Intelligent conference management system based on cloud platform

Abstract

The invention discloses an intelligent conference management system based on a cloud platform, and relates to the technical field of conference management, and the system comprises a conference data collection module, a conference response module, a real-time monitoring module, a response terminal, a knowledge precipitation efficiency evaluation module and a data warehouse. Effective proceeding of the conference is guaranteed, and the problems of low conference efficiency, low decision-making quality and poor communication effect are further avoided; the conference conclusion is converted into the knowledge base document, the conference knowledge base is constructed, the conference knowledge base is monitored, the knowledge sharing range is greatly expanded, and meanwhile data support is provided for follow-up storage and confidentiality of the conference conclusion and conference management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of conference management, and in particular to an intelligent conference management system based on a cloud platform. BACKGROUND

[0002] In the operation system of modern enterprises, conferences are the key link of information circulation, decision-making and team collaboration. From regular meetings of grassroots departments to high-level strategic seminars, conferences run through every stage of enterprise operation. However, inefficient meetings not only waste time and resources, but also may hinder enterprise development. Therefore, scientific and effective conference management is an important guarantee for enterprises to improve operational efficiency and achieve strategic goals.

[0003] Prior art such as the invention patent application with publication number CN118134434B discloses an intelligent conference management system and method based on artificial intelligence, which accurately controls the quality and effect of the conference, improves the conference effect and efficiency, and the invention patent application with publication number CN117078223B discloses an intelligent conference management system based on artificial intelligence, which combines conference room comfort detection analysis and light atmosphere detection and evaluation of each conference link and provides early warning and control, which is conducive to ensuring the smooth completion of the conference and realizing effective supervision of the conference process.

[0004] In combination with the above-mentioned scheme, it can be found that the prior art still has the following deficiencies, which are embodied in the following aspects: traditional conference management mostly adopts post-reporting summary, on the one hand, it lacks real-time dynamic detection and intervention of the conference, which guarantees the effective progress of the conference, further leading to low conference efficiency, low decision-making quality and poor communication effect, on the other hand, it rarely converts conference conclusions into knowledge base documents, constructs conference knowledge base, and monitors the conference knowledge base, which limits the scope of knowledge sharing and makes it difficult to provide data support for the storage, secrecy and management of subsequent conference conclusions. SUMMARY

[0005] The present application aims to provide an intelligent conference management system based on a cloud platform, which solves the problems in the background art.

[0006] To solve the above technical problems, the present application adopts the following technical scheme: the present application provides an intelligent conference management system based on a cloud platform, comprising: a conference data collection module for collecting basic data and performance data of a plurality of historical conferences through a cloud platform, the basic data refers to descriptions and records related to conferences formed in past conference activities of an enterprise or organization, including team coding and conference mode, the performance data refers to actual data related to conference results and decisions generated in the past conference process, a conference management rule library is constructed, the conference management rule library includes a plurality of recommended conference modes corresponding to a plurality of conference themes of team coding.

[0007] The conference response module is configured to collect relevant information of the real-time conference through the cloud platform, and determine a conference mode of the real-time conference according to a conference management rule library.

[0008] The real-time monitoring module is configured to dynamically monitor the conference and evaluate whether the progress of the real-time conference deviates when the real-time conference is in progress, and generate a voice reminder.

[0009] The response terminal is configured to remind the real-time conference by the voice reminder when the real-time conference deviates.

[0010] The knowledge sedimentation efficiency evaluation module is configured to track conversion of conference conclusions into a knowledge base document, construct a conference knowledge base, monitor the conference knowledge base, divide a target value level of a conference theme based on monitoring data of the conference knowledge base, and display the target value level to a decision-making layer.

[0011] The present application has the following advantages: (1) The present application dynamically detects and intervenes in the conference in real time, guarantees effective progress of the conference, and further avoids problems of low conference efficiency, low decision-making quality and poor communication effect.

[0012] (2) The present application converts conference conclusions into a knowledge base document, constructs a conference knowledge base, and monitors the conference knowledge base, greatly expands the knowledge sharing range, and provides data support for subsequent storage, secrecy and management of the conference conclusions. BRIEF DESCRIPTION OF DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of these drawings.

[0014] Figure 1 The present application is a system structure connection schematic diagram. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0016] Referring to Figure 1 The present application provides an intelligent conference management system based on a cloud platform, comprising:

[0017] The meeting data collection module is used to collect basic and performance data of several historical meetings through a cloud platform. The basic data refers to the descriptions and records related to the meetings formed by the enterprise or organization in past meeting activities, including team codes and meeting modes. The performance data refers to the actual data related to meeting results and decisions generated during past meetings. The module constructs a meeting management rule base, which includes recommended meeting modes for several meeting topics corresponding to several team codes.

[0018] The specific meeting topics include morning meetings, weekly meetings, monthly meetings, annual meetings, project meetings, budget meetings, board meetings, and shareholders' meetings.

[0019] In a specific embodiment of the present invention, the basic data includes team coding, meeting mode and meeting minutes. The meeting mode includes online mode, offline mode and hybrid mode. The performance data includes the number of times someone speaks, decision-making speed, agenda completion rate, post-meeting survey satisfaction, cost overrun, and number of times someone leaves the meeting.

[0020] In a specific embodiment of the present invention, the method for constructing the meeting management rule base is as follows: based on the team codes, meeting modes and meeting minutes collected from the basic data of several historical meetings collected by the cloud platform, the meeting modes and meeting minutes of several historical meetings with several team codes are mapped to the meeting modes and meeting minutes of several historical meetings with several team codes. Based on the meeting minutes, the meeting topics of several historical meetings with several team codes are identified, and the effectiveness score of the meeting modes of several historical meetings with several team codes is evaluated through performance data.

[0021] It should be noted that the existing technology for identifying the meeting topics of several historical meetings using several team codes is relatively mature, and will not be elaborated here.

[0022] It should be noted that the evaluation of the effectiveness score of several historical meetings coded by several teams through performance data specifically involves: uniformizing the performance data; the comparison for uniformization can be the appropriate interval corresponding to the performance data defined in the data warehouse; the uniformization standard is the minimum and maximum values ​​of the appropriate intervals corresponding to several performance data; if a performance data is not within the appropriate interval, it is forcibly reset to 0; or it can be the maximum and minimum values ​​of the performance data. For example, for the number of speeches, the maximum and minimum number of speeches are extracted from the effectiveness score of several historical meetings coded by several teams. Then, the number of speeches of several historical meetings coded by several teams is uniformized. All the uniformized performance data are weighted and summed to obtain the effectiveness score, thus obtaining the effectiveness score of several historical meetings coded by several teams.

[0023] The weighted summation mentioned above specifically involves assigning a specific weight to each performance data point, with the weights having positive and negative signs to measure the relationship between the performance data and the effectiveness score. For example, if the number of times someone speaks and the speed of decision-making are directly proportional to the effectiveness score, then the weights for the number of times someone speaks and the speed of decision-making are positive. If the cost overrun and the number of times someone leaves the meeting are inversely proportional to the effectiveness score, then the weights for the cost overrun and the number of times someone leaves the meeting are negative. The specific settings can be configured by the company's meeting management personnel, and no specific restrictions are imposed here.

[0024] The mapping process yields several performance scores for several meeting modes corresponding to various meeting topics for several team codes. A performance score dataset for these meeting modes corresponding to various team codes for various meeting topics is constructed. The dispersion L_S(i,m,p) of the performance score dataset is obtained through dispersion analysis. If [X_G(i,m,p)≥XG′]∧[L_S(i,m,p)≤LS′], then the meeting mode is recorded as the recommended meeting mode for the corresponding meeting topic for that team code. In this formula, X_G(i,m,p) is the average value of the performance score dataset, and XG′ and LS′ are the preset performance score threshold and dispersion threshold, respectively. Based on this, a meeting management rule base is constructed, where i, m, and p are the team code number, meeting topic number, and meeting mode number, respectively.

[0025] It should be noted that the higher the dispersion of the dataset, the greater the discrepancy in the performance evaluation data, and the lower its reliability. Therefore, it is necessary to analyze the dispersion of the performance evaluation datasets for several meeting modes. The dispersion analysis of datasets is a relatively mature technology in the existing field, and will not be elaborated here.

[0026] The meeting response module is used to collect relevant information about real-time meetings through the cloud platform and determine the meeting mode of the real-time meeting based on the meeting management rule library.

[0027] In a specific embodiment of the present invention, the method for determining the meeting mode of a real-time meeting is as follows: Relevant information about the real-time meeting is collected through a cloud platform. This relevant information includes team code, meeting topic, and main meeting content. A recommended meeting mode is retrieved from the meeting management rule base based on the team code and meeting topic of the real-time meeting. If the retrieval is successful, the retrieved recommended meeting mode is used as the meeting mode of the real-time meeting. If the retrieval fails, a set of meeting keywords (a1, a2, ..., aj, ..., ak) is obtained through semantic recognition based on the main meeting content of the real-time meeting. Here, a1, a2, and aj represent the first keyword, the second keyword, and the j-th keyword, respectively, and j is the keyword number, j = 1, 2, ..., k.

[0028] The set of feature keywords corresponding to each meeting mode is extracted from the data warehouse and compared. The similarity between the set of meeting keywords of the real-time meeting and the set of feature keywords of each meeting mode is obtained by set similarity evaluation. If the set of feature keywords of a certain meeting mode has the highest similarity, then the meeting mode corresponding to that meeting mode is taken as the meeting mode of the real-time meeting.

[0029] It should be noted that the set similarity assessment can specifically adopt the Jacquard similarity, cosine similarity or Dice coefficient for evaluation, which are relatively mature and simple in the existing technology, and will not be elaborated here.

[0030] Specifically, the meeting keywords include: mostly middle-aged and elderly people, decision-making, sensitive information, unstable network, high-frequency interaction, mostly young people, training, routine information, stable network, and low-frequency interaction.

[0031] It should be noted that the specific method for obtaining the similarity between the set of meeting keywords of the real-time meeting and the set of feature keywords of each meeting mode through set similarity evaluation is as follows: each keyword in the set of meeting keywords of the real-time meeting is compared with each keyword in the set of feature keywords of each meeting mode. If they are the same, the similarity of the keyword is recorded as 1; otherwise, it is recorded as 0. Thus, the similarity between each keyword of the real-time meeting and the corresponding keyword of each meeting mode is obtained. The similarity between the set of meeting keywords of the real-time meeting and the set of feature keywords of each meeting mode is obtained by weighted summation using the weight index of each keyword in the set of feature keywords of each meeting mode.

[0032] The weight index of each keyword in the set of characteristic keywords for each meeting mode mentioned above specifically reflects the influence of each keyword on the choice of meeting format. The specific weight index can be set by the meeting decision-makers themselves, or a decision matrix can be constructed using the APH (Analytic Hierarchy Process) method. By comparing the importance of dimensions pairwise, the weights can be calculated. This method is relatively mature in the existing technology and will not be elaborated here. The sum of the weight indices of each keyword is 1.

[0033] The real-time monitoring module is used to dynamically monitor the meeting during the real-time meeting, assess whether the progress of the real-time meeting has deviated, and generate voice reminders.

[0034] This invention performs real-time dynamic monitoring and intervention on meetings to ensure their effective conduct and further avoid problems such as low meeting efficiency, poor decision-making quality, and ineffective communication.

[0035] In a specific embodiment of the present invention, the method for dynamically monitoring the meeting during a real-time meeting is as follows: if the meeting mode is online, the meeting is monitored through an online meeting platform and eye-tracking technology.

[0036] If the meeting is conducted offline, the meeting will be monitored through the monitoring system installed in the meeting room.

[0037] If the meeting mode is hybrid, the meeting will be monitored through an online meeting platform, eye-tracking technology, and a meeting room monitoring system.

[0038] It should be noted that the monitoring method used is an existing and relatively simple technology, and will not be elaborated upon here.

[0039] In a specific embodiment of the present invention, the method for evaluating whether the process of the real-time meeting has deviated and generating a voice reminder is as follows: based on the main content of the real-time meeting, the main content of the real-time meeting is decomposed into multiple core node tags and corresponding time nodes, and a role responsibility matrix is ​​preset based on the team member codes of the real-time meeting. The role responsibility matrix is ​​specifically the leading team member code corresponding to each core node tag.

[0040] By dynamically monitoring the real-time meeting, the speech content and duration of several team members are obtained, and the role defocus index of the real-time meeting is calculated. If the role defocus index of the real-time meeting is greater than or equal to the preset role defocus index threshold, it is determined that the process of the real-time meeting has deviated, and a type of voice reminder is generated.

[0041] By dynamically monitoring the real-time meeting, OCR recognition of screen-shared content or uploaded documents is obtained, and a correlation graph between the shared document and the real-time meeting is constructed. The matching degree between the keywords of the shared document in the real-time meeting and the current core node label is calculated using the keyword density calculation formula. If the matching degree is less than the preset matching degree threshold of the shared document, it is determined that the process of the real-time meeting has deviated, and a second type of voice reminder is generated. The keyword density calculation formula is relatively mature in the existing technology and will not be elaborated here.

[0042] By dynamically monitoring real-time meetings and marking actual decision points, and combining the core decision tags of the real-time meeting with the corresponding time nodes, the decision delay duration is obtained, the decision lag cost of the real-time meeting is calculated, and three types of voice reminders are generated.

[0043] By dynamically monitoring the real-time meeting, the set of emotional categories of the speakers is clustered in real time to determine the emotional interference offset judgment value of the real-time meeting. If the emotional interference offset judgment value of the real-time meeting is 1, it is determined that the process of the real-time meeting has deviated, and four types of voice reminders are generated.

[0044] It should be noted that the specific method for determining the emotional interference offset judgment value of the real-time meeting is as follows: based on the set of speaking emotion categories, identify several risky speaking emotion categories, summarize the total number of risky speaking emotion categories, and divide it by the total number of speaking emotion categories to obtain the risky speaking emotion category ratio coefficient. If the risky speaking emotion category ratio coefficient is greater than or equal to the preset risky speaking emotion category ratio coefficient threshold, then the emotional interference offset judgment value of the real-time meeting is recorded as 1; otherwise, it is recorded as 0. The preset risky speaking emotion category ratio coefficient threshold can be set by the meeting administrator, or it can be obtained by training with a large amount of historical data to obtain the maximum tolerable risky speaking emotion category ratio coefficient corresponding to the normal completion of the meeting.

[0045] In a specific embodiment of the present invention, the real-time conference role defocus index is specifically calculated using the formula: S_J=T_no / (T_no+T_yes)*P_Y.

[0046] In the formula, S_J represents the role defocus index of the real-time meeting, T_no and T_yes represent the speaking duration of the non-dominant role and the speaking duration of the dominant role in the real-time meeting, respectively, and P_Y represents the speaking deviation degree of the non-dominant role in the real-time meeting. The speaking duration of the non-dominant role and the speaking duration of the dominant role in the real-time meeting are specifically determined based on the speaking duration encoded by several team members and the role responsibility matrix, and the speaking deviation degree of the non-dominant role in the real-time meeting is specifically determined based on the speaking content encoded by several team members and the role responsibility matrix, combined with semantic analysis.

[0047] In a specific embodiment of the present invention, the decision lag cost of the real-time meeting is specifically calculated using the formula: C_B=T_Y*M*γ.

[0048] In the formula, C_B represents the decision lag cost of the real-time meeting, T_Y is the decision delay duration of the real-time meeting, M is the average hourly wage of the team members in the real-time meeting, and γ is the preset decision urgency weight.

[0049] It should be noted that the average hourly wage of team members is specifically obtained from the enterprise's internal human resource planning table stored in the data warehouse, and the preset decision urgency weight is uploaded by the meeting management personnel.

[0050] The response terminal is used to remind the real-time meeting of any deviations via voice prompts.

[0051] The knowledge accumulation efficiency evaluation module is used to track the transformation of meeting conclusions into knowledge base documents, build the meeting knowledge base, monitor the meeting knowledge base, and classify the target value levels of several meeting topics based on the monitoring data of the meeting knowledge base, and display them to the decision-making level.

[0052] This invention transforms meeting conclusions into knowledge base documents, constructs a meeting knowledge base, and monitors the meeting knowledge base, greatly expanding the scope of knowledge sharing. At the same time, it provides data support for the subsequent storage, confidentiality, and management of meeting conclusions and meetings.

[0053] In a specific embodiment of the present invention, the conference knowledge base includes several knowledge base documents with several team codes corresponding to several conference topics.

[0054] In a specific embodiment of the present invention, the monitoring data based on the meeting knowledge base is divided into target value levels for several meeting topics. The specific division method is as follows: the monitoring data includes the set of cited document types, the set of citing departments, the number of collaboration touchpoints, the innovation conversion rate, the half-life index, and the first trigger duration of several knowledge base documents corresponding to several meeting topics by several team codes. The network centrality and cross-department citation rate of several knowledge base documents are determined according to the set of cited document types and the set of citing departments of several knowledge base documents.

[0055] The number of collaborative touchpoints is specifically the sum of the number of collaborative editing, commenting, and task assignments initiated based on a certain knowledge base document.

[0056] The innovation conversion rate is specifically the proportion of documents derived from a knowledge base that are actually implemented.

[0057] The half-life index specifically refers to the time required for the citation count of a knowledge base document to drop to 50% of its peak value. The first trigger duration specifically refers to the time required for a knowledge base document to be first cited by a decision.

[0058] It should be noted that the specific method for determining the network centrality of several knowledge base documents is as follows: based on the set of cited document types of several knowledge base documents, extract several cited document types, summarize the number of cited document types, and summarize the total number of cited document types in the conference knowledge base. Divide the number of cited document types of several knowledge base documents by the total number of cited document types in the conference knowledge base to obtain the network centrality of several knowledge base documents. The method for determining the cross-departmental citation rate of several knowledge base documents is similar to the analysis method for the network centrality of several knowledge base documents, and will not be elaborated here.

[0059] The value index J_Z(i,m,h) of several knowledge base documents corresponding to several conference topics and several team codes is obtained by uniformizing the network centrality, cross-departmental citation rate, number of collaboration touchpoints, innovation conversion rate, half-life index, and first trigger duration, and then weighting and summing them.

[0060] It should be noted that the specific weights for the weighted summation can be determined by the knowledge base document expert scoring method.

[0061] If J_Z(i,m,h)∈JZ_f, then the knowledge base document is classified into the f-th value level, where JZ_f is the value index interval corresponding to the pre-set f-th value level. Several knowledge base documents of different value levels are obtained, and the conference topics corresponding to the several knowledge base documents are mapped to obtain several conference topics of different value levels. Several different value levels of several conference topics are mapped to obtain several different value levels. The value level with the largest distribution is selected as the target value level of several conference topics.

[0062] In this example, we assume there are three value levels: Level 1, Level 2, and Level 3. Assuming that higher levels correspond to greater value indicators, Level 1 is less important than Level 2, and Level 2 is less important than Level 3. In other words, Level 1 meeting topics are of low value, allowing for shorter document coverage and lower confidentiality levels. When there are many meetings, their urgency for timely execution is low. Level 2 meeting topics are of moderate value, maintaining document coverage and confidentiality levels. When there are many meetings, their urgency for timely execution is moderate. Level 3 meeting topics are of high value, extending document coverage and increasing confidentiality levels. When there are many meetings, their urgency for timely execution is high, which helps reduce the maintenance and management costs of meeting documents while ensuring reasonable management of meeting documents and facilitating data support for subsequent meetings.

[0063] It should be noted that the intelligent meeting management system of the cloud platform of this invention is specifically designed for internal enterprise use. According to relevant laws, regulations, and provisions, to protect enterprise office privacy, this patent application may require obtaining relevant information about enterprise meetings and monitoring the meeting process. Hereby, the applicant solemnly promises to strictly abide by relevant laws, regulations, and provisions, be responsible for the confidentiality of enterprise information, and will not use enterprise information for other purposes. At the same time, the applicant will also take necessary technical and organizational measures to ensure the security and confidentiality of enterprise information. By agreeing to and authorizing the provision of enterprise information, the enterprise also understands and agrees to the above authorization and commitment.

[0064] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A cloud-based intelligent meeting management system, characterized in that, include: The meeting data collection module is used to collect basic and performance data of several historical meetings through the cloud platform and build a meeting management rule base. The meeting management rule base includes recommended meeting modes for several meeting topics corresponding to several team codes. The meeting response module is used to collect relevant information about real-time meetings through the cloud platform and determine the meeting mode of the real-time meeting based on the meeting management rule library; The real-time monitoring module is used to dynamically monitor the meeting during the real-time meeting, assess whether the progress of the real-time meeting has deviated, and generate voice reminders. The response terminal is used to remind the real-time meeting of any deviations via voice prompts. The knowledge accumulation efficiency evaluation module is used to track the transformation of meeting conclusions into knowledge base documents, build the meeting knowledge base, monitor the meeting knowledge base, and classify the target value levels of several meeting topics based on the monitoring data of the meeting knowledge base, and display them to the decision-making level.

2. The intelligent meeting management system based on a cloud platform according to claim 1, characterized in that, The basic data includes team coding, meeting mode, and meeting minutes. The meeting mode includes online mode, offline mode, and hybrid mode. The performance data includes the number of times someone speaks, decision-making speed, agenda completion rate, post-meeting survey satisfaction, cost overrun, and number of times someone leaves the meeting.

3. The intelligent meeting management system based on a cloud platform according to claim 2, characterized in that, The specific method for constructing the meeting management rule base is as follows: Based on the team codes, meeting modes, and meeting minutes collected from several historical meetings on the cloud platform, the meeting modes and meeting minutes of several historical meetings with several team codes are mapped to the meeting minutes. The meeting topics of several historical meetings with several team codes are identified based on the meeting minutes, and the effectiveness score of the meeting modes of several historical meetings with several team codes is evaluated through performance data. The mapping process yields several performance scores for several meeting modes corresponding to various meeting topics for several team codes. A performance score dataset for these meeting modes corresponding to various team codes for various meeting topics is constructed. The dispersion L_S(i,m,p) of the performance score dataset is obtained through dispersion analysis. If [X_G(i,m,p)≥XG′]∧[L_S(i,m,p)≤LS′], then the meeting mode is recorded as the recommended meeting mode for the corresponding meeting topic for that team code. In this formula, X_G(i,m,p) is the average value of the performance score dataset, and XG′ and LS′ are the preset performance score threshold and dispersion threshold, respectively. Based on this, a meeting management rule base is constructed, where i, m, and p are the team code number, meeting topic number, and meeting mode number, respectively.

4. The intelligent meeting management system based on a cloud platform according to claim 3, characterized in that, The specific method for determining the meeting mode of the real-time meeting is as follows: The system collects relevant information about real-time meetings through a cloud platform. This information includes team codes, meeting topics, and main meeting content. Based on the team codes and meeting topics of the real-time meetings, it searches the meeting management rule base for recommended meeting modes. If the search is successful, the recommended meeting mode is used as the meeting mode for the real-time meetings. If the search fails, based on the main meeting content, it obtains a set of meeting keywords (a1, a2, ..., aj, ..., ak) for the real-time meetings through semantic recognition, where a1, a2, and aj represent the first keyword, the second keyword, and the j-th keyword, respectively, and j is the keyword number, j = 1, 2, ..., k. The set of feature keywords corresponding to each meeting mode is extracted from the data warehouse and compared. The similarity between the set of meeting keywords of the real-time meeting and the set of feature keywords of each meeting mode is obtained by set similarity evaluation. If the set of feature keywords of a certain meeting mode has the highest similarity, then the meeting mode corresponding to that meeting mode is taken as the meeting mode of the real-time meeting.

5. The intelligent meeting management system based on a cloud platform according to claim 1, characterized in that, The specific method for dynamically monitoring the meeting during a real-time meeting is as follows: If the meeting is conducted online, the meeting will be monitored using an online meeting platform and eye-tracking technology. If the meeting is conducted offline, the meeting will be monitored through the monitoring system installed in the meeting room. If the meeting mode is hybrid, the meeting will be monitored through an online meeting platform, eye-tracking technology, and a meeting room monitoring system.

6. The intelligent meeting management system based on a cloud platform according to claim 1, characterized in that, The specific method for assessing whether the progress of the real-time meeting has deviated and generating a voice reminder is as follows: Based on the main content of the real-time meeting, the main content of the real-time meeting is decomposed into multiple core node tags and corresponding time nodes. Based on the team member codes of the real-time meeting, a role responsibility matrix is ​​preset. The role responsibility matrix is ​​specifically the leading team member code corresponding to each core node tag. By dynamically monitoring the real-time meeting, the speech content and duration of several team members are obtained, and the role defocus index of the real-time meeting is calculated. If the role defocus index of the real-time meeting is greater than or equal to the preset role defocus index threshold, it is determined that the process of the real-time meeting has deviated, and a type of voice reminder is generated. By dynamically monitoring the real-time meeting, the OCR recognition of screen-shared content or uploaded documents is obtained, and a correlation graph between the shared document and the real-time meeting is constructed. The matching degree between the keywords of the shared document in the real-time meeting and the current core node label is calculated using the keyword density calculation formula. If the matching degree is less than the preset matching degree threshold of the shared document, it is determined that the process of the real-time meeting has deviated, and a second type of voice reminder is generated. By dynamically monitoring real-time meetings, the actual decision points are dynamically marked. Combined with the time nodes corresponding to the core decision tags of the real-time meeting, the decision delay duration is obtained, the decision lag cost of the real-time meeting is calculated, and three types of voice reminders are generated. By dynamically monitoring the real-time meeting, the set of emotional categories of the speakers is clustered in real time to determine the emotional interference offset judgment value of the real-time meeting. If the emotional interference offset judgment value of the real-time meeting is 1, it is determined that the process of the real-time meeting has deviated, and four types of voice reminders are generated.

7. The intelligent meeting management system based on a cloud platform according to claim 6, characterized in that, The specific formula for calculating the role defocus index of the real-time meeting is: S_J=T_no / (T_no+T_yes)*P_Y; In the formula, S_J represents the role defocus index of the real-time meeting, T_no and T_yes represent the speaking duration of the non-dominant role and the speaking duration of the dominant role in the real-time meeting, respectively, and P_Y represents the speaking deviation degree of the non-dominant role in the real-time meeting. The speaking duration of the non-dominant role and the speaking duration of the dominant role in the real-time meeting are specifically determined based on the speaking duration encoded by several team members and the role responsibility matrix, and the speaking deviation degree of the non-dominant role in the real-time meeting is specifically determined based on the speaking content encoded by several team members and the role responsibility matrix, combined with semantic analysis.

8. The intelligent meeting management system based on a cloud platform according to claim 6, characterized in that, The specific formula for calculating the decision lag cost of the real-time meeting is: C_B=T_Y*M*γ; In the formula, C_B represents the decision lag cost of the real-time meeting, T_Y is the decision delay duration of the real-time meeting, M is the average hourly wage of the team members in the real-time meeting, and γ is the preset decision urgency weight.

9. The intelligent meeting management system based on a cloud platform according to claim 1, characterized in that, The conference knowledge base includes several knowledge base documents with several team codes corresponding to several conference topics.

10. The intelligent meeting management system based on a cloud platform according to claim 9, characterized in that, The monitoring data based on the meeting knowledge base is used to classify several meeting topics into target value levels. The specific classification method is as follows: The monitoring data includes the set of cited document types, the set of citing departments, the number of collaboration touchpoints, the innovation conversion rate, the half-life index, and the first trigger duration of several knowledge base documents corresponding to several meeting topics of several teams. The network centrality and cross-department citation rate of several knowledge base documents are determined based on the set of cited document types and the set of citing departments of several knowledge base documents. The value index J_Z(i,m,h) of several knowledge base documents corresponding to several conference topics and several team codes is obtained by uniformizing the network centrality, cross-departmental citation rate, number of collaboration touchpoints, innovation conversion rate, half-life index and first trigger duration, and then weighting and summing them. If J_Z(i,m,h)∈JZ_f, then the knowledge base document is classified into the f-th value level, where JZ_f is the value index interval corresponding to the pre-set f-th value level. Several knowledge base documents of different value levels are obtained, and the conference topics corresponding to the several knowledge base documents are mapped to obtain several conference topics of different value levels. Several different value levels of several conference topics are mapped to obtain several different value levels. The value level with the largest distribution is selected as the target value level of several conference topics.

Images