threshold, the highest-ranked action suggestion candidate would be output as the action suggestion 590.
  • FIG. 6 illustrates an exemplary user interface according to an embodiment.
  • a target user Jim is attending an online meeting with other participants through an online meeting application.
  • the title of the online meeting is “AI Project Progress” .
  • User interface 600 of the online meeting application is shown on a screen of a terminal device of Jim.
  • the rating poll may be created by Jim or suggested according to the embodiments of the present disclosure.
  • the embodiments of the present invention may collect session data of the online meeting and poll data of the rating poll, and provide an action suggestion about launching a feedback poll.
  • a pop-up block 620 may be shown in the user interface 600, which includes an analysis result of the rating poll as shown in block 622, and an action suggestion as shown in block 624.
  • the action suggestion includes an example of the suggested feedback poll. If Jim accepts the action suggestion, he may simply click on a button “Apply” , and then the suggested feedback poll will be sent to other participants automatically.
  • FIG. 7 illustrates an exemplary user interface according to an embodiment.
  • the embodiments of the present invention may collect session data of the online meeting and poll data of the rating poll, and provide an action suggestion about creating a feedback poll through sending an email to Jim automatically, as shown in user interface 700 of an email service.
  • the email may include an analysis result of the rating poll as shown in dotted block 712, and an action suggestion as shown in dotted block 714. If Jim accepts the action suggestion, he may click on a button “Yes” , and then a form service may be triggered for creating a feedback poll.
  • FIG. 8 illustrates an exemplary user interface according to an embodiment.
  • a target user Jim is attending an online meeting “AI Project Progress” with other participants through an online meeting application.
  • User interface 800 of the online meeting application is shown on a screen of a terminal device of Jim.
  • the embodiments of the present invention may collect session data of the online meeting and poll data of the confirmation poll, and provide an action suggestion about sending the document again to Eric and Beth who did not receive the document.
  • a pop-up block 820 may be shown in the user interface 800, which includes an analysis result 822 of the confirmation poll and an action suggestion 824. If Jim accepts the action suggestion, he may click on a button “Yes” , and then the document will be sent to Eric and Beth automatically.
  • FIG. 9 illustrates an exemplary user interface according to an embodiment.
  • a target user Steve edited an email including a poll 910 for collecting opinions of preferred topic from a group of recipients, as shown in user interface 900 of an email service, and then sent the email to the recipients.
  • Steve is editing a new email for replying to the recipients, as shown in the lower half part of FIG. 9.
  • the embodiments of the present invention may collect session data of the email including the poll 910 and poll data of the poll 910, and provide an action suggestion about launching another poll.
  • the action suggestion is provided in the user interface 900, as shown by dotted block 920.
  • the action suggestion may comprise a suggested poll 930 for collecting confirmation of the topic “technical barrier” which is selected most in a poll response to the poll 910. If Steve accepts the action suggestion, he may click on a button “Apply” , and then the poll 930 will be sent to the recipients automatically.
  • FIG. 6 to FIG. 9 merely show several exemplary user interfaces according to the embodiments of the present disclosure, and the embodiments of the present disclosure are not limited to any specific approach for presenting action suggestions to a target user.
  • FIG. 10 illustrates a flowchart of an exemplary method 1000 for providing action suggestion for a communication session according to an embodiment.
  • session insight information may be generated based on session data of the communication session.
  • poll insight information may be generated based on poll data of at least one previous poll associated with the communication session.
  • an action suggestion may be generated based at least on the session insight information and the poll insight information.
  • the session data may comprise at least one item of: session textual content, session time information, session metadata, user operation information and external information.
  • the session insight information may comprise a first action and a first intent predicted for the action suggestion.
  • the generating session insight information may comprise: predicting the first action through performing a series of action analyzing to the session data; and predicting the first intent through performing a series of intent analyzing to the session data.
  • the series of action analyzing may comprise multiple sequential action analyzing stages.
  • Each action analyzing stage may comprise at least one of: obtaining a current predicted action through performing action analyzing to a corresponding item in the session data; if a preceding predicted action output by a preceding action analyzing stage does not meet a predetermined confidence threshold, obtaining a current predicted action through performing action analyzing to a corresponding item in the session data, and replacing the preceding predicted action by the current predicted action; and if a preceding predicted action output by a preceding action analyzing stage meets a predetermined confidence threshold, maintaining the preceding predicted action.
  • the series of intent analyzing may comprise multiple sequential intent analyzing stages.
  • Each intent analyzing stage may comprise at least one of: obtaining a current predicted intent through performing intent analyzing to a corresponding item in the session data; obtaining a current predicted intent through performing intent analyzing to a corresponding item in the session data, and updating a preceding predicted intent output by a preceding intent analyzing stage with the current predicted intent; if a preceding predicted intent output by a preceding intent analyzing stage does not meet a predetermined confidence threshold, obtaining a current predicted intent through performing intent analyzing to a corresponding item in the session data, and replacing the preceding predicted intent by the current predicted intent; and if a preceding predicted intent output by a preceding intent analyzing stage meets a predetermined confidence threshold, maintaining the preceding predicted intent.
  • the poll data may comprise: poll metadata and poll response.
  • the poll insight information may comprise: a second action and a second intent predicted for the action suggestion, and an analysis result.
  • the generating poll insight information may comprise: obtaining at least one intermediate analysis result through performing statistical analyzing to the poll response according to a type of the at least one previous poll; predicting the second action through performing action analyzing to the intermediate analysis result; predicting the second intent through performing intent analyzing to the intermediate analysis result and poll content in the poll metadata; and generating the analysis result according to the intermediate analysis result.
  • the generating poll insight information may comprise: determining whether the intermediate analysis result meets a statistical significance condition; and in response to determining that the intermediate analysis result meets the statistical significance condition, triggering to perform the operations of predicting the second action, predicting the second intent, and generating the analysis result.
  • the at least one previous poll may comprise two or more previous polls.
  • the obtaining at least one intermediate analysis result may comprise: for each previous poll, obtaining an intermediate analysis result through performing statistical analyzing to a poll response of the previous poll according to a type of the previous poll.
  • the generating poll insight information may further comprise: ranking two or more intermediate analysis results corresponding to the two or more previous polls; and for the highest-ranked intermediate analysis result, triggering to perform the operations of predicting the second action, predicting the second intent, and generating the analysis result.
  • the generating an action suggestion may comprise: merging a first action in the session insight information and a second action in the poll insight information into at least one merged action; merging a first intent in the session insight information and a second intent in the poll insight information into at least one merged intent; and generating the action suggestion based at least on the at least one merged action and the at least one merged intent.
  • the generating the action suggestion may comprise: generating at least one action suggestion candidate through searching a predetermined knowledge base according to the at least one merged action and the at least one merged intent; ranking the at least one action suggestion candidate; and outputting the highest-ranked action suggestion candidate as the action suggestion.
  • the at least one action suggestion candidate may be generated further based on session metadata in the session data.
  • the action suggestion may comprise a prompt associated with at least one action of: creating a poll, launching a poll, sending a document, setting up a new communication session, recording an annotation and sending a message.
  • the session insight information may at least comprise a first intent
  • the poll insight information may at least comprise a second intent.
  • the first intent and the second intent may be defined with at least one dimension of: purpose, target, domain, time information and recipients.
  • the method 1000 may further comprise any steps/operations for providing action suggestion for a communication session according to the above embodiments of the present disclosure.
  • FIG. 11 illustrates an exemplary apparatus 1100 for providing action suggestion for a communication session according to an embodiment.
  • the apparatus 1100 may comprise: a session insight information generating module 1110, for generating session insight information based on session data of the communication session; a poll insight information generating module 1120, for generating poll insight information based on poll data of at least one previous poll associated with the communication session; and an action suggestion generating module 1130, for generating an action suggestion based at least on the session insight information and the poll insight information.
  • a session insight information generating module 1110 for generating session insight information based on session data of the communication session
  • a poll insight information generating module 1120 for generating poll insight information based on poll data of at least one previous poll associated with the communication session
  • an action suggestion generating module 1130 for generating an action suggestion based at least on the session insight information and the poll insight information.
  • the apparatus 1100 may further comprise any other modules for performing any steps/operations in the methods for providing action suggestion for a communication session according to the above embodiments of the present disclosure.
  • FIG. 12 illustrates an exemplary apparatus 1200 for providing action suggestion for a communication session according to an embodiment.
  • the apparatus 1200 comprises: at least one processor 1210; and a memory 1220 storing computer-executable instructions.
  • the computer-executable instructions when executed, cause the at least one processor 1210 to: generate session insight information based on session data of the communication session; generate poll insight information based on poll data of at least one previous poll associated with the communication session; and generate an action suggestion based at least on the session insight information and the poll insight information.
  • the session data may comprise at least one item of: session textual content, session time information, session metadata, user operation information and external information.
  • the session insight information may comprise a first action and a first intent predicted for the action suggestion.
  • the poll data may comprise: poll metadata and poll response.
  • the poll insight information may comprise: a second action and a second intent predicted for the action suggestion, and an analysis result.
  • the generating poll insight information may comprise: obtaining at least one intermediate analysis result through performing statistical analyzing to the poll response according to a type of the at least one previous poll; predicting the second action through performing action analyzing to the intermediate analysis result; predicting the second intent through performing intent analyzing to the intermediate analysis result and poll content in the metadata; and generating the analysis result according to the intermediate analysis result.
  • the generating an action suggestion may comprise: merging a first action in the session insight information and a second action in the poll insight information into at least one merged action; merging a first intent in the session insight information and a second intent in the poll insight information into at least one merged intent; and generating the action suggestion based at least on the at least one merged action and the at least one merged intent.
  • the computer-executable instructions when executed, may cause the at least one processor 1210 to perform any other operations in the methods for providing action suggestion for a communication session according to the above embodiments of the present disclosure.
  • the embodiments of the present disclosure propose a computer program product for providing action suggestion for a communication session.
  • the computer program product may comprise a computer program that is executed by at least one processor for: generating session insight information based on session data of the communication session; generating poll insight information based on poll data of at least one previous poll associated with the communication session; and generating an action suggestion based at least on the session insight information and the poll insight information.
  • the computer program may be further executed by the at least one processor for performing any steps/operations in the methods for providing action suggestion for a communication session according to the above embodiments of the present disclosure.
  • the embodiments of the present disclosure may be embodied in a non- transitory computer-readable medium.
  • the non-transitory computer-readable medium may comprise instructions that, when executed, cause one or more processors to perform any steps/operations in the methods for providing action suggestion for a communication session according to the above embodiments of the present disclosure.
  • modules in the apparatuses described above may be implemented in various approaches. These modules may be implemented as hardware, software, or a combination thereof. Moreover, any of these modules may be further functionally divided into sub-modules or combined together.
  • processors have been described in connection with various apparatuses and methods. These processors may be implemented using electronic hardware, computer software, or any combination thereof. Whether such processors are implemented as hardware or software will depend upon the particular application and overall design constraints imposed on the system.
  • a processor, any portion of a processor, or any combination of processors presented in the present disclosure may be implemented with a microprocessor, microcontroller, digital signal processor (DSP) , a field-programmable gate array (FPGA) , a programmable logic device (PLD) , a state machine, gated logic, discrete hardware circuits, and other suitable processing components configured to perform the various functions described throughout the present disclosure.
  • DSP digital signal processor
  • FPGA field-programmable gate array
  • PLD programmable logic device
  • a state machine gated logic, discrete hardware circuits, and other suitable processing components configured to perform the various functions described throughout the present disclosure.
  • the functionality of a processor, any portion of a processor, or any combination of processors presented in the present disclosure may be
  • a computer-readable medium may include, by way of example, memory such as a magnetic storage device (e.g., hard disk, floppy disk, magnetic strip) , an optical disk, a smart card, a flash memory device, random access memory (RAM) , read only memory (ROM) , programmable ROM (PROM) , erasable PROM (EPROM) , electrically erasable PROM (EEPROM) , a register, or a removable disk.
  • RAM random access memory
  • ROM read only memory
  • PROM programmable ROM
  • EPROM erasable PROM
  • EEPROM electrically erasable PROM

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Physics & Mathematics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

La présente divulgation concerne des procédés et des appareils pour fournir une suggestion d'action pour une session de communication. Des informations d'aperçu de session peuvent être générées sur la base de données de session de la session de communication. Des informations d'aperçu d'interrogation peuvent être générées sur la base de données d'interrogation d'au moins une interrogation précédente associée à la session de communication. Une suggestion d'action peut être générée sur la base au moins des informations d'aperçu de session et des informations d'aperçu d'interrogation.
PCT/CN2022/123709 2022-10-05 2022-10-05 Suggestion d'action pour une session de communication WO2024073872A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/123709 WO2024073872A1 (fr) 2022-10-05 2022-10-05 Suggestion d'action pour une session de communication

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/123709 WO2024073872A1 (fr) 2022-10-05 2022-10-05 Suggestion d'action pour une session de communication

Publications (1)

Publication Number Publication Date
WO2024073872A1 true WO2024073872A1 (fr) 2024-04-11

Family

ID=84047644

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/123709 WO2024073872A1 (fr) 2022-10-05 2022-10-05 Suggestion d'action pour une session de communication

Country Status (1)

Country Link
WO (1) WO2024073872A1 (fr)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170063744A1 (en) * 2015-09-02 2017-03-02 International Business Machines Corporation Generating Poll Information from a Chat Session
WO2021216190A1 (fr) * 2020-04-22 2021-10-28 Microsoft Technology Licensing, Llc Fourniture d'assistance pour service de formulaire
US20220138410A1 (en) * 2019-04-30 2022-05-05 Microsoft Technology Licensing, Llc Document auto-completion

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170063744A1 (en) * 2015-09-02 2017-03-02 International Business Machines Corporation Generating Poll Information from a Chat Session
US20220138410A1 (en) * 2019-04-30 2022-05-05 Microsoft Technology Licensing, Llc Document auto-completion
WO2021216190A1 (fr) * 2020-04-22 2021-10-28 Microsoft Technology Licensing, Llc Fourniture d'assistance pour service de formulaire

Similar Documents

Publication Publication Date Title
US11710136B2 (en) Multi-client service system platform
US11087283B2 (en) Method and system for managing, matching, and sourcing employment candidates in a recruitment campaign
EP3467821B1 (fr) Sélection de services de transcription et de traduction et génération de résultats combinés
US10361981B2 (en) Automatic extraction of commitments and requests from communications and content
US20180089163A1 (en) Systems methods and computer-readable storage media for real-time automated conversational agent
CN114556354A (zh) 自动确定和呈现来自事件的个性化动作项
US11880666B2 (en) Generating conversation descriptions using neural networks
Windiatmoko et al. Developing facebook chatbot based on deep learning using rasa framework for university enquiries
CN111276148A (zh) 基于卷积神经网络的回访方法、系统及存储介质
US20240005089A1 (en) Document auto-completion
CN117151662B (zh) 基于ai的岗位信息处理方法、装置、设备及存储介质
CN115735357A (zh) 用于电话会议讨论的投票问题
CN117236909A (zh) 基于ai的招聘邀约方法、装置、设备及存储介质
CN116775815B (zh) 对话数据的处理方法、装置、电子设备及存储介质
US20230144977A1 (en) Providing form service assistance
CN113065850B (zh) 用于智能外呼机器人的话术测试方法及装置
CN113362045A (zh) 会议日程的生成方法、装置、电子设备及可读存储介质
US20210406839A1 (en) Computerized meeting system
WO2024073872A1 (fr) Suggestion d'action pour une session de communication
CN115033675A (zh) 会话方法、会话装置、电子设备及存储介质
Aksyonov et al. Development of a Chatbot for the Subject Area Train Schedule for Telegram
Hrycyk et al. Not so fast, classifier–accuracy and entropy reduction in incremental intent classification
CN111046151A (zh) 一种消息处理方法及装置
US12125045B2 (en) Multi-client service system platform
US20240020481A1 (en) Automatic Detection of Sentiment Based on Electronic Communication

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22798236

Country of ref document: EP

Kind code of ref document: A1